Better handling of JSON requests

Requires fallocate(2) support with FALLOC_FL_PUNCH_HOLE, as
well as a filesystem that supports it (in Linux 3.7,
tmpfs, btrfs, xfs, or ext4)

.coveragerc

@@ -7,4 +7,4 @@
 exclude_lines =
 	pragma: no cover
 	if 0:
-omit = nilmdb/utils/datetime_tz*
+omit = nilmdb/utils/datetime_tz*,nilmdb/scripts,nilmdb/_version.py

.gitattributes

@@ -0,0 +1 @@
+nilmdb/_version.py export-subst

.gitignore

@@ -18,6 +18,10 @@ nilmdb/server/rbtree.so
 dist/
 nilmdb.egg-info/
 
+# This gets generated as needed by setup.py
+MANIFEST.in
+MANIFEST
+
 # Misc
 timeit*out
 

.pylintrc

@@ -0,0 +1,250 @@
+# -*- conf -*-
+[MASTER]
+
+# Specify a configuration file.
+#rcfile=
+
+# Python code to execute, usually for sys.path manipulation such as
+# pygtk.require().
+#init-hook=
+
+# Profiled execution.
+profile=no
+
+# Add files or directories to the blacklist. They should be base names, not
+# paths.
+ignore=datetime_tz
+
+# Pickle collected data for later comparisons.
+persistent=no
+
+# List of plugins (as comma separated values of python modules names) to load,
+# usually to register additional checkers.
+load-plugins=
+
+
+[MESSAGES CONTROL]
+
+# Enable the message, report, category or checker with the given id(s). You can
+# either give multiple identifier separated by comma (,) or put this option
+# multiple time.
+#enable=
+
+# Disable the message, report, category or checker with the given id(s). You
+# can either give multiple identifier separated by comma (,) or put this option
+# multiple time (only on the command line, not in the configuration file where
+# it should appear only once).
+disable=C0111,R0903,R0201,R0914,R0912,W0142,W0703,W0702
+
+
+[REPORTS]
+
+# Set the output format. Available formats are text, parseable, colorized, msvs
+# (visual studio) and html
+output-format=parseable
+
+# Include message's id in output
+include-ids=yes
+
+# Put messages in a separate file for each module / package specified on the
+# command line instead of printing them on stdout. Reports (if any) will be
+# written in a file name "pylint_global.[txt|html]".
+files-output=no
+
+# Tells whether to display a full report or only the messages
+reports=yes
+
+# Python expression which should return a note less than 10 (10 is the highest
+# note). You have access to the variables errors warning, statement which
+# respectively contain the number of errors / warnings messages and the total
+# number of statements analyzed. This is used by the global evaluation report
+# (RP0004).
+evaluation=10.0 - ((float(5 * error + warning + refactor + convention) / statement) * 10)
+
+# Add a comment according to your evaluation note. This is used by the global
+# evaluation report (RP0004).
+comment=no
+
+
+[SIMILARITIES]
+
+# Minimum lines number of a similarity.
+min-similarity-lines=4
+
+# Ignore comments when computing similarities.
+ignore-comments=yes
+
+# Ignore docstrings when computing similarities.
+ignore-docstrings=yes
+
+
+[TYPECHECK]
+
+# Tells whether missing members accessed in mixin class should be ignored. A
+# mixin class is detected if its name ends with "mixin" (case insensitive).
+ignore-mixin-members=yes
+
+# List of classes names for which member attributes should not be checked
+# (useful for classes with attributes dynamically set).
+ignored-classes=SQLObject
+
+# When zope mode is activated, add a predefined set of Zope acquired attributes
+# to generated-members.
+zope=no
+
+# List of members which are set dynamically and missed by pylint inference
+# system, and so shouldn't trigger E0201 when accessed. Python regular
+# expressions are accepted.
+generated-members=REQUEST,acl_users,aq_parent
+
+
+[FORMAT]
+
+# Maximum number of characters on a single line.
+max-line-length=80
+
+# Maximum number of lines in a module
+max-module-lines=1000
+
+# String used as indentation unit. This is usually " " (4 spaces) or "\t" (1
+# tab).
+indent-string='    '
+
+
+[MISCELLANEOUS]
+
+# List of note tags to take in consideration, separated by a comma.
+notes=FIXME,XXX,TODO
+
+
+[VARIABLES]
+
+# Tells whether we should check for unused import in __init__ files.
+init-import=no
+
+# A regular expression matching the beginning of the name of dummy variables
+# (i.e. not used).
+dummy-variables-rgx=_|dummy
+
+# List of additional names supposed to be defined in builtins. Remember that
+# you should avoid to define new builtins when possible.
+additional-builtins=
+
+
+[BASIC]
+
+# Required attributes for module, separated by a comma
+required-attributes=
+
+# List of builtins function names that should not be used, separated by a comma
+bad-functions=apply,input
+
+# Regular expression which should only match correct module names
+module-rgx=(([a-z_][a-z0-9_]*)|([A-Z][a-zA-Z0-9]+))$
+
+# Regular expression which should only match correct module level names
+const-rgx=(([A-Z_][A-Z0-9_]*)|(__.*__)|version)$
+
+# Regular expression which should only match correct class names
+class-rgx=[A-Z_][a-zA-Z0-9]+$
+
+# Regular expression which should only match correct function names
+function-rgx=[a-z_][a-z0-9_]{0,30}$
+
+# Regular expression which should only match correct method names
+method-rgx=[a-z_][a-z0-9_]{0,30}$
+
+# Regular expression which should only match correct instance attribute names
+attr-rgx=[a-z_][a-z0-9_]{0,30}$
+
+# Regular expression which should only match correct argument names
+argument-rgx=[a-z_][a-z0-9_]{0,30}$
+
+# Regular expression which should only match correct variable names
+variable-rgx=[a-z_][a-z0-9_]{0,30}$
+
+# Regular expression which should only match correct list comprehension /
+# generator expression variable names
+inlinevar-rgx=[A-Za-z_][A-Za-z0-9_]*$
+
+# Good variable names which should always be accepted, separated by a comma
+good-names=i,j,k,ex,Run,_
+
+# Bad variable names which should always be refused, separated by a comma
+bad-names=foo,bar,baz,toto,tutu,tata
+
+# Regular expression which should only match functions or classes name which do
+# not require a docstring
+no-docstring-rgx=__.*__
+
+
+[CLASSES]
+
+# List of interface methods to ignore, separated by a comma. This is used for
+# instance to not check methods defines in Zope's Interface base class.
+ignore-iface-methods=isImplementedBy,deferred,extends,names,namesAndDescriptions,queryDescriptionFor,getBases,getDescriptionFor,getDoc,getName,getTaggedValue,getTaggedValueTags,isEqualOrExtendedBy,setTaggedValue,isImplementedByInstancesOf,adaptWith,is_implemented_by
+
+# List of method names used to declare (i.e. assign) instance attributes.
+defining-attr-methods=__init__,__new__,setUp
+
+# List of valid names for the first argument in a class method.
+valid-classmethod-first-arg=cls
+
+
+[DESIGN]
+
+# Maximum number of arguments for function / method
+max-args=5
+
+# Argument names that match this expression will be ignored. Default to name
+# with leading underscore
+ignored-argument-names=_.*
+
+# Maximum number of locals for function / method body
+max-locals=15
+
+# Maximum number of return / yield for function / method body
+max-returns=6
+
+# Maximum number of branch for function / method body
+max-branchs=12
+
+# Maximum number of statements in function / method body
+max-statements=50
+
+# Maximum number of parents for a class (see R0901).
+max-parents=7
+
+# Maximum number of attributes for a class (see R0902).
+max-attributes=7
+
+# Minimum number of public methods for a class (see R0903).
+min-public-methods=2
+
+# Maximum number of public methods for a class (see R0904).
+max-public-methods=20
+
+
+[IMPORTS]
+
+# Deprecated modules which should not be used, separated by a comma
+deprecated-modules=regsub,string,TERMIOS,Bastion,rexec
+
+# Create a graph of every (i.e. internal and external) dependencies in the
+# given file (report RP0402 must not be disabled)
+import-graph=
+
+# Create a graph of external dependencies in the given file (report RP0402 must
+# not be disabled)
+ext-import-graph=
+
+# Create a graph of internal dependencies in the given file (report RP0402 must
+# not be disabled)
+int-import-graph=
+
+
+[EXCEPTIONS]
+
+# Exceptions that will emit a warning when being caught. Defaults to
+# "Exception"
+overgeneral-exceptions=Exception

Makefile

@@ -1,12 +1,42 @@
+# By default, run the tests.
 all: test
 
+version:
+	python setup.py version
+
+build:
+	python setup.py build_ext --inplace
+
+dist: sdist
+sdist:
+	python setup.py sdist
+
+install:
+	python setup.py install
+
+docs:
+	make -C docs
+
 lint:
-	pylint -f parseable nilmdb
+	pylint --rcfile=.pylintrc nilmdb
 
 test:
-	python runtests.py
+ifeq ($(INSIDE_EMACS), t)
+# Use the slightly more flexible script
+	python tests/runtests.py
+else
+# Let setup.py check dependencies, build stuff, and run the test
+	python setup.py nosetests
+endif
 
 clean::
 	find . -name '*pyc' | xargs rm -f
 	rm -f .coverage
 	rm -rf tests/*testdb*
+	rm -rf nilmdb.egg-info/ build/ nilmdb/server/*.so MANIFEST.in
+	make -C docs clean
+
+gitclean::
+	git clean -dXf
+
+.PHONY: all version build dist sdist install docs lint test clean

README.txt

@@ -3,8 +3,24 @@ by Jim Paris <jim@jtan.com>
 
 Prerequisites:
 
-  sudo apt-get install python2.7 python-cherrypy3 python-decorator python-nose python-coverage python-setuptools
+  # Runtime and build environments
+  sudo apt-get install python2.7 python2.7-dev python-setuptools cython
+
+  # Base NilmDB dependencies
+  sudo apt-get install python-cherrypy3 python-decorator python-simplejson
+  sudo apt-get install python-requests python-dateutil python-tz python-psutil
+
+  # Tools for running tests
+  sudo apt-get install python-nose python-coverage
+
+Test:
+  python setup.py nosetests
 
 Install:
 
   python setup.py install
+
+Usage:
+
+  nilmdb-server --help
+  nilmtool --help

nilmdb/__init__.py

@@ -1,4 +1,8 @@
 """Main NilmDB import"""
 
-from server import NilmDB, Server
-from client import Client
+from nilmdb.server import NilmDB, Server
+from nilmdb.client import Client
+
+from nilmdb._version import get_versions
+__version__ = get_versions()['version']
+del get_versions

nilmdb/_version.py

@@ -0,0 +1,197 @@
+
+IN_LONG_VERSION_PY = True
+# This file helps to compute a version number in source trees obtained from
+# git-archive tarball (such as those provided by githubs download-from-tag
+# feature). Distribution tarballs (build by setup.py sdist) and build
+# directories (produced by setup.py build) will contain a much shorter file
+# that just contains the computed version number.
+
+# This file is released into the public domain. Generated by
+# versioneer-0.7+ (https://github.com/warner/python-versioneer)
+
+# these strings will be replaced by git during git-archive
+git_refnames = "$Format:%d$"
+git_full = "$Format:%H$"
+
+
+import subprocess
+import sys
+
+def run_command(args, cwd=None, verbose=False):
+    try:
+        # remember shell=False, so use git.cmd on windows, not just git
+        p = subprocess.Popen(args, stdout=subprocess.PIPE, cwd=cwd)
+    except EnvironmentError:
+        e = sys.exc_info()[1]
+        if verbose:
+            print("unable to run %s" % args[0])
+            print(e)
+        return None
+    stdout = p.communicate()[0].strip()
+    if sys.version >= '3':
+        stdout = stdout.decode()
+    if p.returncode != 0:
+        if verbose:
+            print("unable to run %s (error)" % args[0])
+        return None
+    return stdout
+
+
+import sys
+import re
+import os.path
+
+def get_expanded_variables(versionfile_source):
+    # the code embedded in _version.py can just fetch the value of these
+    # variables. When used from setup.py, we don't want to import
+    # _version.py, so we do it with a regexp instead. This function is not
+    # used from _version.py.
+    variables = {}
+    try:
+        for line in open(versionfile_source,"r").readlines():
+            if line.strip().startswith("git_refnames ="):
+                mo = re.search(r'=\s*"(.*)"', line)
+                if mo:
+                    variables["refnames"] = mo.group(1)
+            if line.strip().startswith("git_full ="):
+                mo = re.search(r'=\s*"(.*)"', line)
+                if mo:
+                    variables["full"] = mo.group(1)
+    except EnvironmentError:
+        pass
+    return variables
+
+def versions_from_expanded_variables(variables, tag_prefix, verbose=False):
+    refnames = variables["refnames"].strip()
+    if refnames.startswith("$Format"):
+        if verbose:
+            print("variables are unexpanded, not using")
+        return {} # unexpanded, so not in an unpacked git-archive tarball
+    refs = set([r.strip() for r in refnames.strip("()").split(",")])
+    for ref in list(refs):
+        if not re.search(r'\d', ref):
+            if verbose:
+                print("discarding '%s', no digits" % ref)
+            refs.discard(ref)
+            # Assume all version tags have a digit. git's %d expansion
+            # behaves like git log --decorate=short and strips out the
+            # refs/heads/ and refs/tags/ prefixes that would let us
+            # distinguish between branches and tags. By ignoring refnames
+            # without digits, we filter out many common branch names like
+            # "release" and "stabilization", as well as "HEAD" and "master".
+    if verbose:
+        print("remaining refs: %s" % ",".join(sorted(refs)))
+    for ref in sorted(refs):
+        # sorting will prefer e.g. "2.0" over "2.0rc1"
+        if ref.startswith(tag_prefix):
+            r = ref[len(tag_prefix):]
+            if verbose:
+                print("picking %s" % r)
+            return { "version": r,
+                     "full": variables["full"].strip() }
+    # no suitable tags, so we use the full revision id
+    if verbose:
+        print("no suitable tags, using full revision id")
+    return { "version": variables["full"].strip(),
+             "full": variables["full"].strip() }
+
+def versions_from_vcs(tag_prefix, versionfile_source, verbose=False):
+    # this runs 'git' from the root of the source tree. That either means
+    # someone ran a setup.py command (and this code is in versioneer.py, so
+    # IN_LONG_VERSION_PY=False, thus the containing directory is the root of
+    # the source tree), or someone ran a project-specific entry point (and
+    # this code is in _version.py, so IN_LONG_VERSION_PY=True, thus the
+    # containing directory is somewhere deeper in the source tree). This only
+    # gets called if the git-archive 'subst' variables were *not* expanded,
+    # and _version.py hasn't already been rewritten with a short version
+    # string, meaning we're inside a checked out source tree.
+
+    try:
+        here = os.path.abspath(__file__)
+    except NameError:
+        # some py2exe/bbfreeze/non-CPython implementations don't do __file__
+        return {} # not always correct
+
+    # versionfile_source is the relative path from the top of the source tree
+    # (where the .git directory might live) to this file. Invert this to find
+    # the root from __file__.
+    root = here
+    if IN_LONG_VERSION_PY:
+        for i in range(len(versionfile_source.split("/"))):
+            root = os.path.dirname(root)
+    else:
+        root = os.path.dirname(here)
+    if not os.path.exists(os.path.join(root, ".git")):
+        if verbose:
+            print("no .git in %s" % root)
+        return {}
+
+    GIT = "git"
+    if sys.platform == "win32":
+        GIT = "git.cmd"
+    stdout = run_command([GIT, "describe", "--tags", "--dirty", "--always"],
+                         cwd=root)
+    if stdout is None:
+        return {}
+    if not stdout.startswith(tag_prefix):
+        if verbose:
+            print("tag '%s' doesn't start with prefix '%s'" % (stdout, tag_prefix))
+        return {}
+    tag = stdout[len(tag_prefix):]
+    stdout = run_command([GIT, "rev-parse", "HEAD"], cwd=root)
+    if stdout is None:
+        return {}
+    full = stdout.strip()
+    if tag.endswith("-dirty"):
+        full += "-dirty"
+    return {"version": tag, "full": full}
+
+
+def versions_from_parentdir(parentdir_prefix, versionfile_source, verbose=False):
+    if IN_LONG_VERSION_PY:
+        # We're running from _version.py. If it's from a source tree
+        # (execute-in-place), we can work upwards to find the root of the
+        # tree, and then check the parent directory for a version string. If
+        # it's in an installed application, there's no hope.
+        try:
+            here = os.path.abspath(__file__)
+        except NameError:
+            # py2exe/bbfreeze/non-CPython don't have __file__
+            return {} # without __file__, we have no hope
+        # versionfile_source is the relative path from the top of the source
+        # tree to _version.py. Invert this to find the root from __file__.
+        root = here
+        for i in range(len(versionfile_source.split("/"))):
+            root = os.path.dirname(root)
+    else:
+        # we're running from versioneer.py, which means we're running from
+        # the setup.py in a source tree. sys.argv[0] is setup.py in the root.
+        here = os.path.abspath(sys.argv[0])
+        root = os.path.dirname(here)
+
+    # Source tarballs conventionally unpack into a directory that includes
+    # both the project name and a version string.
+    dirname = os.path.basename(root)
+    if not dirname.startswith(parentdir_prefix):
+        if verbose:
+            print("guessing rootdir is '%s', but '%s' doesn't start with prefix '%s'" %
+                  (root, dirname, parentdir_prefix))
+        return None
+    return {"version": dirname[len(parentdir_prefix):], "full": ""}
+
+tag_prefix = "nilmdb-"
+parentdir_prefix = "nilmdb-"
+versionfile_source = "nilmdb/_version.py"
+
+def get_versions(default={"version": "unknown", "full": ""}, verbose=False):
+    variables = { "refnames": git_refnames, "full": git_full }
+    ver = versions_from_expanded_variables(variables, tag_prefix, verbose)
+    if not ver:
+        ver = versions_from_vcs(tag_prefix, versionfile_source, verbose)
+    if not ver:
+        ver = versions_from_parentdir(parentdir_prefix, versionfile_source,
+                                      verbose)
+    if not ver:
+        ver = default
+    return ver
+

nilmdb/client/__init__.py

@@ -1,4 +1,4 @@
 """nilmdb.client"""
 
-from .client import Client
-from .errors import *
+from nilmdb.client.client import Client
+from nilmdb.client.errors import ClientError, ServerError, Error

nilmdb/client/client.py

@@ -5,34 +5,45 @@
 import nilmdb
 import nilmdb.utils
 import nilmdb.client.httpclient
-from nilmdb.utils.printf import *
 
 import time
-import sys
-import re
-import os
 import simplejson as json
-import itertools
-
-version = "1.0"
+import contextlib
 
 def float_to_string(f):
-    # Use repr to maintain full precision in the string output.
+    """Use repr to maintain full precision in the string output."""
     return repr(float(f))
 
+def extract_timestamp(line):
+    """Extract just the timestamp from a line of data text"""
+    return float(line.split()[0])
+
 class Client(object):
     """Main client interface to the Nilm database."""
 
-    client_version = version
+    def __init__(self, url, post_json = False):
+        """Initialize client with given URL.  If post_json is true,
+        POST requests are sent with Content-Type 'application/json'
+        instead of the default 'x-www-form-urlencoded'."""
+        self.http = nilmdb.client.httpclient.HTTPClient(url, post_json)
+        self.post_json = post_json
 
-    def __init__(self, url):
-        self.http = nilmdb.client.httpclient.HTTPClient(url)
+    # __enter__/__exit__ allow this class to be a context manager
+    def __enter__(self):
+        return self
 
-    def _json_param(self, data):
+    def __exit__(self, exc_type, exc_value, traceback):
+        self.close()
+
+    def _json_post_param(self, data):
         """Return compact json-encoded version of parameter"""
+        if self.post_json:
+            # If we're posting as JSON, we don't need to encode it further here
+            return data
         return json.dumps(data, separators=(',',':'))
 
     def close(self):
+        """Close the connection; safe to call multiple times"""
         self.http.close()
 
     def geturl(self):
@@ -43,20 +54,19 @@ class Client(object):
         """Return server version"""
         return self.http.get("version")
 
-    def dbpath(self):
-        """Return server database path"""
-        return self.http.get("dbpath")
+    def dbinfo(self):
+        """Return server database info (path, size, free space)
+        as a dictionary."""
+        return self.http.get("dbinfo")
 
-    def dbsize(self):
-        """Return server database size as human readable string"""
-        return self.http.get("dbsize")
-
-    def stream_list(self, path = None, layout = None):
+    def stream_list(self, path = None, layout = None, extent = False):
         params = {}
         if path is not None:
             params["path"] = path
         if layout is not None:
             params["layout"] = layout
+        if extent:
+            params["extent"] = 1
         return self.http.get("stream/list", params)
 
     def stream_get_metadata(self, path, keys = None):
@@ -70,28 +80,28 @@ class Client(object):
         metadata."""
         params = {
             "path": path,
-            "data": self._json_param(data)
+            "data": self._json_post_param(data)
             }
-        return self.http.get("stream/set_metadata", params)
+        return self.http.post("stream/set_metadata", params)
 
     def stream_update_metadata(self, path, data):
         """Update stream metadata from a dictionary"""
         params = {
             "path": path,
-            "data": self._json_param(data)
+            "data": self._json_post_param(data)
             }
-        return self.http.get("stream/update_metadata", params)
+        return self.http.post("stream/update_metadata", params)
 
     def stream_create(self, path, layout):
         """Create a new stream"""
         params = { "path": path,
                    "layout" : layout }
-        return self.http.get("stream/create", params)
+        return self.http.post("stream/create", params)
 
     def stream_destroy(self, path):
         """Delete stream and its contents"""
         params = { "path": path }
-        return self.http.get("stream/destroy", params)
+        return self.http.post("stream/destroy", params)
 
     def stream_remove(self, path, start = None, end = None):
         """Remove data from the specified time range"""
@@ -102,79 +112,47 @@ class Client(object):
             params["start"] = float_to_string(start)
         if end is not None:
             params["end"] = float_to_string(end)
-        return self.http.get("stream/remove", params)
+        return self.http.post("stream/remove", params)
+
+    @contextlib.contextmanager
+    def stream_insert_context(self, path, start = None, end = None):
+        """Return a context manager that allows data to be efficiently
+        inserted into a stream in a piecewise manner.  Data is be provided
+        as single lines, and is aggregated and sent to the server in larger
+        chunks as necessary.  Data lines must match the database layout for
+        the given path, and end with a newline.
+
+        Example:
+          with client.stream_insert_context('/path', start, end) as ctx:
+            ctx.insert_line('1234567890.0 1 2 3 4\\n')
+            ctx.insert_line('1234567891.0 1 2 3 4\\n')
+
+        For more details, see help for nilmdb.client.client.StreamInserter
+
+        This may make multiple requests to the server, if the data is
+        large enough or enough time has passed between insertions.
+        """
+        ctx = StreamInserter(self, path, start, end)
+        yield ctx
+        ctx.finalize()
 
     def stream_insert(self, path, data, start = None, end = None):
-        """Insert data into a stream.  data should be a file-like object
-        that provides ASCII data that matches the database layout for path.
+        """Insert rows of data into a stream.  data should be an
+        iterable object that provides ASCII data that matches the
+        database layout for path.  See stream_insert_context for
+        details on the 'start' and 'end' parameters."""
+        with self.stream_insert_context(path, start, end) as ctx:
+            ctx.insert_iter(data)
+        return ctx.last_response
 
-        start and end are the starting and ending timestamp of this
-        stream; all timestamps t in the data must satisfy 'start <= t
-        < end'.  If left unspecified, 'start' is the timestamp of the
-        first line of data, and 'end' is the timestamp on the last line
-        of data, plus a small delta of 1μs.
-        """
-        params = { "path": path }
-
-        # See design.md for a discussion of how much data to send.
-        # These are soft limits -- actual data might be rounded up.
-        max_data = 1048576
-        max_time = 30
-        end_epsilon = 1e-6
-
-
-        def extract_timestamp(line):
-            return float(line.split()[0])
-
-        def sendit():
-            # If we have more data after this, use the timestamp of
-            # the next line as the end.  Otherwise, use the given
-            # overall end time, or add end_epsilon to the last data
-            # point.
-            if nextline:
-                block_end = extract_timestamp(nextline)
-                if end and block_end > end:
-                    # This is unexpected, but we'll defer to the server
-                    # to return an error in this case.
-                    block_end = end
-            elif end:
-                block_end = end
-            else:
-                block_end = extract_timestamp(line) + end_epsilon
-
-            # Send it
-            params["start"] = float_to_string(block_start)
-            params["end"] = float_to_string(block_end)
-            return self.http.put("stream/insert", block_data, params)
-
-        clock_start = time.time()
-        block_data = ""
-        block_start = start
-        result = None
-        for (line, nextline) in nilmdb.utils.misc.pairwise(data):
-            # If we don't have a starting time, extract it from the first line
-            if block_start is None:
-                block_start = extract_timestamp(line)
-
-            clock_elapsed = time.time() - clock_start
-            block_data += line
-
-            # If we have enough data, or enough time has elapsed,
-            # send this block to the server, and empty things out
-            # for the next block.
-            if (len(block_data) > max_data) or (clock_elapsed > max_time):
-                result = sendit()
-                block_start = None
-                block_data = ""
-                clock_start = time.time()
-
-        # One last block?
-        if len(block_data):
-            result = sendit()
-
-        # Return the most recent JSON result we got back, or None if
-        # we didn't make any requests.
-        return result
+    def stream_insert_block(self, path, block, start, end):
+        """Insert an entire block of data into a stream.  Like
+        stream_insert, except 'block' contains multiple lines of ASCII
+        text and is sent in one single chunk."""
+        params = { "path": path,
+                   "start": float_to_string(start),
+                   "end": float_to_string(end) }
+        return self.http.put("stream/insert", block, params)
 
     def stream_intervals(self, path, start = None, end = None):
         """
@@ -187,7 +165,7 @@ class Client(object):
             params["start"] = float_to_string(start)
         if end is not None:
             params["end"] = float_to_string(end)
-        return self.http.get_gen("stream/intervals", params, retjson = True)
+        return self.http.get_gen("stream/intervals", params)
 
     def stream_extract(self, path, start = None, end = None, count = False):
         """
@@ -195,8 +173,8 @@ class Client(object):
         lines of ASCII-formatted data that matches the database
         layout for the given path.
 
