Add nilmdb.utils.interval.intersection by generalizing set_difference

This will help with implementing nilmrun.

README.txt

@@ -10,6 +10,9 @@ Prerequisites:
   sudo apt-get install python-cherrypy3 python-decorator python-simplejson
   sudo apt-get install python-requests python-dateutil python-tz python-psutil
 
+  # Other dependencies (required by some modules)
+  sudo apt-get install python-numpy
+
   # Tools for running tests
   sudo apt-get install python-nose python-coverage
 
@@ -24,3 +27,5 @@ Usage:
 
   nilmdb-server --help
   nilmtool --help
+
+See docs/wsgi.md for info on setting up a WSGI application in Apache.

docs/design.md

@@ -389,3 +389,52 @@ Possible solutions:
     are always printed as int64 values, and a new format
     "@1234567890123456" is added to the parser for specifying them
     exactly.
+
+Binary interface
+----------------
+
+The ASCII interface is too slow for high-bandwidth processing, like
+sinefits, prep, etc.  A binary interface was added so that you can
+extract the raw binary out of the bulkdata storage.  This binary is
+a little-endian format, e.g. in C a uint16_6 stream would be:
+
+    #include <endian.h>
+    #include <stdint.h>
+    struct {
+        int64_t timestamp_le;
+        uint16_t data_le[6];
+    } __attribute__((packed));
+
+Remember to byteswap (with e.g. `letoh` in C)!
+
+This interface is used by the new `nilmdb.client.numpyclient.NumpyClient`
+class, which is a subclass of the normal `nilmcb.client.client.Client`
+and has all of the same functions.  It adds three new functions:
+
+- `stream_extract_numpy` to extract data as a Numpy array
+
+- `stream_insert_numpy` to insert data as a Numpy array
+
+- `stream_insert_numpy_context` is the context manager for
+  incrementally inserting data
+
+It is significantly faster!  It is about 20 times faster to decimate a
+stream with `nilm-decimate` when the filter code is using the new
+binary/numpy interface.
+
+
+WSGI interface & chunked requests
+---------------------------------
+
+mod_wsgi requires "WSGIChunkedRequest On" to handle
+"Transfer-encoding: Chunked" requests.  However, `/stream/insert`
+doesn't handle this correctly right now, because:
+
+- The `cherrpy.request.body.read()` call needs to be fixed for chunked requests
+
+- We don't want to just buffer endlessly in the server, and it will
+  require some thought on how to handle data in chunks (what to do about
+  interval endpoints).
+
+It is probably better to just keep the endpoint management on the client
+side, so leave "WSGIChunkedRequest off" for now.

docs/wsgi.md

@@ -0,0 +1,32 @@
+WSGI Application in Apache
+--------------------------
+
+Install `apache2` and `libapache2-mod-wsgi`
+
+We'll set up the database server at URL `http://myhost.com/nilmdb`.
+The database will be stored in `/home/nilm/db`, and the process will
+run as user `nilm`, group `nilm`.
+
+First, create a WSGI script `/home/nilm/nilmdb.wsgi` containing:
+
+    import nilmdb.server
+    application = nilmdb.server.wsgi_application("/home/nilm/db", "/nilmdb")
+
+The first parameter is the local filesystem path, and the second
+parameter is the path part of the URL.
+
+Then, set up Apache with a configuration like:
+
+    <VirtualHost>
+        WSGIScriptAlias /nilmdb /home/nilm/nilmdb.wsgi
+        WSGIDaemonProcess nilmdb-procgroup threads=32 user=nilm group=nilm
+        <Location /nilmdb>
+            WSGIProcessGroup nilmdb-procgroup
+            WSGIApplicationGroup nilmdb-appgroup
+
+            # Access control example:
+            Order deny,allow
+            Deny from all
+            Allow from 1.2.3.4
+        </Location>
+    </VirtualHost>

extras/fix-oversize-files.py

@@ -0,0 +1,50 @@
+#!/usr/bin/python
+
+import os
+import sys
+import cPickle as pickle
+import argparse
+import fcntl
+import re
+from nilmdb.client.numpyclient import layout_to_dtype
+
+parser = argparse.ArgumentParser(
+    description = """
+Fix database corruption where binary writes caused too much data to be
+written to the file.  Truncates files to the correct length.  This was
+fixed by b98ff1331a515ad47fd3203615e835b529b039f9.
+""")
+parser.add_argument("path", action="store", help='Database root path')
+parser.add_argument("-y", "--yes", action="store_true", help='Fix them')
+args = parser.parse_args()
+
+lock = os.path.join(args.path, "data.lock")
+with open(lock, "w") as f:
+    fcntl.flock(f.fileno(), fcntl.LOCK_EX | fcntl.LOCK_NB)
+
+    fix = {}
+
+    for (path, dirs, files) in os.walk(args.path):
+        if "_format" in files:
+            with open(os.path.join(path, "_format")) as format:
+                fmt = pickle.load(format)
+                rowsize = layout_to_dtype(fmt["layout"]).itemsize
+                maxsize = rowsize * fmt["rows_per_file"]
+                fix[path] = maxsize
+                if maxsize < 128000000: # sanity check
+                    raise Exception("bad maxsize " + str(maxsize))
+
+    for fixpath in fix:
+        for (path, dirs, files) in os.walk(fixpath):
+            for fn in files:
+                if not re.match("^[0-9a-f]{4,}$", fn):
+                    continue
+                fn = os.path.join(path, fn)
+                size = os.path.getsize(fn)
+                maxsize = fix[fixpath]
+                if size > maxsize:
+                    diff = size - maxsize
+                    print diff, "too big:", fn
+                    if args.yes:
+                        with open(fn, "a+") as dbfile:
+                            dbfile.truncate(maxsize)

extras/nilmtool-bash-completion.sh

@@ -17,4 +17,4 @@ _nilmtool_argcomplete() {
         unset COMPREPLY
     fi
 }
-complete -o nospace -o default -F _nilmtool_argcomplete nilmtool
+complete -o nospace -F _nilmtool_argcomplete nilmtool

nilmdb/client/client.py

@@ -58,6 +58,11 @@ class Client(object):
         return self.http.get("dbinfo")
 
     def stream_list(self, path = None, layout = None, extended = False):
+        """Return a sorted list of [path, layout] lists.  If 'path' or
+        'layout' are specified, only return streams that match those
+        exact values.  If 'extended' is True, the returned lists have
+        extended info, e.g.: [path, layout, extent_min, extent_max,
+        total_rows, total_seconds."""
         params = {}
         if path is not None:
             params["path"] = path
@@ -65,9 +70,11 @@ class Client(object):
             params["layout"] = layout
         if extended:
             params["extended"] = 1
-        return self.http.get("stream/list", params)
+        streams = self.http.get("stream/list", params)
+        return nilmdb.utils.sort.sort_human(streams, key = lambda s: s[0])
 
     def stream_get_metadata(self, path, keys = None):
+        """Get stream metadata"""
         params = { "path": path }
         if keys is not None:
             params["key"] = keys
@@ -97,7 +104,7 @@ class Client(object):
         return self.http.post("stream/create", params)
 
     def stream_destroy(self, path):
-        """Delete stream and its contents"""
+        """Delete stream.  Fails if any data is still present."""
         params = { "path": path }
         return self.http.post("stream/destroy", params)
 
@@ -116,15 +123,19 @@ class Client(object):
             params["start"] = timestamp_to_string(start)
         if end is not None:
             params["end"] = timestamp_to_string(end)
-        return self.http.post("stream/remove", params)
+        total = 0
+        for count in self.http.post_gen("stream/remove", params):
+            total += int(count)
+        return total
 
     @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.
+        inserted into a stream in a piecewise manner.  Data is
+        provided as ASCII lines, and is aggregated and sent to the
+        server in larger or smaller 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:
@@ -136,15 +147,17 @@ class Client(object):
         This may make multiple requests to the server, if the data is
         large enough or enough time has passed between insertions.
         """
-        ctx = StreamInserter(self.http, path, start, end)
+        ctx = StreamInserter(self, path, start, end)
         yield ctx
         ctx.finalize()
+        ctx.destroy()
 
     def stream_insert(self, path, data, start = None, end = None):
         """Insert rows of data into a stream.  data should be a string
         or iterable that provides ASCII data that matches the database
-        layout for path.  See stream_insert_context for details on the
-        'start' and 'end' parameters."""
+        layout for path.  Data is passed through stream_insert_context,
+        so it will be broken into reasonably-sized chunks and
+        start/end will be deduced if missing."""
         with self.stream_insert_context(path, start, end) as ctx:
             if isinstance(data, basestring):
                 ctx.insert(data)
@@ -153,11 +166,28 @@ class Client(object):
                     ctx.insert(chunk)
         return ctx.last_response
 
+    def stream_insert_block(self, path, data, start, end, binary = False):
+        """Insert a single fixed block of data into the stream.  It is
+        sent directly to the server in one block with no further
+        processing.
+
+        If 'binary' is True, provide raw binary data in little-endian
+        format matching the path layout, including an int64 timestamp.
+        Otherwise, provide ASCII data matching the layout."""
+        params = {
+            "path": path,
+            "start": timestamp_to_string(start),
+            "end": timestamp_to_string(end),
+        }
+        if binary:
+            params["binary"] = 1
+        return self.http.put("stream/insert", data, params, binary = binary)
+
     def stream_intervals(self, path, start = None, end = None, diffpath = None):
         """
         Return a generator that yields each stream interval.
 
-        If diffpath is not None, yields only interval ranges that are
+        If 'diffpath' is not None, yields only interval ranges that are
         present in 'path' but not in 'diffpath'.
         """
         params = {
@@ -171,14 +201,23 @@ class Client(object):
             params["end"] = timestamp_to_string(end)
         return self.http.get_gen("stream/intervals", params)
 
-    def stream_extract(self, path, start = None, end = None, count = False):
+    def stream_extract(self, path, start = None, end = None,
+                       count = False, markup = False, binary = False):
         """
         Extract data from a stream.  Returns a generator that yields
         lines of ASCII-formatted data that matches the database
         layout for the given path.
 
-        Specify count = True to return a count of matching data points
+        If 'count' is True, return a count of matching data points
         rather than the actual data.  The output format is unchanged.
+
+        If 'markup' is True, include comments in the returned data
+        that indicate interval starts and ends.
+
+        If 'binary' is True, return chunks of raw binary data, rather
+        than lines of ASCII-formatted data.  Raw binary data is
+        little-endian and matches the database types (including an
+        int64 timestamp).
         """
         params = {
             "path": path,
@@ -189,7 +228,11 @@ class Client(object):
             params["end"] = timestamp_to_string(end)
         if count:
             params["count"] = 1
-        return self.http.get_gen("stream/extract", params)
+        if markup:
+            params["markup"] = 1
+        if binary:
+            params["binary"] = 1
+        return self.http.get_gen("stream/extract", params, binary = binary)
 
     def stream_count(self, path, start = None, end = None):
         """
@@ -238,13 +281,13 @@ class StreamInserter(object):
     _max_data = 2 * 1024 * 1024
     _max_data_after_send = 64 * 1024
 
-    def __init__(self, http, path, start = None, end = None):
-        """'http' is the httpclient object.  'path' is the database
+    def __init__(self, client, path, start, end):
+        """'client' is the client object.  'path' is the database
         path to insert to.  'start' and 'end' are used for the first
-        contiguous interval."""
+        contiguous interval and may be None."""
         self.last_response = None
 
-        self._http = http
+        self._client = client
         self._path = path
 
         # Start and end for the overall contiguous interval we're
@@ -257,6 +300,15 @@ class StreamInserter(object):
         self._block_data = []
         self._block_len = 0
 
+        self.destroyed = False
+
+    def destroy(self):
+        """Ensure this object can't be used again without raising
+        an error"""
+        def error(*args, **kwargs):
+            raise Exception("don't reuse this context object")
+        self._send_block = self.insert = self.finalize = self.send = error
+
     def insert(self, data):
         """Insert a chunk of ASCII formatted data in string form.  The
         overall data must consist of lines terminated by '\\n'."""
@@ -307,6 +359,11 @@ class StreamInserter(object):
         part of a new interval and there may be a gap left in-between."""
         self._send_block(final = True)
 
+    def send(self):
+        """Send any data that we might have buffered up.  Does not affect
+        any other treatment of timestamps or endpoints."""
+        self._send_block(final = False)
+
     def _get_first_noncomment(self, block):
         """Return the (start, end) indices of the first full line in
         block that isn't a comment, or raise IndexError if
@@ -398,7 +455,7 @@ class StreamInserter(object):
             self._interval_start = end_ts
 
         # Double check endpoints
-        if start_ts is None or end_ts is None:
+        if (start_ts is None or end_ts is None) or (start_ts == end_ts):
             # If the block has no non-comment lines, it's OK
             try:
                 self._get_first_noncomment(block)
@@ -407,9 +464,7 @@ class StreamInserter(object):
             raise ClientError("have data to send, but no start/end times")
 
         # Send it
-        params = { "path": self._path,
-                   "start": timestamp_to_string(start_ts),
-                   "end": timestamp_to_string(end_ts) }
-        self.last_response = self._http.put("stream/insert", block, params)
+        self.last_response = self._client.stream_insert_block(
+            self._path, block, start_ts, end_ts, binary = False)
 
         return

nilmdb/client/httpclient.py

@@ -16,7 +16,7 @@ class HTTPClient(object):
         reparsed = urlparse.urlparse(baseurl).geturl()
         if '://' not in reparsed:
             reparsed = urlparse.urlparse("http://" + baseurl).geturl()
-        self.baseurl = reparsed
+        self.baseurl = reparsed.rstrip('/') + '/'
 
         # Build Requests session object, enable SSL verification
         self.session = requests.Session()
@@ -105,12 +105,17 @@ class HTTPClient(object):
         else:
             return self._req("POST", url, None, params)
 
-    def put(self, url, data, params = None):
+    def put(self, url, data, params = None, binary = False):
         """Simple PUT (parameters in URL, data in body)"""
-        return self._req("PUT", url, params, data)
+        if binary:
+            h = { 'Content-type': 'application/octet-stream' }
+        else:
+            h = { 'Content-type': 'text/plain; charset=utf-8' }
+        return self._req("PUT", url, query = params, body = data, headers = h)
 
     # Generator versions that return data one line at a time.
-    def _req_gen(self, method, url, query = None, body = None, headers = None):
+    def _req_gen(self, method, url, query = None, body = None,
+                 headers = None, binary = False):
         """
         Make a request and return a generator that gives back strings
         or JSON decoded lines of the body data, or raise an error if
@@ -118,16 +123,50 @@ class HTTPClient(object):
         """
         (response, isjson) = self._do_req(method, url, query, body,
                                           stream = True, headers = headers)
-        if isjson:
-            for line in response.iter_lines():
+
+        # Like the iter_lines function in Requests, but only splits on
+        # the specified line ending.
+        def lines(source, ending):
+            pending = None
+            for chunk in source:
+                if pending is not None:
+                    chunk = pending + chunk
+                tmp = chunk.split(ending)
+                lines = tmp[:-1]
+                if chunk.endswith(ending):
+                    pending = None
+                else:
+                    pending = tmp[-1]
+                for line in lines:
+                    yield line
+            if pending is not None: # pragma: no cover (missing newline)
+                yield pending
+
+        # Yield the chunks or lines as requested
+        if binary:
+            for chunk in response.iter_content(chunk_size = 65536):
+                yield chunk
+        elif isjson:
+            for line in lines(response.iter_content(chunk_size = 1),
+                              ending = '\r\n'):
                 yield json.loads(line)
         else:
-            for line in response.iter_lines():
+            for line in lines(response.iter_content(chunk_size = 65536),
+                              ending = '\n'):
                 yield line
 
-    def get_gen(self, url, params = None):
+    def get_gen(self, url, params = None, binary = False):
         """Simple GET (parameters in URL) returning a generator"""
-        return self._req_gen("GET", url, params)
+        return self._req_gen("GET", url, params, binary = binary)
+
+    def post_gen(self, url, params = None):
+        """Simple POST (parameters in body) returning a generator"""
+        if self.post_json:
+            return self._req_gen("POST", url, None,
+                                 json.dumps(params),
+                                 { 'Content-type': 'application/json' })
+        else:
+            return self._req_gen("POST", url, None, params)
 
     # Not much use for a POST or PUT generator, since they don't
     # return much data.

nilmdb/client/numpyclient.py

@@ -0,0 +1,258 @@
+# -*- coding: utf-8 -*-
+
+"""Provide a NumpyClient class that is based on normal Client, but has
+additional methods for extracting and inserting data via Numpy arrays."""
+
+import nilmdb.utils
+import nilmdb.client.client
+import nilmdb.client.httpclient
+from nilmdb.client.errors import ClientError
+
+import contextlib
+from nilmdb.utils.time import timestamp_to_string, string_to_timestamp
+
+import numpy
+import cStringIO
+
+def layout_to_dtype(layout):
+    ltype = layout.split('_')[0]
+    lcount = int(layout.split('_')[1])
+    if ltype.startswith('int'):
+        atype = '<i' + str(int(ltype[3:]) / 8)
+    elif ltype.startswith('uint'):
+        atype = '<u' + str(int(ltype[4:]) / 8)
+    elif ltype.startswith('float'):
+        atype = '<f' + str(int(ltype[5:]) / 8)
+    else:
+        raise ValueError("bad layout")
+    return numpy.dtype([('timestamp', '<i8'), ('data', atype, lcount)])
+
+class NumpyClient(nilmdb.client.client.Client):
+    """Subclass of nilmdb.client.Client that adds additional methods for
+    extracting and inserting data via Numpy arrays."""
+
+    def _get_dtype(self, path, layout):
+        if layout is None:
+            streams = self.stream_list(path)
+            if len(streams) != 1:
+                raise ClientError("can't get layout for path: " + path)
+            layout = streams[0][1]
+        return layout_to_dtype(layout)
+
+    def stream_extract_numpy(self, path, start = None, end = None,
+                             layout = None, maxrows = 100000,
+                             structured = False):
+        """
+        Extract data from a stream.  Returns a generator that yields
+        Numpy arrays of up to 'maxrows' of data each.
+
+        If 'layout' is None, it is read using stream_info.
+
+        If 'structured' is False, all data is converted to float64
+        and returned in a flat 2D array.  Otherwise, data is returned
+        as a structured dtype in a 1D array.
+        """
+        dtype = self._get_dtype(path, layout)
+
+        def to_numpy(data):
+            a = numpy.fromstring(data, dtype)
+            if structured:
+                return a
+            return numpy.c_[a['timestamp'], a['data']]
+
+        chunks = []
+        total_len = 0
+        maxsize = dtype.itemsize * maxrows
+        for data in self.stream_extract(path, start, end, binary = True):
+            # Add this block of binary data
+            chunks.append(data)
+            total_len += len(data)
+
+            # See if we have enough to make the requested Numpy array
+            while total_len >= maxsize:
+                assembled = "".join(chunks)
+                total_len -= maxsize
+                chunks = [ assembled[maxsize:] ]
+                block = assembled[:maxsize]
+                yield to_numpy(block)
+
+        if total_len:
+            yield to_numpy("".join(chunks))
+
+    @contextlib.contextmanager
+    def stream_insert_numpy_context(self, path, start = None, end = None,
+                                    layout = None):
+        """Return a context manager that allows data to be efficiently
+        inserted into a stream in a piecewise manner.  Data is
+        provided as Numpy arrays, and is aggregated and sent to the
+        server in larger or smaller chunks as necessary.  Data format
+        must match the database layout for the given path.
+
+        For more details, see help for
+        nilmdb.client.numpyclient.StreamInserterNumpy
+
+        If 'layout' is not None, use it as the layout rather than
+        querying the database.
+        """
+        dtype = self._get_dtype(path, layout)
+        ctx = StreamInserterNumpy(self, path, start, end, dtype)
+        yield ctx
+        ctx.finalize()
+        ctx.destroy()
+
+    def stream_insert_numpy(self, path, data, start = None, end = None,
+                            layout = None):
+        """Insert data into a stream.  data should be a Numpy array
+        which will be passed through stream_insert_numpy_context to
+        break it into chunks etc.  See the help for that function
+        for details."""
+        with self.stream_insert_numpy_context(path, start, end, layout) as ctx:
+            if isinstance(data, numpy.ndarray):
+                ctx.insert(data)
+            else:
+                for chunk in data:
+                    ctx.insert(chunk)
+        return ctx.last_response
+
+class StreamInserterNumpy(nilmdb.client.client.StreamInserter):
+    """Object returned by stream_insert_numpy_context() that manages
+    the insertion of rows of data into a particular path.
+
+    See help for nilmdb.client.client.StreamInserter for details.
+    The only difference is that, instead of ASCII formatted data,
+    this context manager can take Numpy arrays, which are either
+    structured (1D with complex dtype) or flat (2D with simple dtype).
+    """
+
+    # Soft limit of how many bytes to send per HTTP request.
+    _max_data = 2 * 1024 * 1024
+
+    def __init__(self, client, path, start, end, dtype):
+        """
+        'client' is the client object.  'path' is the database path
+        to insert to.  'start' and 'end' are used for the first
+        contiguous interval and may be None.  'dtype' is the Numpy
+        dtype for this stream.
+        """
+        super(StreamInserterNumpy, self).__init__(client, path, start, end)
+        self._dtype = dtype
+
+        # Max rows to send at once
+        self._max_rows = self._max_data // self._dtype.itemsize
+
+        # List of the current arrays we're building up to send
+        self._block_arrays = []
+        self._block_rows = 0
+
+    def insert(self, array):
+        """Insert Numpy data, which must match the layout type."""
+        if type(array) != numpy.ndarray:
+            array = numpy.array(array)
+        if array.ndim == 1:
+            # Already a structured array; just verify the type
+            if array.dtype != self._dtype:
+                raise ValueError("wrong dtype for 1D (structured) array")
+        elif array.ndim == 2:
+            # Convert to structured array
+            sarray = numpy.zeros(array.shape[0], dtype=self._dtype)
+            try:
+                sarray['timestamp'] = array[:,0]
+                # Need the squeeze in case sarray['data'] is 1 dimensional
+                sarray['data'] = numpy.squeeze(array[:,1:])
+            except (IndexError, ValueError):
+                raise ValueError("wrong number of fields for this data type")
+            array = sarray
+        else:
+            raise ValueError("wrong number of dimensions in array")
+
+        length = len(array)
+        maxrows = self._max_rows
+
+        if length == 0:
+            return
+        if length > maxrows:
+            # This is more than twice what we wanted to send, so split
+            # it up.  This is a bit inefficient, but the user really
+            # shouldn't be providing this much data at once.
+            for cut in range(0, length, maxrows):
+                self.insert(array[cut:(cut + maxrows)])
+            return
+
+        # Add this array to our list
+        self._block_arrays.append(array)
+        self._block_rows += length
+
+        # Send if it's too long
+        if self._block_rows >= maxrows:
+            self._send_block(final = False)
+
+    def _send_block(self, final = False):
+        """Send the data current stored up.  One row might be left
+        over if we need its timestamp saved."""
+
+        # Build the full array to send
+        if self._block_rows == 0:
+            array = numpy.zeros(0, dtype = self._dtype)
+        else:
+            array = numpy.hstack(self._block_arrays)
+
+        # Get starting timestamp
+        start_ts = self._interval_start
+        if start_ts is None:
+            # Pull start from the first row
+            try:
+                start_ts = array['timestamp'][0]
+            except IndexError:
+                pass # no timestamp is OK, if we have no data
+
+        # Get ending timestamp
+        if final:
+            # For a final block, the timestamp is either the
+            # user-provided end, or the timestamp of the last line
+            # plus epsilon.
+            end_ts = self._interval_end
+            if end_ts is None:
+                try:
+                    end_ts = array['timestamp'][-1]
+                    end_ts += nilmdb.utils.time.epsilon
+                except IndexError:
+                    pass # no timestamp is OK, if we have no data
+            self._block_arrays = []
+            self._block_rows = 0
+
+            # Next block is completely fresh
+            self._interval_start = None
+            self._interval_end = None
+        else:
+            # An intermediate block.  We need to save the last row
+            # for the next block, and use its timestamp as the ending
+            # timestamp for this one.
+            if len(array) < 2:
+                # Not enough data to send an intermediate block
+                return
+            end_ts = array['timestamp'][-1]
+            if self._interval_end is not None and end_ts > self._interval_end:
+                # User gave us bad endpoints; send it anyway, and let
+                # the server complain so that the error is the same
+                # as if we hadn't done this chunking.
+                end_ts = self._interval_end
+            self._block_arrays = [ array[-1:] ]
+            self._block_rows = 1
+            array = array[:-1]
+
+            # Next block continues where this one ended
+            self._interval_start = end_ts
+
+        # If we have no endpoints, or equal endpoints, it's OK as long
+        # as there's no data to send
+        if (start_ts is None or end_ts is None) or (start_ts == end_ts):
+            if len(array) == 0:
+                return
+            raise ClientError("have data to send, but invalid start/end times")
+
+        # Send it
+        data = array.tostring()
+        self.last_response = self._client.stream_insert_block(
+            self._path, data, start_ts, end_ts, binary = True)
+
+        return

