Add --optimize option to nilmtool intervals

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
@@ -69,6 +74,7 @@ class Client(object):
         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

nilmdb/cmdline/intervals.py

@@ -1,5 +1,6 @@
 from nilmdb.utils.printf import *
 import nilmdb.utils.time
+from nilmdb.utils.interval import Interval
 
 import fnmatch
 import argparse
@@ -42,6 +43,8 @@ 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("-o", "--optimize", action="store_true",
+                       help="Optimize (merge adjacent) intervals")
 
     return cmd
 
@@ -58,9 +61,16 @@ def cmd_intervals(self):
         time_string = nilmdb.utils.time.timestamp_to_human
 
     try:
-           for (start, end) in self.client.stream_intervals(
-               self.args.path, self.args.start, self.args.end, self.args.diff):
-               printf("[ %s -> %s ]\n", time_string(start), time_string(end))
+        intervals = ( Interval(start, end) for (start, end) in
+                      self.client.stream_intervals(self.args.path,
+                                                   self.args.start,
+                                                   self.args.end,
+                                                   self.args.diff) )
+        if self.args.optimize:
+            intervals = nilmdb.utils.interval.optimize(intervals)
+        for i in intervals:
+            printf("[ %s -> %s ]\n", time_string(i.start), time_string(i.end))
+
     except nilmdb.client.ClientError as e:
         self.die("error listing intervals: %s", str(e))
 

nilmdb/utils/interval.py

@@ -58,18 +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.
@@ -84,31 +77,57 @@ 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:
-                out_start = ts
+            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):
     """

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/time.py

@@ -60,7 +60,7 @@ 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).
     """

tests/test_cmdline.py

@@ -1011,6 +1011,18 @@ class TestCmdline(object):
         self.match("[ Thu, 01 Jan 2004 00:00:00.000000 +0000 -"
                    "> Sat, 01 Jan 2005 00:00:00.000000 +0000 ]\n")
 
+        # optimize
+        self.ok("insert -s 01-01-2002 -e 01-01-2004 /diff/1 /dev/null")
+        self.ok("intervals /diff/1")
+        self.match("[ Sat, 01 Jan 2000 00:00:00.000000 +0000 -"
+                   "> Thu, 01 Jan 2004 00:00:00.000000 +0000 ]\n"
+                   "[ Thu, 01 Jan 2004 00:00:00.000000 +0000 -"
+                   "> Sat, 01 Jan 2005 00:00:00.000000 +0000 ]\n")
+        self.ok("intervals /diff/1 --optimize")
+        self.ok("intervals /diff/1 -o")
+        self.match("[ Sat, 01 Jan 2000 00:00:00.000000 +0000 -"
+                   "> Sat, 01 Jan 2005 00:00:00.000000 +0000 ]\n")
+
         self.ok("destroy -R /diff/1")
         self.ok("destroy -R /diff/2")
 

tests/test_interval.py

@@ -234,13 +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)
 
@@ -310,6 +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)
 

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 +106,9 @@ 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)()
+        eq_(list(i), [1,2,3,4,5])
+        eq_(i[3], 3)