-        Specify count=True to just get a count of values rather than
-        the actual data.
+        Specify count = True to return a count of matching data points
+        rather than the actual data.  The output format is unchanged.
         """
         params = {
             "path": path,
@@ -207,5 +185,203 @@ class Client(object):
             params["end"] = float_to_string(end)
         if count:
             params["count"] = 1
+        return self.http.get_gen("stream/extract", params)
 
-        return self.http.get_gen("stream/extract", params, retjson = False)
+    def stream_count(self, path, start = None, end = None):
+        """
+        Return the number of rows of data in the stream that satisfy
+        the given timestamps.
+        """
+        counts = list(self.stream_extract(path, start, end, count = True))
+        return int(counts[0])
+
+class StreamInserter(object):
+    """Object returned by stream_insert_context() that manages
+    the insertion of rows of data into a particular path.
+
+    The basic data flow is that we are filling a contiguous interval
+    on the server, with no gaps, that extends from timestamp 'start'
+    to timestamp 'end'.  Data timestamps satisfy 'start <= t < end'.
+    Data is provided by the user one line at a time with
+    .insert_line() or .insert_iter().
+
+    1. The first inserted line begins a new interval that starts at
+    'start'.  If 'start' is not given, it is deduced from the first
+    line's timestamp.
+
+    2. Subsequent lines go into the same contiguous interval.  As lines
+    are inserted, this routine may make multiple insertion requests to
+    the server, but will structure the timestamps to leave no gaps.
+
+    3. The current contiguous interval can be completed by manually
+    calling .finalize(), which the context manager will also do
+    automatically.  This will send any remaining data to the server,
+    using the 'end' timestamp to end the interval.
+
+    After a .finalize(), inserting new data goes back to step 1.
+
+    .update_start() can be called before step 1 to change the start
+    time for the interval.  .update_end() can be called before step 3
+    to change the end time for the interval.
+    """
+
+    # See design.md for a discussion of how much data to send.
+    # These are soft limits -- actual data might be rounded up.
+    # We send when we have a certain amount of data queued, or
+    # when a certain amount of time has passed since the last send.
+    _max_data = 2 * 1024 * 1024
+    _max_time = 30
+
+    # Delta to add to the final timestamp, if "end" wasn't given
+    _end_epsilon = 1e-6
+
+    def __init__(self, client, path, start = None, end = None):
+        """'http' is the httpclient object.  'path' is the database
+        path to insert to.  'start' and 'end' are used for the first
+        contiguous interval."""
+        self.last_response = None
+
+        self._client = client
+        self._path = path
+
+        # Start and end for the overall contiguous interval we're
+        # filling
+        self._interval_start = start
+        self._interval_end = end
+
+        # Data for the specific block we're building up to send
+        self._block_data = []
+        self._block_len = 0
+        self._block_start = None
+
+        # Time of last request
+        self._last_time = time.time()
+
+        # We keep a buffer of the two most recently inserted lines.
+        # Only the older one actually gets processed; the newer one
+        # is used to "look-ahead" to the next timestamp if we need
+        # to internally split an insertion into two requests.
+        self._line_old = None
+        self._line_new = None
+
+    def insert_iter(self, iter):
+        """Insert all lines of ASCII formatted data from the given
+        iterable.  Lines must be terminated with '\\n'."""
+        for line in iter:
+            self.insert_line(line)
+
+    def insert_line(self, line, allow_intermediate = True):
+        """Insert a single line of ASCII formatted data.  Line
+        must be terminated with '\\n'."""
+        if line and (len(line) < 1 or line[-1] != '\n'):
+            raise ValueError("lines must end in with a newline character")
+
+        # Store this new line, but process the previous (old) one.
+        # This lets us "look ahead" to the next line.
+        self._line_old = self._line_new
+        self._line_new = line
+        if self._line_old is None:
+            return
+
+        # If starting a new block, pull out the timestamp if needed.
+        if self._block_start is None:
+            if self._interval_start is not None:
+                # User provided a start timestamp.  Use it once, then
+                # clear it for the next block.
+                self._block_start = self._interval_start
+                self._interval_start = None
+            else:
+                # Extract timestamp from the first row
+                self._block_start = extract_timestamp(self._line_old)
+
+        # Save the line
+        self._block_data.append(self._line_old)
+        self._block_len += len(self._line_old)
+
+        if allow_intermediate:
+            # Send an intermediate block to the server if needed.
+            elapsed = time.time() - self._last_time
+            if (self._block_len > self._max_data) or (elapsed > self._max_time):
+                self._send_block_intermediate()
+
+    def update_start(self, start):
+        """Update the start time for the next contiguous interval.
+        Call this before starting to insert data for a new interval,
+        for example, after .finalize()"""
+        self._interval_start = start
+
+    def update_end(self, end):
+        """Update the end time for the current contiguous interval.
+        Call this before .finalize()"""
+        self._interval_end = end
+
+    def finalize(self):
+        """Stop filling the current contiguous interval.
+        All outstanding data will be sent, and the interval end
+        time of the interval will be taken from the 'end' argument
+        used when initializing this class, or the most recent
+        value passed to update_end(), or the last timestamp plus
+        a small epsilon value if no other endpoint was provided.
+
+        If more data is inserted after a finalize(), it will become
+        part of a new interval and there may be a gap left in-between."""
+        # Special marker tells insert_line that this is the end
+        self.insert_line(None, allow_intermediate = False)
+
+        if self._block_len > 0:
+            # We have data pending, so send the final block
+            self._send_block_final()
+        elif None not in (self._interval_start, self._interval_end):
+            # We have no data, but enough information to create an
+            # empty interval.
+            self._block_start = self._interval_start
+            self._interval_start = None
+            self._send_block_final()
+        else:
+            # No data, and no timestamps to use to create an empty
+            # interval.
+            pass
+
+        # Make sure both timestamps are emptied for future intervals.
+        self._interval_start = None
+        self._interval_end = None
+
+    def _send_block_intermediate(self):
+        """Send data, when we still have more data to send.
+        Use the timestamp from the next line, so that the blocks
+        are contiguous."""
+        block_end = extract_timestamp(self._line_new)
+        if self._interval_end is not None and block_end > self._interval_end:
+            # Something's fishy -- the timestamp we found is after
+            # the user's specified end.  Limit it here, and the
+            # server will return an error.
+            block_end = self._interval_end
+        self._send_block(block_end)
+
+    def _send_block_final(self):
+        """Send data, when this is the last block for the interval.
+        There is no next line, so figure out the actual interval end
+        using interval_end or end_epsilon."""
+        if self._interval_end is not None:
+            # Use the user's specified end timestamp
+            block_end = self._interval_end
+            # Clear it in case we send more intervals in the future.
+            self._interval_end = None
+        else:
+            # Add an epsilon to the last timestamp we saw
+            block_end = extract_timestamp(self._line_old) + self._end_epsilon
+        self._send_block(block_end)
+
+    def _send_block(self, block_end):
+        """Send current block to the server"""
+        self.last_response = self._client.stream_insert_block(
+            self._path, "".join(self._block_data),
+            self._block_start, block_end)
+
+        # Clear out the block
+        self._block_data = []
+        self._block_len = 0
+        self._block_start = None
+
+        # Note when we sent it
+        self._last_time = time.time()

nilmdb/client/httpclient.py

@@ -2,21 +2,15 @@
 
 import nilmdb
 import nilmdb.utils
-from nilmdb.utils.printf import *
-from nilmdb.client.errors import *
+from nilmdb.client.errors import ClientError, ServerError, Error
 
-import time
-import sys
-import re
-import os
 import simplejson as json
 import urlparse
-import pycurl
-import cStringIO
+import requests
 
 class HTTPClient(object):
     """Class to manage and perform HTTP requests from the client"""
-    def __init__(self, baseurl = ""):
+    def __init__(self, baseurl = "", post_json = False):
         """If baseurl is supplied, all other functions that take
         a URL can be given a relative URL instead."""
         # Verify / clean up URL
@@ -24,28 +18,25 @@ class HTTPClient(object):
         if '://' not in reparsed:
             reparsed = urlparse.urlparse("http://" + baseurl).geturl()
         self.baseurl = reparsed
-        self.curl = pycurl.Curl()
-        self.curl.setopt(pycurl.SSL_VERIFYHOST, 2)
-        self.curl.setopt(pycurl.FOLLOWLOCATION, 1)
-        self.curl.setopt(pycurl.MAXREDIRS, 5)
-        self._setup_url()
 
-    def _setup_url(self, url = "", params = ""):
-        url = urlparse.urljoin(self.baseurl, url)
-        if params:
-            url = urlparse.urljoin(
-                url, "?" + nilmdb.utils.urllib.urlencode(params))
-        self.curl.setopt(pycurl.URL, url)
-        self.url = url
+        # Build Requests session object, enable SSL verification
+        self.session = requests.Session()
+        self.session.verify = True
 
-    def _check_error(self, body = None):
-        code = self.curl.getinfo(pycurl.RESPONSE_CODE)
-        if code == 200:
-            return
-        # Default variables for exception
-        args = { "url" : self.url,
+        # Saved response, so that tests can verify a few things.
+        self._last_response = {}
+
+        # Whether to send application/json POST bodies (versus
+        # x-www-form-urlencoded)
+        self.post_json = post_json
+
+    def _handle_error(self, url, code, body):
+        # Default variables for exception.  We use the entire body as
+        # the default message, in case we can't extract it from a JSON
+        # response.
+        args = { "url" : url,
                  "status" : str(code),
-                 "message" : None,
+                 "message" : body,
                  "traceback" : None }
         try:
             # Fill with server-provided data if we can
@@ -67,133 +58,76 @@ class HTTPClient(object):
             else:
                 raise Error(**args)
 
-    def _req_generator(self, url, params):
-        """
-        Like self._req(), but runs the perform in a separate thread.
-        It returns a generator that spits out arbitrary-sized chunks
-        of the resulting data, instead of using the WRITEFUNCTION
-        callback.
-        """
-        self._setup_url(url, params)
-        self._status = None
-        error_body = ""
-        self._headers = ""
-        def header_callback(data):
-            if self._status is None:
-                self._status = int(data.split(" ")[1])
-            self._headers += data
-        self.curl.setopt(pycurl.HEADERFUNCTION, header_callback)
-        def func(callback):
-            self.curl.setopt(pycurl.WRITEFUNCTION, callback)
-            self.curl.perform()
-        try:
-            with nilmdb.utils.Iteratorizer(func, curl_hack = True) as it:
-                for i in it:
-                    if self._status == 200:
-                        # If we had a 200 response, yield the data to caller.
-                        yield i
-                    else:
-                        # Otherwise, collect it into an error string.
-                        error_body += i
-        except pycurl.error as e:
-            raise ServerError(status = "502 Error",
-                              url = self.url,
-                              message = e[1])
-        # Raise an exception if there was an error
-        self._check_error(error_body)
-
-    def _req(self, url, params):
-        """
-        GET or POST that returns raw data.  Returns the body
-        data as a string, or raises an error if it contained an error.
-        """
-        self._setup_url(url, params)
-        body = cStringIO.StringIO()
-        self.curl.setopt(pycurl.WRITEFUNCTION, body.write)
-        self._headers = ""
-        def header_callback(data):
-            self._headers += data
-        self.curl.setopt(pycurl.HEADERFUNCTION, header_callback)
-        try:
-            self.curl.perform()
-        except pycurl.error as e:
-            raise ServerError(status = "502 Error",
-                              url = self.url,
-                              message = e[1])
-        body_str = body.getvalue()
-        # Raise an exception if there was an error
-        self._check_error(body_str)
-        return body_str
-
     def close(self):
-        self.curl.close()
+        self.session.close()
 
-    def _iterate_lines(self, it):
+    def _do_req(self, method, url, query_data, body_data, stream, headers):
+        url = urlparse.urljoin(self.baseurl, url)
+        try:
+            response = self.session.request(method, url,
+                                            params = query_data,
+                                            data = body_data,
+                                            stream = stream,
+                                            headers = headers)
+        except requests.RequestException as e:
+            raise ServerError(status = "502 Error", url = url,
+                              message = str(e.message))
+        if response.status_code != 200:
+            self._handle_error(url, response.status_code, response.content)
+        self._last_response = response
+        if response.headers["content-type"] in ("application/json",
+                                                "application/x-json-stream"):
+            return (response, True)
+        else:
+            return (response, False)
+
+    # Normal versions that return data directly
+    def _req(self, method, url, query = None, body = None, headers = None):
         """
-        Given an iterator that returns arbitrarily-sized chunks
-        of data, return '\n'-delimited lines of text
+        Make a request and return the body data as a string or parsed
+        JSON object, or raise an error if it contained an error.
         """
-        partial = ""
-        for chunk in it:
-            partial += chunk
-            lines = partial.split("\n")
-            for line in lines[0:-1]:
-                yield line
-            partial = lines[-1]
-        if partial != "":
-            yield partial
+        (response, isjson) = self._do_req(method, url, query, body,
+                                          stream = False, headers = headers)
+        if isjson:
+            return json.loads(response.content)
+        return response.content
 
-    # Non-generator versions
-    def _doreq(self, url, params, retjson):
+    def get(self, url, params = None):
+        """Simple GET (parameters in URL)"""
+        return self._req("GET", url, params, None)
+
+    def post(self, url, params = None):
+        """Simple POST (parameters in body)"""
+        if self.post_json:
+            return self._req("POST", url, None,
+                             json.dumps(params),
+                             { 'Content-type': 'application/json' })
+        else:
+            return self._req("POST", url, None, params)
+
+    def put(self, url, data, params = None):
+        """Simple PUT (parameters in URL, data in body)"""
+        return self._req("PUT", url, params, data)
+
+    # Generator versions that return data one line at a time.
+    def _req_gen(self, method, url, query = None, body = None, headers = None):
         """
-        Perform a request, and return the body.
-
-        url: URL to request (relative to baseurl)
-        params: dictionary of query parameters
-        retjson: expect JSON and return python objects instead of string
+        Make a request and return a generator that gives back strings
+        or JSON decoded lines of the body data, or raise an error if
+        it contained an eror.
         """
-        out = self._req(url, params)
-        if retjson:
-            return json.loads(out)
-        return out
-
-    def get(self, url, params = None, retjson = True):
-        """Simple GET"""
-        self.curl.setopt(pycurl.UPLOAD, 0)
-        return self._doreq(url, params, retjson)
-
-    def put(self, url, postdata, params = None, retjson = True):
-        """Simple PUT"""
-        self.curl.setopt(pycurl.UPLOAD, 1)
-        self._setup_url(url, params)
-        data = cStringIO.StringIO(postdata)
-        self.curl.setopt(pycurl.READFUNCTION, data.read)
-        return self._doreq(url, params, retjson)
-
-    # Generator versions
-    def _doreq_gen(self, url, params, retjson):
-        """
-        Perform a request, and return lines of the body in a generator.
-
-        url: URL to request (relative to baseurl)
-        params: dictionary of query parameters
-        retjson: expect JSON and yield python objects instead of strings
-        """
-        for line in self._iterate_lines(self._req_generator(url, params)):
-            if retjson:
+        (response, isjson) = self._do_req(method, url, query, body,
+                                          stream = True, headers = headers)
+        for line in response.iter_lines():
+            if isjson:
                 yield json.loads(line)
             else:
                 yield line
 
-    def get_gen(self, url, params = None, retjson = True):
-        """Simple GET, returning a generator"""
-        self.curl.setopt(pycurl.UPLOAD, 0)
-        return self._doreq_gen(url, params, retjson)
+    def get_gen(self, url, params = None):
+        """Simple GET (parameters in URL) returning a generator"""
+        return self._req_gen("GET", url, params)
 
-    def put_gen(self, url, postdata, params = None, retjson = True):
-        """Simple PUT, returning a generator"""
-        self.curl.setopt(pycurl.UPLOAD, 1)
-        self._setup_url(url, params)
-        data = cStringIO.StringIO(postdata)
-        self.curl.setopt(pycurl.READFUNCTION, data.read)
-        return self._doreq_gen(url, params, retjson)
+    # Not much use for a POST or PUT generator, since they don't
+    # return much data.

nilmdb/cmdline/__init__.py

@@ -1,3 +1,3 @@
 """nilmdb.cmdline"""
 
-from .cmdline import Cmdline
+from nilmdb.cmdline.cmdline import Cmdline

nilmdb/cmdline/cmdline.py

@@ -3,22 +3,19 @@
 import nilmdb
 from nilmdb.utils.printf import *
 from nilmdb.utils import datetime_tz
+import nilmdb.utils.time
 
-import dateutil.parser
 import sys
-import re
+import os
 import argparse
 from argparse import ArgumentDefaultsHelpFormatter as def_form
 
-version = "1.0"
-
 # Valid subcommands.  Defined in separate files just to break
 # things up -- they're still called with Cmdline as self.
-subcommands = [ "info", "create", "list", "metadata", "insert", "extract",
-                "remove", "destroy" ]
+subcommands = [ "help", "info", "create", "list", "metadata",
+                "insert", "extract", "remove", "destroy" ]
 
-# Import the subcommand modules.  Equivalent way of doing this would be
-# from . import info as cmd_info
+# Import the subcommand modules
 subcmd_mods = {}
 for cmd in subcommands:
     subcmd_mods[cmd] = __import__("nilmdb.cmdline." + cmd, fromlist = [ cmd ])
@@ -30,74 +27,21 @@ class JimArgumentParser(argparse.ArgumentParser):
 
 class Cmdline(object):
 
-    def __init__(self, argv):
-        self.argv = argv
+    def __init__(self, argv = None):
+        self.argv = argv or sys.argv[1:]
         self.client = None
+        self.def_url = os.environ.get("NILMDB_URL", "http://localhost:12380")
+        self.subcmd = {}
 
     def arg_time(self, toparse):
         """Parse a time string argument"""
         try:
-            return self.parse_time(toparse).totimestamp()
+            return nilmdb.utils.time.parse_time(toparse).totimestamp()
         except ValueError as e:
             raise argparse.ArgumentTypeError(sprintf("%s \"%s\"",
                                                      str(e), toparse))
 
-    def parse_time(self, toparse):
-        """
-        Parse a free-form time string and return a datetime_tz object.
-        If the string doesn't contain a timestamp, the current local
-        timezone is assumed (e.g. from the TZ env var).
-        """
-        # If string isn't "now" and doesn't contain at least 4 digits,
-        # consider it invalid.  smartparse might otherwise accept
-        # empty strings and strings with just separators.
-        if toparse != "now" and len(re.findall(r"\d", toparse)) < 4:
-            raise ValueError("not enough digits for a timestamp")
-
-        # Try to just parse the time as given
-        try:
-            return datetime_tz.datetime_tz.smartparse(toparse)
-        except ValueError:
-            pass
-
-        # Try to extract a substring in a condensed format that we expect
-        # to see in a filename or header comment
-        res = re.search(r"(^|[^\d])("            # non-numeric or SOL
-                        r"(199\d|2\d\d\d)"       # year
-                        r"[-/]?"                 # separator
-                        r"(0[1-9]|1[012])"       # month
-                        r"[-/]?"                 # separator
-                        r"([012]\d|3[01])"       # day
-                        r"[-T ]?"                # separator
-                        r"([01]\d|2[0-3])"       # hour
-                        r"[:]?"                  # separator
-                        r"([0-5]\d)"             # minute
-                        r"[:]?"                  # separator
-                        r"([0-5]\d)?"            # second
-                        r"([-+]\d\d\d\d)?"       # timezone
-                        r")", toparse)
-        if res is not None:
-            try:
-                return datetime_tz.datetime_tz.smartparse(res.group(2))
-            except ValueError:
-                pass
-
-        # Could also try to successively parse substrings, but let's
-        # just give up for now.
-        raise ValueError("unable to parse timestamp")
-
-    def time_string(self, timestamp):
-        """
-        Convert a Unix timestamp to a string for printing, using the
-        local timezone for display (e.g. from the TZ env var).
-        """
-        dt = datetime_tz.datetime_tz.fromtimestamp(timestamp)
-        return dt.strftime("%a, %d %b %Y %H:%M:%S.%f %z")
-
     def parser_setup(self):
-        version_string = sprintf("nilmtool %s, client library %s",
-                                 version, nilmdb.Client.client_version)
-
         self.parser = JimArgumentParser(add_help = False,
                                         formatter_class = def_form)
 
@@ -105,22 +49,21 @@ class Cmdline(object):
         group.add_argument("-h", "--help", action='help',
                            help='show this help message and exit')
         group.add_argument("-V", "--version", action="version",
-                           version=version_string)
+                           version = nilmdb.__version__)
 
         group = self.parser.add_argument_group("Server")
         group.add_argument("-u", "--url", action="store",
-                           default="http://localhost:12380/",
+                           default=self.def_url,
                            help="NilmDB server URL (default: %(default)s)")
 
-        sub = self.parser.add_subparsers(title="Commands",
-                                         dest="command",
-                                         description="Specify --help after "
-                                         "the command for command-specific "
-                                         "options.")
+        sub = self.parser.add_subparsers(
+            title="Commands", dest="command",
+            description="Use 'help command' or 'command --help' for more "
+            "details on a particular command.")
 