nilmdb/cmdline/cmdline.py

@@ -10,6 +10,7 @@ import sys
 import os
 import argparse
 from argparse import ArgumentDefaultsHelpFormatter as def_form
+import signal
 
 try: # pragma: no cover
     import argcomplete
@@ -71,17 +72,29 @@ class Complete(object): # pragma: no cover
         path = parsed_args.path
         if not path:
             return []
-        return ( self.escape(k + '=' + v)
-                 for (k,v) in client.stream_get_metadata(path).iteritems()
-                 if k.startswith(prefix) )
-
+        results = []
+        # prefix comes in as UTF-8, but results need to be Unicode,
+        # weird.  Still doesn't work in all cases, but that's bugs in
+        # argcomplete.
+        prefix = nilmdb.utils.unicode.decode(prefix)
+        for (k,v) in client.stream_get_metadata(path).iteritems():
+            kv = self.escape(k + '=' + v)
+            if kv.startswith(prefix):
+                results.append(kv)
+        return results
 
 class Cmdline(object):
 
     def __init__(self, argv = None):
         self.argv = argv or sys.argv[1:]
+        try:
+            # Assume command line arguments are encoded with stdin's encoding,
+            # and reverse it.  Won't be needed in Python 3, but for now..
+            self.argv = [ x.decode(sys.stdin.encoding) for x in self.argv ]
+        except Exception: # pragma: no cover
+            pass
         self.client = None
-        self.def_url = os.environ.get("NILMDB_URL", "http://localhost:12380")
+        self.def_url = os.environ.get("NILMDB_URL", "http://localhost/nilmdb/")
         self.subcmd = {}
         self.complete = Complete()
 
@@ -126,6 +139,13 @@ class Cmdline(object):
         sys.exit(-1)
 
     def run(self):
+        # Set SIGPIPE to its default handler -- we don't need Python
+        # to catch it for us.
+        try:
+            signal.signal(signal.SIGPIPE, signal.SIG_DFL)
+        except ValueError: # pragma: no cover
+            pass
+
         # Clear cached timezone, so that we can pick up timezone changes
         # while running this from the test suite.
         datetime_tz._localtz = None

nilmdb/cmdline/destroy.py

@@ -1,5 +1,6 @@
 from nilmdb.utils.printf import *
 import nilmdb.client
+import fnmatch
 
 from argparse import ArgumentDefaultsHelpFormatter as def_form
 
@@ -7,20 +8,42 @@ def setup(self, sub):
     cmd = sub.add_parser("destroy", help="Delete a stream and all data",
                          formatter_class = def_form,
                          description="""
-                         Destroy the stream at the specified path.  All
-                         data and metadata related to the stream is
-                         permanently deleted.
+                         Destroy the stream at the specified path.
+                         The stream must be empty.  All metadata
+                         related to the stream is permanently deleted.
+
+                         Wildcards and multiple paths are supported.
                          """)
     cmd.set_defaults(handler = cmd_destroy)
+    group = cmd.add_argument_group("Options")
+    group.add_argument("-R", "--remove", action="store_true",
+                       help="Remove all data before destroying stream")
+    group.add_argument("-q", "--quiet", action="store_true",
+                       help="Don't display names when destroying "
+                       "multiple paths")
     group = cmd.add_argument_group("Required arguments")
-    group.add_argument("path",
-                       help="Path of the stream to delete, e.g. /foo/bar",
+    group.add_argument("path", nargs='+',
+                       help="Path of the stream to delete, e.g. /foo/bar/*",
                        ).completer = self.complete.path
     return cmd
 
 def cmd_destroy(self):
     """Destroy stream"""
+    streams = [ s[0] for s in self.client.stream_list() ]
+    paths = []
+    for path in self.args.path:
+        new = fnmatch.filter(streams, path)
+        if not new:
+            self.die("error: no stream matched path: %s", path)
+        paths.extend(new)
+
+    for path in paths:
+        if not self.args.quiet and len(paths) > 1:
+            printf("Destroying %s\n", path)
+
         try:
-        self.client.stream_destroy(self.args.path)
+            if self.args.remove:
+                count = self.client.stream_remove(path)
+            self.client.stream_destroy(path)
         except nilmdb.client.ClientError as e:
             self.die("error destroying stream: %s", str(e))

nilmdb/cmdline/extract.py

@@ -1,6 +1,7 @@
 from __future__ import print_function
 from nilmdb.utils.printf import *
 import nilmdb.client
+import sys
 
 def setup(self, sub):
     cmd = sub.add_parser("extract", help="Extract data",
@@ -24,11 +25,15 @@ def setup(self, sub):
                        ).completer = self.complete.time
 
     group = cmd.add_argument_group("Output format")
+    group.add_argument("-B", "--binary", action="store_true",
+                       help="Raw binary output")
     group.add_argument("-b", "--bare", action="store_true",
                        help="Exclude timestamps from output lines")
     group.add_argument("-a", "--annotate", action="store_true",
                        help="Include comments with some information "
                        "about the stream")
+    group.add_argument("-m", "--markup", action="store_true",
+                       help="Include comments with interval starts and ends")
     group.add_argument("-T", "--timestamp-raw", action="store_true",
                        help="Show raw timestamps in annotated information")
     group.add_argument("-c", "--count", action="store_true",
@@ -40,6 +45,11 @@ def cmd_extract_verify(self):
         if self.args.start > self.args.end:
             self.parser.error("start is after end")
 
+    if self.args.binary:
+        if (self.args.bare or self.args.annotate or self.args.markup or
+            self.args.timestamp_raw or self.args.count):
+            self.parser.error("--binary cannot be combined with other options")
+
 def cmd_extract(self):
     streams = self.client.stream_list(self.args.path)
     if len(streams) != 1:
@@ -58,15 +68,23 @@ def cmd_extract(self):
         printf("# end: %s\n", time_string(self.args.end))
 
     printed = False
+    if self.args.binary:
+        printer = sys.stdout.write
+    else:
+        printer = print
+    bare = self.args.bare
+    count = self.args.count
     for dataline in self.client.stream_extract(self.args.path,
                                                self.args.start,
                                                self.args.end,
-                                               self.args.count):
-        if self.args.bare and not self.args.count:
+                                               self.args.count,
+                                               self.args.markup,
+                                               self.args.binary):
+        if bare and not count:
             # Strip timestamp (first element).  Doesn't make sense
             # if we are only returning a count.
             dataline = ' '.join(dataline.split(' ')[1:])
-        print(dataline)
+        printer(dataline)
         printed = True
     if not printed:
         if self.args.annotate:

nilmdb/cmdline/info.py

@@ -21,5 +21,8 @@ def cmd_info(self):
     printf("Server URL: %s\n", self.client.geturl())
     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"]))
+    for (desc, field) in [("used by NilmDB", "size"),
+                          ("used by other", "other"),
+                          ("reserved", "reserved"),
+                          ("free", "free")]:
+        printf("Server disk space %s: %s\n", desc, human_size(dbinfo[field]))

nilmdb/cmdline/list.py

@@ -10,22 +10,16 @@ def setup(self, sub):
                          formatter_class = def_form,
                          description="""
                          List streams available in the database,
-                         optionally filtering by layout or path.  Wildcards
-                         are accepted.
+                         optionally filtering by path.  Wildcards
+                         are accepted; non-matching paths or wildcards
+                         are ignored.
                          """)
     cmd.set_defaults(verify = cmd_list_verify,
                      handler = cmd_list)
 
     group = cmd.add_argument_group("Stream filtering")
-    group.add_argument("-p", "--path", metavar="PATH", default="*",
-                       help="Match only this path (-p can be omitted)",
+    group.add_argument("path", metavar="PATH", default=["*"], nargs='*',
                        ).completer = self.complete.path
-    group.add_argument("path_positional", default="*",
-                       nargs="?", help=argparse.SUPPRESS,
-                       ).completer = self.complete.path
-    group.add_argument("-l", "--layout", default="*",
-                       help="Match only this stream layout",
-                       ).completer = self.complete.layout
 
     group = cmd.add_argument_group("Interval info")
     group.add_argument("-E", "--ext", action="store_true",
@@ -49,20 +43,12 @@ def setup(self, sub):
     group = cmd.add_argument_group("Misc options")
     group.add_argument("-T", "--timestamp-raw", action="store_true",
                        help="Show raw timestamps when printing times")
+    group.add_argument("-l", "--layout", action="store_true",
+                       help="Show layout type next to path name")
 
     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.
-    got_opt = self.args.path != "*"
-    got_pos = self.args.path_positional != "*"
-    if got_pos:
-        if got_opt:
-            self.parser.error("too many paths specified")
-        else:
-            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 must precede end")
@@ -80,13 +66,16 @@ def cmd_list(self):
     else:
         time_string = nilmdb.utils.time.timestamp_to_human
 
+    for argpath in self.args.path:
         for stream in streams:
             (path, layout, int_min, int_max, rows, time) = stream[:6]
-        if not (fnmatch.fnmatch(path, self.args.path) and
-                fnmatch.fnmatch(layout, self.args.layout)):
+            if not fnmatch.fnmatch(path, argpath):
                 continue
 
+            if self.args.layout:
                 printf("%s %s\n", path, layout)
+            else:
+                printf("%s\n", path)
 
             if self.args.ext:
                 if int_min is None or int_max is None:
@@ -102,7 +91,8 @@ def cmd_list(self):
                 printed = False
                 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))
+                    printf("  [ %s -> %s ]\n",
+                           time_string(start), time_string(end))
                     printed = True
                 if not printed:
                     printf("  (no intervals)\n")

nilmdb/cmdline/metadata.py

@@ -9,7 +9,8 @@ def setup(self, sub):
                          a stream.
                          """,
                          usage="%(prog)s path [-g [key ...] | "
-                         "-s key=value [...] | -u key=value [...]]")
+                         "-s key=value [...] | -u key=value [...]] | "
+                         "-d [key ...]")
     cmd.set_defaults(handler = cmd_metadata)
 
     group = cmd.add_argument_group("Required arguments")
@@ -30,6 +31,9 @@ def setup(self, sub):
                      help="Update metadata using provided "
                      "key=value pairs",
                      ).completer = self.complete.meta_keyval
+    exc.add_argument("-d", "--delete", nargs="*", metavar="key",
+                     help="Delete metadata for specified keys (default all)",
+                     ).completer = self.complete.meta_key
     return cmd
 
 def cmd_metadata(self):
@@ -37,10 +41,10 @@ def cmd_metadata(self):
     if self.args.set is not None or self.args.update is not None:
         # Either set, or update
         if self.args.set is not None:
-            keyvals = self.args.set
+            keyvals = map(nilmdb.utils.unicode.decode, self.args.set)
             handler = self.client.stream_set_metadata
         else:
-            keyvals = self.args.update
+            keyvals = map(nilmdb.utils.unicode.decode, self.args.update)
             handler = self.client.stream_update_metadata
 
         # Extract key=value pairs
@@ -56,15 +60,31 @@ def cmd_metadata(self):
             handler(self.args.path, data)
         except nilmdb.client.ClientError as e:
             self.die("error setting/updating metadata: %s", str(e))
+    elif self.args.delete is not None:
+        # Delete (by setting values to empty strings)
+        keys = None
+        if self.args.delete:
+            keys = map(nilmdb.utils.unicode.decode, self.args.delete)
+        try:
+            data = self.client.stream_get_metadata(self.args.path, keys)
+            for key in data:
+                data[key] = ""
+            self.client.stream_update_metadata(self.args.path, data)
+        except nilmdb.client.ClientError as e:
+            self.die("error deleting metadata: %s", str(e))
     else:
         # Get (or unspecified)
-        keys = self.args.get or None
+        keys = None
+        if self.args.get:
+            keys = map(nilmdb.utils.unicode.decode, self.args.get)
         try:
             data = self.client.stream_get_metadata(self.args.path, keys)
         except nilmdb.client.ClientError as e:
             self.die("error getting metadata: %s", str(e))
         for key, value in sorted(data.items()):
-            # Omit nonexistant keys
+            # Print nonexistant keys as having empty value
             if value is None:
                 value = ""
-            printf("%s=%s\n", key, value)
+            printf("%s=%s\n",
+                   nilmdb.utils.unicode.encode(key),
+                   nilmdb.utils.unicode.encode(value))

nilmdb/cmdline/remove.py

@@ -1,17 +1,19 @@
 from nilmdb.utils.printf import *
 import nilmdb.client
+import fnmatch
 
 def setup(self, sub):
     cmd = sub.add_parser("remove", help="Remove data",
                          description="""
                          Remove all data from a specified time range within a
-                         stream.
+                         stream.  If multiple streams or wildcards are provided,
+                         the same time range is removed from all streams.
                          """)
     cmd.set_defaults(handler = cmd_remove)
 
     group = cmd.add_argument_group("Data selection")
-    group.add_argument("path",
-                       help="Path of stream, e.g. /foo/bar",
+    group.add_argument("path", nargs='+',
+                       help="Path of stream, e.g. /foo/bar/*",
                        ).completer = self.complete.path
     group.add_argument("-s", "--start", required=True,
                        metavar="TIME", type=self.arg_time,
@@ -23,18 +25,31 @@ def setup(self, sub):
                        ).completer = self.complete.time
 
     group = cmd.add_argument_group("Output format")
+    group.add_argument("-q", "--quiet", action="store_true",
+                       help="Don't display names when removing "
+                       "from multiple paths")
     group.add_argument("-c", "--count", action="store_true",
                        help="Output number of data points removed")
     return cmd
 
 def cmd_remove(self):
+    streams = [ s[0] for s in self.client.stream_list() ]
+    paths = []
+    for path in self.args.path:
+        new = fnmatch.filter(streams, path)
+        if not new:
+            self.die("error: no stream matched path: %s", path)
+        paths.extend(new)
+
     try:
-        count = self.client.stream_remove(self.args.path,
+        for path in paths:
+            if not self.args.quiet and len(paths) > 1:
+                printf("Removing from %s\n", path)
+            count = self.client.stream_remove(path,
                                               self.args.start, self.args.end)
+            if self.args.count:
+                printf("%d\n", count);
     except nilmdb.client.ClientError as e:
         self.die("error removing data: %s", str(e))
 
-    if self.args.count:
-        printf("%d\n", count)
-
     return 0

nilmdb/scripts/nilmdb_server.py

@@ -22,7 +22,7 @@ def main():
     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"))
+                       default = "./db")
     group.add_argument('-q', '--quiet', help = 'Silence output',
                        action = 'store_true')
     group.add_argument('-t', '--traceback',

nilmdb/server/__init__.py

@@ -17,5 +17,5 @@ except (ImportError, TypeError): # pragma: no cover
     pass
 
 from nilmdb.server.nilmdb import NilmDB
-from nilmdb.server.server import Server
+from nilmdb.server.server import Server, wsgi_application
 from nilmdb.server.errors import NilmDBError, StreamError, OverlapError

nilmdb/server/bulkdata.py

@@ -14,18 +14,21 @@ import re
 import sys
 import tempfile
 
+import nilmdb.utils.lock
 from . import rocket
 
 # 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
+table_cache_size = 32
+fd_cache_size = 8
 
 @nilmdb.utils.must_close(wrap_verify = False)
 class BulkData(object):
     def __init__(self, basepath, **kwargs):
         self.basepath = basepath
         self.root = os.path.join(self.basepath, "data")
+        self.lock = self.root + ".lock"
+        self.lockfile = None
 
         # Tuneables
         if "file_size" in kwargs:
@@ -44,8 +47,22 @@ class BulkData(object):
         if not os.path.isdir(self.root):
             os.mkdir(self.root)
 
+        # Create the lock
+        self.lockfile = open(self.lock, "w")
+        if not nilmdb.utils.lock.exclusive_lock(self.lockfile):
+            raise IOError('database at "' + self.basepath +
+                          '" is already locked by another process')
+
     def close(self):
         self.getnode.cache_remove_all()
+        if self.lockfile:
+            nilmdb.utils.lock.exclusive_unlock(self.lockfile)
+            self.lockfile.close()
+            try:
+                os.unlink(self.lock)
+            except OSError: # pragma: no cover
+                pass
+            self.lockfile = None
 
     def _encode_filename(self, path):
         # Encode all paths to UTF-8, regardless of sys.getfilesystemencoding(),
@@ -62,7 +79,12 @@ class BulkData(object):
         if Table.exists(ospath):
             raise ValueError("stream already exists at this path")
         if os.path.isdir(ospath):
-            raise ValueError("subdirs of this path already exist")
+            # Look for any files in subdirectories.  Fully empty subdirectories
+            # are OK; they might be there during a rename
+            for (root, dirs, files) in os.walk(ospath):
+                if len(files):
+                    raise ValueError(
+                        "non-empty subdirs of this path already exist")
 
     def _create_parents(self, unicodepath):
         """Verify the path name, and create parent directories if they
@@ -134,7 +156,7 @@ class BulkData(object):
 
             # Open and cache it
             self.getnode(unicodepath)
-        except:
+        except Exception:
             exc_info = sys.exc_info()
             try:
                 os.rmdir(ospath)
@@ -171,7 +193,6 @@ class BulkData(object):
         # Basic checks
         if oldospath == newospath:
             raise ValueError("old and new paths are the same")
-        self._create_check_ospath(newospath)
 
         # Move the table to a temporary location
         tmpdir = tempfile.mkdtemp(prefix = "rename-", dir = self.root)
@@ -179,6 +200,9 @@ class BulkData(object):
         os.rename(oldospath, tmppath)
 
         try:
+            # Check destination path
+            self._create_check_ospath(newospath)
+
             # Create parent dirs for new location
             self._create_parents(newunicodepath)
 
@@ -371,7 +395,7 @@ class Table(object):
             # Try deleting subdir, too
             try:
                 os.rmdir(os.path.join(self.root, subdir))
-            except:
+            except Exception:
                 pass
 
     # Cache open files
@@ -389,12 +413,16 @@ class Table(object):
         return rocket.Rocket(self.layout,
                              os.path.join(self.root, subdir, filename))
 
-    def append_string(self, data, start, end):
+    def append_data(self, data, start, end, binary = False):
         """Parse the formatted string in 'data', according to the
         current layout, and append it to the table.  If any timestamps
         are non-monotonic, or don't fall between 'start' and 'end',
         a ValueError is raised.
 
