Implementing loglinear classification (without training), for CL classification

appengine/findit/crash/results.py

 # Copyright 2016 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 from collections import namedtuple
 
+
+# TODO(wrengr): we should change things to use integers with None as
+# \"infinity\", rather than using floats.
 # TODO(http://crbug.com/644476): this class needs a better name.
 class AnalysisInfo(namedtuple('AnalysisInfo',
     ['min_distance', 'min_distance_frame'])):
   __slots__ = ()
 
 
 # TODO(wrengr): it's not clear why the ``priority`` is stored at all,
 # given that every use in this file discards it. ``Result.file_to_stack_infos``
 # should just store pointers directly to the frames themselves rather
 # than needing this intermediate object.
@@ skipping 120 matching lines @@
     range, then the difference between those points. Otherwise, returns
     zero (because the ranges overlap)."""
   assert end1 >= start1
   assert end2 >= start2
   # There are six possible cases, but in all the cases where the two
   # ranges overlap, the latter two differences will be negative.
   return max(0, start2 - end1, start1 - end2)
 
 
 class MatchResults(dict):
-  """A dict indexing MatchResult with its revision."""
+  """A map from revisions to the MatchResult object for that revision."""
 
   def __init__(self, ignore_cls=None):
     super(MatchResults, self).__init__()
     self._ignore_cls = ignore_cls
 
   def GenerateMatchResults(self, file_path, dep_path,
                            stack_infos, changelogs, blame):
-    """Generates match results.
+    """Compute match results from a list of CLs, and store them.
 
     Match results are generated based on newly found file path, its stack_infos,
     and all the changelogs that touched this file in the dep in regression
     ranges, those reverted changelogs should be ignored.
 
     Args:
       file_path (str): File path of the crashed file.
       dep_path (str): Path of the dependency of the file.
       stack_infos (list): List of stack_info dicts, represents frames of this
         file and the callstack priorities of those frames.
       changelogs (list): List of Changelog objects in the dep in regression
         range which touched the file.
       blame (Blame): Blame of the file.
     """
     for changelog in changelogs:
       if self._ignore_cls and changelog.revision in self._ignore_cls:
         continue
 
       if changelog.revision not in self:
         self[changelog.revision] = MatchResult(changelog, dep_path)
 
       match_result = self[changelog.revision]
       match_result.Update(file_path, stack_infos, blame)