         # Set up subcommands (defined in separate files)
         for cmd in subcommands:
-            subcmd_mods[cmd].setup(self, sub)
+            self.subcmd[cmd] = subcmd_mods[cmd].setup(self, sub)
 
     def die(self, formatstr, *args):
         fprintf(sys.stderr, formatstr + "\n", *args)
@@ -143,7 +86,9 @@ class Cmdline(object):
 
         self.client = nilmdb.Client(self.args.url)
 
-        # Make a test connection to make sure things work
+        # Make a test connection to make sure things work,
+        # unless the particular command requests that we don't.
+        if "no_test_connect" not in self.args:
             try:
                 server_version = self.client.version()
             except nilmdb.client.Error as e:

nilmdb/cmdline/create.py

@@ -1,9 +1,7 @@
 from nilmdb.utils.printf import *
 import nilmdb
 import nilmdb.client
-import textwrap
 
-from argparse import ArgumentDefaultsHelpFormatter as def_form
 from argparse import RawDescriptionHelpFormatter as raw_form
 
 def setup(self, sub):
@@ -28,6 +26,7 @@ Layout types are of the format: type_count
                        help="Path (in database) of new stream, e.g. /foo/bar")
     group.add_argument("layout",
                        help="Layout type for new stream, e.g. float32_8")
+    return cmd
 
 def cmd_create(self):
     """Create new stream"""

nilmdb/cmdline/destroy.py

@@ -16,6 +16,7 @@ def setup(self, sub):
     group = cmd.add_argument_group("Required arguments")
     group.add_argument("path",
                        help="Path of the stream to delete, e.g. /foo/bar")
+    return cmd
 
 def cmd_destroy(self):
     """Destroy stream"""

nilmdb/cmdline/extract.py

@@ -1,8 +1,6 @@
 from __future__ import print_function
 from nilmdb.utils.printf import *
-import nilmdb
 import nilmdb.client
-import sys
 
 def setup(self, sub):
     cmd = sub.add_parser("extract", help="Extract data",
@@ -32,6 +30,7 @@ def setup(self, sub):
                        help="Show raw timestamps in annotated information")
     group.add_argument("-c", "--count", action="store_true",
                        help="Just output a count of matched data points")
+    return cmd
 
 def cmd_extract_verify(self):
     if self.args.start is not None and self.args.end is not None:
@@ -47,7 +46,7 @@ def cmd_extract(self):
     if self.args.timestamp_raw:
         time_string = repr
     else:
-        time_string = self.time_string
+        time_string = nilmdb.utils.time.format_time
 
     if self.args.annotate:
         printf("# path: %s\n", self.args.path)

nilmdb/cmdline/help.py

@@ -0,0 +1,26 @@
+from nilmdb.utils.printf import *
+
+import argparse
+import sys
+
+def setup(self, sub):
+    cmd = sub.add_parser("help", help="Show detailed help for a command",
+                         description="""
+                         Show help for a command. 'help command' is
+                         the same as 'command --help'.
+                         """)
+    cmd.set_defaults(handler = cmd_help)
+    cmd.set_defaults(no_test_connect = True)
+    cmd.add_argument("command", nargs="?",
+                     help="Command to get help about")
+    cmd.add_argument("rest", nargs=argparse.REMAINDER,
+                     help=argparse.SUPPRESS)
+    return cmd
+
+def cmd_help(self):
+    if self.args.command in self.subcmd:
+        self.subcmd[self.args.command].print_help()
+    else:
+        self.parser.print_help()
+
+    return

nilmdb/cmdline/info.py

@@ -1,4 +1,6 @@
+import nilmdb
 from nilmdb.utils.printf import *
+from nilmdb.utils import human_size
 
 from argparse import ArgumentDefaultsHelpFormatter as def_form
 
@@ -10,11 +12,14 @@ def setup(self, sub):
                          version.
                          """)
     cmd.set_defaults(handler = cmd_info)
+    return cmd
 
 def cmd_info(self):
     """Print info about the server"""
-    printf("Client library version: %s\n", self.client.client_version)
+    printf("Client version: %s\n", nilmdb.__version__)
     printf("Server version: %s\n", self.client.version())
     printf("Server URL: %s\n", self.client.geturl())
-    printf("Server database path: %s\n", self.client.dbpath())
-    printf("Server database size: %s\n", self.client.dbsize())
+    dbinfo = self.client.dbinfo()
+    printf("Server database path: %s\n", dbinfo["path"])
+    printf("Server database size: %s\n", human_size(dbinfo["size"]))
+    printf("Server database free space: %s\n", human_size(dbinfo["free"]))

nilmdb/cmdline/insert.py

@@ -2,6 +2,7 @@ from nilmdb.utils.printf import *
 import nilmdb
 import nilmdb.client
 import nilmdb.utils.timestamper as timestamper
+import nilmdb.utils.time
 
 import sys
 
@@ -46,6 +47,7 @@ def setup(self, sub):
                        help="Path of stream, e.g. /foo/bar")
     group.add_argument("file", nargs="*", default=['-'],
                        help="File(s) to insert (default: - (stdin))")
+    return cmd
 
 def cmd_insert(self):
     # Find requested stream
@@ -53,8 +55,6 @@ def cmd_insert(self):
     if len(streams) != 1:
         self.die("error getting stream info for path %s", self.args.path)
 
-    layout = streams[0][1]
-
     if self.args.start and len(self.args.file) != 1:
         self.die("error: --start can only be used with one input file")
 
@@ -75,7 +75,7 @@ def cmd_insert(self):
                 start = self.args.start
             else:
                 try:
-                    start = self.parse_time(filename)
+                    start = nilmdb.utils.time.parse_time(filename)
                 except ValueError:
                     self.die("error extracting time from filename '%s'",
                              filename)
@@ -93,7 +93,7 @@ def cmd_insert(self):
 
         # Insert the data
         try:
-            result = self.client.stream_insert(self.args.path, ts)
+            self.client.stream_insert(self.args.path, ts)
         except nilmdb.client.Error as e:
             # TODO: It would be nice to be able to offer better errors
             # here, particularly in the case of overlap, which just shows

nilmdb/cmdline/list.py

@@ -1,6 +1,5 @@
 from nilmdb.utils.printf import *
-import nilmdb
-import nilmdb.client
+import nilmdb.utils.time
 
 import fnmatch
 import argparse
@@ -25,11 +24,13 @@ def setup(self, sub):
     group.add_argument("-l", "--layout", default="*",
                        help="Match only this stream layout")
 
+    group = cmd.add_argument_group("Interval extent")
+    group.add_argument("-E", "--extent", action="store_true",
+                       help="Show min/max timestamps in this stream")
+
     group = cmd.add_argument_group("Interval details")
     group.add_argument("-d", "--detail", action="store_true",
                        help="Show available data time intervals")
-    group.add_argument("-T", "--timestamp-raw", action="store_true",
-                       help="Show raw timestamps in time intervals")
     group.add_argument("-s", "--start",
                        metavar="TIME", type=self.arg_time,
                        help="Starting timestamp (free-form, inclusive)")
@@ -37,6 +38,12 @@ def setup(self, sub):
                        metavar="TIME", type=self.arg_time,
                        help="Ending timestamp (free-form, noninclusive)")
 
+    group = cmd.add_argument_group("Misc options")
+    group.add_argument("-T", "--timestamp-raw", action="store_true",
+                       help="Show raw timestamps in time intervals or extents")
+
+    return cmd
+
 def cmd_list_verify(self):
     # A hidden "path_positional" argument lets the user leave off the
     # "-p" when specifying the path.  Handle it here.
@@ -49,30 +56,40 @@ def cmd_list_verify(self):
             self.args.path = self.args.path_positional
 
     if self.args.start is not None and self.args.end is not None:
-        if self.args.start > self.args.end:
-            self.parser.error("start is after end")
+        if self.args.start >= self.args.end:
+            self.parser.error("start must precede end")
+
+    if self.args.start is not None or self.args.end is not None:
+        if not self.args.detail:
+            self.parser.error("--start and --end only make sense with --detail")
 
 def cmd_list(self):
     """List available streams"""
-    streams = self.client.stream_list()
+    streams = self.client.stream_list(extent = True)
 
     if self.args.timestamp_raw:
         time_string = repr
     else:
-        time_string = self.time_string
+        time_string = nilmdb.utils.time.format_time
 
-    for (path, layout) in streams:
+    for (path, layout, extent_min, extent_max) in streams:
         if not (fnmatch.fnmatch(path, self.args.path) and
                 fnmatch.fnmatch(layout, self.args.layout)):
             continue
 
         printf("%s %s\n", path, layout)
-        if not self.args.detail:
-            continue
 
+        if self.args.extent:
+            if extent_min is None or extent_max is None:
+                printf("  extent: (no data)\n")
+            else:
+                printf("  extent: %s -> %s\n",
+                       time_string(extent_min), time_string(extent_max))
+
+        if self.args.detail:
             printed = False
-        for (start, end) in self.client.stream_intervals(path, self.args.start,
-                                                         self.args.end):
+            for (start, end) in self.client.stream_intervals(
+                path, self.args.start, self.args.end):
                 printf("  [ %s -> %s ]\n", time_string(start), time_string(end))
                 printed = True
             if not printed:

nilmdb/cmdline/metadata.py

@@ -26,6 +26,7 @@ def setup(self, sub):
     exc.add_argument("-u", "--update", nargs="+", metavar="key=value",
                      help="Update metadata using provided "
                      "key=value pairs")
+    return cmd
 
 def cmd_metadata(self):
     """Manipulate metadata"""

nilmdb/cmdline/remove.py

@@ -1,7 +1,6 @@
 from nilmdb.utils.printf import *
 import nilmdb
 import nilmdb.client
-import sys
 
 def setup(self, sub):
     cmd = sub.add_parser("remove", help="Remove data",
@@ -9,8 +8,7 @@ def setup(self, sub):
                          Remove all data from a specified time range within a
                          stream.
                          """)
-    cmd.set_defaults(verify = cmd_remove_verify,
-                     handler = cmd_remove)
+    cmd.set_defaults(handler = cmd_remove)
 
     group = cmd.add_argument_group("Data selection")
     group.add_argument("path",
@@ -25,11 +23,7 @@ def setup(self, sub):
     group = cmd.add_argument_group("Output format")
     group.add_argument("-c", "--count", action="store_true",
                        help="Output number of data points removed")
-
-def cmd_remove_verify(self):
-    if self.args.start is not None and self.args.end is not None:
-        if self.args.start > self.args.end:
-            self.parser.error("start is after end")
+    return cmd
 
 def cmd_remove(self):
     try:

nilmdb/scripts/__init__.py

@@ -0,0 +1 @@
+# Command line scripts

nilmdb/scripts/nilmdb_server.py

@@ -0,0 +1,91 @@
+#!/usr/bin/python
+
+import nilmdb.server
+import argparse
+import os
+import socket
+
+def main():
+    """Main entry point for the 'nilmdb-server' command line script"""
+
+    parser = argparse.ArgumentParser(
+        description = 'Run the NilmDB server',
+        formatter_class = argparse.ArgumentDefaultsHelpFormatter)
+
+    parser.add_argument("-V", "--version", action="version",
+                        version = nilmdb.__version__)
+
+    group = parser.add_argument_group("Standard options")
+    group.add_argument('-a', '--address',
+                       help = 'Only listen on the given address',
+                       default = '0.0.0.0')
+    group.add_argument('-p', '--port', help = 'Listen on the given port',
+                       type = int, default = 12380)
+    group.add_argument('-d', '--database', help = 'Database directory',
+                       default = os.path.join(os.getcwd(), "db"))
+    group.add_argument('-q', '--quiet', help = 'Silence output',
+                       action = 'store_true')
+    group.add_argument('-n', '--nosync', help = 'Use asynchronous '
+                       'commits for sqlite transactions',
+                       action = 'store_true', default = False)
+    group.add_argument('-t', '--traceback',
+                       help = 'Provide tracebacks in client errors',
+                       action = 'store_true', default = False)
+
+    group = parser.add_argument_group("Debug options")
+    group.add_argument('-y', '--yappi', help = 'Run under yappi profiler and '
+                       'invoke interactive shell afterwards',
+                       action = 'store_true')
+
+    args = parser.parse_args()
+
+    # Create database object.  Needs to be serialized before passing
+    # to the Server.
+    db = nilmdb.utils.serializer_proxy(nilmdb.NilmDB)(args.database,
+                                                      sync = not args.nosync)
+
+    # Configure the server
+    if args.quiet:
+        embedded = True
+    else:
+        embedded = False
+    server = nilmdb.server.Server(db,
+                                  host = args.address,
+                                  port = args.port,
+                                  embedded = embedded,
+                                  force_traceback = args.traceback)
+
+    # Print info
+    if not args.quiet:
+        print "Version: %s" % nilmdb.__version__
+        print "Database: %s" % (os.path.realpath(args.database))
+        if args.address == '0.0.0.0' or args.address == '::':
+            host = socket.getfqdn()
+        else:
+            host = args.address
+        print "Server URL: http://%s:%d/" % ( host, args.port)
+        print "----"
+
+    # Run it
+    if args.yappi:
+        print "Running in yappi"
+        try:
+            import yappi
+            yappi.start()
+            server.start(blocking = True)
+        finally:
+            yappi.stop()
+            yappi.print_stats(sort_type = yappi.SORTTYPE_TTOT, limit = 50)
+            from IPython import embed
+            embed(header = "Use the yappi object to explore further, "
+                  "quit to exit")
+    else:
+        server.start(blocking = True)
+
+    # Clean up
+    if not args.quiet:
+        print "Closing database"
+        db.close()
+
+if __name__ == "__main__":
+    main()

nilmdb/scripts/nilmtool.py

@@ -0,0 +1,10 @@
+#!/usr/bin/python
+
+import nilmdb.cmdline
+
+def main():
+    """Main entry point for the 'nilmtool' command line script"""
+    nilmdb.cmdline.Cmdline().run()
+
+if __name__ == "__main__":
+    main()

nilmdb/server/__init__.py

@@ -1,15 +1,22 @@
 """nilmdb.server"""
 
+from __future__ import absolute_import
+
 # Try to set up pyximport to automatically rebuild Cython modules.  If
 # this doesn't work, it's OK, as long as the modules were built externally.
 # (e.g. python setup.py build_ext --inplace)
-try:
+try: # pragma: no cover
+    import Cython
+    import distutils.version
+    if (distutils.version.LooseVersion(Cython.__version__) <
+        distutils.version.LooseVersion("0.17")): # pragma: no cover
+        raise ImportError("Cython version too old")
     import pyximport
-    pyximport.install()
-    import layout
-except: # pragma: no cover
+    pyximport.install(inplace = True, build_in_temp = False)
+except (ImportError, TypeError): # pragma: no cover
     pass
 
-from .nilmdb import NilmDB
-from .server import Server
-from .errors import *
+import nilmdb.server.layout
+from nilmdb.server.nilmdb import NilmDB
+from nilmdb.server.server import Server
+from nilmdb.server.errors import NilmDBError, StreamError, OverlapError

nilmdb/server/bulkdata.py

@@ -8,19 +8,27 @@ import nilmdb
 from nilmdb.utils.printf import *
 
 import os
-import sys
 import cPickle as pickle
 import struct
-import fnmatch
 import mmap
 import re
 
+# If we have the faulthandler module, use it.  All of the mmap stuff
+# might trigger a SIGSEGV or SIGBUS if we're not careful, and
+# faulthandler will give a traceback in that case.  (the Python
+# interpreter will still die either way).
+try: # pragma: no cover
+    import faulthandler
+    faulthandler.enable()
+except: # pragma: no cover
+    pass
+
 # Up to 256 open file descriptors at any given time.
 # These variables are global so they can be used in the decorator arguments.
 table_cache_size = 16
 fd_cache_size = 16
 
-@nilmdb.utils.must_close(wrap_verify = True)
+@nilmdb.utils.must_close(wrap_verify = False)
 class BulkData(object):
     def __init__(self, basepath, **kwargs):
         self.basepath = basepath
@@ -91,8 +99,7 @@ class BulkData(object):
                 "float32": 'f',
                 "float64": 'd',
                 }
-            for n in range(layout.count):
-                struct_fmt += struct_mapping[layout.datatype]
+            struct_fmt += struct_mapping[layout.datatype] * layout.count
         except KeyError:
             raise ValueError("no such layout, or bad data types")
 
@@ -164,7 +171,67 @@ class BulkData(object):
         ospath = os.path.join(self.root, *elements)
         return Table(ospath)
 
-@nilmdb.utils.must_close(wrap_verify = True)
+@nilmdb.utils.must_close(wrap_verify = False)
+class File(object):
+    """Object representing a single file on disk.  Data can be appended,
+    or the self.mmap handle can be used for random reads."""
+
+    def __init__(self, root, subdir, filename):
+        # Create path if it doesn't exist
+        try:
+            os.mkdir(os.path.join(root, subdir))
+        except OSError:
+            pass
+
+        # Open/create file
+        self._f = open(os.path.join(root, subdir, filename), "a+b", 0)
+
+        # Seek to end, and get size
+        self._f.seek(0, 2)
+        self.size = self._f.tell()
+
+        # Open mmap object
+        self.mmap = None
+        self._mmap_reopen()
+
+    def _mmap_reopen(self):
+        if self.size == 0:
+            # Don't mmap if the file is empty; it would fail
+            pass
+        elif self.mmap is None:
+            # Not opened yet, so open it
+            self.mmap = mmap.mmap(self._f.fileno(), 0)
+        else:
+            # Already opened, so just resize it
+            self.mmap.resize(self.size)
+
+    def close(self):
+        if self.mmap is not None:
+            self.mmap.close()
+        self._f.close()
+
+    def append(self, data): # pragma: no cover (below version used instead)
+        # Write data, flush it, and resize our mmap accordingly
+        self._f.write(data)
+        self._f.flush()
+        self.size += len(data)
+        self._mmap_reopen()
+
+    def append_pack_iter(self, count, packer, dataiter):
+        # An optimized verison of append, to avoid flushing the file
+        # and resizing the mmap after each data point.
+        try:
+            rows = []
+            for i in xrange(count):
+                row = dataiter.next()
+                rows.append(packer(*row))
+            self._f.write("".join(rows))
+        finally:
+            self._f.flush()
+            self.size = self._f.tell()
+            self._mmap_reopen()
+
+@nilmdb.utils.must_close(wrap_verify = False)
 class Table(object):
     """Tools to help access a single table (data at a specific OS path)."""
     # See design.md for design details
@@ -185,12 +252,12 @@ class Table(object):
         packer = struct.Struct(struct_fmt)
         rows_per_file = max(file_size // packer.size, 1)
 
-        format = { "rows_per_file": rows_per_file,
+        fmt = { "rows_per_file": rows_per_file,
                 "files_per_dir": files_per_dir,
                 "struct_fmt": struct_fmt,
                 "version": 1 }
         with open(os.path.join(root, "_format"), "wb") as f:
-            pickle.dump(format, f, 2)
+            pickle.dump(fmt, f, 2)
 
     # Normal methods
     def __init__(self, root):
@@ -199,22 +266,22 @@ class Table(object):
 
         # Load the format and build packer
         with open(os.path.join(self.root, "_format"), "rb") as f:
-            format = pickle.load(f)
+            fmt = pickle.load(f)
 
-        if format["version"] != 1: # pragma: no cover (just future proofing)
-            raise NotImplementedError("version " + format["version"] +
+        if fmt["version"] != 1: # pragma: no cover (just future proofing)
+            raise NotImplementedError("version " + fmt["version"] +
                                       " bulk data store not supported")
 
-        self.rows_per_file = format["rows_per_file"]
-        self.files_per_dir = format["files_per_dir"]
-        self.packer = struct.Struct(format["struct_fmt"])
+        self.rows_per_file = fmt["rows_per_file"]
+        self.files_per_dir = fmt["files_per_dir"]
+        self.packer = struct.Struct(fmt["struct_fmt"])
         self.file_size = self.packer.size * self.rows_per_file
 
         # Find nrows
         self.nrows = self._get_nrows()
 
     def close(self):
-        self.mmap_open.cache_remove_all()
+        self.file_open.cache_remove_all()
 
     # Internal helpers
     def _get_nrows(self):
@@ -278,37 +345,11 @@ class Table(object):
 
     # Cache open files
     @nilmdb.utils.lru_cache(size = fd_cache_size,
-                            keys = slice(0,3), # exclude newsize
-                            onremove = lambda x: x.close())
-    def mmap_open(self, subdir, filename, newsize = None):
+                            onremove = lambda f: f.close())
+    def file_open(self, subdir, filename):
         """Open and map a given 'subdir/filename' (relative to self.root).
-        Will be automatically closed when evicted from the cache.
-
-        If 'newsize' is provided, the file is truncated to the given
-        size before the mapping is returned.  (Note that the LRU cache
-        on this function means the truncate will only happen if the
-        object isn't already cached; mmap.resize should be used too.)"""
-        try:
-            os.mkdir(os.path.join(self.root, subdir))
-        except OSError:
-            pass
-        f = open(os.path.join(self.root, subdir, filename), "a+", 0)
-        if newsize is not None:
-            # mmap can't map a zero-length file, so this allows the
-            # caller to set the filesize between file creation and
-            # mmap.
-            f.truncate(newsize)
-        mm = mmap.mmap(f.fileno(), 0)
-        return mm
-
-    def mmap_open_resize(self, subdir, filename, newsize):
-        """Open and map a given 'subdir/filename' (relative to self.root).
-        The file is resized to the given size."""
-        # Pass new size to mmap_open
-        mm = self.mmap_open(subdir, filename, newsize)
-        # In case we got a cached copy, need to call mm.resize too.
-        mm.resize(newsize)
-        return mm
+        Will be automatically closed when evicted from the cache."""
+        return File(self.root, subdir, filename)
 
     def append(self, data):
         """Append the data and flush it to disk.
@@ -320,14 +361,11 @@ class Table(object):
             (subdir, fname, offset, count) = self._offset_from_row(self.nrows)
             if count > remaining:
                 count = remaining
-            newsize = offset + count * self.packer.size
-            mm = self.mmap_open_resize(subdir, fname, newsize)
-            mm.seek(offset)
+
+            f = self.file_open(subdir, fname)
 
             # Write the data
-            for i in xrange(count):
-                row = dataiter.next()
-                mm.write(self.packer.pack(*row))
+            f.append_pack_iter(count, self.packer.pack, dataiter)
             remaining -= count
             self.nrows += count
 
@@ -354,7 +392,7 @@ class Table(object):
                 (subdir, filename, offset, count) = self._offset_from_row(row)
                 if count > remaining:
                     count = remaining
-                mm = self.mmap_open(subdir, filename)
+                mm = self.file_open(subdir, filename).mmap
                 for i in xrange(count):
                     ret.append(list(self.packer.unpack_from(mm, offset)))
                     offset += self.packer.size
@@ -366,14 +404,22 @@ class Table(object):
         if key < 0 or key >= self.nrows:
             raise IndexError("Index out of range")
         (subdir, filename, offset, count) = self._offset_from_row(key)
-        mm = self.mmap_open(subdir, filename)
+        mm = self.file_open(subdir, filename).mmap
         # unpack_from ignores the mmap object's current seek position
         return list(self.packer.unpack_from(mm, offset))
 
     def _remove_rows(self, subdir, filename, start, stop):
         """Helper to mark specific rows as being removed from a
-        file, and potentially removing or truncating the file itself."""
-        # Import an existing list of deleted rows for this file
+        file, and potentially remove or truncate the file itself."""
+        # Close potentially open file in file_open LRU cache
+        self.file_open.cache_remove(self, subdir, filename)
+
+        # We keep a file like 0000.removed that contains a list of
+        # which rows have been "removed".  Note that we never have to
+        # remove entries from this list, because we never decrease
+        # self.nrows, and so we will never overwrite those locations in the
+        # file.  Only when the list covers the entire extent of the
+        # file will that file be removed.
         datafile = os.path.join(self.root, subdir, filename)
         cachefile = datafile + ".removed"
         try:
@@ -413,8 +459,8 @@ class Table(object):
         # are generally easier if we don't have to special-case that.
         if (len(merged) == 1 and
             merged[0][0] == 0 and merged[0][1] == self.rows_per_file):
-            # Close potentially open file in mmap_open LRU cache
-            self.mmap_open.cache_remove(self, subdir, filename)
+            # Close potentially open file in file_open LRU cache
+            self.file_open.cache_remove(self, subdir, filename)
 
             # Delete files
             os.remove(datafile)
@@ -427,6 +473,14 @@ class Table(object):
             except:
                 pass
         else:
+            # File needs to stick around.  This means we can get
+            # degenerate cases where we have large files containing as
+            # little as one row.  Try to punch a hole in the file,
+            # so that this region doesn't take up filesystem space.
+            offset = start * self.packer.size
+            count = (stop - start) * self.packer.size
+            nilmdb.utils.fallocate.punch_hole(datafile, offset, count)
+
             # Update cache.  Try to do it atomically.
             nilmdb.utils.atomic.replace_file(cachefile,
                                              pickle.dumps(merged, 2))

nilmdb/server/interval.pyx

@@ -36,7 +36,7 @@ cdef class Interval:
         """
         'start' and 'end' are arbitrary floats that represent time
         """
-        if start > end:
+        if start >= end:
             # Explicitly disallow zero-width intervals (since they're half-open)
             raise IntervalError("start %s must precede end %s" % (start, end))
         self.start = float(start)

nilmdb/server/layout.pyx

@@ -4,7 +4,6 @@ import time
 import sys
 import inspect
 import cStringIO
-import numpy as np
 
 cdef enum:
     max_value_count = 64
@@ -42,10 +41,12 @@ class Layout:
 
         if datatype == 'uint16':
             self.parse = self.parse_uint16
-            self.format = self.format_uint16
+            self.format_str = "%.6f" + " %d" * self.count
+            self.format = self.format_generic
         elif datatype == 'float32' or datatype == 'float64':
             self.parse = self.parse_float64
-            self.format = self.format_float64
+            self.format_str = "%.6f" + " %f" * self.count
+            self.format = self.format_generic
         else:
             raise KeyError("invalid type")
 
@@ -57,15 +58,15 @@ class Layout:
         cdef double ts
         # Return doubles even in float32 case, since they're going into
         # a Python array which would upconvert to double anyway.
-        result = []
+        result = [0] * (self.count + 1)
         cdef char *end
         ts = libc.stdlib.strtod(text, &end)
         if end == text:
             raise ValueError("bad timestamp")
-        result.append(ts)
+        result[0] = ts
         for n in range(self.count):
             text = end
-            result.append(libc.stdlib.strtod(text, &end))
+            result[n+1] = libc.stdlib.strtod(text, &end)
             if end == text:
                 raise ValueError("wrong number of values")
         n = 0
@@ -79,18 +80,18 @@ class Layout:
         cdef int n
         cdef double ts
         cdef int v
-        result = []
         cdef char *end
+        result = [0] * (self.count + 1)
         ts = libc.stdlib.strtod(text, &end)
         if end == text:
             raise ValueError("bad timestamp")
-        result.append(ts)
+        result[0] = ts
         for n in range(self.count):
             text = end
             v = libc.stdlib.strtol(text, &end, 10)
             if v < 0 or v > 65535:
                 raise ValueError("value out of range")
-            result.append(v)
+            result[n+1] = v
             if end == text:
                 raise ValueError("wrong number of values")
         n = 0
@@ -101,25 +102,12 @@ class Layout:
         return (ts, result)
 