+        If 'binary' is True, the data should be in raw binary format
+        instead: little-endian, matching the current table's layout,
+        including the int64 timestamp.
+
         If this function succeeds, it returns normally.  Otherwise,
         the table is reverted back to its original state by truncating
         or deleting files as necessary."""
@@ -413,17 +441,26 @@ class Table(object):
                 # Ask the rocket object to parse and append up to "count"
                 # rows of data, verifying things along the way.
                 try:
+                    if binary:
+                        appender = f.append_binary
+                    else:
+                        appender = f.append_string
                     (added_rows, data_offset, last_timestamp, linenum
-                     ) = f.append_string(count, data, data_offset, linenum,
+                     ) = appender(count, data, data_offset, linenum,
                                   start, end, last_timestamp)
                 except rocket.ParseError as e:
                     (linenum, colnum, errtype, obj) = e.args
+                    if binary:
+                        where = "byte %d: " % (linenum)
+                    else:
                         where = "line %d, column %d: " % (linenum, colnum)
                     # Extract out the error line, add column marker
                     try:
+                        if binary:
+                            raise IndexError
                         bad = data.splitlines()[linenum-1]
-                        badptr = ' ' * (colnum - 1) + '^'
-                    except IndexError: # pragma: no cover
+                        bad += '\n' + ' ' * (colnum - 1) + '^'
+                    except IndexError:
                         bad = ""
                     if errtype == rocket.ERR_NON_MONOTONIC:
                         err = "timestamp is not monotonically increasing"
@@ -439,7 +476,7 @@ class Table(object):
                     else:
                         err = str(obj)
                     raise ValueError("error parsing input data: " +
-                                     where + err + "\n" + bad + "\n" + badptr)
+                                     where + err + "\n" + bad)
                 tot_rows += added_rows
         except Exception:
             # Some failure, so try to roll things back by truncating or
@@ -455,7 +492,7 @@ class Table(object):
             # Success, so update self.nrows accordingly
             self.nrows = tot_rows
 
-    def get_data(self, start, stop):
+    def get_data(self, start, stop, binary = False):
         """Extract data corresponding to Python range [n:m],
         and returns a formatted string"""
         if (start is None or
@@ -473,10 +510,13 @@ class Table(object):
             if count > remaining:
                 count = remaining
             f = self.file_open(subdir, filename)
+            if binary:
+                ret.append(f.extract_binary(offset, count))
+            else:
                 ret.append(f.extract_string(offset, count))
             remaining -= count
             row += count
-        return "".join(ret)
+        return b"".join(ret)
 
     def __getitem__(self, row):
         """Extract timestamps from a row, with table[n] notation."""
@@ -504,7 +544,7 @@ class Table(object):
             with open(cachefile, "rb") as f:
                 ranges = pickle.load(f)
             cachefile_present = True
-        except:
+        except Exception:
             ranges = []
             cachefile_present = False
 

nilmdb/server/interval.pyx

@@ -286,22 +286,17 @@ cdef class IntervalSet:
         (potentially) subsetted to make the one that is being
         returned.
         """
-        if not isinstance(interval, Interval):
-            raise TypeError("bad type")
+        if orig:
             for n in self.tree.intersect(interval.start, interval.end):
                 i = n.obj
-            if i:
-                if i.start >= interval.start and i.end <= interval.end:
-                    if orig:
-                        yield (i, i)
-                    else:
-                        yield i
-                else:
                 subset = i.subset(max(i.start, interval.start),
                                   min(i.end, interval.end))
-                    if orig:
                 yield (subset, i)
         else:
+            for n in self.tree.intersect(interval.start, interval.end):
+                i = n.obj
+                subset = i.subset(max(i.start, interval.start),
+                                  min(i.end, interval.end))
                 yield subset
 
     cpdef intersects(self, Interval other):

nilmdb/server/nilmdb.py

@@ -12,6 +12,7 @@ Manages both the SQL database and the table storage backend.
 from __future__ import absolute_import
 import nilmdb.utils
 from nilmdb.utils.printf import *
+from nilmdb.utils.time import timestamp_to_string
 
 from nilmdb.utils.interval import IntervalError
 from nilmdb.server.interval import Interval, DBInterval, IntervalSet
@@ -83,7 +84,18 @@ class NilmDB(object):
     verbose = 0
 
     def __init__(self, basepath, max_results=None,
-                 bulkdata_args=None):
+                 max_removals=None, bulkdata_args=None):
+        """Initialize NilmDB at the given basepath.
+        Other arguments are for debugging / testing:
+
+        'max_results' is the max rows to send in a single
+        stream_intervals or stream_extract response.
+
+        'max_removals' is the max rows to delete at once
+        in stream_move.
+
+        'bulkdata_args' is kwargs for the bulkdata module.
+        """
         if bulkdata_args is None:
             bulkdata_args = {}
 
@@ -94,7 +106,9 @@ class NilmDB(object):
         try:
             os.makedirs(self.basepath)
         except OSError as e:
-            if e.errno != errno.EEXIST:
+            if e.errno != errno.EEXIST: # pragma: no cover
+                # (no coverage, because it's hard to trigger this case
+                # if tests are run as root)
                 raise IOError("can't create tree " + self.basepath)
 
         # Our data goes inside it
@@ -105,19 +119,20 @@ class NilmDB(object):
         self.con = sqlite3.connect(sqlfilename, check_same_thread = True)
         try:
             self._sql_schema_update()
-        finally: # pragma: no cover
+        except Exception: # pragma: no cover
             self.data.close()
+            raise
 
         # See big comment at top about the performance implications of this
         self.con.execute("PRAGMA synchronous=NORMAL")
         self.con.execute("PRAGMA journal_mode=WAL")
 
         # Approximate largest number of elements that we want to send
-        # in a single reply (for stream_intervals, stream_extract)
-        if max_results:
-            self.max_results = max_results
-        else:
-            self.max_results = 16384
+        # in a single reply (for stream_intervals, stream_extract).
+        self.max_results = max_results or 16384
+
+        # Remove up to this many rows per call to stream_remove.
+        self.max_removals = max_removals or 1048576
 
     def get_basepath(self):
         return self.basepath
@@ -161,7 +176,7 @@ class NilmDB(object):
             raise NilmDBError("start must precede end")
         return (start, end)
 
-    @nilmdb.utils.lru_cache(size = 16)
+    @nilmdb.utils.lru_cache(size = 64)
     def _get_intervals(self, stream_id):
         """
         Return a mutable IntervalSet corresponding to the given stream ID.
@@ -334,14 +349,14 @@ class NilmDB(object):
 
         Returns (intervals, restart) tuple.
 
-        intervals is a list of [start,end] timestamps of all intervals
+        'intervals' is a list of [start,end] timestamps of all intervals
         that exist for path, between start and end.
 
-        restart, if nonzero, means that there were too many results to
-        return in a single request.  The data is complete from the
-        starting timestamp to the point at which it was truncated,
-        and a new request with a start time of 'restart' will fetch
-        the next block of data.
+        'restart', if not None, means that there were too many results
+        to return in a single request.  The data is complete from the
+        starting timestamp to the point at which it was truncated, and
+        a new request with a start time of 'restart' will fetch the
+        next block of data.
         """
         stream_id = self._stream_id(path)
         intervals = self._get_intervals(stream_id)
@@ -363,7 +378,7 @@ class NilmDB(object):
                 break
             result.append([i.start, i.end])
         else:
-            restart = 0
+            restart = None
         return (result, restart)
 
     def stream_create(self, path, layout_name):
@@ -439,28 +454,37 @@ class NilmDB(object):
                         (newpath, stream_id))
 
     def stream_destroy(self, path):
-        """Fully remove a table and all of its data from the database.
-        No way to undo it!  Metadata is removed."""
+        """Fully remove a table from the database.  Fails if there are
+        any intervals data present; remove them first.  Metadata is
+        also removed."""
         stream_id = self._stream_id(path)
 
-        # Delete the cached interval data (if it was cached)
+        # Verify that no intervals are present, and clear the cache
+        iset = self._get_intervals(stream_id)
+        if len(iset):
+            raise NilmDBError("all intervals must be removed before "
+                              "destroying a stream")
         self._get_intervals.cache_remove(self, stream_id)
 
-        # Delete the data
+        # Delete the bulkdata storage
         self.data.destroy(path)
 
-        # Delete metadata, stream, intervals
+        # Delete metadata, stream, intervals (should be none)
         with self.con as con:
             con.execute("DELETE FROM metadata WHERE stream_id=?", (stream_id,))
             con.execute("DELETE FROM ranges WHERE stream_id=?", (stream_id,))
             con.execute("DELETE FROM streams WHERE id=?", (stream_id,))
 
-    def stream_insert(self, path, start, end, data):
+    def stream_insert(self, path, start, end, data, binary = False):
         """Insert new data into the database.
            path: Path at which to add the data
            start: Starting timestamp
            end: Ending timestamp
            data: Textual data, formatted according to the layout of path
+
+           'binary', if True, means that 'data' is raw binary:
+           little-endian, matching the current table's layout,
+           including the int64 timestamp.
            """
         # First check for basic overlap using timestamp info given.
         stream_id = self._stream_id(path)
@@ -474,7 +498,7 @@ class NilmDB(object):
         # there are any parse errors.
         table = self.data.getnode(path)
         row_start = table.nrows
-        table.append_string(data, start, end)
+        table.append_data(data, start, end, binary)
         row_end = table.nrows
 
         # Insert the record into the sql database.
@@ -517,23 +541,31 @@ class NilmDB(object):
                                   dbinterval.db_startpos,
                                   dbinterval.db_endpos)
 
-    def stream_extract(self, path, start = None, end = None, count = False):
+    def stream_extract(self, path, start = None, end = None,
+                       count = False, markup = False, binary = False):
         """
         Returns (data, restart) tuple.
 
-        data is ASCII-formatted data from the database, formatted
+        'data' is ASCII-formatted data from the database, formatted
         according to the layout of the stream.
 
-        restart, if nonzero, means that there were too many results to
+        'restart', if not None, means that there were too many results to
         return in a single request.  The data is complete from the
         starting timestamp to the point at which it was truncated,
         and a new request with a start time of 'restart' will fetch
         the next block of data.
 
-        count, if true, means to not return raw data, but just the count
+        'count', if true, means to not return raw data, but just the count
         of rows that would have been returned.  This is much faster
         than actually fetching the data.  It is not limited by
         max_results.
+
+        'markup', if true, indicates that returned data should be
+        marked with a comment denoting when a particular interval
+        starts, and another comment when an interval ends.
+
+        'binary', if true, means to return raw binary rather than
+        ASCII-formatted data.
         """
         stream_id = self._stream_id(path)
         table = self.data.getnode(path)
@@ -543,7 +575,9 @@ class NilmDB(object):
         result = []
         matched = 0
         remaining = self.max_results
-        restart = 0
+        restart = None
+        if binary and (markup or count):
+            raise NilmDBError("binary mode can't be used with markup or count")
         for interval in intervals.intersection(requested):
             # Reading single rows from the table is too slow, so
             # we use two bisections to find both the starting and
@@ -562,14 +596,26 @@ class NilmDB(object):
                 row_end = row_max
                 restart = table[row_max]
 
+            # Add markup
+            if markup:
+                result.append("# interval-start " +
+                              timestamp_to_string(interval.start) + "\n")
+
             # Gather these results up
-            result.append(table.get_data(row_start, row_end))
+            result.append(table.get_data(row_start, row_end, binary))
 
             # Count them
             remaining -= row_end - row_start
 
-            if restart:
+            # Add markup, and exit if restart is set.
+            if restart is not None:
+                if markup:
+                    result.append("# interval-end " +
+                                  timestamp_to_string(restart) + "\n")
                 break
+            if markup:
+                result.append("# interval-end " +
+                              timestamp_to_string(interval.end) + "\n")
 
         if count:
             return matched
@@ -578,9 +624,17 @@ class NilmDB(object):
     def stream_remove(self, path, start = None, end = None):
         """
         Remove data from the specified time interval within a stream.
-        Removes all data in the interval [start, end), and intervals
-        are truncated or split appropriately.  Returns the number of
-        data points removed.
+
+        Removes data in the interval [start, end), and intervals are
+        truncated or split appropriately.
+
+        Returns a (removed, restart) tuple.
+
+        'removed' is the number of data points that were removed.
+
+        'restart', if not None, means there were too many rows to
+        remove in a single request.  This function should be called
+        again with a start time of 'restart' to complete the removal.
         """
         stream_id = self._stream_id(path)
         table = self.data.getnode(path)
@@ -588,6 +642,8 @@ class NilmDB(object):
         (start, end) = self._check_user_times(start, end)
         to_remove = Interval(start, end)
         removed = 0
+        remaining = self.max_removals
+        restart = None
 
         # Can't remove intervals from within the iterator, so we need to
         # remember what's currently in the intersection now.
@@ -598,6 +654,13 @@ class NilmDB(object):
             row_start = self._find_start(table, dbint)
             row_end = self._find_end(table, dbint)
 
+            # Shorten it if we'll hit the maximum number of removals
+            row_max = row_start + remaining
+            if row_max < row_end:
+                row_end = row_max
+                dbint.end = table[row_max]
+                restart = dbint.end
+
             # Adjust the DBInterval to match the newly found ends
             dbint.db_start = dbint.start
             dbint.db_end = dbint.end
@@ -612,5 +675,9 @@ class NilmDB(object):
 
             # Count how many were removed
             removed += row_end - row_start
+            remaining -= row_end - row_start
 
-        return removed
+            if restart is not None:
+                break
+
+        return (removed, restart)

nilmdb/server/rocket.c

@@ -5,6 +5,9 @@
 #include <ctype.h>
 #include <stdint.h>
 
+#define __STDC_FORMAT_MACROS
+#include <inttypes.h>
+
 /* Values missing from stdint.h */
 #define UINT8_MIN 0
 #define UINT16_MIN 0
@@ -19,16 +22,9 @@
 
 typedef int64_t timestamp_t;
 
-/* This code probably needs to be double-checked for the case where
-   sizeof(long) != 8, so enforce that here with something that will
-   fail at build time.  We assume that the python integer type can
-   hold an int64_t. */
-const static char __long_ok[1 - 2*!(sizeof(int64_t) ==
-				    sizeof(long int))] = { 0 };
-
 /* Somewhat arbitrary, just so we can use fixed sizes for strings
    etc. */
-static const int MAX_LAYOUT_COUNT = 128;
+static const int MAX_LAYOUT_COUNT = 1024;
 
 /* Error object and constants */
 static PyObject *ParseError;
@@ -58,7 +54,7 @@ static PyObject *raise_str(int line, int col, int code, const char *string)
 static PyObject *raise_int(int line, int col, int code, int64_t num)
 {
 	PyObject *o;
-	o = Py_BuildValue("(iiil)", line, col, code, num);
+	o = Py_BuildValue("(iiiL)", line, col, code, (long long)num);
 	if (o != NULL) {
 		PyErr_SetObject(ParseError, o);
 		Py_DECREF(o);
@@ -249,11 +245,11 @@ static PyObject *Rocket_get_file_size(Rocket *self)
 /****
  * Append from string
  */
-static inline long int strtol10(const char *nptr, char **endptr) {
-	return strtol(nptr, endptr, 10);
+static inline long int strtoll10(const char *nptr, char **endptr) {
+	return strtoll(nptr, endptr, 10);
 }
-static inline long int strtoul10(const char *nptr, char **endptr) {
-	return strtoul(nptr, endptr, 10);
+static inline long int strtoull10(const char *nptr, char **endptr) {
+	return strtoull(nptr, endptr, 10);
 }
 
 /* .append_string(count, data, offset, linenum, start, end, last_timestamp) */
@@ -264,6 +260,7 @@ static PyObject *Rocket_append_string(Rocket *self, PyObject *args)
 	int offset;
 	const char *linestart;
 	int linenum;
+        long long ll1, ll2, ll3;
 	timestamp_t start;
 	timestamp_t end;
 	timestamp_t last_timestamp;
@@ -280,10 +277,13 @@ static PyObject *Rocket_append_string(Rocket *self, PyObject *args)
 	   but we need the null termination for strto*.  If we had
 	   strnto* that took a length, we could use t# and not require
 	   a copy. */
-	if (!PyArg_ParseTuple(args, "isiilll:append_string", &count,
+	if (!PyArg_ParseTuple(args, "isiiLLL:append_string", &count,
 			      &data, &offset, &linenum,
-			      &start, &end, &last_timestamp))
+			      &ll1, &ll2, &ll3))
 		return NULL;
+        start = ll1;
+        end = ll2;
+        last_timestamp = ll3;
 
 	/* Skip spaces, but don't skip over a newline. */
 #define SKIP_BLANK(buf) do {			\
@@ -372,14 +372,14 @@ static PyObject *Rocket_append_string(Rocket *self, PyObject *args)
 				goto extra_data_on_line;		\
 			break
 
-			CS(INT8,   strtol10,  t64.i, t8.i,  t8.u,         , 1);
-			CS(UINT8,  strtoul10, t64.u, t8.u,  t8.u,         , 1);
-			CS(INT16,  strtol10,  t64.i, t16.i, t16.u, le16toh, 2);
-			CS(UINT16, strtoul10, t64.u, t16.u, t16.u, le16toh, 2);
-			CS(INT32,  strtol10,  t64.i, t32.i, t32.u, le32toh, 4);
-			CS(UINT32, strtoul10, t64.u, t32.u, t32.u, le32toh, 4);
-			CS(INT64,  strtol10,  t64.i, t64.i, t64.u, le64toh, 8);
-			CS(UINT64, strtoul10, t64.u, t64.u, t64.u, le64toh, 8);
+			CS(INT8,   strtoll10,  t64.i, t8.i,  t8.u,         , 1);
+			CS(UINT8,  strtoull10, t64.u, t8.u,  t8.u,         , 1);
+			CS(INT16,  strtoll10,  t64.i, t16.i, t16.u, le16toh, 2);
+			CS(UINT16, strtoull10, t64.u, t16.u, t16.u, le16toh, 2);
+			CS(INT32,  strtoll10,  t64.i, t32.i, t32.u, le32toh, 4);
+			CS(UINT32, strtoull10, t64.u, t32.u, t32.u, le32toh, 4);
+			CS(INT64,  strtoll10,  t64.i, t64.i, t64.u, le64toh, 8);
+			CS(UINT64, strtoull10, t64.u, t64.u, t64.u, le64toh, 8);
 			CS(FLOAT32, strtod,   t64.d, t32.f, t32.u, le32toh, 4);
 			CS(FLOAT64, strtod,   t64.d, t64.d, t64.u, le64toh, 8);
 #undef CS
@@ -397,7 +397,8 @@ static PyObject *Rocket_append_string(Rocket *self, PyObject *args)
 	/* Build return value and return */
 	offset = buf - data;
 	PyObject *o;
-	o = Py_BuildValue("(iili)", written, offset, last_timestamp, linenum);
+	o = Py_BuildValue("(iiLi)", written, offset,
+                          (long long)last_timestamp, linenum);
 	return o;
 err:
 	PyErr_SetFromErrno(PyExc_OSError);
@@ -419,6 +420,72 @@ extra_data_on_line:
 			 ERR_OTHER, "extra data on line");
 }
 
+/****
+ * Append from binary data
+ */
+
+/* .append_binary(count, data, offset, linenum, start, end, last_timestamp) */
+static PyObject *Rocket_append_binary(Rocket *self, PyObject *args)
+{
+        int count;
+	const uint8_t *data;
+        int data_len;
+        int linenum;
+	int offset;
+        long long ll1, ll2, ll3;
+	timestamp_t start;
+	timestamp_t end;
+	timestamp_t last_timestamp;
+
+	if (!PyArg_ParseTuple(args, "it#iiLLL:append_binary",
+                              &count, &data, &data_len, &offset,
+                              &linenum, &ll1, &ll2, &ll3))
+		return NULL;
+        start = ll1;
+        end = ll2;
+        last_timestamp = ll3;
+
+        /* Advance to offset */
+        if (offset > data_len)
+                return raise_str(0, 0, ERR_OTHER, "bad offset");
+        data += offset;
+        data_len -= offset;
+
+        /* Figure out max number of rows to insert */
+        int rows = data_len / self->binary_size;
+        if (rows > count)
+                rows = count;
+
+        /* Check timestamps */
+        timestamp_t ts;
+	int i;
+        for (i = 0; i < rows; i++) {
+                /* Read raw timestamp, byteswap if needed */
+                memcpy(&ts, &data[i * self->binary_size], 8);
+                ts = le64toh(ts);
+
+                /* Check limits */
+                if (ts <= last_timestamp)
+                        return raise_int(i, 0, ERR_NON_MONOTONIC, ts);
+                last_timestamp = ts;
+                if (ts < start || ts >= end)
+                        return raise_int(i, 0, ERR_OUT_OF_INTERVAL, ts);
+        }
+
+        /* Write binary data */
+        if (fwrite(data, self->binary_size, rows, self->file) != rows) {
+                PyErr_SetFromErrno(PyExc_OSError);
+                return NULL;
+        }
+	fflush(self->file);
+
+	/* Build return value and return */
+	PyObject *o;
+	o = Py_BuildValue("(iiLi)", rows, offset + rows * self->binary_size,
+                          (long long)last_timestamp, linenum);
+	return o;
+}
+
 /****
  * Extract to string
  */
@@ -472,7 +539,7 @@ static PyObject *Rocket_extract_string(Rocket *self, PyObject *args)
 		if (fread(&t64.u, 8, 1, self->file) != 1)
 			goto err;
 		t64.u = le64toh(t64.u);
-		ret = sprintf(&str[len], "%ld", t64.i);
+		ret = sprintf(&str[len], "%" PRId64, t64.i);
 		if (ret <= 0)
 			goto err;
 		len += ret;
@@ -484,7 +551,7 @@ static PyObject *Rocket_extract_string(Rocket *self, PyObject *args)
 			/* read and format in a loop */			\
 			for (i = 0; i < self->layout_count; i++) {	\
 				if (fread(&disktype, bytes,		\
-					  1, self->file) < 0)		\
+					  1, self->file) != 1)		\
 					goto err;			\
 				disktype = letoh(disktype);		\
 				ret = sprintf(&str[len], " " fmt,	\
@@ -494,14 +561,14 @@ static PyObject *Rocket_extract_string(Rocket *self, PyObject *args)
 				len += ret;				\
 			}						\
 			break
-			CASE(INT8,   "%hhd",   t8.i,  t8.u,         , 1);
-			CASE(UINT8,  "%hhu",   t8.u,  t8.u,         , 1);
-			CASE(INT16,  "%hd",    t16.i, t16.u, le16toh, 2);
-			CASE(UINT16, "%hu",    t16.u, t16.u, le16toh, 2);
-			CASE(INT32,  "%d",     t32.i, t32.u, le32toh, 4);
-			CASE(UINT32, "%u",     t32.u, t32.u, le32toh, 4);
-			CASE(INT64,  "%ld",    t64.i, t64.u, le64toh, 8);
-			CASE(UINT64, "%lu",    t64.u, t64.u, le64toh, 8);
+			CASE(INT8,   "%" PRId8,  t8.i,  t8.u,         , 1);
+			CASE(UINT8,  "%" PRIu8,  t8.u,  t8.u,         , 1);
+			CASE(INT16,  "%" PRId16, t16.i, t16.u, le16toh, 2);
+			CASE(UINT16, "%" PRIu16, t16.u, t16.u, le16toh, 2);
+			CASE(INT32,  "%" PRId32, t32.i, t32.u, le32toh, 4);
+			CASE(UINT32, "%" PRIu32, t32.u, t32.u, le32toh, 4);
+			CASE(INT64,  "%" PRId64, t64.i, t64.u, le64toh, 8);
+			CASE(UINT64, "%" PRIu64, t64.u, t64.u, le64toh, 8);
 			/* These next two are a bit debatable.  floats
 			   are 6-9 significant figures, so we print 7.
 			   Doubles are 15-19, so we print 17.  This is
@@ -527,6 +594,46 @@ err:
 	return NULL;
 }
 
+/****
+ * Extract to binary string containing raw little-endian binary data
+ */
+static PyObject *Rocket_extract_binary(Rocket *self, PyObject *args)
+{
+	long count;
+	long offset;
+
+	if (!PyArg_ParseTuple(args, "ll", &offset, &count))
+		return NULL;
+	if (!self->file) {
+		PyErr_SetString(PyExc_Exception, "no file");
+		return NULL;
+	}
+	/* Seek to target location */
+	if (fseek(self->file, offset, SEEK_SET) < 0) {
+		PyErr_SetFromErrno(PyExc_OSError);
+		return NULL;
+	}
+
+        uint8_t *str;
+        int len = count * self->binary_size;
+        str = malloc(len);
+        if (str == NULL) {
+                PyErr_SetFromErrno(PyExc_OSError);
+                return NULL;
+        }
+
+        /* Data in the file is already in the desired little-endian
+           binary format, so just read it directly. */
+        if (fread(str, self->binary_size, count, self->file) != count) {
+                free(str);
+                PyErr_SetFromErrno(PyExc_OSError);
+                return NULL;
+        }
+
+	PyObject *pystr = PyBytes_FromStringAndSize((char *)str, len);
+	free(str);
+	return pystr;
+}
 
 /****
  * Extract timestamp
@@ -551,7 +658,7 @@ static PyObject *Rocket_extract_timestamp(Rocket *self, PyObject *args)
 
 	/* Convert and return */
 	t64.u = le64toh(t64.u);
-	return Py_BuildValue("l", t64.i);
+	return Py_BuildValue("L", (long long)t64.i);
 }
 
 /****
@@ -571,11 +678,13 @@ static PyMemberDef Rocket_members[] = {
 };
 
 static PyMethodDef Rocket_methods[] = {
-	{ "close", (PyCFunction)Rocket_close, METH_NOARGS,
+	{ "close",
+          (PyCFunction)Rocket_close, METH_NOARGS,
 	  "close(self)\n\n"
 	  "Close file handle" },
 
-	{ "append_string", (PyCFunction)Rocket_append_string, METH_VARARGS,
+	{ "append_string",
+          (PyCFunction)Rocket_append_string, METH_VARARGS,
 	  "append_string(self, count, data, offset, line, start, end, ts)\n\n"
           "Parse string and append data.\n"
 	  "\n"
@@ -590,16 +699,46 @@ static PyMethodDef Rocket_methods[] = {
 	  "Raises ParseError if timestamps are non-monotonic, outside\n"
 	  "the start/end interval etc.\n"
 	  "\n"
-          "On success, return a tuple with three values:\n"
+          "On success, return a tuple:\n"
           "  added_rows: how many rows were added from the file\n"
           "  data_offset: current offset into the data string\n"
-          "  last_timestamp: last timestamp we parsed" },
+          "  last_timestamp: last timestamp we parsed\n"
+          "  linenum: current line number" },
 
-	{ "extract_string", (PyCFunction)Rocket_extract_string, METH_VARARGS,
+	{ "append_binary",
+	  (PyCFunction)Rocket_append_binary, METH_VARARGS,
+	  "append_binary(self, count, data, offset, line, start, end, ts)\n\n"
+          "Append binary data, which must match the data layout.\n"
+	  "\n"
+	  "  count: maximum number of rows to add\n"
+          "  data: binary data\n"
+          "  offset: byte offset into data to start adding\n"
+          "  line: current line number (unused)\n"
+          "  start: starting timestamp for interval\n"
+          "  end: end timestamp for interval\n"
+          "  ts: last timestamp that was previously parsed\n"
+	  "\n"
+	  "Raises ParseError if timestamps are non-monotonic, outside\n"
+	  "the start/end interval etc.\n"
+	  "\n"
+          "On success, return a tuple:\n"
+          "  added_rows: how many rows were added from the file\n"
+          "  data_offset: current offset into the data string\n"
+          "  last_timestamp: last timestamp we parsed\n"
+          "  linenum: current line number (copied from argument)" },
+
+	{ "extract_string",
+          (PyCFunction)Rocket_extract_string, METH_VARARGS,
 	  "extract_string(self, offset, count)\n\n"
 	  "Extract count rows of data from the file at offset offset.\n"
 	  "Return an ascii formatted string according to the layout" },
 
+	{ "extract_binary",
+	  (PyCFunction)Rocket_extract_binary, METH_VARARGS,
+	  "extract_binary(self, offset, count)\n\n"
+	  "Extract count rows of data from the file at offset offset.\n"
+	  "Return a raw binary string of data matching the data layout." },
+
 	{ "extract_timestamp",
 	  (PyCFunction)Rocket_extract_timestamp, METH_VARARGS,
 	  "extract_timestamp(self, offset)\n\n"

nilmdb/server/server.py

@@ -11,130 +11,32 @@ from nilmdb.utils.time import string_to_timestamp
 import cherrypy
 import sys
 import os
+import socket
 import simplejson as json
 import decorator
 import psutil
+import traceback
+
+from nilmdb.server.serverutil import (
+    chunked_response,
+    response_type,
+    workaround_cp_bug_1200,
+    exception_to_httperror,
+    CORS_allow,
+    json_to_request_params,
+    json_error_page,
+    cherrypy_start,
+    cherrypy_stop,
+    bool_param,
+    )
+
+# Add CORS_allow tool
+cherrypy.tools.CORS_allow = cherrypy.Tool('on_start_resource', CORS_allow)
 
 class NilmApp(object):
     def __init__(self, db):
         self.db = db
 
-# Decorators
-def chunked_response(func):
-    """Decorator to enable chunked responses."""
-    # Set this to False to get better tracebacks from some requests
-    # (/stream/extract, /stream/intervals).
-    func._cp_config = { 'response.stream': True }
-    return func
-
-def response_type(content_type):
-    """Return a decorator-generating function that sets the
-    response type to the specified string."""
-    def wrapper(func, *args, **kwargs):
-        cherrypy.response.headers['Content-Type'] = content_type
-        return func(*args, **kwargs)
-    return decorator.decorator(wrapper)
-
-@decorator.decorator
-def workaround_cp_bug_1200(func, *args, **kwargs): # pragma: no cover
-    """Decorator to work around CherryPy bug #1200 in a response
-    generator.
-
-    Even if chunked responses are disabled, LookupError or
-    UnicodeError exceptions may still be swallowed by CherryPy due to
-    bug #1200.  This throws them as generic Exceptions instead so that
-    they make it through.
-    """
-    exc_info = None
-    try:
-        for val in func(*args, **kwargs):
-            yield val
-    except (LookupError, UnicodeError):
-        # Re-raise it, but maintain the original traceback
-        exc_info = sys.exc_info()
-        new_exc = Exception(exc_info[0].__name__ + ": " + str(exc_info[1]))
-        raise new_exc, None, exc_info[2]
-    finally:
-        del exc_info
-
-def exception_to_httperror(*expected):
-    """Return a decorator-generating function that catches expected
-    errors and throws a HTTPError describing it instead.
-
-        @exception_to_httperror(NilmDBError, ValueError)
-        def foo():
-            pass
-    """
-    def wrapper(func, *args, **kwargs):
-        exc_info = None
-        try:
-            return func(*args, **kwargs)
-        except expected:
-            # Re-raise it, but maintain the original traceback
-            exc_info = sys.exc_info()
-            new_exc = cherrypy.HTTPError("400 Bad Request", str(exc_info[1]))
-            raise new_exc, None, exc_info[2]
-        finally:
-            del exc_info
-    # We need to preserve the function's argspecs for CherryPy to
-    # handle argument errors correctly.  Decorator.decorator takes
-    # 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"""
@@ -145,7 +47,10 @@ class Root(NilmApp):
     # /
     @cherrypy.expose
     def index(self):
-        raise cherrypy.NotFound()
+        cherrypy.response.headers['Content-Type'] = 'text/plain'
+        msg = sprintf("This is NilmDB version %s, running on host %s.\n",
+                      nilmdb.__version__, socket.getfqdn())
+        return msg
 
     # /favicon.ico
     @cherrypy.expose
@@ -165,13 +70,32 @@ class Root(NilmApp):
         """Return a dictionary with the database path,
         size of the database in bytes, and free disk space in bytes"""
         path = self.db.get_basepath()
+        usage = psutil.disk_usage(path)
+        dbsize = nilmdb.utils.du(path)
         return { "path": path,
-                 "size": nilmdb.utils.du(path),
-                 "free": psutil.disk_usage(path).free }
+                 "size": dbsize,
+                 "other": usage.used - dbsize,
+                 "reserved": usage.total - usage.used - usage.free,
+                 "free": usage.free }
 
 class Stream(NilmApp):
     """Stream-specific operations"""
 
+    # Helpers
+    def _get_times(self, start_param, end_param):
+        (start, end) = (None, None)
+        if start_param is not None:
+            start = string_to_timestamp(start_param)
+        if end_param is not None:
+            end = string_to_timestamp(end_param)
+        if start is not None and end is not None:
+            if start >= end:
+                raise cherrypy.HTTPError(
+                    "400 Bad Request",
+                    sprintf("start must precede end (%s >= %s)",
+                            start_param, end_param))
+        return (start, end)
+
     # /stream/list
     # /stream/list?layout=float32_8
     # /stream/list?path=/newton/prep&extended=1
@@ -182,10 +106,10 @@ class Stream(NilmApp):
         layout parameter, just list streams that match the given path
         or layout.
 
-        If extent is not given, returns a list of lists containing
-        the path and layout: [ path, layout ]
+        If extended is missing or zero, returns a list of lists
+        containing the path and layout: [ path, layout ]
 
-        If extended is provided, returns a list of lists containing
+        If extended is true, returns a list of lists containing
         extended info: [ path, layout, extent_min, extent_max,
         total_rows, total_seconds ].  More data may be added.
         """
@@ -210,7 +134,7 @@ class Stream(NilmApp):
     @exception_to_httperror(NilmDBError)
     @cherrypy.tools.CORS_allow(methods = ["POST"])
     def destroy(self, path):
-        """Delete a stream and its associated data."""
+        """Delete a stream.  Fails if any data is still present."""
         return self.db.stream_destroy(path)
 
     # /stream/rename?oldpath=/newton/prep&newpath=/newton/prep/1
@@ -288,32 +212,42 @@ class Stream(NilmApp):
     @cherrypy.tools.json_out()
     @exception_to_httperror(NilmDBError, ValueError)
     @cherrypy.tools.CORS_allow(methods = ["PUT"])
-    def insert(self, path, start, end):
+    def insert(self, path, start, end, binary = False):
         """
         Insert new data into the database.  Provide textual data
         (matching the path's layout) as a HTTP PUT.
+
+        If 'binary' is True, expect raw binary data, rather than lines
+        of ASCII-formatted data.  Raw binary data is always
+        little-endian and matches the database types (including an
+        int64 timestamp).
         """
+        binary = bool_param(binary)
+
         # Important that we always read the input before throwing any
         # errors, to keep lengths happy for persistent connections.
         # 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()
 
+        # Verify content type for binary data
+        content_type = cherrypy.request.headers.get('content-type')
+        if binary and content_type:
+            if content_type != "application/octet-stream":
+                raise cherrypy.HTTPError("400", "Content type must be "
+                                         "application/octet-stream for "
+                                         "binary data, not " + content_type)
+
         # Check path and get layout
-        streams = self.db.stream_list(path = path)
-        if len(streams) != 1:
-            raise cherrypy.HTTPError("404 Not Found", "No such stream")
+        if len(self.db.stream_list(path = path)) != 1:
+            raise cherrypy.HTTPError("404", "No such stream: " + path)
 
         # Check limits
-        start = string_to_timestamp(start)
-        end = string_to_timestamp(end)
-        if start >= end:
-            raise cherrypy.HTTPError("400 Bad Request",
-                                     "start must precede end")
+        (start, end) = self._get_times(start, end)
 
         # Pass the data directly to nilmdb, which will parse it and
         # raise a ValueError if there are any problems.
-        self.db.stream_insert(path, start, end, body)
+        self.db.stream_insert(path, start, end, body, binary)
 
         # Done
         return
@@ -322,24 +256,34 @@ class Stream(NilmApp):
     # /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"])
+    @chunked_response
+    @response_type("application/x-json-stream")
     def remove(self, path, start = None, end = None):
         """
         Remove data from the backend database.  Removes all data in
-        the interval [start, end).  Returns the number of data points
-        removed.
+        the interval [start, end).
+
+        Returns the number of data points removed.  Since this is a potentially
+        long-running operation, multiple numbers may be returned as the
+        data gets removed from the backend database.  The total number of
+        points removed is the sum of all of these numbers.
         """
-        if start is not None:
-            start = string_to_timestamp(start)
-        if end is not None:
-            end = string_to_timestamp(end)
-        if start is not None and end is not None:
-            if start >= end:
-                raise cherrypy.HTTPError("400 Bad Request",
-                                         "start must precede end")
-        return self.db.stream_remove(path, start, end)
+        (start, end) = self._get_times(start, end)
+
+        if len(self.db.stream_list(path = path)) != 1:
+            raise cherrypy.HTTPError("404", "No such stream: " + path)
+
+        @workaround_cp_bug_1200
+        def content(start, end):
+            # Note: disable chunked responses to see tracebacks from here.
+            while True:
+                (removed, restart) = self.db.stream_remove(path, start, end)
+                yield json.dumps(removed) + "\r\n"
+                if restart is None:
+                    break
+                start = restart
+        return content(start, end)
 
     # /stream/intervals?path=/newton/prep
     # /stream/intervals?path=/newton/prep&start=1234567890.0&end=1234567899.0
@@ -362,15 +306,7 @@ class Stream(NilmApp):
         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 = string_to_timestamp(start)
-        if end is not None:
-            end = string_to_timestamp(end)
-
-        if start is not None and end is not None:
-            if start >= end:
-                raise cherrypy.HTTPError("400 Bad Request",
-                                         "start must precede end")
+        (start, end) = self._get_times(start, end)
 
         if len(self.db.stream_list(path = path)) != 1:
             raise cherrypy.HTTPError("404", "No such stream: " + path)
@@ -386,7 +322,7 @@ class Stream(NilmApp):
                                                            diffpath)
                 response = ''.join([ json.dumps(i) + "\r\n" for i in ints ])
                 yield response
-                if restart == 0:
+                if restart is None:
                     break
                 start = restart
         return content(start, end)
@@ -394,48 +330,62 @@ class Stream(NilmApp):
     # /stream/extract?path=/newton/prep&start=1234567890.0&end=1234567899.0
     @cherrypy.expose
     @chunked_response
-    @response_type("text/plain")
-    def extract(self, path, start = None, end = None, count = False):
+    def extract(self, path, start = None, end = None,
+                count = False, markup = False, binary = False):
         """
         Extract data from backend database.  Streams the resulting
         entries as ASCII text lines separated by newlines.  This may
         make multiple requests to the nilmdb backend to avoid causing
         it to block for too long.
 
-        Add count=True to return a count rather than actual data.
-        """
-        if start is not None:
-            start = string_to_timestamp(start)
-        if end is not None:
-            end = string_to_timestamp(end)
+        If 'count' is True, returns a count rather than actual data.
 
-        # Check parameters
-        if start is not None and end is not None:
-            if start >= end:
-                raise cherrypy.HTTPError("400 Bad Request",
-                                         "start must precede end")
+        If 'markup' is True, adds comments to the stream denoting each
+        interval's start and end timestamp.
+
+        If 'binary' is True, return raw binary data, rather than lines
+        of ASCII-formatted data.  Raw binary data is always
+        little-endian and matches the database types (including an
+        int64 timestamp).
+        """
+        binary = bool_param(binary)
+        markup = bool_param(markup)
+        count = bool_param(count)
+
+        (start, end) = self._get_times(start, end)
 
         # Check path and get layout
-        streams = self.db.stream_list(path = path)
-        if len(streams) != 1:
-            raise cherrypy.HTTPError("404 Not Found", "No such stream")
+        if len(self.db.stream_list(path = path)) != 1:
+            raise cherrypy.HTTPError("404", "No such stream: " + path)
+
+        if binary:
+            content_type = "application/octet-stream"
+            if markup or count:
+                raise cherrypy.HTTPError("400", "can't mix binary and "
+                                         "markup or count modes")
+        else:
+            content_type = "text/plain"
+        cherrypy.response.headers['Content-Type'] = content_type
 
         @workaround_cp_bug_1200
-        def content(start, end, count):
+        def content(start, end):
             # Note: disable chunked responses to see tracebacks from here.
             if count:
-                matched = self.db.stream_extract(path, start, end, count)
+                matched = self.db.stream_extract(path, start, end,
+                                                 count = True)
                 yield sprintf("%d\n", matched)
                 return
 
             while True:
-                (data, restart) = self.db.stream_extract(path, start, end)
+                (data, restart) = self.db.stream_extract(
+                    path, start, end, count = False,
+                    markup = markup, binary = binary)
                 yield data
 
-                if restart == 0:
+                if restart is None:
                     return
                 start = restart
-        return content(start, end, count)
+        return content(start, end)
 
 class Exiter(object):
     """App that exits the server, for testing"""
@@ -453,7 +403,8 @@ class Server(object):
                  stoppable = False,       # whether /exit URL exists
                  embedded = True,         # hide diagnostics and output, etc
                  fast_shutdown = False,   # don't wait for clients to disconn.
-                 force_traceback = False  # include traceback in all errors
+                 force_traceback = False, # include traceback in all errors
+                 basepath = '',           # base URL path for cherrypy.tree
                  ):
         # Save server version, just for verification during tests
         self.version = nilmdb.__version__
@@ -513,7 +464,7 @@ class Server(object):
         if stoppable:
             root.exit = Exiter()
         cherrypy.tree.apps = {}
-        cherrypy.tree.mount(root, "/", config = { "/" : app_config })
+        cherrypy.tree.mount(root, basepath, config = { "/" : app_config })
 
         # Shutdowns normally wait for clients to disconnect.  To speed
         # up tests, set fast_shutdown = True
@@ -523,69 +474,68 @@ class Server(object):
         else:
             cherrypy.server.shutdown_timeout = 5
 
+        # Set up the WSGI application pointer for external programs
+        self.wsgi_application = cherrypy.tree
+
     def json_error_page(self, status, message, traceback, version):
         """Return a custom error page in JSON so the client can parse it"""
-        errordata = { "status" : status,
-                      "message" : message,
-                      "traceback" : traceback }
-        # Don't send a traceback if the error was 400-499 (client's fault)
-        try:
-            code = int(status.split()[0])
-            if not self.force_traceback:
-                if code >= 400 and code <= 499:
-                    errordata["traceback"] = ""
-        except Exception: # pragma: no cover
-            pass
-        # Override the response type, which was previously set to text/html
-        cherrypy.serving.response.headers['Content-Type'] = (
-            "application/json;charset=utf-8" )
-        # Undo the HTML escaping that cherrypy's get_error_page function applies
-        # (cherrypy issue 1135)
-        for k, v in errordata.iteritems():
-            v = v.replace("&lt;","<")
-            v = v.replace("&gt;",">")
-            v = v.replace("&amp;","&")
-            errordata[k] = v
-        return json.dumps(errordata, separators=(',',':'))
+        return json_error_page(status, message, traceback, version,
+                               self.force_traceback)
 
     def start(self, blocking = False, event = None):
-
-        if not self.embedded: # pragma: no cover
-            # Handle signals nicely
-            if hasattr(cherrypy.engine, "signal_handler"):
-                cherrypy.engine.signal_handler.subscribe()
-            if hasattr(cherrypy.engine, "console_control_handler"):
-                cherrypy.engine.console_control_handler.subscribe()
-
-        # Cherrypy stupidly calls os._exit(70) when it can't bind the
-        # port.  At least try to print a reasonable error and continue
-        # in this case, rather than just dying silently (as we would
-        # otherwise do in embedded mode)
-        real_exit = os._exit
-        def fake_exit(code): # pragma: no cover
-            if code == os.EX_SOFTWARE:
-                fprintf(sys.stderr, "error: CherryPy called os._exit!\n")
-            else:
-                real_exit(code)
-        os._exit = fake_exit
-        cherrypy.engine.start()
-        os._exit = real_exit
-
-        # Signal that the engine has started successfully
-        if event is not None:
-            event.set()
-
-        if blocking:
-            try:
-                cherrypy.engine.wait(cherrypy.engine.states.EXITING,
-                                     interval = 0.1, channel = 'main')
-            except (KeyboardInterrupt, IOError): # pragma: no cover
-                cherrypy.engine.log('Keyboard Interrupt: shutting down bus')
-                cherrypy.engine.exit()
-            except SystemExit: # pragma: no cover
-                cherrypy.engine.log('SystemExit raised: shutting down bus')
-                cherrypy.engine.exit()
-                raise
+        cherrypy_start(blocking, event, self.embedded)
 
     def stop(self):
-        cherrypy.engine.exit()
+        cherrypy_stop()
+
+# Use a single global nilmdb.server.NilmDB and nilmdb.server.Server
+# instance since the database can only be opened once.  For this to
+# work, the web server must use only a single process and single
+# Python interpreter.  Multiple threads are OK.
+_wsgi_server = None
+def wsgi_application(dbpath, basepath): # pragma: no cover
+    """Return a WSGI application object with a database at the
+    specified path.
+
+    'dbpath' is a filesystem location, e.g. /home/nilm/db
+
+    'basepath' is the URL path of the application base, which
+    is the same as the first argument to Apache's WSGIScriptAlias
+    directive.
+    """
+    def application(environ, start_response):
+        global _wsgi_server
+        if _wsgi_server is None:
+            # Try to start the server
+            try:
+                db = nilmdb.utils.serializer_proxy(nilmdb.server.NilmDB)(dbpath)
+                _wsgi_server = nilmdb.server.Server(
+                    db, embedded = True,
+                    basepath = basepath.rstrip('/'))
+            except Exception:
+                # Build an error message on failure
+                import pprint
+                err = sprintf("Initializing database at path '%s' failed:\n\n",
+                              dbpath)
+                err += traceback.format_exc()
+                try:
+                    import pwd
+                    import grp
+                    err += sprintf("\nRunning as: uid=%d (%s), gid=%d (%s) "
+                                   "on host %s, pid %d\n",
+                                   os.getuid(), pwd.getpwuid(os.getuid())[0],
+                                   os.getgid(), grp.getgrgid(os.getgid())[0],
+                                   socket.gethostname(), os.getpid())
+                except ImportError:
+                    pass
+                err += sprintf("\nEnvironment:\n%s\n", pprint.pformat(environ))
+        if _wsgi_server is None:
+            # Serve up the error with our own mini WSGI app.
+            headers = [ ('Content-type', 'text/plain'),
+                        ('Content-length', str(len(err))) ]
+            start_response("500 Internal Server Error", headers)
+            return [err]
+
+        # Call the normal application
+        return _wsgi_server.wsgi_application(environ, start_response)
+    return application

nilmdb/server/serverutil.py

@@ -0,0 +1,214 @@
+"""Miscellaneous decorators and other helpers for running a CherryPy
+server"""
+
+import cherrypy
+import sys
+import os
+import decorator
+import simplejson as json
+
+# Helper to parse parameters into booleans
+def bool_param(s):
+    """Return a bool indicating whether parameter 's' was True or False,
+    supporting a few different types for 's'."""
+    try:
+        ss = s.lower()
+        if ss in [ "0", "false", "f", "no", "n" ]:
+            return False
+        if ss in [ "1", "true", "t", "yes", "y" ]:
+            return True
+    except Exception:
+        return bool(s)
+    raise cherrypy.HTTPError("400 Bad Request",
+                             "can't parse parameter: " + ss)
+
+# Decorators
+def chunked_response(func):
+    """Decorator to enable chunked responses."""
+    # Set this to False to get better tracebacks from some requests
+    # (/stream/extract, /stream/intervals).
+    func._cp_config = { 'response.stream': True }
+    return func
+
+def response_type(content_type):
+    """Return a decorator-generating function that sets the
+    response type to the specified string."""
+    def wrapper(func, *args, **kwargs):
+        cherrypy.response.headers['Content-Type'] = content_type
+        return func(*args, **kwargs)
+    return decorator.decorator(wrapper)
+
+@decorator.decorator
+def workaround_cp_bug_1200(func, *args, **kwargs): # pragma: no cover
+    """Decorator to work around CherryPy bug #1200 in a response
+    generator.
+
+    Even if chunked responses are disabled, LookupError or
+    UnicodeError exceptions may still be swallowed by CherryPy due to
+    bug #1200.  This throws them as generic Exceptions instead so that
+    they make it through.
+    """
+    exc_info = None
+    try:
+        for val in func(*args, **kwargs):
+            yield val
+    except (LookupError, UnicodeError):
+        # Re-raise it, but maintain the original traceback
+        exc_info = sys.exc_info()
+        new_exc = Exception(exc_info[0].__name__ + ": " + str(exc_info[1]))
+        raise new_exc, None, exc_info[2]
+    finally:
+        del exc_info
+
+def exception_to_httperror(*expected):
+    """Return a decorator-generating function that catches expected
+    errors and throws a HTTPError describing it instead.
+
+        @exception_to_httperror(NilmDBError, ValueError)
+        def foo():
+            pass
+    """
+    def wrapper(func, *args, **kwargs):
+        exc_info = None
+        try:
+            return func(*args, **kwargs)
+        except expected:
+            # Re-raise it, but maintain the original traceback
+            exc_info = sys.exc_info()
+            new_exc = cherrypy.HTTPError("400 Bad Request", str(exc_info[1]))
+            raise new_exc, None, exc_info[2]
+        finally:
+            del exc_info
+    # We need to preserve the function's argspecs for CherryPy to
+    # handle argument errors correctly.  Decorator.decorator takes
+    # 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.
+
+    Add this to CherryPy with:
+    cherrypy.tools.CORS_allow = cherrypy.Tool('on_start_resource', CORS_allow)
+    """
+    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)
+
+
+# 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)
+
+# Used as an "error_page.default" handler
+def json_error_page(status, message, traceback, version,
+                    force_traceback = False):
+    """Return a custom error page in JSON so the client can parse it"""
+    errordata = { "status" : status,
+                  "message" : message,
+                  "traceback" : traceback }
+    # Don't send a traceback if the error was 400-499 (client's fault)
+    try:
+        code = int(status.split()[0])
+        if not force_traceback:
+            if code >= 400 and code <= 499:
+                errordata["traceback"] = ""
+    except Exception: # pragma: no cover
+        pass
+    # Override the response type, which was previously set to text/html
+    cherrypy.serving.response.headers['Content-Type'] = (
+        "application/json;charset=utf-8" )
+    # Undo the HTML escaping that cherrypy's get_error_page function applies
+    # (cherrypy issue 1135)
+    for k, v in errordata.iteritems():
+        v = v.replace("&lt;","<")
+        v = v.replace("&gt;",">")
+        v = v.replace("&amp;","&")
+        errordata[k] = v
+    return json.dumps(errordata, separators=(',',':'))
+
+# Start/stop CherryPy standalone server
+def cherrypy_start(blocking = False, event = False, embedded = False):
+    """Start the CherryPy server, handling errors and signals
+    somewhat gracefully."""
+
+    if not embedded: # pragma: no cover
+        # Handle signals nicely
+        if hasattr(cherrypy.engine, "signal_handler"):
+            cherrypy.engine.signal_handler.subscribe()
+        if hasattr(cherrypy.engine, "console_control_handler"):
+            cherrypy.engine.console_control_handler.subscribe()
+
+    # Cherrypy stupidly calls os._exit(70) when it can't bind the
+    # port.  At least try to print a reasonable error and continue
+    # in this case, rather than just dying silently (as we would
+    # otherwise do in embedded mode)
+    real_exit = os._exit
+    def fake_exit(code): # pragma: no cover
+        if code == os.EX_SOFTWARE:
+            fprintf(sys.stderr, "error: CherryPy called os._exit!\n")
+        else:
+            real_exit(code)
+    os._exit = fake_exit
+    cherrypy.engine.start()
+    os._exit = real_exit
+
+    # Signal that the engine has started successfully
+    if event is not None:
+        event.set()
+
+    if blocking:
+        try:
+            cherrypy.engine.wait(cherrypy.engine.states.EXITING,
+                                 interval = 0.1, channel = 'main')
+        except (KeyboardInterrupt, IOError): # pragma: no cover
+            cherrypy.engine.log('Keyboard Interrupt: shutting down bus')
+            cherrypy.engine.exit()
+        except SystemExit: # pragma: no cover
+            cherrypy.engine.log('SystemExit raised: shutting down bus')
+            cherrypy.engine.exit()
+            raise
+
+# Stop CherryPy server
+def cherrypy_stop():
+    cherrypy.engine.exit()

nilmdb/utils/__init__.py

@@ -1,7 +1,7 @@
 """NilmDB utilities"""
 