     # Formatters
-    def format_float64(self, d):
+    def format_generic(self, d):
         n = len(d) - 1
         if n != self.count:
             raise ValueError("wrong number of values for layout type: "
                              "got %d, wanted %d" % (n, self.count))
-        s = "%.6f" % d[0]
-        for i in range(n):
-            s += " %f" % d[i+1]
-        return s + "\n"
-
-    def format_uint16(self, d):
-        n = len(d) - 1
-        if n != self.count:
-            raise ValueError("wrong number of values for layout type: "
-                             "got %d, wanted %d" % (n, self.count))
-        s = "%.6f" % d[0]
-        for i in range(n):
-            s += " %d" % d[i+1]
-        return s + "\n"
+        return (self.format_str % tuple(d)) + "\n"
 
 # Get a layout by name
 def get_named(typestring):
@@ -154,7 +142,7 @@ class Parser(object):
         layout, into an internal data structure suitable for a
         pytables 'table.append(parser.data)'.
         """
-        cdef double last_ts = 0, ts
+        cdef double last_ts = -1e12, ts
         cdef int n = 0, i
         cdef char *line
 

nilmdb/server/nilmdb.py

@@ -15,11 +15,9 @@ from nilmdb.utils.printf import *
 from nilmdb.server.interval import (Interval, DBInterval,
                                     IntervalSet, IntervalError)
 from nilmdb.server import bulkdata
-from nilmdb.server.errors import *
+from nilmdb.server.errors import NilmDBError, StreamError, OverlapError
 
 import sqlite3
-import time
-import sys
 import os
 import errno
 import bisect
@@ -80,7 +78,10 @@ class NilmDB(object):
     verbose = 0
 
     def __init__(self, basepath, sync=True, max_results=None,
-                 bulkdata_args={}):
+                 bulkdata_args=None):
+        if bulkdata_args is None:
+            bulkdata_args = {}
+
         # set up path
         self.basepath = os.path.abspath(basepath)
 
@@ -96,12 +97,7 @@ class NilmDB(object):
 
         # SQLite database too
         sqlfilename = os.path.join(self.basepath, "data.sql")
-        # We use check_same_thread = False, assuming that the rest
-        # of the code (e.g. Server) will be smart and not access this
-        # database from multiple threads simultaneously.  Otherwise
-        # false positives will occur when the database is only opened
-        # in one thread, and only accessed in another.
-        self.con = sqlite3.connect(sqlfilename, check_same_thread = False)
+        self.con = sqlite3.connect(sqlfilename, check_same_thread = True)
         self._sql_schema_update()
 
         # See big comment at top about the performance implications of this
@@ -141,6 +137,15 @@ class NilmDB(object):
             with self.con:
                 cur.execute("PRAGMA user_version = {v:d}".format(v=version))
 
+    def _check_user_times(self, start, end):
+        if start is None:
+            start = -1e12
+        if end is None:
+            end = 1e12
+        if start >= end:
+            raise NilmDBError("start must precede end")
+        return (start, end)
+
     @nilmdb.utils.lru_cache(size = 16)
     def _get_intervals(self, stream_id):
         """
@@ -156,7 +161,7 @@ class NilmDB(object):
                 iset += DBInterval(start_time, end_time,
                                    start_time, end_time,
                                    start_pos, end_pos)
-        except IntervalError as e: # pragma: no cover
+        except IntervalError: # pragma: no cover
             raise NilmDBError("unexpected overlap in ranges table!")
 
         return iset
@@ -264,28 +269,39 @@ class NilmDB(object):
 
         return
 
-    def stream_list(self, path = None, layout = None):
-        """Return list of [path, layout] lists of all streams
-        in the database.
+    def stream_list(self, path = None, layout = None, extent = False):
+        """Return list of lists of all streams in the database.
 
         If path is specified, include only streams with a path that
         matches the given string.
 
         If layout is specified, include only streams with a layout
         that matches the given string.
-        """
-        where = "WHERE 1=1"
-        params = ()
-        if layout:
-            where += " AND layout=?"
-            params += (layout,)
-        if path:
-            where += " AND path=?"
-            params += (path,)
-        result = self.con.execute("SELECT path, layout "
-                                  "FROM streams " + where, params).fetchall()
 
-        return sorted(list(x) for x in result)
+        If extent = False, returns a list of lists containing
+        the path and layout: [ path, layout ]
+
+        If extent = True, returns a list of lists containing the
+        path, layout, and min/max extent of the data:
+        [ path, layout, extent_min, extent_max ]
+        """
+        params = ()
+        query = "SELECT streams.path, streams.layout"
+        if extent:
+            query += ", min(ranges.start_time), max(ranges.end_time)"
+        query += " FROM streams"
+        if extent:
+            query += " LEFT JOIN ranges ON streams.id = ranges.stream_id"
+        query += " WHERE 1=1"
+        if layout is not None:
+            query += " AND streams.layout=?"
+            params += (layout,)
+        if path is not None:
+            query += " AND streams.path=?"
+            params += (path,)
+        query += " GROUP BY streams.id ORDER BY streams.path"
+        result = self.con.execute(query, params).fetchall()
+        return [ list(x) for x in result ]
 
     def stream_intervals(self, path, start = None, end = None):
         """
@@ -302,7 +318,8 @@ class NilmDB(object):
         """
         stream_id = self._stream_id(path)
         intervals = self._get_intervals(stream_id)
-        requested = Interval(start or 0, end or 1e12)
+        (start, end) = self._check_user_times(start, end)
+        requested = Interval(start, end)
         result = []
         for n, i in enumerate(intervals.intersection(requested)):
             if n >= self.max_results:
@@ -395,7 +412,7 @@ class NilmDB(object):
            path: Path at which to add the data
            start: Starting timestamp
            end: Ending timestamp
-           data: Rows of data, to be passed to PyTable's table.append
+           data: Rows of data, to be passed to bulkdata table.append
                  method.  E.g. nilmdb.layout.Parser.data
            """
         # First check for basic overlap using timestamp info given.
@@ -416,7 +433,7 @@ class NilmDB(object):
         self._add_interval(stream_id, interval, row_start, row_end)
 
         # And that's all
-        return "ok"
+        return
 
     def _find_start(self, table, dbinterval):
         """
@@ -474,7 +491,8 @@ class NilmDB(object):
         stream_id = self._stream_id(path)
         table = self.data.getnode(path)
         intervals = self._get_intervals(stream_id)
-        requested = Interval(start or 0, end or 1e12)
+        (start, end) = self._check_user_times(start, end)
+        requested = Interval(start, end)
         result = []
         matched = 0
         remaining = self.max_results
@@ -520,12 +538,10 @@ class NilmDB(object):
         stream_id = self._stream_id(path)
         table = self.data.getnode(path)
         intervals = self._get_intervals(stream_id)
-        to_remove = Interval(start or 0, end or 1e12)
+        (start, end) = self._check_user_times(start, end)
+        to_remove = Interval(start, end)
         removed = 0
 
-        if start == end:
-            return 0
-
         # Can't remove intervals from within the iterator, so we need to
         # remember what's currently in the intersection now.
         all_candidates = list(intervals.intersection(to_remove, orig = True))

nilmdb/server/server.py

@@ -5,29 +5,20 @@
 from __future__ import absolute_import
 import nilmdb
 from nilmdb.utils.printf import *
-from nilmdb.server.errors import *
+from nilmdb.server.errors import NilmDBError
 
 import cherrypy
 import sys
-import time
 import os
 import simplejson as json
 import decorator
 import traceback
-
-try:
-    import cherrypy
-    cherrypy.tools.json_out
-except: # pragma: no cover
-    sys.stderr.write("Cherrypy 3.2+ required\n")
-    sys.exit(1)
+import psutil
 
 class NilmApp(object):
     def __init__(self, db):
         self.db = db
 
-version = "1.2"
-
 # Decorators
 def chunked_response(func):
     """Decorator to enable chunked responses."""
@@ -57,7 +48,7 @@ def workaround_cp_bug_1200(func, *args, **kwargs): # pragma: no cover
     try:
         for val in func(*args, **kwargs):
             yield val
-    except (LookupError, UnicodeError) as e:
+    except (LookupError, UnicodeError):
         raise Exception("bug workaround; real exception is:\n" +
                         traceback.format_exc())
 
@@ -80,13 +71,66 @@ def exception_to_httperror(*expected):
     # care of that.
     return decorator.decorator(wrapper)
 
+# Custom CherryPy tools
+
+def CORS_allow(methods):
+    """This does several things:
+
+    Handles CORS preflight requests.
+    Adds Allow: header to all requests.
+    Raise 405 if request.method not in method.
+
+    It is similar to cherrypy.tools.allow, with the CORS stuff added.
+    """
+    request = cherrypy.request.headers
+    response = cherrypy.response.headers
+
+    if not isinstance(methods, (tuple, list)): # pragma: no cover
+        methods = [ methods ]
+    methods = [ m.upper() for m in methods if m ]
+    if not methods: # pragma: no cover
+        methods = [ 'GET', 'HEAD' ]
+    elif 'GET' in methods and 'HEAD' not in methods: # pragma: no cover
+        methods.append('HEAD')
+    response['Allow'] = ', '.join(methods)
+
+    # Allow all origins
+    if 'Origin' in request:
+        response['Access-Control-Allow-Origin'] = request['Origin']
+
+    # If it's a CORS request, send response.
+    request_method = request.get("Access-Control-Request-Method", None)
+    request_headers = request.get("Access-Control-Request-Headers", None)
+    if (cherrypy.request.method == "OPTIONS" and
+        request_method and request_headers):
+        response['Access-Control-Allow-Headers'] = request_headers
+        response['Access-Control-Allow-Methods'] = ', '.join(methods)
+        # Try to stop further processing and return a 200 OK
+        cherrypy.response.status = "200 OK"
+        cherrypy.response.body = ""
+        cherrypy.request.handler = lambda: ""
+        return
+
+    # Reject methods that were not explicitly allowed
+    if cherrypy.request.method not in methods:
+        raise cherrypy.HTTPError(405)
+
+cherrypy.tools.CORS_allow = cherrypy.Tool('on_start_resource', CORS_allow)
+
+# Helper for json_in tool to process JSON data into normal request
+# parameters.
+def json_to_request_params(body):
+    cherrypy.lib.jsontools.json_processor(body)
+    if not isinstance(cherrypy.request.json, dict):
+        raise cherrypy.HTTPError(415)
+    cherrypy.request.params.update(cherrypy.request.json)
+
 # CherryPy apps
 class Root(NilmApp):
     """Root application for NILM database"""
 
-    def __init__(self, db, version):
+    def __init__(self, db):
         super(Root, self).__init__(db)
-        self.server_version = version
 
     # /
     @cherrypy.expose
@@ -102,38 +146,47 @@ class Root(NilmApp):
     @cherrypy.expose
     @cherrypy.tools.json_out()
     def version(self):
-        return self.server_version
+        return nilmdb.__version__
 
-    # /dbpath
+    # /dbinfo
     @cherrypy.expose
     @cherrypy.tools.json_out()
-    def dbpath(self):
-        return self.db.get_basepath()
-
-    # /dbsize
-    @cherrypy.expose
-    @cherrypy.tools.json_out()
-    def dbsize(self):
-        return nilmdb.utils.du(self.db.get_basepath())
+    def dbinfo(self):
+        """Return a dictionary with the database path,
+        size of the database in bytes, and free disk space in bytes"""
+        path = self.db.get_basepath()
+        return { "path": path,
+                 "size": nilmdb.utils.du(path),
+                 "free": psutil.disk_usage(path).free }
 
 class Stream(NilmApp):
     """Stream-specific operations"""
 
     # /stream/list
     # /stream/list?layout=PrepData
-    # /stream/list?path=/newton/prep
+    # /stream/list?path=/newton/prep&extent=1
     @cherrypy.expose
     @cherrypy.tools.json_out()
-    def list(self, path = None, layout = None):
+    def list(self, path = None, layout = None, extent = None):
         """List all streams in the database.  With optional path or
         layout parameter, just list streams that match the given path
-        or layout"""
-        return self.db.stream_list(path, layout)
+        or layout.
+
+        If extent is not given, returns a list of lists containing
+        the path and layout: [ path, layout ]
+
+        If extent is provided, returns a list of lists containing the
+        path, layout, and min/max extent of the data:
+        [ path, layout, extent_min, extent_max ]
+        """
+        return self.db.stream_list(path, layout, bool(extent))
 
     # /stream/create?path=/newton/prep&layout=PrepData
     @cherrypy.expose
+    @cherrypy.tools.json_in()
     @cherrypy.tools.json_out()
     @exception_to_httperror(NilmDBError, ValueError)
+    @cherrypy.tools.CORS_allow(methods = ["POST"])
     def create(self, path, layout):
         """Create a new stream in the database.  Provide path
         and one of the nilmdb.layout.layouts keys.
@@ -142,8 +195,10 @@ class Stream(NilmApp):
 
     # /stream/destroy?path=/newton/prep
     @cherrypy.expose
+    @cherrypy.tools.json_in()
     @cherrypy.tools.json_out()
     @exception_to_httperror(NilmDBError)
+    @cherrypy.tools.CORS_allow(methods = ["POST"])
     def destroy(self, path):
         """Delete a stream and its associated data."""
         return self.db.stream_destroy(path)
@@ -174,31 +229,41 @@ class Stream(NilmApp):
 
     # /stream/set_metadata?path=/newton/prep&data=<json>
     @cherrypy.expose
+    @cherrypy.tools.json_in()
     @cherrypy.tools.json_out()
-    @exception_to_httperror(NilmDBError, LookupError, TypeError)
+    @exception_to_httperror(NilmDBError, LookupError)
+    @cherrypy.tools.CORS_allow(methods = ["POST"])
     def set_metadata(self, path, data):
-        """Set metadata for the named stream, replacing any
-        existing metadata.  Data should be a json-encoded
-        dictionary"""
-        data_dict = json.loads(data)
-        self.db.stream_set_metadata(path, data_dict)
-        return "ok"
+        """Set metadata for the named stream, replacing any existing
+        metadata.  Data can be json-encoded or a plain dictionary (if
+        it was sent as application/json to begin with)"""
+        if not isinstance(data, dict):
+            try:
+                data = dict(json.loads(data))
+            except TypeError as e:
+                raise NilmDBError("can't parse 'data' parameter: " + e.message)
+        self.db.stream_set_metadata(path, data)
 
     # /stream/update_metadata?path=/newton/prep&data=<json>
     @cherrypy.expose
+    @cherrypy.tools.json_in()
     @cherrypy.tools.json_out()
-    @exception_to_httperror(NilmDBError, LookupError, TypeError)
+    @exception_to_httperror(NilmDBError, LookupError)
+    @cherrypy.tools.CORS_allow(methods = ["POST"])
     def update_metadata(self, path, data):
         """Update metadata for the named stream.  Data
         should be a json-encoded dictionary"""
-        data_dict = json.loads(data)
-        self.db.stream_update_metadata(path, data_dict)
-        return "ok"
+        if not isinstance(data, dict):
+            try:
+                data = dict(json.loads(data))
+            except TypeError as e:
+                raise NilmDBError("can't parse 'data' parameter: " + e.message)
+        self.db.stream_update_metadata(path, data)
 
     # /stream/insert?path=/newton/prep
     @cherrypy.expose
     @cherrypy.tools.json_out()
-    #@cherrypy.tools.disable_prb()
+    @cherrypy.tools.CORS_allow(methods = ["PUT"])
     def insert(self, path, start, end):
         """
         Insert new data into the database.  Provide textual data
@@ -206,12 +271,9 @@ class Stream(NilmApp):
         """
         # Important that we always read the input before throwing any
         # errors, to keep lengths happy for persistent connections.
-        # However, CherryPy 3.2.2 has a bug where this fails for GET
-        # requests, so catch that. (issue #1134)
-        try:
+        # Note that CherryPy 3.2.2 has a bug where this fails for GET
+        # requests, if we ever want to handle those (issue #1134)
         body = cherrypy.request.body.read()
-        except TypeError:
-            raise cherrypy.HTTPError("400 Bad Request", "No request body")
 
         # Check path and get layout
         streams = self.db.stream_list(path = path)
@@ -228,37 +290,37 @@ class Stream(NilmApp):
                                      "error parsing input data: " +
                                      e.message)
 
-        if (not parser.min_timestamp or not parser.max_timestamp or
-            not len(parser.data)):
-            raise cherrypy.HTTPError("400 Bad Request",
-                                     "no data provided")
-
         # Check limits
         start = float(start)
         end = float(end)
-        if parser.min_timestamp < start:
+        if start >= end:
+            raise cherrypy.HTTPError("400 Bad Request",
+                                     "start must precede end")
+        if parser.min_timestamp is not None and parser.min_timestamp < start:
             raise cherrypy.HTTPError("400 Bad Request", "Data timestamp " +
                                      repr(parser.min_timestamp) +
                                      " < start time " + repr(start))
-        if parser.max_timestamp >= end:
+        if parser.max_timestamp is not None and parser.max_timestamp >= end:
             raise cherrypy.HTTPError("400 Bad Request", "Data timestamp " +
                                      repr(parser.max_timestamp) +
                                      " >= end time " + repr(end))
 
         # Now do the nilmdb insert, passing it the parser full of data.
         try:
-            result = self.db.stream_insert(path, start, end, parser.data)
+            self.db.stream_insert(path, start, end, parser.data)
         except NilmDBError as e:
             raise cherrypy.HTTPError("400 Bad Request", e.message)
 
         # Done
-        return "ok"
+        return
 
     # /stream/remove?path=/newton/prep
     # /stream/remove?path=/newton/prep&start=1234567890.0&end=1234567899.0
     @cherrypy.expose
+    @cherrypy.tools.json_in()
     @cherrypy.tools.json_out()
     @exception_to_httperror(NilmDBError)
+    @cherrypy.tools.CORS_allow(methods = ["POST"])
     def remove(self, path, start = None, end = None):
         """
         Remove data from the backend database.  Removes all data in
@@ -270,26 +332,25 @@ class Stream(NilmApp):
         if end is not None:
             end = float(end)
         if start is not None and end is not None:
-            if end < start:
+            if start >= end:
                 raise cherrypy.HTTPError("400 Bad Request",
-                                         "end before start")
+                                         "start must precede end")
         return self.db.stream_remove(path, start, end)
 
     # /stream/intervals?path=/newton/prep
     # /stream/intervals?path=/newton/prep&start=1234567890.0&end=1234567899.0
     @cherrypy.expose
     @chunked_response
-    @response_type("text/plain")
+    @response_type("application/x-json-stream")
     def intervals(self, path, start = None, end = None):
         """
         Get intervals from backend database.  Streams the resulting
-        intervals as JSON strings separated by newlines.  This may
+        intervals as JSON strings separated by CR LF pairs.  This may
         make multiple requests to the nilmdb backend to avoid causing
         it to block for too long.
 
-        Note that the response type is set to 'text/plain' even
-        though we're sending back JSON; this is because we're not
-        really returning a single JSON object.
+        Note that the response type is the non-standard
+        'application/x-json-stream' for lack of a better option.
         """
         if start is not None:
             start = float(start)
@@ -297,9 +358,9 @@ class Stream(NilmApp):
             end = float(end)
 
         if start is not None and end is not None:
-            if end < start:
+            if start >= end:
                 raise cherrypy.HTTPError("400 Bad Request",
-                                         "end before start")
+                                         "start must precede end")
 
         streams = self.db.stream_list(path = path)
         if len(streams) != 1:
@@ -309,8 +370,8 @@ class Stream(NilmApp):
         def content(start, end):
             # Note: disable chunked responses to see tracebacks from here.
             while True:
-                (intervals, restart) = self.db.stream_intervals(path,start,end)
-                response = ''.join([ json.dumps(i) + "\n" for i in intervals ])
+                (ints, restart) = self.db.stream_intervals(path, start, end)
+                response = ''.join([ json.dumps(i) + "\r\n" for i in ints ])
                 yield response
                 if restart == 0:
                     break
@@ -337,9 +398,9 @@ class Stream(NilmApp):
 
         # Check parameters
         if start is not None and end is not None:
-            if end < start:
+            if start >= end:
                 raise cherrypy.HTTPError("400 Bad Request",
-                                         "end before start")
+                                         "start must precede end")
 
         # Check path and get layout
         streams = self.db.stream_list(path = path)
@@ -387,31 +448,51 @@ class Server(object):
                  fast_shutdown = False,   # don't wait for clients to disconn.
                  force_traceback = False  # include traceback in all errors
                  ):
-        self.version = version
+        # Save server version, just for verification during tests
+        self.version = nilmdb.__version__
 
-        # Need to wrap DB object in a serializer because we'll call
-        # into it from separate threads.
         self.embedded = embedded
-        self.db = nilmdb.utils.Serializer(db)
+        self.db = db
+        if not getattr(db, "_thread_safe", None):
+            raise KeyError("Database object " + str(db) + " doesn't claim "
+                           "to be thread safe.  You should pass "
+                           "nilmdb.utils.serializer_proxy(NilmDB)(args) "
+                           "rather than NilmDB(args).")
+
+        # Build up global server configuration
         cherrypy.config.update({
             'server.socket_host': host,
             'server.socket_port': port,
             'engine.autoreload_on': False,
-            'server.max_request_body_size': 4*1024*1024,
-            'error_page.default': self.json_error_page,
+            'server.max_request_body_size': 8*1024*1024,
             })
         if self.embedded:
             cherrypy.config.update({ 'environment': 'embedded' })
 
-        # Send a permissive Access-Control-Allow-Origin (CORS) header
-        # with all responses so that browsers can send cross-domain
-        # requests to this server.
-        cherrypy.config.update({ 'response.headers.Access-Control-Allow-Origin':
-                                 '*' })
+        # Build up application specific configuration
+        app_config = {}
+        app_config.update({
+            'error_page.default': self.json_error_page,
+            })
+
+        # Some default headers to just help identify that things are working
+        app_config.update({ 'response.headers.X-Jim-Is-Awesome': 'yeah' })
+
+        # Set up Cross-Origin Resource Sharing (CORS) handler so we
+        # can correctly respond to browsers' CORS preflight requests.
+        # This also limits verbs to GET and HEAD by default.
+        app_config.update({ 'tools.CORS_allow.on': True,
+                            'tools.CORS_allow.methods': ['GET', 'HEAD'] })
+
+        # Configure the 'json_in' tool to also allow other content-types
+        # (like x-www-form-urlencoded), and to treat JSON as a dict that
+        # fills requests.param.
+        app_config.update({ 'tools.json_in.force': False,
+                            'tools.json_in.processor': json_to_request_params })
 
         # Send tracebacks in error responses.  They're hidden by the
         # error_page function for client errors (code 400-499).
-        cherrypy.config.update({ 'request.show_tracebacks' : True })
+        app_config.update({ 'request.show_tracebacks' : True })
         self.force_traceback = force_traceback
 
         # Patch CherryPy error handler to never pad out error messages.
@@ -419,11 +500,13 @@ class Server(object):
         # error messages.
         cherrypy._cperror._ie_friendly_error_sizes = {}
 
-        cherrypy.tree.apps = {}
-        cherrypy.tree.mount(Root(self.db, self.version), "/")
-        cherrypy.tree.mount(Stream(self.db), "/stream")
+        # Build up the application and mount it
+        root = Root(self.db)
+        root.stream = Stream(self.db)
         if stoppable:
-            cherrypy.tree.mount(Exiter(), "/exit")
+            root.exit = Exiter()
+        cherrypy.tree.apps = {}
+        cherrypy.tree.mount(root, "/", config = { "/" : app_config })
 
         # Shutdowns normally wait for clients to disconnect.  To speed
         # up tests, set fast_shutdown = True
@@ -444,7 +527,7 @@ class Server(object):
             if not self.force_traceback:
                 if code >= 400 and code <= 499:
                     errordata["traceback"] = ""
-        except Exception as e: # pragma: no cover
+        except Exception: # pragma: no cover
             pass
         # Override the response type, which was previously set to text/html
         cherrypy.serving.response.headers['Content-Type'] = (

nilmdb/utils/__init__.py

@@ -1,11 +1,11 @@
 """NilmDB utilities"""
 
-from .timer import Timer
-from .iteratorizer import Iteratorizer
-from .serializer import Serializer
-from .lrucache import lru_cache
-from .diskusage import du
-from .mustclose import must_close
-from .urllib import urlencode
-from . import misc
-from . import atomic
+from nilmdb.utils.timer import Timer
+from nilmdb.utils.iteratorizer import Iteratorizer
+from nilmdb.utils.serializer import serializer_proxy
+from nilmdb.utils.lrucache import lru_cache
+from nilmdb.utils.diskusage import du, human_size
+from nilmdb.utils.mustclose import must_close
+from nilmdb.utils import atomic
+import nilmdb.utils.threadsafety
+import nilmdb.utils.fallocate

nilmdb/utils/diskusage.py

@@ -1,7 +1,7 @@
 import os
 from math import log
 
-def sizeof_fmt(num):
+def human_size(num):
     """Human friendly file size"""
     unit_list = zip(['bytes', 'kiB', 'MiB', 'GiB', 'TiB'], [0, 0, 1, 2, 2])
     if num > 1:
@@ -15,15 +15,11 @@ def sizeof_fmt(num):
     if num == 1: # pragma: no cover
         return '1 byte'
 
-def du_bytes(path):
+def du(path):
     """Like du -sb, returns total size of path in bytes."""
     size = os.path.getsize(path)
     if os.path.isdir(path):
-        for file in os.listdir(path):
-            filepath = os.path.join(path, file)
-            size += du_bytes(filepath)
+        for thisfile in os.listdir(path):
+            filepath = os.path.join(path, thisfile)
+            size += du(filepath)
     return size
-
-def du(path):
-    """Like du -sh, returns total size of path as a human-readable string."""
-    return sizeof_fmt(du_bytes(path))

nilmdb/utils/fallocate.py

@@ -0,0 +1,49 @@
+# Implementation of hole punching via fallocate, if the OS
+# and filesystem support it.
+
+try:
+    import os
+    import ctypes
+    import ctypes.util
+
+    def make_fallocate():
+        libc_name = ctypes.util.find_library('c')
+        libc = ctypes.CDLL(libc_name, use_errno=True)
+
+        _fallocate = libc.fallocate
+        _fallocate.restype = ctypes.c_int
+        _fallocate.argtypes = [ ctypes.c_int, ctypes.c_int,
+                                ctypes.c_int64, ctypes.c_int64 ]
+
+        del libc
+        del libc_name
+
+        def fallocate(fd, mode, offset, len_):
+            res = _fallocate(fd, mode, offset, len_)
+            if res != 0: # pragma: no cover
+                errno = ctypes.get_errno()
+                raise IOError(errno, os.strerror(errno))
+        return fallocate
+
+    fallocate = make_fallocate()
+    del make_fallocate
+except Exception: # pragma: no cover
+    fallocate = None
+
+FALLOC_FL_KEEP_SIZE = 0x01
+FALLOC_FL_PUNCH_HOLE = 0x02
+
+def punch_hole(filename, offset, length, ignore_errors = True):
+    """Punch a hole in the file.  This isn't well supported, so errors
+    are ignored by default."""
+    try:
+        if fallocate is None: # pragma: no cover
+            raise IOError("fallocate not available")
+        with open(filename, "r+") as f:
+            fallocate(f.fileno(),
+                      FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
+                      offset, length)
+    except IOError: # pragma: no cover
+        if ignore_errors:
+            return
+        raise

nilmdb/utils/iteratorizer.py

@@ -16,6 +16,7 @@ class IteratorizerThread(threading.Thread):
         callback (provided by this class) as an argument
         """
         threading.Thread.__init__(self)
+        self.name = "Iteratorizer-" + function.__name__ + "-" + self.name
         self.function = function
         self.queue = queue
         self.die = False
@@ -95,5 +96,5 @@ def Iteratorizer(function, curl_hack = False):
         while thread.isAlive():
             try:
                 queue.get(True, 0.01)
-            except:
+            except: # pragma: no cover
                 pass

nilmdb/utils/lrucache.py

@@ -5,7 +5,6 @@
 
 import collections
 import decorator
-import warnings
 
 def lru_cache(size = 10, onremove = None, keys = slice(None)):
     """Least-recently-used cache decorator.

nilmdb/utils/misc.py

@@ -1,8 +0,0 @@
-import itertools
-
-def pairwise(iterable):
-    "s -> (s0,s1), (s1,s2), ..., (sn,None)"
-    a, b = itertools.tee(iterable)
-    next(b, None)
-    return itertools.izip_longest(a, b)
-

nilmdb/utils/mustclose.py

@@ -12,15 +12,12 @@ def must_close(errorfile = sys.stderr, wrap_verify = False):
     already been called."""
     def class_decorator(cls):
 
-        # Helper to replace a class method with a wrapper function,
-        # while maintaining argument specs etc.
-        def wrap_class_method(wrapper_func):
-            method = wrapper_func.__name__
-            if method in cls.__dict__:
-                orig = getattr(cls, method).im_func
-            else:
-                orig = lambda self: None
-            setattr(cls, method, decorator.decorator(wrapper_func, orig))
+        def wrap_class_method(wrapper):
+            try:
+                orig = getattr(cls, wrapper.__name__).im_func
+            except:
+                orig = lambda x: None
+            setattr(cls, wrapper.__name__, decorator.decorator(wrapper, orig))
 
         @wrap_class_method
         def __init__(orig, self, *args, **kwargs):
@@ -38,6 +35,7 @@ def must_close(errorfile = sys.stderr, wrap_verify = False):
 
         @wrap_class_method
         def close(orig, self, *args, **kwargs):
+            if "_must_close" in self.__dict__:
                 del self._must_close
             return orig(self, *args, **kwargs)
 

nilmdb/utils/serializer.py

@@ -1,6 +1,10 @@
 import Queue
 import threading
 import sys
+import decorator
+import inspect
+import types
+import functools
 
 # This file provides a class that will wrap an object and serialize
 # all calls to its methods.  All calls to that object will be queued
@@ -12,8 +16,9 @@ import sys
 class SerializerThread(threading.Thread):
     """Thread that retrieves call information from the queue, makes the
     call, and returns the results."""
-    def __init__(self, call_queue):
+    def __init__(self, classname, call_queue):
         threading.Thread.__init__(self)
+        self.name = "Serializer-" + classname + "-" + self.name
         self.call_queue = call_queue
 
     def run(self):
@@ -22,51 +27,83 @@ class SerializerThread(threading.Thread):
             # Terminate if result_queue is None
             if result_queue is None:
                 return
+            exception = None
+            result = None
             try:
                 result = func(*args, **kwargs) # wrapped
             except:
-                result_queue.put((sys.exc_info(), None))
-            else:
-                result_queue.put((None, result))
+                exception = sys.exc_info()
+            # Ensure we delete these before returning a result, so
+            # we don't unncessarily hold onto a reference while
+            # we're waiting for the next call.
+            del func, args, kwargs
+            result_queue.put((exception, result))
+            del exception, result
 
-class WrapCall(object):
-    """Wrap a callable using the given queues"""
+def serializer_proxy(obj_or_type):
+    """Wrap the given object or type in a SerializerObjectProxy.
 
-    def __init__(self, call_queue, result_queue, func):
+    Returns a SerializerObjectProxy object that proxies all method
+    calls to the object, as well as attribute retrievals.
+
+    The proxied requests, including instantiation, are performed in a
+    single thread and serialized between caller threads.
+    """
+    class SerializerCallProxy(object):
+        def __init__(self, call_queue, func, objectproxy):
             self.call_queue = call_queue
-        self.result_queue = result_queue
             self.func = func
-
+            # Need to hold a reference to object proxy so it doesn't
+            # go away (and kill the thread) until after get called.
+            self.objectproxy = objectproxy
         def __call__(self, *args, **kwargs):
-        self.call_queue.put((self.result_queue, self.func, args, kwargs))
-        ( exc_info, result ) = self.result_queue.get()
+            result_queue = Queue.Queue()
+            self.call_queue.put((result_queue, self.func, args, kwargs))
+            ( exc_info, result ) = result_queue.get()
             if exc_info is None:
                 return result
             else:
                 raise exc_info[0], exc_info[1], exc_info[2]
 
-class WrapObject(object):
-    """Wrap all calls to methods in a target object with WrapCall"""
-
-    def __init__(self, target):
-        self.__wrap_target = target
-        self.__wrap_call_queue = Queue.Queue()
-        self.__wrap_serializer = SerializerThread(self.__wrap_call_queue)
-        self.__wrap_serializer.daemon = True
-        self.__wrap_serializer.start()
+    class SerializerObjectProxy(object):
+        def __init__(self, obj_or_type, *args, **kwargs):
+            self.__object = obj_or_type
+            try:
+                if type(obj_or_type) in (types.TypeType, types.ClassType):
+                    classname = obj_or_type.__name__
+                else:
+                    classname = obj_or_type.__class__.__name__
+            except AttributeError: # pragma: no cover
+                classname = "???"
+            self.__call_queue = Queue.Queue()
+            self.__thread = SerializerThread(classname, self.__call_queue)
+            self.__thread.daemon = True
+            self.__thread.start()
+            self._thread_safe = True
 
         def __getattr__(self, key):
-        """Wrap methods of self.__wrap_target in a WrapCall instance"""
-        func = getattr(self.__wrap_target, key)
-        if not callable(func):
-            raise TypeError("Can't serialize attribute %r (type: %s)"
-                            % (key, type(func)))
-        result_queue = Queue.Queue()
-        return WrapCall(self.__wrap_call_queue, result_queue, func)
+            if key.startswith("_SerializerObjectProxy__"): # pragma: no cover
+                raise AttributeError
+            attr = getattr(self.__object, key)
+            if not callable(attr):
+                getter = SerializerCallProxy(self.__call_queue, getattr, self)
+                return getter(self.__object, key)
+            r = SerializerCallProxy(self.__call_queue, attr, self)
+            return r
+
+        def __call__(self, *args, **kwargs):
+            """Call this to instantiate the type, if a type was passed
+            to serializer_proxy.  Otherwise, pass the call through."""
+            ret = SerializerCallProxy(self.__call_queue,
+                                      self.__object, self)(*args, **kwargs)
+            if type(self.__object) in (types.TypeType, types.ClassType):
+                # Instantiation
+                self.__object = ret
+                return self
+            return ret
 
         def __del__(self):
-        self.__wrap_call_queue.put((None, None, None, None))
-        self.__wrap_serializer.join()
+            self.__call_queue.put((None, None, None, None))
+            self.__thread.join()
 
-# Just an alias
-Serializer = WrapObject
+    return SerializerObjectProxy(obj_or_type)

nilmdb/utils/threadsafety.py

@@ -0,0 +1,109 @@
+from nilmdb.utils.printf import *
+import threading
+import warnings
+import types
+
+def verify_proxy(obj_or_type, exception = False, check_thread = True,
+                 check_concurrent = True):
+    """Wrap the given object or type in a VerifyObjectProxy.
+
+    Returns a VerifyObjectProxy that proxies all method calls to the
+    given object, as well as attribute retrievals.
+
+    When calling methods, the following checks are performed.  If
+    exception is True, an exception is raised.  Otherwise, a warning
+    is printed.
+
+    check_thread = True     # Warn/fail if two different threads call methods.
+    check_concurrent = True # Warn/fail if two functions are concurrently
+                            # run through this proxy
+    """
+    class Namespace(object):
+        pass
+    class VerifyCallProxy(object):
+        def __init__(self, func, parent_namespace):
+            self.func = func
+            self.parent_namespace = parent_namespace
+
+        def __call__(self, *args, **kwargs):
+            p = self.parent_namespace
+            this = threading.current_thread()
+            try:
+                callee = self.func.__name__
+            except AttributeError:
+                callee = "???"
+
+            if p.thread is None:
+                p.thread = this
+                p.thread_callee = callee
+
+            if check_thread and p.thread != this:
+                err = sprintf("unsafe threading: %s called %s.%s,"
+                              " but %s called %s.%s",
+                              p.thread.name, p.classname, p.thread_callee,
+                              this.name, p.classname, callee)
+                if exception:
+                    raise AssertionError(err)
+                else: # pragma: no cover
+                    warnings.warn(err)
+
+            need_concur_unlock = False
+            if check_concurrent:
+                if p.concur_lock.acquire(False) == False:
+                    err = sprintf("unsafe concurrency: %s called %s.%s "
+                                  "while %s is still in %s.%s",
+                                  this.name, p.classname, callee,
+                                  p.concur_tname, p.classname, p.concur_callee)
+                    if exception:
+                        raise AssertionError(err)
+                    else: # pragma: no cover
+                        warnings.warn(err)
+                else:
+                    p.concur_tname = this.name
+                    p.concur_callee = callee
+                    need_concur_unlock = True
+
+            try:
+                ret = self.func(*args, **kwargs)
+            finally:
+                if need_concur_unlock:
+                    p.concur_lock.release()
+            return ret
+
+    class VerifyObjectProxy(object):
+        def __init__(self, obj_or_type, *args, **kwargs):
+            p = Namespace()
+            self.__ns = p
+            p.thread = None
+            p.thread_callee = None
+            p.concur_lock = threading.Lock()
+            p.concur_tname = None
+            p.concur_callee = None
+            self.__obj = obj_or_type
+            try:
+                if type(obj_or_type) in (types.TypeType, types.ClassType):
+                    p.classname = self.__obj.__name__
+                else:
+                    p.classname = self.__obj.__class__.__name__
+            except AttributeError: # pragma: no cover
+                p.classname = "???"
+
+        def __getattr__(self, key):
+            if key.startswith("_VerifyObjectProxy__"): # pragma: no cover
+                raise AttributeError
+            attr = getattr(self.__obj, key)
+            if not callable(attr):
+                return VerifyCallProxy(getattr, self.__ns)(self.__obj, key)
+            return VerifyCallProxy(attr, self.__ns)
+
+        def __call__(self, *args, **kwargs):
+            """Call this to instantiate the type, if a type was passed
+            to verify_proxy.  Otherwise, pass the call through."""
+            ret = VerifyCallProxy(self.__obj, self.__ns)(*args, **kwargs)
+            if type(self.__obj) in (types.TypeType, types.ClassType):
+                # Instantiation
+                self.__obj = ret
+                return self
+            return ret
+
+    return VerifyObjectProxy(obj_or_type)

nilmdb/utils/time.py

@@ -0,0 +1,54 @@
+from nilmdb.utils import datetime_tz
+import re
+
+def parse_time(toparse):
+    """
+    Parse a free-form time string and return a datetime_tz object.
+    If the string doesn't contain a timestamp, the current local
+    timezone is assumed (e.g. from the TZ env var).
+    """
+    # If string isn't "now" and doesn't contain at least 4 digits,
+    # consider it invalid.  smartparse might otherwise accept
+    # empty strings and strings with just separators.
+    if toparse != "now" and len(re.findall(r"\d", toparse)) < 4:
+        raise ValueError("not enough digits for a timestamp")
+
+    # Try to just parse the time as given
+    try:
+        return datetime_tz.datetime_tz.smartparse(toparse)
+    except ValueError:
+        pass
+
+    # Try to extract a substring in a condensed format that we expect
+    # to see in a filename or header comment
+    res = re.search(r"(^|[^\d])("            # non-numeric or SOL
+                    r"(199\d|2\d\d\d)"       # year
+                    r"[-/]?"                 # separator
+                    r"(0[1-9]|1[012])"       # month
+                    r"[-/]?"                 # separator
+                    r"([012]\d|3[01])"       # day
+                    r"[-T ]?"                # separator
+                    r"([01]\d|2[0-3])"       # hour
+                    r"[:]?"                  # separator
+                    r"([0-5]\d)"             # minute
+                    r"[:]?"                  # separator
+                    r"([0-5]\d)?"            # second
+                    r"([-+]\d\d\d\d)?"       # timezone
+                    r")", toparse)
+    if res is not None:
+        try:
+            return datetime_tz.datetime_tz.smartparse(res.group(2))
+        except ValueError:
+            pass
+
+    # Could also try to successively parse substrings, but let's
+    # just give up for now.
+    raise ValueError("unable to parse timestamp")
+
+def format_time(timestamp):
+    """
+    Convert a Unix timestamp to a string for printing, using the
+    local timezone for display (e.g. from the TZ env var).
+    """
+    dt = datetime_tz.datetime_tz.fromtimestamp(timestamp)
+    return dt.strftime("%a, %d %b %Y %H:%M:%S.%f %z")

nilmdb/utils/timer.py

@@ -2,10 +2,11 @@
 
 # Simple timer to time a block of code, for optimization debugging
 # use like:
-#   with nilmdb.Timer("flush"):
+#   with nilmdb.utils.Timer("flush"):
 #       foo.flush()
 
 from __future__ import print_function
+from __future__ import absolute_import
 import contextlib
 import time
 

nilmdb/utils/timestamper.py

@@ -3,19 +3,16 @@
 from nilmdb.utils.printf import *
 from nilmdb.utils import datetime_tz
 
-import time
-import os
-
 class Timestamper(object):
     """A file-like object that adds timestamps to lines of an input file."""
-    def __init__(self, file, ts_iter):
+    def __init__(self, infile, ts_iter):
         """file: filename, or another file-like object
            ts_iter: iterator that returns a timestamp string for
            each line of the file"""
-        if isinstance(file, basestring):
-            self.file = open(file, "r")
+        if isinstance(infile, basestring):
+            self.file = open(infile, "r")
         else:
-            self.file = file
+            self.file = infile
         self.ts_iter = ts_iter
 
     def close(self):
@@ -54,7 +51,7 @@ class Timestamper(object):
 
 class TimestamperRate(Timestamper):
     """Timestamper that uses a start time and a fixed rate"""
-    def __init__(self, file, start, rate, end = None):
+    def __init__(self, infile, start, rate, end = None):
         """
         file: file name or object
 
@@ -76,7 +73,7 @@ class TimestamperRate(Timestamper):
         # Handle case where we're passed a datetime or datetime_tz object
         if "totimestamp" in dir(start):
             start = start.totimestamp()
-        Timestamper.__init__(self, file, iterator(start, rate, end))
+        Timestamper.__init__(self, infile, iterator(start, rate, end))
         self.start = start
         self.rate = rate
     def __str__(self):
@@ -87,21 +84,21 @@ class TimestamperRate(Timestamper):
 
 class TimestamperNow(Timestamper):
     """Timestamper that uses current time"""
-    def __init__(self, file):
+    def __init__(self, infile):
         def iterator():
             while True:
                 now = datetime_tz.datetime_tz.utcnow().totimestamp()
                 yield sprintf("%.6f ", now)
-        Timestamper.__init__(self, file, iterator())
+        Timestamper.__init__(self, infile, iterator())
     def __str__(self):
         return "TimestamperNow(...)"
 
 class TimestamperNull(Timestamper):
     """Timestamper that adds nothing to each line"""
-    def __init__(self, file):
+    def __init__(self, infile):
         def iterator():
             while True:
                 yield ""
-        Timestamper.__init__(self, file, iterator())
+        Timestamper.__init__(self, infile, iterator())
     def __str__(self):
         return "TimestamperNull(...)"

nilmdb/utils/urllib.py

@@ -1,37 +0,0 @@
-from __future__ import absolute_import
-from urllib import quote_plus, _is_unicode
-
-# urllib.urlencode insists on encoding Unicode as ASCII.  This is based
-# on that function, except we always encode it as UTF-8 instead.
-
-def urlencode(query):
-    """Encode a dictionary into a URL query string.
-
-    If any values in the query arg are sequences, each sequence
-    element is converted to a separate parameter.
-    """
-
-    query = query.items()
-
-    l = []
-    for k, v in query:
-        k = quote_plus(str(k))
-        if isinstance(v, str):
-            v = quote_plus(v)
-            l.append(k + '=' + v)
-        elif _is_unicode(v):
-            v = quote_plus(v.encode("utf-8","strict"))
-            l.append(k + '=' + v)
-        else:
-            try:
-                # is this a sufficient test for sequence-ness?
-                len(v)
-            except TypeError:
-                # not a sequence
-                v = quote_plus(str(v))
-                l.append(k + '=' + v)
-            else:
-                # loop over the sequence
-                for elt in v:
-                    l.append(k + '=' + quote_plus(str(elt)))
-    return '&'.join(l)

nilmtool.py

@@ -1,6 +0,0 @@
-#!/usr/bin/python
-
-import nilmdb
-import sys
-
-nilmdb.cmdline.Cmdline(sys.argv[1:]).run()

runserver.py

@@ -1,35 +0,0 @@
-#!/usr/bin/python
-
-import nilmdb
-import argparse
-
-formatter = argparse.ArgumentDefaultsHelpFormatter
-parser = argparse.ArgumentParser(description='Run the NILM server',
-                                 formatter_class = formatter)
-parser.add_argument('-p', '--port', help='Port number', type=int, default=12380)
-parser.add_argument('-d', '--database', help='Database directory', default="db")
-parser.add_argument('-y', '--yappi', help='Run with yappi profiler',
-                    action='store_true')
-args = parser.parse_args()
-
-# Start web app on a custom port
-db = nilmdb.NilmDB(args.database)
-server = nilmdb.Server(db, host = "127.0.0.1",
-                       port = args.port,
-                       embedded = False)
-
-
-if args.yappi:
-    print "Running in yappi"
-    try:
-        import yappi
-        yappi.start()
-        server.start(blocking = True)
-    finally:
-        yappi.stop()
-        print "Try: yappi.print_stats(sort_type=yappi.SORTTYPE_TTOT,limit=50)"
-        from IPython import embed
-        embed()
-else:
-    server.start(blocking = True)
-db.close()

setup.cfg

@@ -20,6 +20,7 @@ cover-erase=1
 stop=1
 verbosity=2
 tests=tests
+#tests=tests/test_threadsafety.py
 #tests=tests/test_bulkdata.py
 #tests=tests/test_mustclose.py
 #tests=tests/test_lrucache.py

setup.py

@@ -1,5 +1,11 @@
 #!/usr/bin/python
 
+# To release a new version, tag it:
+#   git tag -a nilmdb-1.1 -m "Version 1.1"
+#   git push --tags
+# Then just package it up:
+#   python setup.py sdist
+
 # This is supposed to be using Distribute:
 #
 #   distutils provides a "setup" method.
@@ -9,32 +15,107 @@
 # So we don't really know if this is using the old setuptools or the
 # Distribute-provided version of setuptools.
 