+from __future__ import absolute_import
 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
@@ -12,3 +12,6 @@ import nilmdb.utils.fallocate
 import nilmdb.utils.time
 import nilmdb.utils.iterator
 import nilmdb.utils.interval
+import nilmdb.utils.lock
+import nilmdb.utils.sort
+import nilmdb.utils.unicode

nilmdb/utils/diskusage.py

@@ -1,4 +1,5 @@
 import os
+import errno
 from math import log
 
 def human_size(num):
@@ -16,10 +17,17 @@ def human_size(num):
         return '1 byte'
 
 def du(path):
-    """Like du -sb, returns total size of path in bytes."""
+    """Like du -sb, returns total size of path in bytes.  Ignore
+    errors that might occur if we encounter broken symlinks or
+    files in the process of being removed."""
+    try:
         size = os.path.getsize(path)
         if os.path.isdir(path):
             for thisfile in os.listdir(path):
                 filepath = os.path.join(path, thisfile)
                 size += du(filepath)
         return size
+    except OSError as e: # pragma: no cover
+        if e.errno != errno.ENOENT:
+            raise
+        return 0

nilmdb/utils/interval.py

@@ -1,5 +1,6 @@
 """Interval.  Like nilmdb.server.interval, but re-implemented here
-in plain Python so clients have easier access to it.
+in plain Python so clients have easier access to it, and with a few
+helper functions.
 
 Intervals are half-open, ie. they include data points with timestamps
 [start, end)
@@ -34,6 +35,10 @@ class Interval:
         return ("[" + nilmdb.utils.time.timestamp_to_string(self.start) +
                 " -> " + nilmdb.utils.time.timestamp_to_string(self.end) + ")")
 
+    def human_string(self):
+        return ("[ " + nilmdb.utils.time.timestamp_to_human(self.start) +
+                " -> " + nilmdb.utils.time.timestamp_to_human(self.end) + " ]")
+
     def __cmp__(self, other):
         """Compare two intervals.  If non-equal, order by start then end"""
         return cmp(self.start, other.start) or cmp(self.end, other.end)
@@ -53,18 +58,11 @@ class Interval:
             raise IntervalError("not a subset")
         return Interval(start, end)
 
-def set_difference(a, b):
-    """
-    Compute the difference (a \\ b) between the intervals in 'a' and
-    the intervals in 'b'; i.e., the ranges that are present in 'self'
-    but not 'other'.
-
-    'a' and 'b' must both be iterables.
-
-    Returns a generator that yields each interval in turn.
-    Output intervals are built as subsets of the intervals in the
-    first argument (a).
-    """
+def _interval_math_helper(a, b, op, subset = True):
+    """Helper for set_difference, intersection functions,
+    to compute interval subsets based on a math operator on ranges
+    present in A and B.  Subsets are computed from A, or new intervals
+    are generated if subset = False."""
     # Iterate through all starts and ends in sorted order.  Add a
     # tag to the iterator so that we can figure out which one they
     # were, after sorting.
@@ -79,28 +77,71 @@ def set_difference(a, b):
     # At each point, evaluate which type of end it is, to determine
     # how to build up the output intervals.
     a_interval = None
-    b_interval = None
+    in_a = False
+    in_b = False
     out_start = None
     for (ts, k, i) in nilmdb.utils.iterator.imerge(a_iter, b_iter):
         if k == 0:
-            # start a interval
             a_interval = i
-            if b_interval is None:
-                out_start = ts
+            in_a = True
         elif k == 1:
-            # start b interval
-            b_interval = i
-            if out_start is not None and out_start != ts:
-                yield a_interval.subset(out_start, ts)
-            out_start = None
+            in_b = True
         elif k == 2:
-            # end a interval
-            if out_start is not None and out_start != ts:
-                yield a_interval.subset(out_start, ts)
-            out_start = None
-            a_interval = None
+            in_a = False
         elif k == 3:
-            # end b interval
-            b_interval = None
-            if a_interval:
+            in_b = False
+        include = op(in_a, in_b)
+        if include and out_start is None:
             out_start = ts
+        elif not include:
+            if out_start is not None and out_start != ts:
+                if subset:
+                    yield a_interval.subset(out_start, ts)
+                else:
+                    yield Interval(out_start, ts)
+            out_start = None
+
+def set_difference(a, b):
+    """
+    Compute the difference (a \\ b) between the intervals in 'a' and
+    the intervals in 'b'; i.e., the ranges that are present in 'self'
+    but not 'other'.
+
+    'a' and 'b' must both be iterables.
+
+    Returns a generator that yields each interval in turn.
+    Output intervals are built as subsets of the intervals in the
+    first argument (a).
+    """
+    return _interval_math_helper(a, b, (lambda a, b: a and not b))
+
+def intersection(a, b):
+    """
+    Compute the intersection between the intervals in 'a' and the
+    intervals in 'b'; i.e., the ranges that are present in both 'a'
+    and 'b'.
+
+    'a' and 'b' must both be iterables.
+
+    Returns a generator that yields each interval in turn.
+    Output intervals are built as subsets of the intervals in the
+    first argument (a).
+    """
+    return _interval_math_helper(a, b, (lambda a, b: a and b))
+
+def optimize(it):
+    """
+    Given an iterable 'it' with intervals, optimize them by joining
+    together intervals that are adjacent in time, and return a generator
+    that yields the new intervals.
+    """
+    saved_int = None
+    for interval in it:
+        if saved_int is not None:
+            if saved_int.end == interval.start:
+                interval.start = saved_int.start
+            else:
+                yield saved_int
+        saved_int = interval
+    if saved_int is not None:
+        yield saved_int

nilmdb/utils/iteratorizer.py

@@ -1,100 +0,0 @@
-import Queue
-import threading
-import sys
-import contextlib
-
-# This file provides a context manager that converts a function
-# that takes a callback into a generator that returns an iterable.
-# This is done by running the function in a new thread.
-
-# Based partially on http://stackoverflow.com/questions/9968592/
-
-class IteratorizerThread(threading.Thread):
-    def __init__(self, queue, function, curl_hack):
-        """
-        function: function to execute, which takes the
-        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
-        self.curl_hack = curl_hack
-
-    def callback(self, data):
-        try:
-            if self.die:
-                raise Exception() # trigger termination
-            self.queue.put((1, data))
-        except:
-            if self.curl_hack:
-                # We can't raise exceptions, because the pycurl
-                # extension module will unconditionally print the
-                # exception itself, and not pass it up to the caller.
-                # Instead, just return a value that tells curl to
-                # abort.  (-1 would be best, in case we were given 0
-                # bytes, but the extension doesn't support that).
-                self.queue.put((2, sys.exc_info()))
-                return 0
-            raise
-
-    def run(self):
-        try:
-            result = self.function(self.callback)
-        except:
-            self.queue.put((2, sys.exc_info()))
-        else:
-            self.queue.put((0, result))
-
-@contextlib.contextmanager
-def Iteratorizer(function, curl_hack = False):
-    """
-    Context manager that takes a function expecting a callback,
-    and provides an iterable that yields the values passed to that
-    callback instead.
-
-    function: function to execute, which takes a callback
-    (provided by this context manager) as an argument
-
-        with iteratorizer(func) as it:
-            for i in it:
-                print 'callback was passed:', i
-        print 'function returned:', it.retval
-    """
-    queue = Queue.Queue(maxsize = 1)
-    thread = IteratorizerThread(queue, function, curl_hack)
-    thread.daemon = True
-    thread.start()
-
-    class iteratorizer_gen(object):
-        def __init__(self, queue):
-            self.queue = queue
-            self.retval = None
-
-        def __iter__(self):
-            return self
-
-        def next(self):
-            (typ, data) = self.queue.get()
-            if typ == 0:
-                # function has returned
-                self.retval = data
-                raise StopIteration
-            elif typ == 1:
-                # data is available
-                return data
-            else:
-                # callback raised an exception
-                raise data[0], data[1], data[2]
-
-    try:
-        yield iteratorizer_gen(queue)
-    finally:
-        # Ask the thread to die, if it's still running.
-        thread.die = True
-        while thread.isAlive():
-            try:
-                queue.get(True, 0.01)
-            except: # pragma: no cover
-                pass

nilmdb/utils/lock.py

@@ -0,0 +1,33 @@
+# File locking
+
+import warnings
+
+try:
+    import fcntl
+    import errno
+
+    def exclusive_lock(f):
+        """Acquire an exclusive lock.  Returns True on successful
+        lock, or False on error."""
+        try:
+            fcntl.flock(f.fileno(), fcntl.LOCK_EX | fcntl.LOCK_NB)
+        except IOError as e:
+            if e.errno in (errno.EACCES, errno.EAGAIN):
+                return False
+            else: # pragma: no cover
+                raise
+        return True
+
+    def exclusive_unlock(f):
+        """Release an exclusive lock."""
+        fcntl.flock(f.fileno(), fcntl.LOCK_UN)
+
+except ImportError: # pragma: no cover
+    def exclusive_lock(f):
+        """Dummy lock function -- does not lock!"""
+        warnings.warn("Pretending to lock " + str(f))
+        return True
+
+    def exclusive_unlock(f):
+        """Release an exclusive lock."""
+        return

nilmdb/utils/mustclose.py

@@ -15,7 +15,7 @@ def must_close(errorfile = sys.stderr, wrap_verify = False):
         def wrap_class_method(wrapper):
             try:
                 orig = getattr(cls, wrapper.__name__).im_func
-            except:
+            except Exception:
                 orig = lambda x: None
             setattr(cls, wrapper.__name__, decorator.decorator(wrapper, orig))
 

nilmdb/utils/serializer.py

@@ -91,6 +91,20 @@ def serializer_proxy(obj_or_type):
             r = SerializerCallProxy(self.__call_queue, attr, self)
             return r
 
+        # For an interable object, on __iter__(), save the object's
+        # iterator and return this proxy.  On next(), call the object's
+        # iterator through this proxy.
+        def __iter__(self):
+            attr = getattr(self.__object, "__iter__")
+            self.__iter = SerializerCallProxy(self.__call_queue, attr, self)()
+            return self
+        def next(self):
+            return SerializerCallProxy(self.__call_queue,
+                                       self.__iter.next, self)()
+
+        def __getitem__(self, key):
+            return self.__getattr__("__getitem__")(key)
+
         def __call__(self, *args, **kwargs):
             """Call this to instantiate the type, if a type was passed
             to serializer_proxy.  Otherwise, pass the call through."""

nilmdb/utils/sort.py

@@ -0,0 +1,18 @@
+import re
+
+def sort_human(items, key = None):
+    """Human-friendly sort (/stream/2 before /stream/10)"""
+    def to_num(val):
+        try:
+            return int(val)
+        except Exception:
+            return val
+
+    def human_key(text):
+        if key:
+            text = key(text)
+        # Break into character and numeric chunks.
+        chunks = re.split(r'([0-9]+)', text)
+        return [ to_num(c) for c in chunks ]
+
+    return sorted(items, key = human_key)

nilmdb/utils/time.py

@@ -1,10 +1,12 @@
+from __future__ import absolute_import
+
 from nilmdb.utils import datetime_tz
 import re
 import time
 
 # Range
 min_timestamp = (-2**63)
-max_timestamp = (2**62 - 1)
+max_timestamp = (2**63 - 1)
 
 # Smallest representable step
 epsilon = 1
@@ -30,6 +32,10 @@ def timestamp_to_human(timestamp):
     """Convert a timestamp (integer microseconds since epoch) to a
     human-readable string, using the local timezone for display
     (e.g. from the TZ env var)."""
+    if timestamp == min_timestamp:
+        return "(minimum)"
+    if timestamp == max_timestamp:
+        return "(maximum)"
     dt = datetime_tz.datetime_tz.fromtimestamp(timestamp_to_unix(timestamp))
     return dt.strftime("%a, %d %b %Y %H:%M:%S.%f %z")
 
@@ -54,10 +60,23 @@ def rate_to_period(hz, cycles = 1):
 def parse_time(toparse):
     """
     Parse a free-form time string and return a nilmdb timestamp
-    (integer seconds since epoch).  If the string doesn't contain a
+    (integer microseconds since epoch).  If the string doesn't contain a
     timestamp, the current local timezone is assumed (e.g. from the TZ
     env var).
     """
+    if toparse == "min":
+        return min_timestamp
+    if toparse == "max":
+        return max_timestamp
+
+    # If it starts with @, treat it as a NILM timestamp
+    # (integer microseconds since epoch)
+    try:
+        if toparse[0] == '@':
+            return int(toparse[1:])
+    except (ValueError, KeyError, IndexError):
+        pass
+
     # 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.
@@ -71,14 +90,6 @@ def parse_time(toparse):
     except (ValueError, OverflowError):
         pass
 
-    # If it starts with @, treat it as a NILM timestamp
-    # (integer microseconds since epoch)
-    try:
-        if toparse[0] == '@':
-            return int(toparse[1:])
-    except (ValueError, KeyError):
-        pass
-
     # If it's parseable as a float, treat it as a Unix or NILM
     # timestamp based on its range.
     try:

nilmdb/utils/unicode.py

@@ -0,0 +1,29 @@
+import sys
+
+if sys.version_info[0] >= 3: # pragma: no cover (future Python3 compat)
+    text_type = str
+else:
+    text_type = unicode
+
+def encode(u):
+    """Try to encode something from Unicode to a string using the
+    default encoding.  If it fails, try encoding as UTF-8."""
+    if not isinstance(u, text_type):
+        return u
+    try:
+        return u.encode()
+    except UnicodeEncodeError:
+        return u.encode("utf-8")
+
+def decode(s):
+    """Try to decode someting from string to Unicode using the
+    default encoding.  If it fails, try decoding as UTF-8."""
+    if isinstance(s, text_type):
+        return s
+    try:
+        return s.decode()
+    except UnicodeDecodeError:
+        try:
+            return s.decode("utf-8")
+        except UnicodeDecodeError:
+            return s # best we can do

setup.py

@@ -39,7 +39,7 @@ 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
+except Exception: pass
 
 # Use Cython if it's new enough, otherwise use preexisting C files.
 cython_modules = [ 'nilmdb.server.interval',
@@ -107,13 +107,13 @@ setup(name='nilmdb',
       author_email = 'jim@jtan.com',
       tests_require = [ 'nose',
                         'coverage',
+                        'numpy',
                         ],
       setup_requires = [ 'distribute',
                          ],
       install_requires = [ 'decorator',
                            'cherrypy >= 3.2',
                            'simplejson',
-                           'pycurl',
                            'python-dateutil',
                            'pytz',
                            'psutil >= 0.3.0',

tests/data/extract-8

@@ -0,0 +1,28 @@
+# interval-start 1332496919900000
+1332496919900000 2.523050e+05 2.254020e+05 4.779410e+03 3.638030e+03 8.138070e+03 4.334460e+03 1.083780e+03 3.743730e+03
+1332496919908333 2.551190e+05 2.237870e+05 5.965640e+03 2.076350e+03 9.468790e+03 3.693880e+03 1.247860e+03 3.393680e+03
+1332496919916667 2.616370e+05 2.247980e+05 4.848970e+03 2.315620e+03 9.323300e+03 4.225460e+03 1.805780e+03 2.593050e+03
+1332496919925000 2.606460e+05 2.251300e+05 3.061360e+03 3.951840e+03 7.662910e+03 5.341410e+03 1.986520e+03 2.276780e+03
+1332496919933333 2.559710e+05 2.235030e+05 4.096030e+03 3.296970e+03 7.827080e+03 5.452120e+03 2.492520e+03 2.929450e+03
+1332496919941667 2.579260e+05 2.217080e+05 5.472320e+03 1.555700e+03 8.495760e+03 4.491140e+03 2.379780e+03 3.741710e+03
+1332496919950000 2.610180e+05 2.242350e+05 4.669770e+03 1.876190e+03 8.366680e+03 3.677510e+03 9.021690e+02 3.549040e+03
+1332496919958333 2.569150e+05 2.274650e+05 2.785070e+03 3.751930e+03 7.440320e+03 3.964860e+03 -3.227860e+02 2.460890e+03
+1332496919966667 2.509510e+05 2.262000e+05 3.772710e+03 3.131950e+03 8.159860e+03 4.539860e+03 7.375190e+02 2.126750e+03
+1332496919975000 2.556710e+05 2.223720e+05 5.826200e+03 8.715560e+02 9.120240e+03 4.545110e+03 2.804310e+03 2.721000e+03
+1332496919983333 2.649730e+05 2.214860e+05 5.839130e+03 4.659180e+02 8.628300e+03 3.934870e+03 2.972490e+03 3.773730e+03
+1332496919991667 2.652170e+05 2.233920e+05 3.718770e+03 2.834970e+03 7.209900e+03 3.460260e+03 1.324930e+03 4.075960e+03
+# interval-end 1332496919991668
+# interval-start 1332496920000000
+1332496920000000 2.564370e+05 2.244300e+05 4.011610e+03 3.475340e+03 7.495890e+03 3.388940e+03 2.613970e+02 3.731260e+03
+1332496920008333 2.539630e+05 2.241670e+05 5.621070e+03 1.548010e+03 9.165170e+03 3.522930e+03 1.058930e+03 2.996960e+03
+1332496920016667 2.585080e+05 2.249300e+05 6.011400e+03 8.188660e+02 9.039950e+03 4.482440e+03 2.490390e+03 2.679340e+03
+1332496920025000 2.596270e+05 2.260220e+05 4.474500e+03 2.423020e+03 7.414190e+03 5.071970e+03 2.439380e+03 2.962960e+03
+1332496920033333 2.551870e+05 2.246320e+05 4.738570e+03 3.398040e+03 7.395120e+03 4.726450e+03 1.839030e+03 3.393530e+03
+1332496920041667 2.571020e+05 2.216230e+05 6.144130e+03 1.441090e+03 8.756480e+03 3.495320e+03 1.869940e+03 3.752530e+03
+1332496920050000 2.636530e+05 2.217700e+05 6.221770e+03 7.389620e+02 9.547600e+03 2.666820e+03 1.462660e+03 3.332570e+03
+1332496920058333 2.636130e+05 2.252560e+05 4.477120e+03 2.437450e+03 8.510210e+03 3.855630e+03 9.594420e+02 2.387180e+03
+1332496920066667 2.553500e+05 2.262640e+05 4.283720e+03 3.923940e+03 7.912470e+03 5.466520e+03 1.284990e+03 2.093720e+03
+1332496920075000 2.527270e+05 2.246090e+05 5.851930e+03 2.491980e+03 8.540630e+03 5.623050e+03 2.339780e+03 3.007140e+03
+1332496920083333 2.584750e+05 2.235780e+05 5.924870e+03 1.394480e+03 8.779620e+03 4.544180e+03 2.132030e+03 3.849760e+03
+1332496920091667 2.615630e+05 2.246090e+05 4.336140e+03 2.455750e+03 8.055380e+03 3.469110e+03 6.278730e+02 3.664200e+03
+# interval-end 1332496920100000

tests/data/timestamped

@@ -0,0 +1,8 @@
+-10000000000 2.61246e+05  2.22735e+05  4.60340e+03  2.58221e+03  8.42804e+03  3.41890e+03  9.57898e+02  4.00585e+03  
+-100000000 2.61246e+05  2.22735e+05  4.60340e+03  2.58221e+03  8.42804e+03  3.41890e+03  9.57898e+02  4.00585e+03  
+-100000 2.61246e+05  2.22735e+05  4.60340e+03  2.58221e+03  8.42804e+03  3.41890e+03  9.57898e+02  4.00585e+03  
+-1000 2.61246e+05  2.22735e+05  4.60340e+03  2.58221e+03  8.42804e+03  3.41890e+03  9.57898e+02  4.00585e+03  
+1 2.61246e+05  2.22735e+05  4.60340e+03  2.58221e+03  8.42804e+03  3.41890e+03  9.57898e+02  4.00585e+03  
+1000 2.61246e+05  2.22735e+05  4.60340e+03  2.58221e+03  8.42804e+03  3.41890e+03  9.57898e+02  4.00585e+03  
+1000000 2.61246e+05  2.22735e+05  4.60340e+03  2.58221e+03  8.42804e+03  3.41890e+03  9.57898e+02  4.00585e+03  
+1000000000 2.61246e+05  2.22735e+05  4.60340e+03  2.58221e+03  8.42804e+03  3.41890e+03  9.57898e+02  4.00585e+03  

tests/runtests.py

@@ -24,7 +24,7 @@ class JimOrderPlugin(nose.plugins.Plugin):
                     name, workingDir=loader.workingDir)
                 try:
                     order = os.path.join(addr.filename, "test.order")
-                except:
+                except Exception:
                     order = None
                 if order and os.path.exists(order):
                     files = []

tests/test.order

@@ -4,7 +4,6 @@ test_lrucache.py
 test_mustclose.py
 
 test_serializer.py
-test_iteratorizer.py
 
 test_timestamper.py
 test_rbtree.py
@@ -13,6 +12,7 @@ test_interval.py
 test_bulkdata.py
 test_nilmdb.py
 test_client.py
+test_numpyclient.py
 test_cmdline.py
 
 test_*.py

tests/test_bulkdata.py

@@ -30,6 +30,11 @@ class TestBulkData(object):
         else:
             data = BulkData(db, file_size = size, files_per_dir = files)
 