+import traceback
+import sys
+import os
+
+try:
     from setuptools import setup, find_packages
     from distutils.extension import Extension
+    import distutils.version
+except ImportError:
+    traceback.print_exc()
+    print "Please install the prerequisites listed in README.txt"
+    sys.exit(1)
 
-from Cython.Build import cythonize
+# Versioneer manages version numbers from git tags.
+# https://github.com/warner/python-versioneer
+import versioneer
+versioneer.versionfile_source = 'nilmdb/_version.py'
+versioneer.versionfile_build = 'nilmdb/_version.py'
+versioneer.tag_prefix = 'nilmdb-'
+versioneer.parentdir_prefix = 'nilmdb-'
 
 # Hack to workaround logging/multiprocessing issue:
 # https://groups.google.com/d/msg/nose-users/fnJ-kAUbYHQ/_UsLN786ygcJ
 try: import multiprocessing
 except: pass
 
-# Build cython modules.
-cython_modules = cythonize("**/*.pyx")
+# Use Cython if it's new enough, otherwise use preexisting C files.
+cython_modules = [ 'nilmdb.server.interval',
+                   'nilmdb.server.layout',
+                   'nilmdb.server.rbtree' ]
+try:
+    import Cython
+    from Cython.Build import cythonize
+    if (distutils.version.LooseVersion(Cython.__version__) <
+        distutils.version.LooseVersion("0.16")):
+        print "Cython version", Cython.__version__, "is too old; not using it."
+        raise ImportError()
+    use_cython = True
+except ImportError:
+    use_cython = False
+
+ext_modules = []
+for modulename in cython_modules:
+    filename = modulename.replace('.','/')
+    if use_cython:
+        ext_modules.extend(cythonize(filename + ".pyx"))
+    else:
+        cfile = filename + ".c"
+        if not os.path.exists(cfile):
+            raise Exception("Missing source file " + cfile + ".  "
+                            "Try installing cython >= 0.16.")
+        ext_modules.append(Extension(modulename, [ cfile ]))
+
+# We need a MANIFEST.in.  Generate it here rather than polluting the
+# repository with yet another setup-related file.
+with open("MANIFEST.in", "w") as m:
+    m.write("""
+# Root
+include README.txt
+include setup.cfg
+include setup.py
+include versioneer.py
+include Makefile
+include .coveragerc
+include .pylintrc
+
+# Cython files -- include source.
+recursive-include nilmdb/server *.pyx *.pyxdep *.pxd
+
+# Tests
+recursive-include tests *.py
+recursive-include tests/data *
+include tests/test.order
+
+# Docs
+recursive-include docs Makefile *.md
+""")
 
 # Run setup
 setup(name='nilmdb',
-      version = '1.0',
+      version = versioneer.get_version(),
+      cmdclass = versioneer.get_cmdclass(),
       url = 'https://git.jim.sh/jim/lees/nilmdb.git',
       author = 'Jim Paris',
+      description = "NILM Database",
+      long_description = "NILM Database",
+      license = "Proprietary",
       author_email = 'jim@jtan.com',
       tests_require = [ 'nose',
                         'coverage',
                         ],
-      setup_requires = [ 'cython',
+      setup_requires = [ 'distribute',
                          ],
-      install_requires = [ 'distribute',
-                           'decorator',
+      install_requires = [ 'decorator',
+                           'cherrypy >= 3.2',
+                           'simplejson',
+                           'pycurl',
+                           'python-dateutil',
+                           'pytz',
+                           'psutil >= 0.3.0',
+                           'requests >= 1.1.0, < 2.0.0',
                            ],
       packages = [ 'nilmdb',
                    'nilmdb.utils',
@@ -42,7 +123,14 @@ setup(name='nilmdb',
                    'nilmdb.server',
                    'nilmdb.client',
                    'nilmdb.cmdline',
+                   'nilmdb.scripts',
                    ],
-      ext_modules = cython_modules,
+      entry_points = {
+          'console_scripts': [
+              'nilmtool = nilmdb.scripts.nilmtool:main',
+              'nilmdb-server = nilmdb.scripts.nilmdb_server:main',
+              ],
+          },
+      ext_modules = ext_modules,
       zip_safe = False,
       )

tests/runtests.py

@@ -6,6 +6,9 @@ import sys
 import glob
 from collections import OrderedDict
 
+# Change into parent dir
+os.chdir(os.path.dirname(os.path.realpath(__file__)) + "/..")
+
 class JimOrderPlugin(nose.plugins.Plugin):
     """When searching for tests and encountering a directory that
     contains a 'test.order' file, run tests listed in that file, in the

tests/test.order

@@ -1,4 +1,5 @@
 test_printf.py
+test_threadsafety.py
 test_lrucache.py
 test_mustclose.py
 

tests/test_client.py

@@ -6,6 +6,7 @@ from nilmdb.utils import timestamper
 from nilmdb.client import ClientError, ServerError
 from nilmdb.utils import datetime_tz
 
+from nose.plugins.skip import SkipTest
 from nose.tools import *
 from nose.tools import assert_raises
 import itertools
@@ -18,10 +19,13 @@ import simplejson as json
 import unittest
 import warnings
 import resource
+import time
+import re
 
 from testutil.helpers import *
 
 testdb = "tests/client-testdb"
+testurl = "http://localhost:32180/"
 
 def setup_module():
     global test_server, test_db
@@ -29,9 +33,9 @@ def setup_module():
     recursive_unlink(testdb)
 
     # Start web app on a custom port
-    test_db = nilmdb.NilmDB(testdb, sync = False)
+    test_db = nilmdb.utils.serializer_proxy(nilmdb.NilmDB)(testdb, sync = False)
     test_server = nilmdb.Server(test_db, host = "127.0.0.1",
-                                port = 12380, stoppable = False,
+                                port = 32180, stoppable = False,
                                 fast_shutdown = True,
                                 force_traceback = False)
     test_server.start(blocking = False)
@@ -44,39 +48,38 @@ def teardown_module():
 
 class TestClient(object):
 
-    def test_client_1_basic(self):
+    def test_client_01_basic(self):
         # Test a fake host
         client = nilmdb.Client(url = "http://localhost:1/")
         with assert_raises(nilmdb.client.ServerError):
             client.version()
-
-        # Trigger same error with a PUT request
-        client = nilmdb.Client(url = "http://localhost:1/")
-        with assert_raises(nilmdb.client.ServerError):
-            client.version()
+        client.close()
 
         # Then a fake URL on a real host
-        client = nilmdb.Client(url = "http://localhost:12380/fake/")
+        client = nilmdb.Client(url = "http://localhost:32180/fake/")
         with assert_raises(nilmdb.client.ClientError):
             client.version()
+        client.close()
 
         # Now a real URL with no http:// prefix
-        client = nilmdb.Client(url = "localhost:12380")
+        client = nilmdb.Client(url = "localhost:32180")
         version = client.version()
+        client.close()
 
         # Now use the real URL
-        client = nilmdb.Client(url = "http://localhost:12380/")
+        client = nilmdb.Client(url = testurl)
         version = client.version()
-        eq_(distutils.version.StrictVersion(version),
-            distutils.version.StrictVersion(test_server.version))
+        eq_(distutils.version.LooseVersion(version),
+            distutils.version.LooseVersion(test_server.version))
 
         # Bad URLs should give 404, not 500
         with assert_raises(ClientError):
             client.http.get("/stream/create")
+        client.close()
 
-    def test_client_2_createlist(self):
+    def test_client_02_createlist(self):
         # Basic stream tests, like those in test_nilmdb:test_stream
-        client = nilmdb.Client(url = "http://localhost:12380/")
+        client = nilmdb.Client(url = testurl)
 
         # Database starts empty
         eq_(client.stream_list(), [])
@@ -90,6 +93,15 @@ class TestClient(object):
         with assert_raises(ClientError):
             client.stream_create("/newton/prep", "NoSuchLayout")
 
+        # Bad method types
+        with assert_raises(ClientError):
+            client.http.put("/stream/list","")
+        # Try a bunch of times to make sure the request body is getting consumed
+        for x in range(10):
+            with assert_raises(ClientError):
+                client.http.post("/stream/list")
+        client = nilmdb.Client(url = testurl)
+
         # Create three streams
         client.stream_create("/newton/prep", "PrepData")
         client.stream_create("/newton/raw", "RawData")
@@ -101,8 +113,10 @@ class TestClient(object):
                                     ["/newton/zzz/rawnotch", "RawNotchedData"]
                                     ])
         # Match just one type or one path
-        eq_(client.stream_list(layout="RawData"), [ ["/newton/raw", "RawData"] ])
-        eq_(client.stream_list(path="/newton/raw"), [ ["/newton/raw", "RawData"] ])
+        eq_(client.stream_list(layout="RawData"),
+            [ ["/newton/raw", "RawData"] ])
+        eq_(client.stream_list(path="/newton/raw"),
+            [ ["/newton/raw", "RawData"] ])
 
         # Try messing with resource limits to trigger errors and get
         # more coverage.  Here, make it so we can only create files 1
@@ -114,9 +128,10 @@ class TestClient(object):
             client.stream_create("/newton/hello", "RawData")
         resource.setrlimit(resource.RLIMIT_FSIZE, limit)
 
+        client.close()
 
-    def test_client_3_metadata(self):
-        client = nilmdb.Client(url = "http://localhost:12380/")
+    def test_client_03_metadata(self):
+        client = nilmdb.Client(url = testurl)
 
         # Set / get metadata
         eq_(client.stream_get_metadata("/newton/prep"), {})
@@ -131,9 +146,10 @@ class TestClient(object):
         client.stream_update_metadata("/newton/raw", meta3)
         eq_(client.stream_get_metadata("/newton/prep"), meta1)
         eq_(client.stream_get_metadata("/newton/raw"), meta1)
-        eq_(client.stream_get_metadata("/newton/raw", [ "description" ] ), meta2)
-        eq_(client.stream_get_metadata("/newton/raw", [ "description",
-                                                        "v_scale" ] ), meta1)
+        eq_(client.stream_get_metadata("/newton/raw",
+                                       [ "description" ] ), meta2)
+        eq_(client.stream_get_metadata("/newton/raw",
+                                       [ "description", "v_scale" ] ), meta1)
 
         # missing key
         eq_(client.stream_get_metadata("/newton/raw", "descr"),
@@ -146,9 +162,14 @@ class TestClient(object):
             client.stream_set_metadata("/newton/prep", [1,2,3])
         with assert_raises(ClientError):
             client.stream_update_metadata("/newton/prep", [1,2,3])
+        client.close()
 
-    def test_client_4_insert(self):
-        client = nilmdb.Client(url = "http://localhost:12380/")
+    def test_client_04_insert(self):
+        client = nilmdb.Client(url = testurl)
+
+        # Limit _max_data to 1 MB, since our test file is 1.5 MB
+        old_max_data = nilmdb.client.client.StreamInserter._max_data
+        nilmdb.client.client.StreamInserter._max_data = 1 * 1024 * 1024
 
         datetime_tz.localtz_set("America/New_York")
 
@@ -177,12 +198,33 @@ class TestClient(object):
         result = client.stream_insert("/newton/prep", data)
         eq_(result, None)
 
-        # Try forcing a server request with empty data
+        # It's OK to insert an empty interval
+        client.http.put("stream/insert", "", { "path": "/newton/prep",
+                                               "start": 1, "end": 2 })
+        eq_(list(client.stream_intervals("/newton/prep")), [[1, 2]])
+        client.stream_remove("/newton/prep")
+        eq_(list(client.stream_intervals("/newton/prep")), [])
+
+        # Timestamps can be negative too
+        client.http.put("stream/insert", "", { "path": "/newton/prep",
+                                               "start": -2, "end": -1 })
+        eq_(list(client.stream_intervals("/newton/prep")), [[-2, -1]])
+        client.stream_remove("/newton/prep")
+        eq_(list(client.stream_intervals("/newton/prep")), [])
+
+        # Intervals that end at zero shouldn't be any different
+        client.http.put("stream/insert", "", { "path": "/newton/prep",
+                                               "start": -1, "end": 0 })
+        eq_(list(client.stream_intervals("/newton/prep")), [[-1, 0]])
+        client.stream_remove("/newton/prep")
+        eq_(list(client.stream_intervals("/newton/prep")), [])
+
+        # Try forcing a server request with equal start and end
         with assert_raises(ClientError) as e:
             client.http.put("stream/insert", "", { "path": "/newton/prep",
                                                    "start": 0, "end": 0 })
         in_("400 Bad Request", str(e.exception))
-        in_("no data provided", str(e.exception))
+        in_("start must precede end", str(e.exception))
 
         # Specify start/end (starts too late)
         data = timestamper.TimestamperRate(testfile, start, 120)
@@ -201,14 +243,14 @@ class TestClient(object):
         in_("400 Bad Request", str(e.exception))
         # Client chunks the input, so the exact timestamp here might change
         # if the chunk positions change.
-        in_("Data timestamp 1332511271.016667 >= end time 1332511201.0",
-            str(e.exception))
+        assert(re.search("Data timestamp 13325[0-9]+\.[0-9]+ "
+                         ">= end time 1332511201.0", str(e.exception))
+               is not None)
 
         # Now do the real load
         data = timestamper.TimestamperRate(testfile, start, 120)
         result = client.stream_insert("/newton/prep", data,
                                       start, start + 119.999777)
-        eq_(result, "ok")
 
         # Verify the intervals.  Should be just one, even if the data
         # was inserted in chunks, due to nilmdb interval concatenation.
@@ -222,20 +264,28 @@ class TestClient(object):
         in_("400 Bad Request", str(e.exception))
         in_("verlap", str(e.exception))
 
-    def test_client_5_extractremove(self):
-        # Misc tests for extract and remove.  Most of them are in test_cmdline.
-        client = nilmdb.Client(url = "http://localhost:12380/")
+        nilmdb.client.client.StreamInserter._max_data = old_max_data
+        client.close()
 
-        for x in client.stream_extract("/newton/prep", 123, 123):
+    def test_client_05_extractremove(self):
+        # Misc tests for extract and remove.  Most of them are in test_cmdline.
+        client = nilmdb.Client(url = testurl)
+
+        for x in client.stream_extract("/newton/prep", 999123, 999124):
             raise AssertionError("shouldn't be any data for this request")
 
         with assert_raises(ClientError) as e:
             client.stream_remove("/newton/prep", 123, 120)
 
-    def test_client_6_generators(self):
+        # Test count
+        eq_(client.stream_count("/newton/prep"), 14400)
+
+        client.close()
+
+    def test_client_06_generators(self):
         # A lot of the client functionality is already tested by test_cmdline,
         # but this gets a bit more coverage that cmdline misses.
-        client = nilmdb.Client(url = "http://localhost:12380/")
+        client = nilmdb.Client(url = testurl)
 
         # Trigger a client error in generator
         start = datetime_tz.datetime_tz.smartparse("20120323T2000")
@@ -246,7 +296,7 @@ class TestClient(object):
                          start.totimestamp(),
                          end.totimestamp()).next()
             in_("400 Bad Request", str(e.exception))
-            in_("end before start", str(e.exception))
+            in_("start must precede end", str(e.exception))
 
         # Trigger a curl error in generator
         with assert_raises(ServerError) as e:
@@ -256,24 +306,6 @@ class TestClient(object):
         with assert_raises(ServerError) as e:
             client.http.get_gen("http://nosuchurl/").next()
 
-        # Check non-json version of string output
-        eq_(json.loads(client.http.get("/stream/list",retjson=False)),
-            client.http.get("/stream/list",retjson=True))
-
-        # Check non-json version of generator output
-        for (a, b) in itertools.izip(
-            client.http.get_gen("/stream/list",retjson=False),
-            client.http.get_gen("/stream/list",retjson=True)):
-            eq_(json.loads(a), b)
-
-        # Check PUT with generator out
-        with assert_raises(ClientError) as e:
-            client.http.put_gen("stream/insert", "",
-                                { "path": "/newton/prep",
-                                  "start": 0, "end": 0 }).next()
-        in_("400 Bad Request", str(e.exception))
-        in_("no data provided", str(e.exception))
-
         # Check 404 for missing streams
         for function in [ client.stream_intervals, client.stream_extract ]:
             with assert_raises(ClientError) as e:
@@ -281,48 +313,54 @@ class TestClient(object):
             in_("404 Not Found", str(e.exception))
             in_("No such stream", str(e.exception))
 
-    def test_client_7_headers(self):
+        client.close()
+
+    def test_client_07_headers(self):
         # Make sure that /stream/intervals and /stream/extract
         # properly return streaming, chunked, text/plain response.
         # Pokes around in client.http internals a bit to look at the
         # response headers.
 
-        client = nilmdb.Client(url = "http://localhost:12380/")
+        client = nilmdb.Client(url = testurl)
         http = client.http
 
-        # Use a warning rather than returning a test failure, so that we can
-        # still disable chunked responses for debugging.
+        # Use a warning rather than returning a test failure for the
+        # transfer-encoding, so that we can still disable chunked
+        # responses for debugging.
+
+        def headers():
+            h = ""
+            for (k, v) in http._last_response.headers.items():
+                h += k + ": " + v + "\n"
+            return h.lower()
 
         # Intervals
-        x = http.get("stream/intervals", { "path": "/newton/prep" },
-                            retjson=False)
-        lines_(x, 1)
-        if "Transfer-Encoding: chunked" not in http._headers:
+        x = http.get("stream/intervals", { "path": "/newton/prep" })
+        if "transfer-encoding: chunked" not in headers():
             warnings.warn("Non-chunked HTTP response for /stream/intervals")
-        if "Content-Type: text/plain;charset=utf-8" not in http._headers:
-            raise AssertionError("/stream/intervals is not text/plain:\n" +
-                                 http._headers)
+        if "content-type: application/x-json-stream" not in headers():
+            raise AssertionError("/stream/intervals content type "
+                                 "is not application/x-json-stream:\n" +
+                                 headers())
 
         # Extract
         x = http.get("stream/extract",
                             { "path": "/newton/prep",
                               "start": "123",
-                              "end": "123" }, retjson=False)
-        if "Transfer-Encoding: chunked" not in http._headers:
+                              "end": "124" })
+        if "transfer-encoding: chunked" not in headers():
             warnings.warn("Non-chunked HTTP response for /stream/extract")
-        if "Content-Type: text/plain;charset=utf-8" not in http._headers:
+        if "content-type: text/plain;charset=utf-8" not in headers():
             raise AssertionError("/stream/extract is not text/plain:\n" +
-                                 http._headers)
+                                 headers())
 
-        # Make sure Access-Control-Allow-Origin gets set
-        if "Access-Control-Allow-Origin: " not in http._headers:
-            raise AssertionError("No Access-Control-Allow-Origin (CORS) "
-                                 "header in /stream/extract response:\n" +
-                                 http._headers)
+        client.close()
 
-    def test_client_8_unicode(self):
+    def test_client_08_unicode(self):
+        # Try both with and without posting JSON
+        for post_json in (False, True):
             # Basic Unicode tests
-        client = nilmdb.Client(url = "http://localhost:12380/")
+            client = nilmdb.Client(url = testurl, post_json = post_json)
 
             # Delete streams that exist
             for stream in client.stream_list():
@@ -356,3 +394,209 @@ class TestClient(object):
             eq_(client.stream_get_metadata(raw[0]), meta1)
             eq_(client.stream_get_metadata(raw[0], [ "alpha" ]), meta2)
             eq_(client.stream_get_metadata(raw[0], [ "alpha", "β" ]), meta1)
+
+            client.close()
+
+    def test_client_09_closing(self):
+        # Make sure we actually close sockets correctly.  New
+        # connections will block for a while if they're not, since the
+        # server will stop accepting new connections.
+        for test in [1, 2]:
+            start = time.time()
+            for i in range(50):
+                if time.time() - start > 15:
+                    raise AssertionError("Connections seem to be blocking... "
+                                         "probably not closing properly.")
+                if test == 1:
+                    # explicit close
+                    client = nilmdb.Client(url = testurl)
+                    with assert_raises(ClientError) as e:
+                        client.stream_remove("/newton/prep", 123, 120)
+                    client.close() # remove this to see the failure
+                elif test == 2:
+                    # use the context manager
+                    with nilmdb.Client(url = testurl) as c:
+                        with assert_raises(ClientError) as e:
+                            c.stream_remove("/newton/prep", 123, 120)
+
+    def test_client_10_context(self):
+        # Test using the client's stream insertion context manager to
+        # insert data.
+        client = nilmdb.Client(testurl)
+
+        client.stream_create("/context/test", "uint16_1")
+        with client.stream_insert_context("/context/test") as ctx:
+            # override _max_data to trigger frequent server updates
+            ctx._max_data = 15
+
+            with assert_raises(ValueError):
+                ctx.insert_line("100 1")
+
+            ctx.insert_line("100 1\n")
+            ctx.insert_iter([ "101 1\n",
+                              "102 1\n",
+                              "103 1\n" ])
+            ctx.insert_line("104 1\n")
+            ctx.insert_line("105 1\n")
+            ctx.finalize()
+
+            ctx.insert_line("106 1\n")
+            ctx.update_end(106.5)
+            ctx.finalize()
+            ctx.update_start(106.8)
+            ctx.insert_line("107 1\n")
+            ctx.insert_line("108 1\n")
+            ctx.insert_line("109 1\n")
+            ctx.insert_line("110 1\n")
+            ctx.insert_line("111 1\n")
+            ctx.update_end(113)
+            ctx.insert_line("112 1\n")
+            ctx.update_end(114)
+            ctx.insert_line("113 1\n")
+            ctx.update_end(115)
+            ctx.insert_line("114 1\n")
+            ctx.finalize()
+
+        with assert_raises(ClientError):
+            with client.stream_insert_context("/context/test", 100, 200) as ctx:
+                ctx.insert_line("115 1\n")
+
+        with assert_raises(ClientError):
+            with client.stream_insert_context("/context/test", 200, 300) as ctx:
+                ctx.insert_line("115 1\n")
+
+        with client.stream_insert_context("/context/test", 200, 300) as ctx:
+            # make sure our override wasn't permanent
+            ne_(ctx._max_data, 15)
+            ctx.insert_line("225 1\n")
+            ctx.finalize()
+
+        eq_(list(client.stream_intervals("/context/test")),
+            [ [ 100, 105.000001 ],
+              [ 106, 106.5 ],
+              [ 106.8, 115 ],
+              [ 200, 300 ] ])
+
+        client.stream_destroy("/context/test")
+        client.close()
+
+    def test_client_11_emptyintervals(self):
+        # Empty intervals are ok!  If recording detection events
+        # by inserting rows into the database, we want to be able to
+        # have an interval where no events occurred.  Test them here.
+        client = nilmdb.Client(testurl)
+        client.stream_create("/empty/test", "uint16_1")
+
+        def info():
+            result = []
+            for interval in list(client.stream_intervals("/empty/test")):
+                result.append((client.stream_count("/empty/test", *interval),
+                               interval))
+            return result
+
+        eq_(info(), [])
+
+        # Insert a region with just a few points
+        with client.stream_insert_context("/empty/test") as ctx:
+            ctx.update_start(100)
+            ctx.insert_line("140 1\n")
+            ctx.insert_line("150 1\n")
+            ctx.insert_line("160 1\n")
+            ctx.update_end(200)
+            ctx.finalize()
+
+        eq_(info(), [(3, [100, 200])])
+
+        # Delete chunk, which will leave one data point and two intervals
+        client.stream_remove("/empty/test", 145, 175)
+        eq_(info(), [(1, [100, 145]),
+                     (0, [175, 200])])
+
+        # Try also creating a completely empty interval from scratch,
+        # in a few different ways.
+        client.stream_insert_block("/empty/test", "", 300, 350)
+        client.stream_insert("/empty/test", [], 400, 450)
+        with client.stream_insert_context("/empty/test", 500, 550):
+            pass
+
+        # If enough timestamps aren't provided, empty streams won't be created.
+        client.stream_insert("/empty/test", [])
+        with client.stream_insert_context("/empty/test"):
+            pass
+        client.stream_insert("/empty/test", [], start = 600)
+        with client.stream_insert_context("/empty/test", start = 700):
+            pass
+        client.stream_insert("/empty/test", [], end = 850)
+        with client.stream_insert_context("/empty/test", end = 950):
+            pass
+
+        # Try various things that might cause problems
+        with client.stream_insert_context("/empty/test", 1000, 1050):
+            ctx.finalize() # inserts [1000, 1050]
+            ctx.finalize() # nothing
+            ctx.finalize() # nothing
+            ctx.insert_line("1100 1\n")
+            ctx.finalize() # inserts [1100, 1100.000001]
+            ctx.update_start(1199)
+            ctx.insert_line("1200 1\n")
+            ctx.update_end(1250)
+            ctx.finalize() # inserts [1199, 1250]
+            ctx.update_start(1299)
+            ctx.finalize() # nothing
+            ctx.update_end(1350)
+            ctx.finalize() # nothing
+            ctx.update_start(1400)
+            ctx.update_end(1450)
+            ctx.finalize()
+            # implicit last finalize inserts [1400, 1450]
+
+        # Check everything
+        eq_(info(), [(1, [100, 145]),
+                     (0, [175, 200]),
+                     (0, [300, 350]),
+                     (0, [400, 450]),
+                     (0, [500, 550]),
+                     (0, [1000, 1050]),
+                     (1, [1100, 1100.000001]),
+                     (1, [1199, 1250]),
+                     (0, [1400, 1450]),
+                     ])
+
+        # Clean up
+        client.stream_destroy("/empty/test")
+        client.close()
+
+    def test_client_12_persistent(self):
+        # Check that connections are persistent when they should be.
+        # This is pretty hard to test; we have to poke deep into
+        # the Requests library.
+        with nilmdb.Client(url = testurl) as c:
+            def connections():
+                try:
+                    poolmanager = c.http._last_response.connection.poolmanager
+                    pool = poolmanager.pools[('http','localhost',32180)]
+                    return (pool.num_connections, pool.num_requests)
+                except:
+                    raise SkipTest("can't get connection info")
+
+            # First request makes a connection
+            c.stream_create("/persist/test", "uint16_1")
+            eq_(connections(), (1, 1))
+
+            # Non-generator
+            c.stream_list("/persist/test")
+            eq_(connections(), (1, 2))
+            c.stream_list("/persist/test")
+            eq_(connections(), (1, 3))
+
+            # Generators
+            for x in c.stream_intervals("/persist/test"):
+                pass
+            eq_(connections(), (1, 4))
+            for x in c.stream_intervals("/persist/test"):
+                pass
+            eq_(connections(), (1, 5))
+
+            # Clean up
+            c.stream_destroy("/persist/test")
+            eq_(connections(), (1, 6))

tests/test_cmdline.py

@@ -11,12 +11,7 @@ from nose.tools import assert_raises
 import itertools
 import os
 import re
-import shutil
 import sys
-import threading
-import urllib2
-from urllib2 import urlopen, HTTPError
-import Queue
 import StringIO
 import shlex
 