+        # Try opening it again (should result in locking error)
+        with assert_raises(IOError) as e:
+            data2 = BulkData(db)
+        in_("already locked by another process", str(e.exception))
+
         # create empty
         with assert_raises(ValueError):
             data.create("/foo", "uint16_8")
@@ -64,9 +69,9 @@ class TestBulkData(object):
         raw = []
         for i in range(1000):
             raw.append("%d 1 2 3 4 5 6 7 8\n" % (10000 + i))
-        node.append_string("".join(raw[0:1]), 0, 50000)
-        node.append_string("".join(raw[1:100]), 0, 50000)
-        node.append_string("".join(raw[100:]), 0, 50000)
+        node.append_data("".join(raw[0:1]), 0, 50000)
+        node.append_data("".join(raw[1:100]), 0, 50000)
+        node.append_data("".join(raw[100:]), 0, 50000)
 
         misc_slices = [ 0, 100, slice(None), slice(0), slice(10),
                         slice(5,10), slice(3,None), slice(3,-3),
@@ -80,7 +85,7 @@ class TestBulkData(object):
         # Extract misc slices while appending, to make sure the
         # data isn't being added in the middle of the file
         for s in [2, slice(1,5), 2, slice(1,5)]:
-            node.append_string("0 0 0 0 0 0 0 0 0\n", 0, 50000)
+            node.append_data("0 0 0 0 0 0 0 0 0\n", 0, 50000)
             raw.append("0 0 0 0 0 0 0 0 0\n")
             eq_(get_node_slice(s), raw[s])
 

tests/test_client.py

@@ -23,6 +23,7 @@ import warnings
 import resource
 import time
 import re
+import struct
 
 from testutil.helpers import *
 
@@ -104,16 +105,19 @@ class TestClient(object):
                 client.http.post("/stream/list")
         client = nilmdb.client.Client(url = testurl)
 
-        # Create three streams
+        # Create four streams
         client.stream_create("/newton/prep", "float32_8")
         client.stream_create("/newton/raw", "uint16_6")
-        client.stream_create("/newton/zzz/rawnotch", "uint16_9")
+        client.stream_create("/newton/zzz/rawnotch2", "uint16_9")
+        client.stream_create("/newton/zzz/rawnotch11", "uint16_9")
 
-        # Verify we got 3 streams
+        # Verify we got 4 streams in the right order
         eq_(client.stream_list(), [ ["/newton/prep", "float32_8"],
                                     ["/newton/raw", "uint16_6"],
-                                    ["/newton/zzz/rawnotch", "uint16_9"]
+                                    ["/newton/zzz/rawnotch2", "uint16_9"],
+                                    ["/newton/zzz/rawnotch11", "uint16_9"]
                                     ])
+
         # Match just one type or one path
         eq_(client.stream_list(layout="uint16_6"),
             [ ["/newton/raw", "uint16_6"] ])
@@ -238,6 +242,22 @@ class TestClient(object):
         in_("400 Bad Request", str(e.exception))
         in_("start must precede end", str(e.exception))
 
+        # Good content type
+        with assert_raises(ClientError) as e:
+            client.http.put("stream/insert", "",
+                            { "path": "xxxx", "start": 0, "end": 1,
+                              "binary": 1 },
+                            binary = True)
+        in_("No such stream", str(e.exception))
+
+        # Bad content type
+        with assert_raises(ClientError) as e:
+            client.http.put("stream/insert", "",
+                            { "path": "xxxx", "start": 0, "end": 1,
+                              "binary": 1 },
+                            binary = False)
+        in_("Content type must be application/octet-stream", str(e.exception))
+
         # Specify start/end (starts too late)
         data = timestamper.TimestamperRate(testfile, start, 120)
         with assert_raises(ClientError) as e:
@@ -293,6 +313,27 @@ class TestClient(object):
         # Test count
         eq_(client.stream_count("/newton/prep"), 14400)
 
+        # Test binary output
+        with assert_raises(ClientError) as e:
+            list(client.stream_extract("/newton/prep",
+                                       markup = True, binary = True))
+        with assert_raises(ClientError) as e:
+            list(client.stream_extract("/newton/prep",
+                                       count = True, binary = True))
+        data = "".join(client.stream_extract("/newton/prep", binary = True))
+        # Quick check using struct
+        unpacker = struct.Struct("<qffffffff")
+        out = []
+        for i in range(14400):
+            out.append(unpacker.unpack_from(data, i * unpacker.size))
+        eq_(out[0], (1332511200000000, 266568.0, 224029.0, 5161.39990234375,
+                     2525.169921875, 8350.83984375, 3724.699951171875,
+                     1355.3399658203125, 2039.0))
+
+        # Just get some coverage
+        with assert_raises(ClientError) as e:
+            client.http.post("/stream/remove", { "path": "/none" })
+
         client.close()
 
     def test_client_06_generators(self):
@@ -311,11 +352,7 @@ class TestClient(object):
 
         # Trigger a curl error in generator
         with assert_raises(ServerError) as e:
-            client.http.get_gen("http://nosuchurl/").next()
-
-        # Trigger a curl error in generator
-        with assert_raises(ServerError) as e:
-            client.http.get_gen("http://nosuchurl/").next()
+            client.http.get_gen("http://nosuchurl.example.com./").next()
 
         # Check 404 for missing streams
         for function in [ client.stream_intervals, client.stream_extract ]:
@@ -355,16 +392,38 @@ class TestClient(object):
                                  headers())
 
         # Extract
-        x = http.get("stream/extract",
-                            { "path": "/newton/prep",
-                              "start": "123",
-                              "end": "124" })
+        x = http.get("stream/extract", { "path": "/newton/prep",
+                                         "start": "123", "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 headers():
             raise AssertionError("/stream/extract is not text/plain:\n" +
                                  headers())
 
+        x = http.get("stream/extract", { "path": "/newton/prep",
+                                         "start": "123", "end": "124",
+                                         "binary": "1" })
+        if "transfer-encoding: chunked" not in headers():
+            warnings.warn("Non-chunked HTTP response for /stream/extract")
+        if "content-type: application/octet-stream" not in headers():
+            raise AssertionError("/stream/extract is not binary:\n" +
+                                 headers())
+
+        # Make sure a binary of "0" is really off
+        x = http.get("stream/extract", { "path": "/newton/prep",
+                                         "start": "123", "end": "124",
+                                         "binary": "0" })
+        if "content-type: application/octet-stream" in headers():
+                    raise AssertionError("/stream/extract is not text:\n" +
+                                         headers())
+
+        # Invalid parameters
+        with assert_raises(ClientError) as e:
+            x = http.get("stream/extract", { "path": "/newton/prep",
+                                             "start": "123", "end": "124",
+                                             "binary": "asdfasfd" })
+        in_("can't parse parameter", str(e.exception))
+
         client.close()
 
     def test_client_08_unicode(self):
@@ -375,6 +434,7 @@ class TestClient(object):
 
             # Delete streams that exist
             for stream in client.stream_list():
+                client.stream_remove(stream[0])
                 client.stream_destroy(stream[0])
 
             # Database is empty
@@ -440,72 +500,80 @@ class TestClient(object):
             # override _max_data to trigger frequent server updates
             ctx._max_data = 15
 
-            ctx.insert("100 1\n")
+            ctx.insert("1000 1\n")
 
-            ctx.insert("101 ")
-            ctx.insert("1\n102 1")
+            ctx.insert("1010 ")
+            ctx.insert("1\n1020 1")
             ctx.insert("")
-            ctx.insert("\n103 1\n")
+            ctx.insert("\n1030 1\n")
 
-            ctx.insert("104 1\n")
+            ctx.insert("1040 1\n")
             ctx.insert("# hello\n")
             ctx.insert("   # hello\n")
-            ctx.insert("  105 1\n")
+            ctx.insert("  1050 1\n")
             ctx.finalize()
 
-            ctx.insert("107 1\n")
-            ctx.update_end(108)
+            ctx.insert("1070 1\n")
+            ctx.update_end(1080)
             ctx.finalize()
-            ctx.update_start(109)
-            ctx.insert("110 1\n")
-            ctx.insert("111 1\n")
-            ctx.insert("112 1\n")
-            ctx.insert("113 1\n")
-            ctx.insert("114 1\n")
-            ctx.update_end(116)
-            ctx.insert("115 1\n")
-            ctx.update_end(117)
-            ctx.insert("116 1\n")
-            ctx.update_end(118)
-            ctx.insert("117 1" +
+            ctx.update_start(1090)
+            ctx.insert("1100 1\n")
+            ctx.insert("1110 1\n")
+            ctx.send()
+            ctx.insert("1120 1\n")
+            ctx.insert("1130 1\n")
+            ctx.insert("1140 1\n")
+            ctx.update_end(1160)
+            ctx.insert("1150 1\n")
+            ctx.update_end(1170)
+            ctx.insert("1160 1\n")
+            ctx.update_end(1180)
+            ctx.insert("1170 1" +
                        " # this is super long" * 100 +
                        "\n")
             ctx.finalize()
             ctx.insert("# this is super long" * 100)
 
         with assert_raises(ClientError):
-            with client.stream_insert_context("/context/test", 100, 200) as ctx:
-                ctx.insert("118 1\n")
+            with client.stream_insert_context("/context/test",
+                                              1000, 2000) as ctx:
+                ctx.insert("1180 1\n")
 
         with assert_raises(ClientError):
-            with client.stream_insert_context("/context/test", 200, 300) as ctx:
-                ctx.insert("118 1\n")
+            with client.stream_insert_context("/context/test",
+                                              2000, 3000) as ctx:
+                ctx.insert("1180 1\n")
 
         with assert_raises(ClientError):
             with client.stream_insert_context("/context/test") as ctx:
                 ctx.insert("bogus data\n")
 
-        with client.stream_insert_context("/context/test", 200, 300) as ctx:
+        with client.stream_insert_context("/context/test", 2000, 3000) as ctx:
             # make sure our override wasn't permanent
             ne_(ctx._max_data, 15)
-            ctx.insert("225 1\n")
+            ctx.insert("2250 1\n")
             ctx.finalize()
 
         with assert_raises(ClientError):
-            with client.stream_insert_context("/context/test", 300, 400) as ctx:
-                ctx.insert("301 1\n")
-                ctx.insert("302 2\n")
-                ctx.insert("303 3\n")
-                ctx.insert("304 4\n")
-                ctx.insert("304 4\n") # non-monotonic after a few lines
+            with client.stream_insert_context("/context/test",
+                                              3000, 4000) as ctx:
+                ctx.insert("3010 1\n")
+                ctx.insert("3020 2\n")
+                ctx.insert("3030 3\n")
+                ctx.insert("3040 4\n")
+                ctx.insert("3040 4\n") # non-monotonic after a few lines
                 ctx.finalize()
 
         eq_(list(client.stream_intervals("/context/test")),
-            [ [ 100, 106 ],
-              [ 107, 108 ],
-              [ 109, 118 ],
-              [ 200, 300 ] ])
+            [ [ 1000, 1051 ],
+              [ 1070, 1080 ],
+              [ 1090, 1180 ],
+              [ 2000, 3000 ] ])
 
+        # destroy stream (try without removing data first)
+        with assert_raises(ClientError):
+            client.stream_destroy("/context/test")
+        client.stream_remove("/context/test")
         client.stream_destroy("/context/test")
         client.close()
 
@@ -559,8 +627,12 @@ class TestClient(object):
         with client.stream_insert_context("/empty/test", end = 950):
             pass
 
+        # Equal start and end is OK as long as there's no data
+        with client.stream_insert_context("/empty/test", start=9, end=9):
+            pass
+
         # Try various things that might cause problems
-        with client.stream_insert_context("/empty/test", 1000, 1050):
+        with client.stream_insert_context("/empty/test", 1000, 1050) as ctx:
             ctx.finalize() # inserts [1000, 1050]
             ctx.finalize() # nothing
             ctx.finalize() # nothing
@@ -600,6 +672,7 @@ class TestClient(object):
                      ])
 
         # Clean up
+        client.stream_remove("/empty/test")
         client.stream_destroy("/empty/test")
         client.close()
 
@@ -613,7 +686,7 @@ class TestClient(object):
                     poolmanager = c.http._last_response.connection.poolmanager
                     pool = poolmanager.pools[('http','localhost',32180)]
                     return (pool.num_connections, pool.num_requests)
-                except:
+                except Exception:
                     raise SkipTest("can't get connection info")
 
             # First request makes a connection
@@ -635,8 +708,9 @@ class TestClient(object):
             eq_(connections(), (1, 5))
 
             # Clean up
+            c.stream_remove("/persist/test")
             c.stream_destroy("/persist/test")
-            eq_(connections(), (1, 6))
+            eq_(connections(), (1, 7))
 
     def test_client_13_timestamp_rounding(self):
         # Test potentially bad timestamps (due to floating point
@@ -661,5 +735,6 @@ class TestClient(object):
                 # Server will round this and give an error on finalize()
                 ctx.insert("299999999.99 1\n")
 
+        client.stream_remove("/rounding/test")
         client.stream_destroy("/rounding/test")
         client.close()

tests/test_cmdline.py

@@ -21,12 +21,13 @@ from testutil.helpers import *
 
 testdb = "tests/cmdline-testdb"
 
-def server_start(max_results = None, bulkdata_args = {}):
+def server_start(max_results = None, max_removals = None, bulkdata_args = {}):
     global test_server, test_db
     # Start web app on a custom port
     test_db = nilmdb.utils.serializer_proxy(nilmdb.server.NilmDB)(
         testdb,
         max_results = max_results,
+        max_removals = max_removals,
         bulkdata_args = bulkdata_args)
     test_server = nilmdb.server.Server(test_db, host = "127.0.0.1",
                                        port = 32180, stoppable = False,
@@ -87,7 +88,7 @@ class TestCmdline(object):
                 sys.exit(0)
             except SystemExit as e:
                 exitcode = e.code
-        captured = outfile.getvalue()
+        captured = nilmdb.utils.unicode.decode(outfile.getvalue())
         self.captured = captured
         self.exitcode = exitcode
 
@@ -233,6 +234,8 @@ class TestCmdline(object):
         eq_(parse_time("1333648800.0"), test)
         eq_(parse_time("1333648800000000"), test)
         eq_(parse_time("@1333648800000000"), test)
+        eq_(parse_time("min"), nilmdb.utils.time.min_timestamp)
+        eq_(parse_time("max"), nilmdb.utils.time.max_timestamp)
         with assert_raises(ValueError):
             parse_time("@hashtag12345")
 
@@ -242,8 +245,10 @@ class TestCmdline(object):
         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")
+        self.contain("Server disk space used by NilmDB")
+        self.contain("Server disk space used by other")
+        self.contain("Server disk space reserved")
+        self.contain("Server disk space free")
 
     def test_04_createlist(self):
         # Basic stream tests, like those in test_client.
@@ -297,38 +302,19 @@ class TestCmdline(object):
 
         # Verify we got those 3 streams and they're returned in
         # alphabetical order.
-        self.ok("list")
+        self.ok("list -l")
         self.match("/newton/prep float32_8\n"
                    "/newton/raw uint16_6\n"
                    "/newton/zzz/rawnotch uint16_9\n")
 
         # Match just one type or one path.  Also check
         # that --path is optional
-        self.ok("list --path /newton/raw")
-        self.match("/newton/raw uint16_6\n")
-
-        self.ok("list /newton/raw")
-        self.match("/newton/raw uint16_6\n")
-
-        self.fail("list -p /newton/raw /newton/raw")
-        self.contain("too many paths")
-
-        self.ok("list --layout uint16_6")
+        self.ok("list --layout /newton/raw")
         self.match("/newton/raw uint16_6\n")
 
         # Wildcard matches
-        self.ok("list --layout uint16*")
-        self.match("/newton/raw uint16_6\n"
-                   "/newton/zzz/rawnotch uint16_9\n")
-
-        self.ok("list --path *zzz* --layout uint16*")
-        self.match("/newton/zzz/rawnotch uint16_9\n")
-
-        self.ok("list *zzz* --layout uint16*")
-        self.match("/newton/zzz/rawnotch uint16_9\n")
-
-        self.ok("list --path *zzz* --layout float32*")
-        self.match("")
+        self.ok("list *zzz*")
+        self.match("/newton/zzz/rawnotch\n")
 
         # reversed range
         self.fail("list /newton/prep --start 2020-01-01 --end 2000-01-01")
@@ -366,6 +352,8 @@ class TestCmdline(object):
         self.contain("No stream at path")
         self.fail("metadata /newton/nosuchstream --set foo=bar")
         self.contain("No stream at path")
+        self.fail("metadata /newton/nosuchstream --delete")
+        self.contain("No stream at path")
 
         self.ok("metadata /newton/prep")
         self.match("description=The Data\nv_scale=1.234\n")
@@ -391,6 +379,19 @@ class TestCmdline(object):
         self.fail("metadata /newton/nosuchpath")
         self.contain("No stream at path /newton/nosuchpath")
 
+        self.ok("metadata /newton/prep --delete")
+        self.ok("metadata /newton/prep --get")
+        self.match("")
+        self.ok("metadata /newton/prep --set "
+                "'description=The Data' "
+                "v_scale=1.234")
+        self.ok("metadata /newton/prep --delete v_scale")
+        self.ok("metadata /newton/prep --get")
+        self.match("description=The Data\n")
+        self.ok("metadata /newton/prep --set description=")
+        self.ok("metadata /newton/prep --get")
+        self.match("")
+
     def test_06_insert(self):
         self.ok("insert --help")
 
@@ -474,33 +475,40 @@ class TestCmdline(object):
         # bad start time
         self.fail("insert -t -r 120 --start 'whatever' /newton/prep /dev/null")
 
+        # Test negative times
+        self.ok("insert --start @-10000000000 --end @1000000001 /newton/prep"
+                " tests/data/timestamped")
+        self.ok("extract -c /newton/prep --start min --end @1000000001")
+        self.match("8\n")
+        self.ok("remove /newton/prep --start min --end @1000000001")
+
     def test_07_detail_extended(self):
         # Just count the number of lines, it's probably fine
         self.ok("list --detail")
         lines_(self.captured, 8)
 
-        self.ok("list --detail --path *prep")
+        self.ok("list --detail *prep")
         lines_(self.captured, 4)
 
-        self.ok("list --detail --path *prep --start='23 Mar 2012 10:02'")
+        self.ok("list --detail *prep --start='23 Mar 2012 10:02'")
         lines_(self.captured, 3)
 
-        self.ok("list --detail --path *prep --start='23 Mar 2012 10:05'")
+        self.ok("list --detail *prep --start='23 Mar 2012 10:05'")
         lines_(self.captured, 2)
 
-        self.ok("list --detail --path *prep --start='23 Mar 2012 10:05:15'")
+        self.ok("list --detail *prep --start='23 Mar 2012 10:05:15'")
         lines_(self.captured, 2)
         self.contain("10:05:15.000")
 
-        self.ok("list --detail --path *prep --start='23 Mar 2012 10:05:15.50'")
+        self.ok("list --detail *prep --start='23 Mar 2012 10:05:15.50'")
         lines_(self.captured, 2)
         self.contain("10:05:15.500")
 
-        self.ok("list --detail --path *prep --start='23 Mar 2012 19:05:15.50'")
+        self.ok("list --detail *prep --start='23 Mar 2012 19:05:15.50'")
         lines_(self.captured, 2)
         self.contain("no intervals")
 
-        self.ok("list --detail --path *prep --start='23 Mar 2012 10:05:15.50'"
+        self.ok("list --detail *prep --start='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")
@@ -509,15 +517,15 @@ class TestCmdline(object):
         lines_(self.captured, 8)
 
         # Verify the "raw timestamp" output
-        self.ok("list --detail --path *prep --timestamp-raw "
+        self.ok("list --detail *prep --timestamp-raw "
                 "--start='23 Mar 2012 10:05:15.50'")
         lines_(self.captured, 2)
         self.contain("[ 1332497115500000 -> 1332497160000000 ]")
 
         # bad time
-        self.fail("list --detail --path *prep -T --start='9332497115.612'")
+        self.fail("list --detail *prep -T --start='9332497115.612'")
         # good time
-        self.ok("list --detail --path *prep -T --start='1332497115.612'")
+        self.ok("list --detail *prep -T --start='1332497115.612'")
         lines_(self.captured, 2)
         self.contain("[ 1332497115612000 -> 1332497160000000 ]")
 
@@ -593,13 +601,28 @@ class TestCmdline(object):
         test(6, "10:00:30", "10:00:31", extra="-b")
         test(7, "10:00:30", "10:00:30.999", extra="-a -T")
         test(7, "10:00:30", "10:00:30.999", extra="-a --timestamp-raw")
+        test(8, "10:01:59.9", "10:02:00.1", extra="--markup")
+        test(8, "10:01:59.9", "10:02:00.1", extra="-m")
 
         # all data put in by tests
-        self.ok("extract -a /newton/prep --start 2000-01-01 --end 2020-01-01")
+        self.ok("extract -a /newton/prep --start min --end max")
         lines_(self.captured, 43204)
         self.ok("extract -c /newton/prep --start 2000-01-01 --end 2020-01-01")
         self.match("43200\n")
 
+        # test binary mode
+        self.fail("extract -c -B /newton/prep -s min -e max")
+        self.contain("binary cannot be combined")
+        self.fail("extract -m -B /newton/prep -s min -e max")
+        self.contain("binary cannot be combined")
+        self.ok("extract -B /newton/prep -s min -e max")
+        eq_(len(self.captured), 43200 * (8 + 8*4))
+
+        # markup for 3 intervals, plus extra markup lines whenever we had
+        # a "restart" from the nilmdb.stream_extract function
+        self.ok("extract -m /newton/prep --start 2000-01-01 --end 2020-01-01")
+        lines_(self.captured, 43210)
+
     def test_09_truncated(self):
         # Test truncated responses by overriding the nilmdb max_results
         server_stop()
@@ -614,7 +637,7 @@ class TestCmdline(object):
 
         # Try nonexistent stream
         self.fail("remove /no/such/foo --start 2000-01-01 --end 2020-01-01")
-        self.contain("No stream at path")
+        self.contain("no stream matched path")
 
         # empty or backward ranges return errors
         self.fail("remove /newton/prep --start 2020-01-01 --end 2000-01-01")
@@ -642,9 +665,14 @@ class TestCmdline(object):
                 "--start '23 Mar 2022 20:00:30' " +
                 "--end '23 Mar 2022 20:00:31'")
         self.match("0\n")
+        self.ok("remove -c /newton/prep /newton/pre* " +
+                "--start '23 Mar 2022 20:00:30' " +
+                "--end '23 Mar 2022 20:00:31'")
+        self.match("Removing from /newton/prep\n0\n" +
+                   "Removing from /newton/prep\n0\n")
 
         # Make sure we have the data we expect
-        self.ok("list --detail /newton/prep")
+        self.ok("list -l --detail /newton/prep")
         self.match("/newton/prep float32_8\n" +
                    "  [ Fri, 23 Mar 2012 10:00:00.000000 +0000"
                    " -> Fri, 23 Mar 2012 10:01:59.991668 +0000 ]\n"
@@ -679,7 +707,7 @@ class TestCmdline(object):
         self.match("24000\n")
 
         # See the missing chunks in list output
-        self.ok("list --detail /newton/prep")
+        self.ok("list --layout --detail /newton/prep")
         self.match("/newton/prep float32_8\n" +
                    "  [ Fri, 23 Mar 2012 10:00:00.000000 +0000"
                    " -> Fri, 23 Mar 2012 10:00:05.000000 +0000 ]\n"
@@ -693,17 +721,15 @@ class TestCmdline(object):
         # Remove all data, verify it's missing
         self.ok("remove /newton/prep --start 2000-01-01 --end 2020-01-01")
         self.match("")  # no count requested this time
-        self.ok("list --detail /newton/prep")
+        self.ok("list -l --detail /newton/prep")
         self.match("/newton/prep float32_8\n" +
                    "  (no intervals)\n")
 
         # Reinsert some data, to verify that no overlaps with deleted
         # data are reported
-        os.environ['TZ'] = "UTC"
-        self.ok("insert --timestamp -f --rate 120 /newton/prep "
-                "tests/data/prep-20120323T1000")
-        self.ok("insert -t --filename --rate 120 /newton/prep "
-                "tests/data/prep-20120323T1002")
+        for minute in ["0", "2"]:
+            self.ok("insert --timestamp -f --rate 120 /newton/prep"
+                    " tests/data/prep-20120323T100" + minute)
 
     def test_11_destroy(self):
         # Delete records
@@ -713,13 +739,16 @@ class TestCmdline(object):
         self.contain("too few arguments")
 
         self.fail("destroy /no/such/stream")
-        self.contain("No stream at path")
+        self.contain("no stream matched path")
+
+        self.fail("destroy -R /no/such/stream")
+        self.contain("no stream matched path")
 
         self.fail("destroy asdfasdf")
-        self.contain("No stream at path")
+        self.contain("no stream matched path")
 
         # From previous tests, we have:
-        self.ok("list")
+        self.ok("list -l")
         self.match("/newton/prep float32_8\n"
                    "/newton/raw uint16_6\n"
                    "/newton/zzz/rawnotch uint16_9\n")
@@ -728,19 +757,26 @@ class TestCmdline(object):
         self.ok("list --detail")
         lines_(self.captured, 7)
 
-        # Delete some
-        self.ok("destroy /newton/prep")
-        self.ok("list")
+        # Fail to destroy because intervals still present
+        self.fail("destroy /newton/prep")
+        self.contain("all intervals must be removed")
+        self.ok("list --detail")
+        lines_(self.captured, 7)
+
+        # Destroy for real
+        self.ok("destroy -R /n*/prep")
+        self.ok("list -l")
         self.match("/newton/raw uint16_6\n"
                    "/newton/zzz/rawnotch uint16_9\n")
 
         self.ok("destroy /newton/zzz/rawnotch")
-        self.ok("list")
+        self.ok("list -l")
         self.match("/newton/raw uint16_6\n")
 
         self.ok("destroy /newton/raw")
         self.ok("create /newton/raw uint16_6")
-        self.ok("destroy /newton/raw")
+        # Specify --remove with no data
+        self.ok("destroy --remove /newton/raw")
         self.ok("list")
         self.match("")
 
@@ -753,18 +789,17 @@ class TestCmdline(object):
             self.ok("list")
             self.contain(path)
             # Make sure it was created empty
-            self.ok("list --detail --path " + path)
+            self.ok("list --detail " + path)
             self.contain("(no intervals)")
 
     def test_12_unicode(self):
         # Unicode paths.
         self.ok("destroy /newton/asdf/qwer")
-        self.ok("destroy /newton/prep")
-        self.ok("destroy /newton/raw")
+        self.ok("destroy /newton/prep /newton/raw")
         self.ok("destroy /newton/zzz")
 
         self.ok(u"create /düsseldorf/raw uint16_6")
-        self.ok("list --detail")
+        self.ok("list -l --detail")
         self.contain(u"/düsseldorf/raw uint16_6")
         self.contain("(no intervals)")
 
@@ -815,7 +850,7 @@ class TestCmdline(object):
 
         # Now recreate the data one more time and make sure there are
         # fewer files.
-        self.ok("destroy /newton/prep")
+        self.ok("destroy --remove /newton/prep")
         self.fail("destroy /newton/prep") # already destroyed
         self.ok("create /newton/prep float32_8")
         os.environ['TZ'] = "UTC"
@@ -826,14 +861,16 @@ class TestCmdline(object):
         for (dirpath, dirnames, filenames) in os.walk(testdb):
             nfiles += len(filenames)
         lt_(nfiles, 50)
-        self.ok("destroy /newton/prep") # destroy again
+        self.ok("destroy -R /newton/prep") # destroy again
 
     def test_14_remove_files(self):
         # Test BulkData's ability to remove when data is split into
         # multiple files.  Should be a fairly comprehensive test of
         # remove functionality.
+        # Also limit max_removals, to cover more functionality.
         server_stop()
-        server_start(bulkdata_args = { "file_size" : 920, # 23 rows per file
+        server_start(max_removals = 4321,
+                     bulkdata_args = { "file_size" : 920, # 23 rows per file
                                        "files_per_dir" : 3 })
 
         # Insert data.  Just for fun, insert out of order
@@ -848,7 +885,7 @@ class TestCmdline(object):
         du_before = nilmdb.utils.diskusage.du(testdb)
 
         # Make sure we have the data we expect
-        self.ok("list --detail")
+        self.ok("list -l --detail")
         self.match("/newton/prep float32_8\n" +
                    "  [ Fri, 23 Mar 2012 10:00:00.000000 +0000"
                    " -> Fri, 23 Mar 2012 10:01:59.991668 +0000 ]\n"
@@ -884,7 +921,7 @@ class TestCmdline(object):
         self.match("3600\n")
 
         # See the missing chunks in list output
-        self.ok("list --detail")
+        self.ok("list -l --detail")
         self.match("/newton/prep float32_8\n" +
                    "  [ Fri, 23 Mar 2012 10:00:00.000000 +0000"
                    " -> Fri, 23 Mar 2012 10:00:05.000000 +0000 ]\n"
@@ -974,8 +1011,8 @@ class TestCmdline(object):
         self.match("[ Thu, 01 Jan 2004 00:00:00.000000 +0000 -"
                    "> Sat, 01 Jan 2005 00:00:00.000000 +0000 ]\n")
 
-        self.ok("destroy /diff/1")
-        self.ok("destroy /diff/2")
+        self.ok("destroy -R /diff/1")
+        self.ok("destroy -R /diff/2")
 
     def test_16_rename(self):
         # Test renaming.  Force file size smaller so we get more files
@@ -1008,7 +1045,7 @@ class TestCmdline(object):
             else:
                 raise AssertionError("data not found at " + seek)
             # Verify "list" output
-            self.ok("list")
+            self.ok("list -l")
             self.match("/" + "/".join(components) + " float32_8\n")
 
         # Lots of renames
@@ -1018,10 +1055,12 @@ class TestCmdline(object):
         self.contain("old and new paths are the same")
         check_path("newton", "prep")
         self.fail("rename /newton/prep /newton")
-        self.contain("subdirs of this path already exist")
+        self.contain("path must contain at least one folder")
         self.fail("rename /newton/prep /newton/prep/")
         self.contain("invalid path")
-        self.ok("rename /newton/prep /newton/foo")
+        self.ok("rename /newton/prep /newton/foo/1")
+        check_path("newton", "foo", "1")
+        self.ok("rename /newton/foo/1 /newton/foo")
         check_path("newton", "foo")
         self.ok("rename /newton/foo /totally/different/thing")
         check_path("totally", "different", "thing")
@@ -1039,7 +1078,7 @@ class TestCmdline(object):
         self.fail("rename /foo/bar /xxx/yyy/zzz/www")
         self.contain("path is subdir of existing node")
         self.ok("rename /foo/bar /xxx/yyy/mmm")
-        self.ok("destroy /xxx/yyy/zzz")
+        self.ok("destroy -R /xxx/yyy/zzz")
         check_path("xxx", "yyy", "mmm")
 
         # Extract it at the final path
@@ -1047,7 +1086,7 @@ class TestCmdline(object):
                 "--end '2012-03-23 10:04:01'")
         eq_(self.captured, extract_before)
 
-        self.ok("destroy /xxx/yyy/mmm")
+        self.ok("destroy -R /xxx/yyy/mmm")
 
         # Make sure temporary rename dirs weren't left around
         for (dirpath, dirnames, filenames) in os.walk(testdb):

tests/test_interval.py

@@ -59,6 +59,14 @@ class TestInterval:
         self.test_interval_intersect()
         Interval = NilmdbInterval
 
+        # Other helpers in nilmdb.utils.interval
+        i = [ UtilsInterval(1,2), UtilsInterval(2,3), UtilsInterval(4,5) ]
+        eq_(list(nilmdb.utils.interval.optimize(i)),
+            [ UtilsInterval(1,3), UtilsInterval(4,5) ])
+        eq_(UtilsInterval(1234567890123456, 1234567890654321).human_string(),
+            "[ Fri, 13 Feb 2009 18:31:30.123456 -0500 -> " +
+            "Fri, 13 Feb 2009 18:31:30.654321 -0500 ]")
+
     def test_interval(self):
         # Test Interval class
         os.environ['TZ'] = "America/New_York"
@@ -226,13 +234,16 @@ class TestInterval:
             x = makeset("[--)") & 1234
 
         def do_test(a, b, c, d):
-            # a & b == c
+            # a & b == c (using nilmdb.server.interval)
             ab = IntervalSet()
             for x in b:
                 for i in (a & x):
                     ab += i
             eq_(ab,c)
 
+            # a & b == c (using nilmdb.utils.interval)
+            eq_(IntervalSet(nilmdb.utils.interval.intersection(a,b)), c)
+
             # a \ b == d
             eq_(IntervalSet(nilmdb.utils.interval.set_difference(a,b)), d)
 
@@ -302,6 +313,17 @@ class TestInterval:
         eq_(nilmdb.utils.interval.set_difference(
             a.intersection(list(c)[0]), b.intersection(list(c)[0])), d)
 
+        # Fill out test coverage for non-subsets
+        def diff2(a,b, subset):
+            return nilmdb.utils.interval._interval_math_helper(
+                a, b, (lambda a, b: b and not a), subset=subset)
+        with assert_raises(nilmdb.utils.interval.IntervalError):
+            list(diff2(a,b,True))
+        list(diff2(a,b,False))
+
+        # Empty second set
+        eq_(nilmdb.utils.interval.set_difference(a, IntervalSet()), a)
+
         # Empty second set
         eq_(nilmdb.utils.interval.set_difference(a, IntervalSet()), a)
 
@@ -385,7 +407,6 @@ class TestIntervalSpeed:
     def test_interval_speed(self):
         import yappi
         import time
-        import testutil.aplotter as aplotter
         import random
         import math
 
@@ -406,6 +427,5 @@ class TestIntervalSpeed:
                    speed/j,
                    speed / (j*math.log(j))) # should be constant
             speeds[j] = speed
-        aplotter.plot(speeds.keys(), speeds.values(), plot_slope=True)
         yappi.stop()
         yappi.print_stats(sort_type=yappi.SORTTYPE_TTOT, limit=10)

tests/test_iteratorizer.py

@@ -1,61 +0,0 @@
-import nilmdb
-from nilmdb.utils.printf import *
-
-import nose
-from nose.tools import *
-from nose.tools import assert_raises
-import threading
-import time
-
-from testutil.helpers import *
-
-def func_with_callback(a, b, callback):
-    callback(a)
-    callback(b)
-    callback(a+b)
-    return "return value"
-
-class TestIteratorizer(object):
-    def test(self):
-
-        # First try it with a normal callback
-        self.result = ""
-        def cb(x):
-            self.result += str(x)
-        func_with_callback(1, 2, cb)
-        eq_(self.result, "123")
-
-        # Now make it an iterator
-        result = ""
-        f = lambda x: func_with_callback(1, 2, x)
-        with nilmdb.utils.Iteratorizer(f) as it:
-            for i in it:
-                result += str(i)
-        eq_(result, "123")
-        eq_(it.retval, "return value")
-
-        # Make sure things work when an exception occurs
-        result = ""
-        with nilmdb.utils.Iteratorizer(
-            lambda x: func_with_callback(1, "a", x)) as it:
-            with assert_raises(TypeError) as e:
-                for i in it:
-                    result += str(i)
-        eq_(result, "1a")
-
-        # Now try to trigger the case where we stop iterating
-        # mid-generator, and expect the iteratorizer to clean up after
-        # itself.  This doesn't have a particular result in the test,
-        # but gains coverage.
-        def foo():
-            with nilmdb.utils.Iteratorizer(f) as it:
-                it.next()
-        foo()
-        eq_(it.retval, None)
-
-        # Do the same thing when the curl hack is applied
-        def foo():
-            with nilmdb.utils.Iteratorizer(f, curl_hack = True) as it:
-                it.next()
-        foo()
-        eq_(it.retval, None)

tests/test_nilmdb.py

@@ -28,9 +28,6 @@ class Test00Nilmdb(object):  # named 00 so it runs first
     def test_NilmDB(self):
         recursive_unlink(testdb)
 
-        with assert_raises(IOError):
-            nilmdb.server.NilmDB("/nonexistant-db/foo")
-
         db = nilmdb.server.NilmDB(testdb)
         db.close()
         db = nilmdb.server.NilmDB(testdb)
@@ -93,13 +90,16 @@ 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
+        # fill in some misc. test coverage
         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)
 
+        with assert_raises(nilmdb.server.NilmDBError):
+            db.stream_extract("/newton/prep", count = True, binary = True)
+
         db.close()
 
 class TestBlockingServer(object):
@@ -157,11 +157,14 @@ class TestServer(object):
 
     def test_server(self):
         # Make sure we can't force an exit, and test other 404 errors
-        for url in [ "/exit", "/", "/favicon.ico" ]:
+        for url in [ "/exit", "/favicon.ico" ]:
             with assert_raises(HTTPError) as e:
                 geturl(url)
             eq_(e.exception.code, 404)
 
+        # Root page
+        in_("This is NilmDB", geturl("/"))
+
         # Check version
         eq_(distutils.version.LooseVersion(getjson("/version")),
             distutils.version.LooseVersion(nilmdb.__version__))

tests/test_numpyclient.py

@@ -0,0 +1,373 @@
+# -*- coding: utf-8 -*-
+
+import nilmdb.server
+import nilmdb.client
+import nilmdb.client.numpyclient
+
+from nilmdb.utils.printf import *
+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
+import distutils.version
+
+from testutil.helpers import *
+
+import numpy as np
+
+testdb = "tests/numpyclient-testdb"
+testurl = "http://localhost:32180/"
+
+def setup_module():
+    global test_server, test_db
+    # Clear out DB
+    recursive_unlink(testdb)
+
+    # Start web app on a custom port
+    test_db = nilmdb.utils.serializer_proxy(nilmdb.server.NilmDB)(
+        testdb, bulkdata_args = { "file_size" : 16384,
+                                  "files_per_dir" : 3 } )
+
+    test_server = nilmdb.server.Server(test_db, host = "127.0.0.1",
+                                       port = 32180, stoppable = False,
+                                       fast_shutdown = True,
+                                       force_traceback = True)
+    test_server.start(blocking = False)
+
+def teardown_module():
+    global test_server, test_db
+    # Close web app
+    test_server.stop()
+    test_db.close()
+
+class TestNumpyClient(object):
+
+    def test_numpyclient_01_basic(self):
+        # Test basic connection
+        client = nilmdb.client.numpyclient.NumpyClient(url = testurl)
+        version = client.version()
+        eq_(distutils.version.LooseVersion(version),
+            distutils.version.LooseVersion(test_server.version))
+
+        # Verify subclassing
+        assert(isinstance(client, nilmdb.client.Client))
+
+        # Layouts
+        for layout in "int8_t", "something_8", "integer_1":
+            with assert_raises(ValueError):
+                for x in client.stream_extract_numpy("/foo", layout=layout):
+                    pass
+        for layout in "int8_1", "uint8_30", "int16_20", "float64_100":
+            with assert_raises(ClientError) as e:
+                for x in client.stream_extract_numpy("/foo", layout=layout):
+                    pass
+            in_("No such stream", str(e.exception))
+
+        with assert_raises(ClientError) as e:
+            for x in client.stream_extract_numpy("/foo"):
+                pass
+        in_("can't get layout for path", str(e.exception))
+
+        client.close()
+
+    def test_numpyclient_02_extract(self):
+        client = nilmdb.client.numpyclient.NumpyClient(url = testurl)
+
+        # Insert some data as text
+        client.stream_create("/newton/prep", "float32_8")
+        testfile = "tests/data/prep-20120323T1000"
+        start = nilmdb.utils.time.parse_time("20120323T1000")
+        rate = 120
+        data = timestamper.TimestamperRate(testfile, start, rate)
+        result = client.stream_insert("/newton/prep", data,
+                                      start, start + 119999777)
+
+        # Extract Numpy arrays
+        array = None
+        pieces = 0
+        for chunk in client.stream_extract_numpy("/newton/prep", maxrows=1000):
+            pieces += 1
+            if array is not None:
+                array = np.vstack((array, chunk))
+            else:
+                array = chunk
+        eq_(array.shape, (14400, 9))
+        eq_(pieces, 15)
+
+        # Try structured
+        s = list(client.stream_extract_numpy("/newton/prep", structured = True))
+        assert(np.array_equal(np.c_[s[0]['timestamp'], s[0]['data']], array))
+
+        # Compare.  Will be close but not exact because the conversion
+        # to and from ASCII was lossy.
+        data = timestamper.TimestamperRate(testfile, start, rate)
+        actual = np.fromstring(" ".join(data), sep=' ').reshape(14400, 9)
+        assert(np.allclose(array, actual))
+
+        client.close()
+
+    def test_numpyclient_03_insert(self):
+        client = nilmdb.client.numpyclient.NumpyClient(url = testurl)
+
+        # Limit _max_data just to get better coverage
+        old_max_data = nilmdb.client.numpyclient.StreamInserterNumpy._max_data
+        nilmdb.client.numpyclient.StreamInserterNumpy._max_data = 100000
+
+        client.stream_create("/test/1", "uint16_1")
+        client.stream_insert_numpy("/test/1",
+                                   np.array([[0, 1],
+                                             [1, 2],
+                                             [2, 3],
+                                             [3, 4]]))
+
+        # Wrong number of dimensions
+        with assert_raises(ValueError) as e:
+            client.stream_insert_numpy("/test/1",
+                                       np.array([[[0, 1],
+                                                  [1, 2]],
+                                                 [[3, 4],
+                                                  [4, 5]]]))
+        in_("wrong number of dimensions", str(e.exception))
+
+        # Wrong number of fields
+        with assert_raises(ValueError) as e:
+            client.stream_insert_numpy("/test/1",
+                                       np.array([[0, 1, 2],
+                                                 [1, 2, 3],
+                                                 [3, 4, 5],
+                                                 [4, 5, 6]]))
+        in_("wrong number of fields", str(e.exception))
+
+        # Unstructured
+        client.stream_create("/test/2", "float32_8")
+        client.stream_insert_numpy(
+            "/test/2",
+            client.stream_extract_numpy(
+                "/newton/prep", structured = False, maxrows = 1000))
+
+        # Structured, and specifying layout
+        client.stream_create("/test/3", "float32_8")
+        client.stream_insert_numpy(
+            path = "/test/3", layout = "float32_8",
+            data = client.stream_extract_numpy(
+                "/newton/prep", structured = True, maxrows = 1000))
+
+        # Structured, specifying wrong layout
+        client.stream_create("/test/4", "float32_8")
+        with assert_raises(ValueError) as e:
+            client.stream_insert_numpy(
+                "/test/4", layout = "uint16_1",
+                data = client.stream_extract_numpy(
+                    "/newton/prep", structured = True, maxrows = 1000))
+        in_("wrong dtype", str(e.exception))
+
+        # Unstructured, and specifying wrong layout
+        client.stream_create("/test/5", "float32_8")
+        with assert_raises(ClientError) as e:
+            client.stream_insert_numpy(
+                "/test/5", layout = "uint16_8",
+                data = client.stream_extract_numpy(
+                    "/newton/prep", structured = False, maxrows = 1000))
+        # timestamps will be screwy here, because data will be parsed wrong
+        in_("error parsing input data", str(e.exception))
+
+        # Make sure the /newton/prep copies are identical
+        a = np.vstack(client.stream_extract_numpy("/newton/prep"))
+        b = np.vstack(client.stream_extract_numpy("/test/2"))
+        c = np.vstack(client.stream_extract_numpy("/test/3"))
+        assert(np.array_equal(a,b))
+        assert(np.array_equal(a,c))
+
+        # Make sure none of the files are greater than 16384 bytes as
+        # we configured with the bulkdata_args above.
+        datapath = os.path.join(testdb, "data")
+        for (dirpath, dirnames, filenames) in os.walk(datapath):
+            for f in filenames:
+                fn = os.path.join(dirpath, f)
+                size = os.path.getsize(fn)
+                if size > 16384:
+                    raise AssertionError(sprintf("%s is too big: %d > %d\n",
+                                                 fn, size, 16384))
+
+        nilmdb.client.numpyclient.StreamInserterNumpy._max_data = old_max_data
+        client.close()
+
+    def test_numpyclient_04_context(self):
+        # Like test_client_context, but with Numpy data
+        client = nilmdb.client.numpyclient.NumpyClient(testurl)
+
+        client.stream_create("/context/test", "uint16_1")
+        with client.stream_insert_numpy_context("/context/test") as ctx:
+            # override _max_rows to trigger frequent server updates
+            ctx._max_rows = 2
+            ctx.insert([[1000, 1]])
+            ctx.insert([[1010, 1], [1020, 1], [1030, 1]])
+            ctx.insert([[1040, 1], [1050, 1]])
+            ctx.finalize()
+            ctx.insert([[1070, 1]])
+            ctx.update_end(1080)
+            ctx.finalize()
+            ctx.update_start(1090)
+            ctx.insert([[1100, 1]])
+            ctx.insert([[1110, 1]])
+            ctx.send()
+            ctx.insert([[1120, 1], [1130, 1], [1140, 1]])
+            ctx.update_end(1160)
+            ctx.insert([[1150, 1]])
+            ctx.update_end(1170)
+            ctx.insert([[1160, 1]])
+            ctx.update_end(1180)
+            ctx.insert([[1170, 123456789.0]])
+            ctx.finalize()
+            ctx.insert(np.zeros((0,2)))
+
+        with assert_raises(ClientError):
+            with client.stream_insert_numpy_context("/context/test",
+                                                    1000, 2000) as ctx:
+                ctx.insert([[1180, 1]])
+
+        with assert_raises(ClientError):
+            with client.stream_insert_numpy_context("/context/test",
+                                                    2000, 3000) as ctx:
+                ctx._max_rows = 2
+                ctx.insert([[3180, 1]])
+                ctx.insert([[3181, 1]])
+
+        with client.stream_insert_numpy_context("/context/test",
+                                                2000, 3000) as ctx:
+            # make sure our override wasn't permanent
+            ne_(ctx._max_rows, 2)
+            ctx.insert([[2250, 1]])
+            ctx.finalize()
+
+        with assert_raises(ClientError):
+            with client.stream_insert_numpy_context("/context/test",
+                                                    3000, 4000) as ctx:
+                ctx.insert([[3010, 1]])
+                ctx.insert([[3020, 2]])
+                ctx.insert([[3030, 3]])
+                ctx.insert([[3040, 4]])
+                ctx.insert([[3040, 4]]) # non-monotonic after a few lines
+                ctx.finalize()
+
+        eq_(list(client.stream_intervals("/context/test")),
+            [ [ 1000, 1051 ],
+              [ 1070, 1080 ],
+              [ 1090, 1180 ],
+              [ 2000, 3000 ] ])
+
+        client.stream_remove("/context/test")
+        client.stream_destroy("/context/test")
+        client.close()
+
+    def test_numpyclient_05_emptyintervals(self):
+        # Like test_client_emptyintervals, with insert_numpy_context
+        client = nilmdb.client.numpyclient.NumpyClient(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_numpy_context("/empty/test") as ctx:
+            ctx.update_start(100)
+            ctx.insert([[140, 1]])
+            ctx.insert([[150, 1]])
+            ctx.insert([[160, 1]])
+            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("/empty/test", "", 300, 350)
+        client.stream_insert("/empty/test", [], 400, 450)
+        with client.stream_insert_numpy_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_numpy_context("/empty/test"):
+            pass
+        client.stream_insert("/empty/test", [], start = 600)
+        with client.stream_insert_numpy_context("/empty/test", start = 700):
+            pass
+        client.stream_insert("/empty/test", [], end = 850)
+        with client.stream_insert_numpy_context("/empty/test", end = 950):
+            pass
+
+        # Equal start and end is OK as long as there's no data
+        with assert_raises(ClientError) as e:
+            with client.stream_insert_numpy_context("/empty/test",
+                                                    start=9, end=9) as ctx:
+                ctx.insert([[9, 9]])
+                ctx.finalize()
+        in_("have data to send, but invalid start/end times", str(e.exception))
+
+        with client.stream_insert_numpy_context("/empty/test",
+                                                start=9, end=9) as ctx:
+            pass
+
+        # reusing a context object is bad
+        with assert_raises(Exception) as e:
+            ctx.insert([[9, 9]])
+
+        # Try various things that might cause problems
+        with client.stream_insert_numpy_context("/empty/test",
+                                                1000, 1050) as ctx:
+            ctx.finalize() # inserts [1000, 1050]
+            ctx.finalize() # nothing
+            ctx.finalize() # nothing
+            ctx.insert([[1100, 1]])
+            ctx.finalize() # inserts [1100, 1101]
+            ctx.update_start(1199)
+            ctx.insert([[1200, 1]])
+            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.insert(np.zeros((0,2)))
+            ctx.update_end(1450)
+            ctx.finalize()
+            ctx.update_start(1500)
+            ctx.insert(np.zeros((0,2)))
+            ctx.update_end(1550)
+            ctx.finalize()
+            ctx.insert(np.zeros((0,2)))
+            ctx.insert(np.zeros((0,2)))
+            ctx.insert(np.zeros((0,2)))
+            ctx.finalize()
+
+        # Check everything
+        eq_(info(), [(1, [100, 145]),
+                     (0, [175, 200]),
+                     (0, [300, 350]),
+                     (0, [400, 450]),
+                     (0, [500, 550]),
+                     (0, [1000, 1050]),
+                     (1, [1100, 1101]),
+                     (1, [1199, 1250]),
+                     (0, [1400, 1450]),
+                     (0, [1500, 1550]),
+                     ])
+
+        # Clean up
+        client.stream_remove("/empty/test")
+        client.stream_destroy("/empty/test")
+        client.close()