@@ -27,11 +22,12 @@ testdb = "tests/cmdline-testdb"
 def server_start(max_results = None, bulkdata_args = {}):
     global test_server, test_db
     # Start web app on a custom port
-    test_db = nilmdb.NilmDB(testdb, sync = False,
+    test_db = nilmdb.utils.serializer_proxy(nilmdb.NilmDB)(
+        testdb, sync = False,
         max_results = max_results,
         bulkdata_args = bulkdata_args)
     test_server = nilmdb.Server(test_db, host = "127.0.0.1",
-                                port = 12380, stoppable = False,
+                                port = 32180, stoppable = False,
                                 fast_shutdown = True,
                                 force_traceback = False)
     test_server.start(blocking = False)
@@ -63,6 +59,7 @@ class TestCmdline(object):
         passing the given input.  Returns a tuple with the output and
         exit code"""
         # printf("TZ=UTC ./nilmtool.py %s\n", arg_string)
+        os.environ['NILMDB_URL'] = "http://localhost:32180/"
         class stdio_wrapper:
             def __init__(self, stdin, stdout, stderr):
                 self.io = (stdin, stdout, stderr)
@@ -162,18 +159,18 @@ class TestCmdline(object):
 
         # try some URL constructions
         self.fail("--url http://nosuchurl/ info")
-        self.contain("Couldn't resolve host 'nosuchurl'")
+        self.contain("error connecting to server")
 
         self.fail("--url nosuchurl info")
-        self.contain("Couldn't resolve host 'nosuchurl'")
+        self.contain("error connecting to server")
 
         self.fail("-u nosuchurl/foo info")
-        self.contain("Couldn't resolve host 'nosuchurl'")
+        self.contain("error connecting to server")
 
-        self.fail("-u localhost:0 info")
-        self.contain("couldn't connect to host")
+        self.fail("-u localhost:1 info")
+        self.contain("error connecting to server")
 
-        self.ok("-u localhost:12380 info")
+        self.ok("-u localhost:32180 info")
         self.ok("info")
 
         # Duplicated arguments should fail, but this isn't implemented
@@ -191,14 +188,46 @@ class TestCmdline(object):
             self.fail("extract --start 2000-01-01 --start 2001-01-02")
             self.contain("duplicated argument")
 
-    def test_02_info(self):
+        # Verify that "help command" and "command --help" are identical
+        # for all commands.
+        self.fail("")
+        m = re.search(r"{(.*)}", self.captured)
+        for command in [""] + m.group(1).split(','):
+            self.ok(command + " --help")
+            cap1 = self.captured
+            self.ok("help " + command)
+            cap2 = self.captured
+            self.ok("help " + command + " asdf --url --zxcv -")
+            cap3 = self.captured
+            eq_(cap1, cap2)
+            eq_(cap2, cap3)
+
+    def test_02_parsetime(self):
+        os.environ['TZ'] = "America/New_York"
+        test = datetime_tz.datetime_tz.now()
+        parse_time = nilmdb.utils.time.parse_time
+        eq_(parse_time(str(test)), test)
+        test = datetime_tz.datetime_tz.smartparse("20120405 1400-0400")
+        eq_(parse_time("hi there 20120405 1400-0400 testing! 123"), test)
+        eq_(parse_time("20120405 1800 UTC"), test)
+        eq_(parse_time("20120405 1400-0400 UTC"), test)
+        for badtime in [ "20120405 1400-9999", "hello", "-", "", "4:00" ]:
+            with assert_raises(ValueError):
+                x = parse_time(badtime)
+        x = parse_time("now")
+        eq_(parse_time("snapshot-20120405-140000.raw.gz"), test)
+        eq_(parse_time("prep-20120405T1400"), test)
+
+    def test_03_info(self):
         self.ok("info")
-        self.contain("Server URL: http://localhost:12380/")
+        self.contain("Server URL: http://localhost:32180/")
+        self.contain("Client version: " + nilmdb.__version__)
         self.contain("Server version: " + test_server.version)
         self.contain("Server database path")
         self.contain("Server database size")
+        self.contain("Server database free space")
 
-    def test_03_createlist(self):
+    def test_04_createlist(self):
         # Basic stream tests, like those in test_client.
 
         # No streams
@@ -272,9 +301,9 @@ class TestCmdline(object):
 
         # reversed range
         self.fail("list /newton/prep --start 2020-01-01 --end 2000-01-01")
-        self.contain("start is after end")
+        self.contain("start must precede end")
 
-    def test_04_metadata(self):
+    def test_05_metadata(self):
         # Set / get metadata
         self.fail("metadata")
         self.fail("metadata --get")
@@ -331,22 +360,6 @@ class TestCmdline(object):
         self.fail("metadata /newton/nosuchpath")
         self.contain("No stream at path /newton/nosuchpath")
 
-    def test_05_parsetime(self):
-        os.environ['TZ'] = "America/New_York"
-        cmd = nilmdb.cmdline.Cmdline(None)
-        test = datetime_tz.datetime_tz.now()
-        eq_(cmd.parse_time(str(test)), test)
-        test = datetime_tz.datetime_tz.smartparse("20120405 1400-0400")
-        eq_(cmd.parse_time("hi there 20120405 1400-0400 testing! 123"), test)
-        eq_(cmd.parse_time("20120405 1800 UTC"), test)
-        eq_(cmd.parse_time("20120405 1400-0400 UTC"), test)
-        for badtime in [ "20120405 1400-9999", "hello", "-", "", "4:00" ]:
-            with assert_raises(ValueError):
-                x = cmd.parse_time(badtime)
-        x = cmd.parse_time("now")
-        eq_(cmd.parse_time("snapshot-20120405-140000.raw.gz"), test)
-        eq_(cmd.parse_time("prep-20120405T1400"), test)
-
     def test_06_insert(self):
         self.ok("insert --help")
 
@@ -415,7 +428,7 @@ class TestCmdline(object):
         # bad start time
         self.fail("insert --rate 120 --start 'whatever' /newton/prep /dev/null")
 
-    def test_07_detail(self):
+    def test_07_detail_extent(self):
         # Just count the number of lines, it's probably fine
         self.ok("list --detail")
         lines_(self.captured, 8)
@@ -442,7 +455,7 @@ class TestCmdline(object):
         self.contain("no intervals")
 
         self.ok("list --detail --path *prep --start='23 Mar 2012 10:05:15.50'"
-                + " --end='23 Mar 2012 10:05:15.50'")
+                + " --end='23 Mar 2012 10:05:15.51'")
         lines_(self.captured, 2)
         self.contain("10:05:15.500")
 
@@ -460,6 +473,18 @@ class TestCmdline(object):
         lines_(self.captured, 2)
         self.contain("[ 1332497115.612 -> 1332497159.991668 ]")
 
+        # Check --extent output
+        self.ok("list --extent")
+        lines_(self.captured, 6)
+
+        self.ok("list -E -T")
+        self.contain("  extent: 1332496800 -> 1332497159.991668")
+        self.contain("  extent: (no data)")
+
+        # Misc
+        self.fail("list --extent --start='23 Mar 2012 10:05:15.50'")
+        self.contain("--start and --end only make sense with --detail")
+
     def test_08_extract(self):
         # nonexistent stream
         self.fail("extract /no/such/foo --start 2000-01-01 --end 2020-01-01")
@@ -471,29 +496,29 @@ class TestCmdline(object):
 
         # empty ranges return error 2
         self.fail("extract -a /newton/prep " +
-                  "--start '23 Mar 2012 10:00:30' " +
-                  "--end '23 Mar 2012 10:00:30'",
+                  "--start '23 Mar 2012 20:00:30' " +
+                  "--end '23 Mar 2012 20:00:31'",
                   exitcode = 2, require_error = False)
         self.contain("no data")
         self.fail("extract -a /newton/prep " +
-                  "--start '23 Mar 2012 10:00:30.000001' " +
-                  "--end '23 Mar 2012 10:00:30.000001'",
+                  "--start '23 Mar 2012 20:00:30.000001' " +
+                  "--end '23 Mar 2012 20:00:30.000002'",
                   exitcode = 2, require_error = False)
         self.contain("no data")
         self.fail("extract -a /newton/prep " +
                   "--start '23 Mar 2022 10:00:30' " +
-                  "--end '23 Mar 2022 10:00:30'",
+                  "--end '23 Mar 2022 10:00:31'",
                   exitcode = 2, require_error = False)
         self.contain("no data")
 
         # but are ok if we're just counting results
         self.ok("extract --count /newton/prep " +
-                "--start '23 Mar 2012 10:00:30' " +
-                "--end '23 Mar 2012 10:00:30'")
+                "--start '23 Mar 2012 20:00:30' " +
+                "--end '23 Mar 2012 20:00:31'")
         self.match("0\n")
         self.ok("extract -c /newton/prep " +
-                "--start '23 Mar 2012 10:00:30.000001' " +
-                "--end '23 Mar 2012 10:00:30.000001'")
+                "--start '23 Mar 2012 20:00:30.000001' " +
+                "--end '23 Mar 2012 20:00:30.000002'")
         self.match("0\n")
 
         # Check various dumps against stored copies of how they should appear
@@ -540,31 +565,31 @@ class TestCmdline(object):
         self.fail("remove /no/such/foo --start 2000-01-01 --end 2020-01-01")
         self.contain("No stream at path")
 
+        # empty or backward ranges return errors
         self.fail("remove /newton/prep --start 2020-01-01 --end 2000-01-01")
-        self.contain("start is after end")
+        self.contain("start must precede end")
 
-        # empty ranges return success, backwards ranges return error
-        self.ok("remove /newton/prep " +
+        self.fail("remove /newton/prep " +
                   "--start '23 Mar 2012 10:00:30' " +
                   "--end '23 Mar 2012 10:00:30'")
-        self.match("")
-        self.ok("remove /newton/prep " +
+        self.contain("start must precede end")
+        self.fail("remove /newton/prep " +
                   "--start '23 Mar 2012 10:00:30.000001' " +
                   "--end '23 Mar 2012 10:00:30.000001'")
-        self.match("")
-        self.ok("remove /newton/prep " +
+        self.contain("start must precede end")
+        self.fail("remove /newton/prep " +
                   "--start '23 Mar 2022 10:00:30' " +
                   "--end '23 Mar 2022 10:00:30'")
-        self.match("")
+        self.contain("start must precede end")
 
         # Verbose
         self.ok("remove -c /newton/prep " +
-                "--start '23 Mar 2012 10:00:30' " +
-                "--end '23 Mar 2012 10:00:30'")
+                "--start '23 Mar 2022 20:00:30' " +
+                "--end '23 Mar 2022 20:00:31'")
         self.match("0\n")
         self.ok("remove --count /newton/prep " +
-                "--start '23 Mar 2012 10:00:30' " +
-                "--end '23 Mar 2012 10:00:30'")
+                "--start '23 Mar 2022 20:00:30' " +
+                "--end '23 Mar 2022 20:00:31'")
         self.match("0\n")
 
         # Make sure we have the data we expect
@@ -765,7 +790,7 @@ class TestCmdline(object):
                 "tests/data/prep-20120323T1000")
 
         # Should take up about 2.8 MB here (including directory entries)
-        du_before = nilmdb.utils.diskusage.du_bytes(testdb)
+        du_before = nilmdb.utils.diskusage.du(testdb)
 
         # Make sure we have the data we expect
         self.ok("list --detail")
@@ -815,7 +840,7 @@ class TestCmdline(object):
 
         # We have 1/8 of the data that we had before, so the file size
         # should have dropped below 1/4 of what it used to be
-        du_after = nilmdb.utils.diskusage.du_bytes(testdb)
+        du_after = nilmdb.utils.diskusage.du(testdb)
         lt_(du_after, (du_before / 4))
 
         # Remove anything that came from the 10:02 data file

tests/test_interval.py

@@ -55,7 +55,7 @@ class TestInterval:
                          for x in [ "03/24/2012", "03/25/2012", "03/26/2012" ] ]
 
         # basic construction
-        i = Interval(d1, d1)
+        i = Interval(d1, d2)
         i = Interval(d1, d3)
         eq_(i.start, d1)
         eq_(i.end, d3)
@@ -77,8 +77,8 @@ class TestInterval:
         assert(Interval(d1, d3) > Interval(d1, d2))
         assert(Interval(d1, d2) < Interval(d2, d3))
         assert(Interval(d1, d3) < Interval(d2, d3))
-        assert(Interval(d2, d2) > Interval(d1, d3))
-        assert(Interval(d3, d3) == Interval(d3, d3))
+        assert(Interval(d2, d2+0.01) > Interval(d1, d3))
+        assert(Interval(d3, d3+0.01) == Interval(d3, d3+0.01))
         #with assert_raises(TypeError): # was AttributeError, that's wrong
         #    x = (i == 123)
 
@@ -293,7 +293,7 @@ class TestIntervalDB:
         # actual start, end can be a subset
         a = DBInterval(150, 200, 100, 200, 10000, 20000)
         b = DBInterval(100, 150, 100, 200, 10000, 20000)
-        c = DBInterval(150, 150, 100, 200, 10000, 20000)
+        c = DBInterval(150, 160, 100, 200, 10000, 20000)
 
         # Make a set of DBIntervals
         iseta = IntervalSet([a, b])

tests/test_layout.py

@@ -9,14 +9,7 @@ from nose.tools import assert_raises
 import distutils.version
 import itertools
 import os
-import shutil
 import sys
-import cherrypy
-import threading
-import urllib2
-from urllib2 import urlopen, HTTPError
-import Queue
-import cStringIO
 import random
 import unittest
 
@@ -246,7 +239,7 @@ class TestLayoutSpeed:
                 parser = Parser(layout)
                 formatter = Formatter(layout)
                 parser.parse(data)
-                data = formatter.format(parser.data)
+                formatter.format(parser.data)
             elapsed = time.time() - start
             printf("roundtrip %s: %d ms, %.1f μs/row, %d rows/sec\n",
                    layout,
@@ -264,3 +257,8 @@ class TestLayoutSpeed:
             return [ sprintf("%d", random.randint(0,65535))
                      for x in range(10) ]
         do_speedtest("uint16_10", datagen)
+
+        def datagen():
+            return [ sprintf("%d", random.randint(0,65535))
+                     for x in range(6) ]
+        do_speedtest("uint16_6", datagen)

tests/test_mustclose.py

@@ -34,6 +34,10 @@ class Bar:
     def __del__(self):
         fprintf(err, "Deleting\n")
 
+    @classmethod
+    def baz(self):
+        fprintf(err, "Baz\n")
+
     def close(self):
         fprintf(err, "Closing\n")
 

tests/test_nilmdb.py

@@ -6,15 +6,15 @@ import distutils.version
 import simplejson as json
 import itertools
 import os
-import shutil
 import sys
-import cherrypy
 import threading
 import urllib2
 from urllib2 import urlopen, HTTPError
-import Queue
 import cStringIO
 import time
+import requests
+
+from nilmdb.utils import serializer_proxy
 
 testdb = "tests/testdb"
 
@@ -93,19 +93,31 @@ class Test00Nilmdb(object):  # named 00 so it runs first
         eq_(db.stream_get_metadata("/newton/prep"), meta1)
         eq_(db.stream_get_metadata("/newton/raw"), meta1)
 
+        # fill in some test coverage for start >= end
+        with assert_raises(nilmdb.server.NilmDBError):
+            db.stream_remove("/newton/prep", 0, 0)
+        with assert_raises(nilmdb.server.NilmDBError):
+            db.stream_remove("/newton/prep", 1, 0)
+        db.stream_remove("/newton/prep", 0, 1)
+
         db.close()
 
 class TestBlockingServer(object):
     def setUp(self):
-        self.db = nilmdb.NilmDB(testdb, sync=False)
+        self.db = serializer_proxy(nilmdb.NilmDB)(testdb, sync=False)
 
     def tearDown(self):
         self.db.close()
 
     def test_blocking_server(self):
+        # Server should fail if the database doesn't have a "_thread_safe"
+        # property.
+        with assert_raises(KeyError):
+            nilmdb.Server(object())
+
         # Start web app on a custom port
         self.server = nilmdb.Server(self.db, host = "127.0.0.1",
-                                    port = 12380, stoppable = True)
+                                    port = 32180, stoppable = True)
 
         # Run it
         event = threading.Event()
@@ -117,13 +129,13 @@ class TestBlockingServer(object):
             raise AssertionError("server didn't start in 10 seconds")
 
         # Send request to exit.
-        req = urlopen("http://127.0.0.1:12380/exit/", timeout = 1)
+        req = urlopen("http://127.0.0.1:32180/exit/", timeout = 1)
 
         # Wait for it
         thread.join()
 
 def geturl(path):
-    req = urlopen("http://127.0.0.1:12380" + path, timeout = 10)
+    req = urlopen("http://127.0.0.1:32180" + path, timeout = 10)
     return req.read()
 
 def getjson(path):
@@ -133,9 +145,9 @@ class TestServer(object):
 
     def setUp(self):
         # Start web app on a custom port
-        self.db = nilmdb.NilmDB(testdb, sync=False)
+        self.db = serializer_proxy(nilmdb.NilmDB)(testdb, sync=False)
         self.server = nilmdb.Server(self.db, host = "127.0.0.1",
-                                    port = 12380, stoppable = False)
+                                    port = 32180, stoppable = False)
         self.server.start(blocking = False)
 
     def tearDown(self):
@@ -151,8 +163,8 @@ class TestServer(object):
             eq_(e.exception.code, 404)
 
         # Check version
-        eq_(distutils.version.StrictVersion(getjson("/version")),
-            distutils.version.StrictVersion(self.server.version))
+        eq_(distutils.version.LooseVersion(getjson("/version")),
+            distutils.version.LooseVersion(nilmdb.__version__))
 
     def test_stream_list(self):
         # Known streams that got populated by an earlier test (test_nilmdb)
@@ -195,11 +207,50 @@ class TestServer(object):
                        "&key=foo")
         eq_(data, {'foo': None})
 
+    def test_cors_headers(self):
+        # Test that CORS headers are being set correctly
 
-    def test_insert(self):
-        # GET instead of POST (no body)
-        # (actual POST test is done by client code)
-        with assert_raises(HTTPError) as e:
-            getjson("/stream/insert?path=/newton/prep&start=0&end=0")
-        eq_(e.exception.code, 400)
+        # Normal GET should send simple response
+        url = "http://127.0.0.1:32180/stream/list"
+        r = requests.get(url, headers = { "Origin": "http://google.com/" })
+        eq_(r.status_code, 200)
+        if "access-control-allow-origin" not in r.headers:
+            raise AssertionError("No Access-Control-Allow-Origin (CORS) "
+                                 "header in response:\n", r.headers)
+        eq_(r.headers["access-control-allow-origin"], "http://google.com/")
 
+        # OPTIONS without CORS preflight headers should result in 405
+        r = requests.options(url, headers = {
+            "Origin": "http://google.com/",
+            })
+        eq_(r.status_code, 405)
+
+        # OPTIONS with preflight headers should give preflight response
+        r = requests.options(url, headers = {
+            "Origin": "http://google.com/",
+            "Access-Control-Request-Method": "POST",
+            "Access-Control-Request-Headers": "X-Custom",
+            })
+        eq_(r.status_code, 200)
+        if "access-control-allow-origin" not in r.headers:
+            raise AssertionError("No Access-Control-Allow-Origin (CORS) "
+                                 "header in response:\n", r.headers)
+        eq_(r.headers["access-control-allow-methods"], "GET, HEAD")
+        eq_(r.headers["access-control-allow-headers"], "X-Custom")
+
+    def test_post_bodies(self):
+        # Test JSON post bodies
+        r = requests.post("http://127.0.0.1:32180/stream/set_metadata",
+                          headers = { "Content-Type": "application/json" },
+                          data = '{"hello": 1}')
+        eq_(r.status_code, 404) # wrong parameters
+
+        r = requests.post("http://127.0.0.1:32180/stream/set_metadata",
+                          headers = { "Content-Type": "application/json" },
+                          data = '["hello"]')
+        eq_(r.status_code, 415) # not a dict
+
+        r = requests.post("http://127.0.0.1:32180/stream/set_metadata",
+                          headers = { "Content-Type": "application/json" },
+                          data = '[hello]')
+        eq_(r.status_code, 400) # badly formatted JSON

tests/test_serializer.py

@@ -9,16 +9,28 @@ import time
 
 from testutil.helpers import *
 
-#raise nose.exc.SkipTest("Skip these")
-
 class Foo(object):
     val = 0
 
+    def __init__(self, asdf = "asdf"):
+        self.init_thread = threading.current_thread().name
+
+    @classmethod
+    def foo(self):
+        pass
+
     def fail(self):
         raise Exception("you asked me to do this")
 
     def test(self, debug = False):
+        self.tester(debug)
+
+    def t(self):
+        pass
+
+    def tester(self, debug = False):
         # purposely not thread-safe
+        self.test_thread = threading.current_thread().name
         oldval = self.val
         newval = oldval + 1
         time.sleep(0.05)
@@ -46,27 +58,29 @@ class Base(object):
             t.join()
         self.verify_result()
 
+    def verify_result(self):
+        eq_(self.foo.val, 20)
+        eq_(self.foo.init_thread, self.foo.test_thread)
+
 class TestUnserialized(Base):
     def setUp(self):
         self.foo = Foo()
 
     def verify_result(self):
         # This should have failed to increment properly
-        assert(self.foo.val != 20)
+        ne_(self.foo.val, 20)
+        # Init and tests ran in different threads
+        ne_(self.foo.init_thread, self.foo.test_thread)
 
-class TestSerialized(Base):
+class TestSerializer(Base):
     def setUp(self):
-        self.realfoo = Foo()
-        self.foo = nilmdb.utils.Serializer(self.realfoo)
+        self.foo = nilmdb.utils.serializer_proxy(Foo)("qwer")
 
-    def tearDown(self):
-        del self.foo
-
-    def verify_result(self):
-        # This should have worked
-        eq_(self.realfoo.val, 20)
-
-    def test_attribute(self):
-        # Can't wrap attributes yet
-        with assert_raises(TypeError):
-            self.foo.val
+    def test_multi(self):
+        sp = nilmdb.utils.serializer_proxy
+        sp(Foo("x")).t()
+        sp(sp(Foo)("x")).t()
+        sp(sp(Foo))("x").t()
+        sp(sp(Foo("x"))).t()
+        sp(sp(Foo)("x")).t()
+        sp(sp(Foo))("x").t()

tests/test_threadsafety.py

@@ -0,0 +1,96 @@
+import nilmdb
+from nilmdb.utils.printf import *
+
+import nose
+from nose.tools import *
+from nose.tools import assert_raises
+
+from testutil.helpers import *
+import threading
+
+class Thread(threading.Thread):
+    def __init__(self, target):
+        self.target = target
+        threading.Thread.__init__(self)
+
+    def run(self):
+        try:
+            self.target()
+        except AssertionError as e:
+            self.error = e
+        else:
+            self.error = None
+
+class Test():
+    def __init__(self):
+        self.test = 1234
+
+    @classmethod
+    def asdf(cls):
+        pass
+
+    def foo(self, exception = False, reenter = False):
+        if exception:
+            raise Exception()
+        self.bar(reenter)
+
+    def bar(self, reenter):
+        if reenter:
+            self.foo()
+        return 123
+
+    def baz_threaded(self, target):
+        t = Thread(target)
+        t.start()
+        t.join()
+        return t
+
+    def baz(self, target):
+        target()
+
+class TestThreadSafety(object):
+    def tryit(self, c, threading_ok, concurrent_ok):
+        eq_(c.test, 1234)
+        c.foo()
+        t = Thread(c.foo)
+        t.start()
+        t.join()
+        if threading_ok and t.error:
+            raise Exception("got unexpected error: " + str(t.error))
+        if not threading_ok and not t.error:
+            raise Exception("failed to get expected error")
+        try:
+            c.baz(c.foo)
+        except AssertionError as e:
+            if concurrent_ok:
+                raise Exception("got unexpected error: " + str(e))
+        else:
+            if not concurrent_ok:
+                raise Exception("failed to get expected error")
+        t = c.baz_threaded(c.foo)
+        if (concurrent_ok and threading_ok) and t.error:
+            raise Exception("got unexpected error: " + str(t.error))
+        if not (concurrent_ok and threading_ok) and not t.error:
+            raise Exception("failed to get expected error")
+
+    def test(self):
+        proxy = nilmdb.utils.threadsafety.verify_proxy
+        self.tryit(Test(), True, True)
+        self.tryit(proxy(Test(), True, True, True), False, False)
+        self.tryit(proxy(Test(), True, True, False), False, True)
+        self.tryit(proxy(Test(), True, False, True), True, False)
+        self.tryit(proxy(Test(), True, False, False), True, True)
+        self.tryit(proxy(Test, True, True, True)(), False, False)
+        self.tryit(proxy(Test, True, True, False)(), False, True)
+        self.tryit(proxy(Test, True, False, True)(), True, False)
+        self.tryit(proxy(Test, True, False, False)(), True, True)
+
+        proxy(proxy(proxy(Test))()).foo()
+
+        c = proxy(Test())
+        c.foo()
+        try:
+            c.foo(exception = True)
+        except Exception:
+            pass
+        c.foo()

versioneer.py

@@ -0,0 +1,655 @@
+#! /usr/bin/python
+
+"""versioneer.py
+
+(like a rocketeer, but for versions)
+
+* https://github.com/warner/python-versioneer
+* Brian Warner
+* License: Public Domain
+* Version: 0.7+
+
+This file helps distutils-based projects manage their version number by just
+creating version-control tags.
+
+For developers who work from a VCS-generated tree (e.g. 'git clone' etc),
+each 'setup.py version', 'setup.py build', 'setup.py sdist' will compute a
+version number by asking your version-control tool about the current
+checkout. The version number will be written into a generated _version.py
+file of your choosing, where it can be included by your __init__.py
+
+For users who work from a VCS-generated tarball (e.g. 'git archive'), it will
+compute a version number by looking at the name of the directory created when
+te tarball is unpacked. This conventionally includes both the name of the
+project and a version number.
+
+For users who work from a tarball built by 'setup.py sdist', it will get a
+version number from a previously-generated _version.py file.
+
+As a result, loading code directly from the source tree will not result in a
+real version. If you want real versions from VCS trees (where you frequently
+update from the upstream repository, or do new development), you will need to
+do a 'setup.py version' after each update, and load code from the build/
+directory.
+
+You need to provide this code with a few configuration values:
+
+ versionfile_source:
+    A project-relative pathname into which the generated version strings
+    should be written. This is usually a _version.py next to your project's
+    main __init__.py file. If your project uses src/myproject/__init__.py,
+    this should be 'src/myproject/_version.py'. This file should be checked
+    in to your VCS as usual: the copy created below by 'setup.py
+    update_files' will include code that parses expanded VCS keywords in
+    generated tarballs. The 'build' and 'sdist' commands will replace it with
+    a copy that has just the calculated version string.
+
+ versionfile_build:
+    Like versionfile_source, but relative to the build directory instead of
+    the source directory. These will differ when your setup.py uses
+    'package_dir='. If you have package_dir={'myproject': 'src/myproject'},
+    then you will probably have versionfile_build='myproject/_version.py' and
+    versionfile_source='src/myproject/_version.py'.
+
+ tag_prefix: a string, like 'PROJECTNAME-', which appears at the start of all
+             VCS tags. If your tags look like 'myproject-1.2.0', then you
+             should use tag_prefix='myproject-'. If you use unprefixed tags
+             like '1.2.0', this should be an empty string.
+
+ parentdir_prefix: a string, frequently the same as tag_prefix, which
+                   appears at the start of all unpacked tarball filenames. If
+                   your tarball unpacks into 'myproject-1.2.0', this should
+                   be 'myproject-'.
+
+To use it:
+
+ 1: include this file in the top level of your project
+ 2: make the following changes to the top of your setup.py:
+     import versioneer
+     versioneer.versionfile_source = 'src/myproject/_version.py'
+     versioneer.versionfile_build = 'myproject/_version.py'
+     versioneer.tag_prefix = '' # tags are like 1.2.0
+     versioneer.parentdir_prefix = 'myproject-' # dirname like 'myproject-1.2.0'
+ 3: add the following arguments to the setup() call in your setup.py:
+     version=versioneer.get_version(),
+     cmdclass=versioneer.get_cmdclass(),
+ 4: run 'setup.py update_files', which will create _version.py, and will
+    append the following to your __init__.py:
+     from _version import __version__
+ 5: modify your MANIFEST.in to include versioneer.py
+ 6: add both versioneer.py and the generated _version.py to your VCS
+"""
+
+import os, sys, re
+from distutils.core import Command
+from distutils.command.sdist import sdist as _sdist
+from distutils.command.build_py import build_py as _build_py
+
+versionfile_source = None
+versionfile_build = None
+tag_prefix = None
+parentdir_prefix = None
+
+VCS = "git"
+IN_LONG_VERSION_PY = False
+
+
+LONG_VERSION_PY = '''
+IN_LONG_VERSION_PY = True
+# This file helps to compute a version number in source trees obtained from
+# git-archive tarball (such as those provided by githubs download-from-tag
+# feature). Distribution tarballs (build by setup.py sdist) and build
+# directories (produced by setup.py build) will contain a much shorter file
+# that just contains the computed version number.
+
+# This file is released into the public domain. Generated by
+# versioneer-0.7+ (https://github.com/warner/python-versioneer)
+
+# these strings will be replaced by git during git-archive
+git_refnames = "%(DOLLAR)sFormat:%%d%(DOLLAR)s"
+git_full = "%(DOLLAR)sFormat:%%H%(DOLLAR)s"
+
+
+import subprocess
+import sys
+
+def run_command(args, cwd=None, verbose=False):
+    try:
+        # remember shell=False, so use git.cmd on windows, not just git
+        p = subprocess.Popen(args, stdout=subprocess.PIPE, cwd=cwd)
+    except EnvironmentError:
+        e = sys.exc_info()[1]
+        if verbose:
+            print("unable to run %%s" %% args[0])
+            print(e)
+        return None
+    stdout = p.communicate()[0].strip()
+    if sys.version >= '3':
+        stdout = stdout.decode()
+    if p.returncode != 0:
+        if verbose:
+            print("unable to run %%s (error)" %% args[0])
+        return None
+    return stdout
+
+
+import sys
+import re
+import os.path
+
+def get_expanded_variables(versionfile_source):
+    # the code embedded in _version.py can just fetch the value of these
+    # variables. When used from setup.py, we don't want to import
+    # _version.py, so we do it with a regexp instead. This function is not
+    # used from _version.py.
+    variables = {}
+    try:
+        for line in open(versionfile_source,"r").readlines():
+            if line.strip().startswith("git_refnames ="):
+                mo = re.search(r'=\s*"(.*)"', line)
+                if mo:
+                    variables["refnames"] = mo.group(1)
+            if line.strip().startswith("git_full ="):
+                mo = re.search(r'=\s*"(.*)"', line)
+                if mo:
+                    variables["full"] = mo.group(1)
+    except EnvironmentError:
+        pass
+    return variables
+
+def versions_from_expanded_variables(variables, tag_prefix, verbose=False):
+    refnames = variables["refnames"].strip()
+    if refnames.startswith("$Format"):
+        if verbose:
+            print("variables are unexpanded, not using")
+        return {} # unexpanded, so not in an unpacked git-archive tarball
+    refs = set([r.strip() for r in refnames.strip("()").split(",")])
+    for ref in list(refs):
+        if not re.search(r'\d', ref):
+            if verbose:
+                print("discarding '%%s', no digits" %% ref)
+            refs.discard(ref)
+            # Assume all version tags have a digit. git's %%d expansion
+            # behaves like git log --decorate=short and strips out the
+            # refs/heads/ and refs/tags/ prefixes that would let us
+            # distinguish between branches and tags. By ignoring refnames
+            # without digits, we filter out many common branch names like
+            # "release" and "stabilization", as well as "HEAD" and "master".
+    if verbose:
+        print("remaining refs: %%s" %% ",".join(sorted(refs)))
+    for ref in sorted(refs):
+        # sorting will prefer e.g. "2.0" over "2.0rc1"
+        if ref.startswith(tag_prefix):
+            r = ref[len(tag_prefix):]
+            if verbose:
+                print("picking %%s" %% r)
+            return { "version": r,
+                     "full": variables["full"].strip() }
+    # no suitable tags, so we use the full revision id
+    if verbose:
+        print("no suitable tags, using full revision id")
+    return { "version": variables["full"].strip(),
+             "full": variables["full"].strip() }
+
+def versions_from_vcs(tag_prefix, versionfile_source, verbose=False):
+    # this runs 'git' from the root of the source tree. That either means
+    # someone ran a setup.py command (and this code is in versioneer.py, so
+    # IN_LONG_VERSION_PY=False, thus the containing directory is the root of
+    # the source tree), or someone ran a project-specific entry point (and
+    # this code is in _version.py, so IN_LONG_VERSION_PY=True, thus the
+    # containing directory is somewhere deeper in the source tree). This only
+    # gets called if the git-archive 'subst' variables were *not* expanded,
+    # and _version.py hasn't already been rewritten with a short version
+    # string, meaning we're inside a checked out source tree.
+
+    try:
+        here = os.path.abspath(__file__)
+    except NameError:
+        # some py2exe/bbfreeze/non-CPython implementations don't do __file__
+        return {} # not always correct
+
+    # versionfile_source is the relative path from the top of the source tree
+    # (where the .git directory might live) to this file. Invert this to find
+    # the root from __file__.
+    root = here
+    if IN_LONG_VERSION_PY:
+        for i in range(len(versionfile_source.split("/"))):
+            root = os.path.dirname(root)
+    else:
+        root = os.path.dirname(here)
+    if not os.path.exists(os.path.join(root, ".git")):
+        if verbose:
+            print("no .git in %%s" %% root)
+        return {}
+
+    GIT = "git"
+    if sys.platform == "win32":
+        GIT = "git.cmd"
+    stdout = run_command([GIT, "describe", "--tags", "--dirty", "--always"],
+                         cwd=root)
+    if stdout is None:
+        return {}
+    if not stdout.startswith(tag_prefix):
+        if verbose:
+            print("tag '%%s' doesn't start with prefix '%%s'" %% (stdout, tag_prefix))
+        return {}
+    tag = stdout[len(tag_prefix):]
+    stdout = run_command([GIT, "rev-parse", "HEAD"], cwd=root)
+    if stdout is None:
+        return {}
+    full = stdout.strip()
+    if tag.endswith("-dirty"):
+        full += "-dirty"
+    return {"version": tag, "full": full}
+
+
+def versions_from_parentdir(parentdir_prefix, versionfile_source, verbose=False):
+    if IN_LONG_VERSION_PY:
+        # We're running from _version.py. If it's from a source tree
+        # (execute-in-place), we can work upwards to find the root of the
+        # tree, and then check the parent directory for a version string. If
+        # it's in an installed application, there's no hope.
+        try:
+            here = os.path.abspath(__file__)
+        except NameError:
+            # py2exe/bbfreeze/non-CPython don't have __file__
+            return {} # without __file__, we have no hope
+        # versionfile_source is the relative path from the top of the source
+        # tree to _version.py. Invert this to find the root from __file__.
+        root = here
+        for i in range(len(versionfile_source.split("/"))):
+            root = os.path.dirname(root)
+    else:
+        # we're running from versioneer.py, which means we're running from
+        # the setup.py in a source tree. sys.argv[0] is setup.py in the root.
+        here = os.path.abspath(sys.argv[0])
+        root = os.path.dirname(here)
+
+    # Source tarballs conventionally unpack into a directory that includes
+    # both the project name and a version string.
+    dirname = os.path.basename(root)
+    if not dirname.startswith(parentdir_prefix):
+        if verbose:
+            print("guessing rootdir is '%%s', but '%%s' doesn't start with prefix '%%s'" %%
+                  (root, dirname, parentdir_prefix))
+        return None
+    return {"version": dirname[len(parentdir_prefix):], "full": ""}
+
+tag_prefix = "%(TAG_PREFIX)s"
+parentdir_prefix = "%(PARENTDIR_PREFIX)s"
+versionfile_source = "%(VERSIONFILE_SOURCE)s"
+
+def get_versions(default={"version": "unknown", "full": ""}, verbose=False):
+    variables = { "refnames": git_refnames, "full": git_full }
+    ver = versions_from_expanded_variables(variables, tag_prefix, verbose)
+    if not ver:
+        ver = versions_from_vcs(tag_prefix, versionfile_source, verbose)
+    if not ver:
+        ver = versions_from_parentdir(parentdir_prefix, versionfile_source,
+                                      verbose)
+    if not ver:
+        ver = default
+    return ver
+
+'''
+
+
+import subprocess
+import sys
+
+def run_command(args, cwd=None, verbose=False):
+    try:
+        # remember shell=False, so use git.cmd on windows, not just git
+        p = subprocess.Popen(args, stdout=subprocess.PIPE, cwd=cwd)
+    except EnvironmentError:
+        e = sys.exc_info()[1]
+        if verbose:
+            print("unable to run %s" % args[0])
+            print(e)
+        return None
+    stdout = p.communicate()[0].strip()
+    if sys.version >= '3':
+        stdout = stdout.decode()
+    if p.returncode != 0:
+        if verbose:
+            print("unable to run %s (error)" % args[0])
+        return None
+    return stdout
+
+
+import sys
+import re
+import os.path
+
+def get_expanded_variables(versionfile_source):
+    # the code embedded in _version.py can just fetch the value of these
+    # variables. When used from setup.py, we don't want to import
+    # _version.py, so we do it with a regexp instead. This function is not
+    # used from _version.py.
+    variables = {}
+    try:
+        for line in open(versionfile_source,"r").readlines():
+            if line.strip().startswith("git_refnames ="):
+                mo = re.search(r'=\s*"(.*)"', line)
+                if mo:
+                    variables["refnames"] = mo.group(1)
+            if line.strip().startswith("git_full ="):
+                mo = re.search(r'=\s*"(.*)"', line)
+                if mo:
+                    variables["full"] = mo.group(1)
+    except EnvironmentError:
+        pass
+    return variables
+
+def versions_from_expanded_variables(variables, tag_prefix, verbose=False):
+    refnames = variables["refnames"].strip()
+    if refnames.startswith("$Format"):
+        if verbose:
+            print("variables are unexpanded, not using")
+        return {} # unexpanded, so not in an unpacked git-archive tarball
+    refs = set([r.strip() for r in refnames.strip("()").split(",")])
+    for ref in list(refs):
+        if not re.search(r'\d', ref):
+            if verbose:
+                print("discarding '%s', no digits" % ref)
+            refs.discard(ref)
+            # Assume all version tags have a digit. git's %d expansion
+            # behaves like git log --decorate=short and strips out the
+            # refs/heads/ and refs/tags/ prefixes that would let us
+            # distinguish between branches and tags. By ignoring refnames
+            # without digits, we filter out many common branch names like
+            # "release" and "stabilization", as well as "HEAD" and "master".
+    if verbose:
+        print("remaining refs: %s" % ",".join(sorted(refs)))
+    for ref in sorted(refs):
+        # sorting will prefer e.g. "2.0" over "2.0rc1"
+        if ref.startswith(tag_prefix):
+            r = ref[len(tag_prefix):]
+            if verbose:
+                print("picking %s" % r)
+            return { "version": r,
+                     "full": variables["full"].strip() }
+    # no suitable tags, so we use the full revision id
+    if verbose:
+        print("no suitable tags, using full revision id")
+    return { "version": variables["full"].strip(),
+             "full": variables["full"].strip() }
+
+def versions_from_vcs(tag_prefix, versionfile_source, verbose=False):
+    # this runs 'git' from the root of the source tree. That either means
+    # someone ran a setup.py command (and this code is in versioneer.py, so
+    # IN_LONG_VERSION_PY=False, thus the containing directory is the root of
+    # the source tree), or someone ran a project-specific entry point (and
+    # this code is in _version.py, so IN_LONG_VERSION_PY=True, thus the
+    # containing directory is somewhere deeper in the source tree). This only
+    # gets called if the git-archive 'subst' variables were *not* expanded,
+    # and _version.py hasn't already been rewritten with a short version
+    # string, meaning we're inside a checked out source tree.
+
+    try:
+        here = os.path.abspath(__file__)
+    except NameError:
+        # some py2exe/bbfreeze/non-CPython implementations don't do __file__
+        return {} # not always correct
+
+    # versionfile_source is the relative path from the top of the source tree
+    # (where the .git directory might live) to this file. Invert this to find
+    # the root from __file__.
+    root = here
+    if IN_LONG_VERSION_PY:
+        for i in range(len(versionfile_source.split("/"))):
+            root = os.path.dirname(root)
+    else:
+        root = os.path.dirname(here)
+    if not os.path.exists(os.path.join(root, ".git")):
+        if verbose:
+            print("no .git in %s" % root)
+        return {}
+
+    GIT = "git"
+    if sys.platform == "win32":
+        GIT = "git.cmd"
+    stdout = run_command([GIT, "describe", "--tags", "--dirty", "--always"],
+                         cwd=root)
+    if stdout is None:
+        return {}
+    if not stdout.startswith(tag_prefix):
+        if verbose:
+            print("tag '%s' doesn't start with prefix '%s'" % (stdout, tag_prefix))
+        return {}
+    tag = stdout[len(tag_prefix):]
+    stdout = run_command([GIT, "rev-parse", "HEAD"], cwd=root)
+    if stdout is None:
+        return {}
+    full = stdout.strip()
+    if tag.endswith("-dirty"):
+        full += "-dirty"
+    return {"version": tag, "full": full}
+
+
+def versions_from_parentdir(parentdir_prefix, versionfile_source, verbose=False):
+    if IN_LONG_VERSION_PY:
+        # We're running from _version.py. If it's from a source tree
+        # (execute-in-place), we can work upwards to find the root of the
+        # tree, and then check the parent directory for a version string. If
+        # it's in an installed application, there's no hope.
+        try:
+            here = os.path.abspath(__file__)
+        except NameError:
+            # py2exe/bbfreeze/non-CPython don't have __file__
+            return {} # without __file__, we have no hope
+        # versionfile_source is the relative path from the top of the source
+        # tree to _version.py. Invert this to find the root from __file__.
+        root = here
+        for i in range(len(versionfile_source.split("/"))):
+            root = os.path.dirname(root)
+    else:
+        # we're running from versioneer.py, which means we're running from
+        # the setup.py in a source tree. sys.argv[0] is setup.py in the root.
+        here = os.path.abspath(sys.argv[0])
+        root = os.path.dirname(here)
+
+    # Source tarballs conventionally unpack into a directory that includes
+    # both the project name and a version string.
+    dirname = os.path.basename(root)
+    if not dirname.startswith(parentdir_prefix):
+        if verbose:
+            print("guessing rootdir is '%s', but '%s' doesn't start with prefix '%s'" %
+                  (root, dirname, parentdir_prefix))
+        return None
+    return {"version": dirname[len(parentdir_prefix):], "full": ""}
+
+import sys
+
+def do_vcs_install(versionfile_source, ipy):
+    GIT = "git"
+    if sys.platform == "win32":
+        GIT = "git.cmd"
+    run_command([GIT, "add", "versioneer.py"])
+    run_command([GIT, "add", versionfile_source])
+    run_command([GIT, "add", ipy])
+    present = False
+    try:
+        f = open(".gitattributes", "r")
+        for line in f.readlines():
+            if line.strip().startswith(versionfile_source):
+                if "export-subst" in line.strip().split()[1:]:
+                    present = True
+        f.close()
+    except EnvironmentError:
+        pass    
+    if not present:
+        f = open(".gitattributes", "a+")
+        f.write("%s export-subst\n" % versionfile_source)
+        f.close()
+        run_command([GIT, "add", ".gitattributes"])
+    
+
+SHORT_VERSION_PY = """
+# This file was generated by 'versioneer.py' (0.7+) from
+# revision-control system data, or from the parent directory name of an
+# unpacked source archive. Distribution tarballs contain a pre-generated copy
+# of this file.
+
+version_version = '%(version)s'
+version_full = '%(full)s'
+def get_versions(default={}, verbose=False):
+    return {'version': version_version, 'full': version_full}
+
+"""
+
+DEFAULT = {"version": "unknown", "full": "unknown"}
+
+def versions_from_file(filename):
+    versions = {}
+    try:
+        f = open(filename)
+    except EnvironmentError:
+        return versions
+    for line in f.readlines():
+        mo = re.match("version_version = '([^']+)'", line)
+        if mo:
+            versions["version"] = mo.group(1)
+        mo = re.match("version_full = '([^']+)'", line)
+        if mo:
+            versions["full"] = mo.group(1)
+    return versions
+
+def write_to_version_file(filename, versions):
+    f = open(filename, "w")
+    f.write(SHORT_VERSION_PY % versions)
+    f.close()
+    print("set %s to '%s'" % (filename, versions["version"]))
+
+
+def get_best_versions(versionfile, tag_prefix, parentdir_prefix,
+                      default=DEFAULT, verbose=False):
+    # returns dict with two keys: 'version' and 'full'
+    #
+    # extract version from first of _version.py, 'git describe', parentdir.
+    # This is meant to work for developers using a source checkout, for users
+    # of a tarball created by 'setup.py sdist', and for users of a
+    # tarball/zipball created by 'git archive' or github's download-from-tag
+    # feature.
+
+    variables = get_expanded_variables(versionfile_source)
+    if variables:
+        ver = versions_from_expanded_variables(variables, tag_prefix)
+        if ver:
+            if verbose: print("got version from expanded variable %s" % ver)
+            return ver
+
+    ver = versions_from_file(versionfile)
+    if ver:
+        if verbose: print("got version from file %s %s" % (versionfile, ver))
+        return ver
+
+    ver = versions_from_vcs(tag_prefix, versionfile_source, verbose)
+    if ver:
+        if verbose: print("got version from git %s" % ver)
+        return ver
+
+    ver = versions_from_parentdir(parentdir_prefix, versionfile_source, verbose)
+    if ver:
+        if verbose: print("got version from parentdir %s" % ver)
+        return ver
+
+    if verbose: print("got version from default %s" % ver)
+    return default
+
+def get_versions(default=DEFAULT, verbose=False):
+    assert versionfile_source is not None, "please set versioneer.versionfile_source"
+    assert tag_prefix is not None, "please set versioneer.tag_prefix"
+    assert parentdir_prefix is not None, "please set versioneer.parentdir_prefix"
+    return get_best_versions(versionfile_source, tag_prefix, parentdir_prefix,
+                             default=default, verbose=verbose)
+def get_version(verbose=False):
+    return get_versions(verbose=verbose)["version"]
+
+class cmd_version(Command):
+    description = "report generated version string"
+    user_options = []
+    boolean_options = []
+    def initialize_options(self):
+        pass
+    def finalize_options(self):
+        pass
+    def run(self):
+        ver = get_version(verbose=True)
+        print("Version is currently: %s" % ver)
+
+class cmd_build_py(_build_py):
+    def run(self):
+        versions = get_versions(verbose=True)
+        _build_py.run(self)
+        # now locate _version.py in the new build/ directory and replace it
+        # with an updated value
+        target_versionfile = os.path.join(self.build_lib, versionfile_build)
+        print("UPDATING %s" % target_versionfile)
+        os.unlink(target_versionfile)
+        f = open(target_versionfile, "w")
+        f.write(SHORT_VERSION_PY % versions)
+        f.close()
+
+class cmd_sdist(_sdist):
+    def run(self):
+        versions = get_versions(verbose=True)
+        self._versioneer_generated_versions = versions
+        # unless we update this, the command will keep using the old version
+        self.distribution.metadata.version = versions["version"]
+        return _sdist.run(self)
+
+    def make_release_tree(self, base_dir, files):
+        _sdist.make_release_tree(self, base_dir, files)
+        # now locate _version.py in the new base_dir directory (remembering
+        # that it may be a hardlink) and replace it with an updated value
+        target_versionfile = os.path.join(base_dir, versionfile_source)
+        print("UPDATING %s" % target_versionfile)
+        os.unlink(target_versionfile)
+        f = open(target_versionfile, "w")
+        f.write(SHORT_VERSION_PY % self._versioneer_generated_versions)
+        f.close()
+
+INIT_PY_SNIPPET = """
+from ._version import get_versions
+__version__ = get_versions()['version']
+del get_versions
+"""
+
+class cmd_update_files(Command):
+    description = "modify __init__.py and create _version.py"
+    user_options = []
+    boolean_options = []
+    def initialize_options(self):
+        pass
+    def finalize_options(self):
+        pass
+    def run(self):
+        ipy = os.path.join(os.path.dirname(versionfile_source), "__init__.py")
+        print(" creating %s" % versionfile_source)
+        f = open(versionfile_source, "w")
+        f.write(LONG_VERSION_PY % {"DOLLAR": "$",
+                                   "TAG_PREFIX": tag_prefix,
+                                   "PARENTDIR_PREFIX": parentdir_prefix,
+                                   "VERSIONFILE_SOURCE": versionfile_source,
+                                   })
+        f.close()
+        try:
+            old = open(ipy, "r").read()
+        except EnvironmentError:
+            old = ""
+        if INIT_PY_SNIPPET not in old:
+            print(" appending to %s" % ipy)
+            f = open(ipy, "a")
+            f.write(INIT_PY_SNIPPET)
+            f.close()
+        else:
+            print(" %s unmodified" % ipy)
+        do_vcs_install(versionfile_source, ipy)
+
+def get_cmdclass():
+    return {'version': cmd_version,
+            'update_files': cmd_update_files,
+            'build_py': cmd_build_py,
+            'sdist': cmd_sdist,
+            }