tests/test_printf.py

@@ -18,7 +18,7 @@ class TestPrintf(object):
             printf("hello, world: %d", 123)
             fprintf(test2, "hello too: %d", 123)
             test3 = sprintf("hello three: %d", 123)
-        except:
+        except Exception:
             sys.stdout = old_stdout
             raise
         sys.stdout = old_stdout

tests/test_serializer.py

@@ -62,6 +62,28 @@ class Base(object):
         eq_(self.foo.val, 20)
         eq_(self.foo.init_thread, self.foo.test_thread)
 
+class ListLike(object):
+    def __init__(self):
+        self.thread = threading.current_thread().name
+        self.foo = 0
+
+    def __iter__(self):
+        eq_(threading.current_thread().name, self.thread)
+        self.foo = 0
+        return self
+
+    def __getitem__(self, key):
+        eq_(threading.current_thread().name, self.thread)
+        return key
+
+    def next(self):
+        eq_(threading.current_thread().name, self.thread)
+        if self.foo < 5:
+            self.foo += 1
+            return self.foo
+        else:
+            raise StopIteration
+
 class TestUnserialized(Base):
     def setUp(self):
         self.foo = Foo()
@@ -84,3 +106,10 @@ class TestSerializer(Base):
         sp(sp(Foo("x"))).t()
         sp(sp(Foo)("x")).t()
         sp(sp(Foo))("x").t()
+
+    def test_iter(self):
+        sp = nilmdb.utils.serializer_proxy
+        i = sp(ListLike)()
+        print iter(i)
+        eq_(list(i), [1,2,3,4,5])
+        eq_(i[3], 3)

tests/testutil/aplotter.py

@@ -1,419 +0,0 @@
-
-#-----------------------------------------------
-#aplotter.py - ascii art function plotter
-#Copyright (c) 2006, Imri Goldberg
-#All rights reserved.
-#
-#Redistribution and use in source and binary forms,
-#with or without modification, are permitted provided
-#that the following conditions are met:
-#
-#    * Redistributions of source code must retain the
-#		above copyright notice, this list of conditions
-#		and the following disclaimer.
-#    * Redistributions in binary form must reproduce the
-#		above copyright notice, this list of conditions
-#		and the following disclaimer in the documentation
-#		and/or other materials provided with the distribution.
-#    * Neither the name of the <ORGANIZATION> nor the names of
-#		its contributors may be used to endorse or promote products
-#		derived from this software without specific prior written permission.
-#
-#THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
-#AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
-#IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
-#ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
-#LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
-#DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
-#SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
-#CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
-#OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-#OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-#-----------------------------------------------
-
-import math
-
-
-EPSILON = 0.000001
-
-def transposed(mat):
-	result = []
-	for i in xrange(len(mat[0])):
-		result.append([x[i] for x in mat])
-	return result
-
-def y_reversed(mat):    
-	result = []
-	for i in range(len(mat)):
-		result.append(list(reversed(mat[i])))
-	return result
-
-def sign(x):
-	if 0<x:
-		return 1
-	if 0 == x:
-		return 0
-	return -1
-
-class Plotter(object):
-
-	class PlotData(object):
-		def __init__(self, x_size, y_size, min_x, max_x, min_y, max_y, x_mod, y_mod):
-			self.x_size = x_size
-			self.y_size = y_size
-			self.min_x = min_x
-			self.max_x = max_x
-			self.min_y = min_y
-			self.max_y = max_y
-			self.x_mod = x_mod
-			self.y_mod = y_mod
-
-			self.x_step = float(max_x - min_x)/float(self.x_size)
-			self.y_step = float(max_y - min_y)/float(self.y_size)
-			self.inv_x_step = 1/self.x_step
-			self.inv_y_step = 1/self.y_step
-
-			self.ratio = self.y_step / self.x_step
-		def __repr__(self):
-			s = "size: %s, bl: %s, tr: %s, step: %s" % ((self.x_size, self.y_size), (self.min_x, self.min_y), (self.max_x, self.max_y),
-														 (self.x_step, self.y_step))
-			return s
-	
-	def __init__(self, **kwargs):
-
-		self.x_size = kwargs.get("x_size", 80)
-		self.y_size = kwargs.get("y_size", 20)
-
-		self.will_draw_axes = kwargs.get("draw_axes", True)
-
-		self.new_line = kwargs.get("newline", "\n")
-
-		self.dot = kwargs.get("dot", "*")
-
-		self.plot_slope = kwargs.get("plot_slope", True)
-
-		self.x_margin = kwargs.get("x_margin", 0.05)
-		self.y_margin = kwargs.get("y_margin", 0.1)
-
-		self.will_plot_labels = kwargs.get("plot_labels", True)
-
-	@staticmethod
-	def get_symbol_by_slope(slope, default_symbol):
-		draw_symbol = default_symbol
-		if slope > math.tan(3*math.pi/8):
-			draw_symbol = "|"
-		elif slope > math.tan(math.pi/8) and slope < math.tan(3*math.pi/8):
-			draw_symbol = "/"
-		elif abs(slope) < math.tan(math.pi/8):
-			draw_symbol = "-"
-		elif slope < math.tan(-math.pi/8) and slope > math.tan(-3*math.pi/8):
-			draw_symbol = "\\"
-		elif slope < math.tan(-3*math.pi/8):
-			draw_symbol = "|"
-		return draw_symbol    
-
-
-	def plot_labels(self, output_buffer, plot_data):
-		if plot_data.y_size < 2:
-			return
-
-		margin_factor = 1
-
-		do_plot_x_label = True
-		do_plot_y_label = True
-
-		x_str = "%+g"
-		if plot_data.x_size < 16:
-			do_plot_x_label = False
-		elif plot_data.x_size < 23:
-			x_str = "%+.2g" 
-
-		y_str = "%+g"    
-		if plot_data.x_size < 8:
-			do_plot_y_label = False
-		elif plot_data.x_size < 11:
-			y_str = "%+.2g"
-			
-		act_min_x = (plot_data.min_x + plot_data.x_mod*margin_factor)
-		act_max_x = (plot_data.max_x - plot_data.x_mod*margin_factor)
-		act_min_y = (plot_data.min_y + plot_data.y_mod*margin_factor)
-		act_max_y = (plot_data.max_y - plot_data.y_mod*margin_factor)
-
-		if abs(act_min_x) < 1:
-			min_x_str = "%+.2g" % act_min_x
-		else:
-			min_x_str = x_str % act_min_x
-
-		if abs(act_max_x) < 1:
-			max_x_str = "%+.2g" % act_max_x
-		else:
-			max_x_str = x_str % act_max_x
-		
-		if abs(act_min_y) < 1:
-			min_y_str = "%+.2g" % act_min_y
-		else:
-			min_y_str = y_str % act_min_y
-
-		if abs(act_max_y) < 1:
-			max_y_str = "%+.2g" % act_max_y
-		else:
-			max_y_str = y_str % act_max_y
-						 
-		min_x_coord = self.get_coord(act_min_x,plot_data.min_x,plot_data.x_step)
-		max_x_coord = self.get_coord(act_max_x,plot_data.min_x,plot_data.x_step)
-		min_y_coord = self.get_coord(act_min_y,plot_data.min_y,plot_data.y_step)
-		max_y_coord = self.get_coord(act_max_y,plot_data.min_y,plot_data.y_step)
-								 
-
-		#print plot_data
-		
-		y_zero_coord = self.get_coord(0, plot_data.min_y, plot_data.y_step)
-
-		#if plot_data.min_x < 0 and plot_data.max_x > 0:
-		x_zero_coord = self.get_coord(0, plot_data.min_x, plot_data.x_step)
-		#else:
-		
-		#pass
-
-		output_buffer[x_zero_coord][min_y_coord] = "+"
-		output_buffer[x_zero_coord][max_y_coord] = "+"
-		output_buffer[min_x_coord][y_zero_coord] = "+"
-		output_buffer[max_x_coord][y_zero_coord] = "+"
-
-		if do_plot_x_label:        
-
-			for i,c in enumerate(min_x_str):
-				output_buffer[min_x_coord+i][y_zero_coord-1] = c
-			for i,c in enumerate(max_x_str):
-				output_buffer[max_x_coord+i-len(max_x_str)][y_zero_coord-1] = c
-
-		if do_plot_y_label:
-
-			for i,c in enumerate(max_y_str):
-				output_buffer[x_zero_coord+i][max_y_coord] = c
-			for i,c in enumerate(min_y_str):
-				output_buffer[x_zero_coord+i][min_y_coord] = c
-			
-
-		
-		
-	
-	def plot_data(self, xy_seq, output_buffer, plot_data):
-		if self.plot_slope:
-			xy_seq = list(xy_seq)
-			#sort according to the x coord
-			xy_seq.sort(key = lambda c: c[0])
-			prev_p = xy_seq[0]
-			e_xy_seq = enumerate(xy_seq)
-			e_xy_seq.next()
-			for i,(x,y) in e_xy_seq:
-				draw_symbol = self.dot
-				line_drawn = self.plot_line(prev_p, (x,y), output_buffer, plot_data)
-				prev_p = (x,y)
-				if not line_drawn:
-					if i > 0 and i < len(xy_seq)-1:
-						px,py = xy_seq[i-1]
-						nx,ny = xy_seq[i+1]
-
-						if abs(nx-px) > EPSILON:
-							slope = (1.0/plot_data.ratio)*(ny-py)/(nx-px)
-							draw_symbol = self.get_symbol_by_slope(slope, draw_symbol)
-					if x < plot_data.min_x or x >= plot_data.max_x or y < plot_data.min_y or y >= plot_data.max_y:
-						continue
-					
-					x_coord = self.get_coord(x, plot_data.min_x, plot_data.x_step)
-					y_coord = self.get_coord(y, plot_data.min_y, plot_data.y_step)            
-					if x_coord >= 0 and x_coord < len(output_buffer) and y_coord >= 0 and y_coord < len(output_buffer[0]):
-						if self.draw_axes:
-							if y_coord == self.get_coord(0, plot_data.min_y, plot_data.y_step) and draw_symbol == "-":
-								draw_symbol = "="
-						output_buffer[x_coord][y_coord] = draw_symbol
-		else:
-			for x,y in xy_seq:
-				if x < plot_data.min_x or x >= plot_data.max_x or y < plot_data.min_y or y >= plot_data.max_y:
-					continue
-				x_coord = self.get_coord(x, plot_data.min_x, plot_data.x_step)
-				y_coord = self.get_coord(y, plot_data.min_y, plot_data.y_step)
-				if x_coord >= 0 and x_coord < len(output_buffer) and y_coord > 0 and y_coord < len(output_buffer[0]):
-					output_buffer[x_coord][y_coord] = self.dot
-
-
-	def plot_line(self, start, end, output_buffer, plot_data):
-
-		start_coord = self.get_coord(start[0], plot_data.min_x, plot_data.x_step), self.get_coord(start[1], plot_data.min_y, plot_data.y_step)
-		end_coord = self.get_coord(end[0], plot_data.min_x, plot_data.x_step), self.get_coord(end[1], plot_data.min_y, plot_data.y_step)
-
-		x0,y0 = start_coord
-		x1,y1 = end_coord
-		if (x0,y0) == (x1,y1):
-			return True    
-		
-		clipped_line = clip_line(start, end, (plot_data.min_x, plot_data.min_y), (plot_data.max_x, plot_data.max_y))
-		if clipped_line != None:
-			start,end = clipped_line
-		else:
-			return False
-		start_coord = self.get_coord(start[0], plot_data.min_x, plot_data.x_step), self.get_coord(start[1], plot_data.min_y, plot_data.y_step)
-		end_coord = self.get_coord(end[0], plot_data.min_x, plot_data.x_step), self.get_coord(end[1], plot_data.min_y, plot_data.y_step)
-
-		x0,y0 = start_coord
-		x1,y1 = end_coord
-		if (x0,y0) == (x1,y1):
-			return True
-		x_zero_coord = self.get_coord(0, plot_data.min_x, plot_data.x_step)
-		y_zero_coord = self.get_coord(0, plot_data.min_y, plot_data.y_step)   
-
-		if start[0]-end[0] == 0:
-			draw_symbol = "|"
-		else:
-			slope = (1.0/plot_data.ratio)*(end[1]-start[1])/(end[0]-start[0])
-			draw_symbol = self.get_symbol_by_slope(slope, self.dot)
-		try:
-
-			delta = x1-x0, y1-y0
-			if abs(delta[0])>abs(delta[1]):
-				s = sign(delta[0])
-				slope = float(delta[1])/delta[0]
-				for i in range(0,abs(int(delta[0]))):
-					cur_draw_symbol = draw_symbol
-					x = i*s
-					cur_y = int(y0+slope*x)
-					if self.draw_axes and cur_y == y_zero_coord and draw_symbol == "-":
-						cur_draw_symbol = "="
-					output_buffer[x0+x][cur_y] = cur_draw_symbol
-				
-				
-			else:
-				s = sign(delta[1])
-				slope = float(delta[0])/delta[1]
-				for i in range(0,abs(int(delta[1]))):
-					y = i*s
-					cur_draw_symbol = draw_symbol
-					cur_y = y0+y
-					if self.draw_axes and cur_y == y_zero_coord and draw_symbol == "-":
-						cur_draw_symbol = "="
-					output_buffer[int(x0+slope*y)][cur_y] = cur_draw_symbol
-		except:
-			print start, end
-			print start_coord, end_coord
-			print plot_data
-			raise
-
-		return False            
-		
-		
-	def plot_single(self, seq, min_x = None, max_x = None, min_y = None, max_y = None):
-		return self.plot_double(range(len(seq)),seq, min_x, max_x, min_y, max_y)
-		
-
-
-
-	def plot_double(self, x_seq, y_seq, min_x = None, max_x = None, min_y = None, max_y = None):
-		if min_x == None:
-			min_x = min(x_seq)
-		if max_x == None:
-			max_x = max(x_seq)
-		if min_y == None:
-			min_y = min(y_seq)
-		if max_y == None:
-			max_y = max(y_seq)
-
-		if max_y == min_y:
-			max_y += 1
-
-		x_mod = (max_x-min_x)*self.x_margin
-		y_mod = (max_y-min_y)*self.y_margin
-		min_x-=x_mod
-		max_x+=x_mod
-		min_y-=y_mod
-		max_y+=y_mod
-
-
-		plot_data = self.PlotData(self.x_size, self.y_size, min_x, max_x, min_y, max_y, x_mod, y_mod)
-
-		output_buffer = [[" "]*self.y_size for i in range(self.x_size)]
-
-		if self.will_draw_axes:
-			self.draw_axes(output_buffer, plot_data)
-
-		self.plot_data(zip(x_seq, y_seq), output_buffer, plot_data)
-
-		if self.will_plot_labels:
-			self.plot_labels(output_buffer, plot_data)
-
-		trans_result = transposed(y_reversed(output_buffer))
-
-		result = self.new_line.join(["".join(row) for row in trans_result])
-		return result
-
-	def draw_axes(self, output_buffer, plot_data):
-		
-		
-		draw_x = False
-		draw_y = False
-
-		if plot_data.min_x <= 0 and plot_data.max_x > 0:
-			draw_y = True
-			zero_x = self.get_coord(0, plot_data.min_x, plot_data.x_step)
-			for y in xrange(plot_data.y_size):
-				output_buffer[zero_x][y] = "|"
-				
-		if plot_data.min_y <= 0 and plot_data.max_y > 0:
-			draw_x = True
-			zero_y = self.get_coord(0, plot_data.min_y, plot_data.y_step)    
-			for x in xrange(plot_data.x_size):
-				output_buffer[x][zero_y] = "-"
-
-		if draw_x and draw_y:
-			output_buffer[zero_x][zero_y] = "+"
-		
-		
-	@staticmethod
-	def get_coord(val, min, step):
-		result = int((val - min)/step)
-		return result
-
-def clip_line(line_pt_1, line_pt_2, rect_bottom_left, rect_top_right):
-	ts = [0.0,1.0]
-	if line_pt_1[0] == line_pt_2[0]:
-		return ((line_pt_1[0], max(min(line_pt_1[1], line_pt_2[1]), rect_bottom_left[1])),
-				(line_pt_1[0], min(max(line_pt_1[1], line_pt_2[1]), rect_top_right[1])))
-	if line_pt_1[1] == line_pt_2[1]:
-		return ((max(min(line_pt_1[0], line_pt_2[0]), rect_bottom_left[0]), line_pt_1[1]),
-				(min(max(line_pt_1[0], line_pt_2[0]), rect_top_right[0]), line_pt_1[1]))
-
-	if ((rect_bottom_left[0] <= line_pt_1[0] and line_pt_1[0] < rect_top_right[0]) and
-		(rect_bottom_left[1] <= line_pt_1[1] and line_pt_1[1] < rect_top_right[1]) and
-		(rect_bottom_left[0] <= line_pt_2[0] and line_pt_2[0] < rect_top_right[0]) and
-		(rect_bottom_left[1] <= line_pt_2[1] and line_pt_2[1] < rect_top_right[1])):
-		return line_pt_1, line_pt_2
-
-	ts.append( float(rect_bottom_left[0]-line_pt_1[0])/(line_pt_2[0]-line_pt_1[0]) )
-	ts.append( float(rect_top_right[0]-line_pt_1[0])/(line_pt_2[0]-line_pt_1[0]) )
-	ts.append( float(rect_bottom_left[1]-line_pt_1[1])/(line_pt_2[1]-line_pt_1[1]) )
-	ts.append( float(rect_top_right[1]-line_pt_1[1])/(line_pt_2[1]-line_pt_1[1]) )
-	
-	ts.sort()
-	if ts[2] < 0 or ts[2] >= 1 or ts[3] < 0 or ts[2]>= 1:
-		return None
-	result = [(pt_1 + t*(pt_2-pt_1)) for t in (ts[2],ts[3]) for (pt_1, pt_2) in zip(line_pt_1, line_pt_2)]
-	return (result[0],result[1]), (result[2], result[3])
-	
-
-
-def plot(*args,**flags):
-	limit_flags_names = set(["min_x","min_y","max_x","max_y"])
-	limit_flags = dict([(n,flags[n]) for n in limit_flags_names & set(flags)])
-	settting_flags = dict([(n,flags[n]) for n in set(flags) - limit_flags_names])
-	
-	if len(args) == 1:
-		p = Plotter(**settting_flags)
-		print p.plot_single(args[0],**limit_flags)
-	elif len(args) == 2:
-		p = Plotter(**settting_flags)
-		print p.plot_double(args[0],args[1],**limit_flags)
-	else:
-		raise NotImplementedError("can't draw multiple graphs yet")
-	
-__all__ = ["Plotter","plot"]
-