[Findit] Initial code of findit for crash. Add scorers to apply heuristic rules.

appengine/findit/crash/results.py

Index: appengine/findit/crash/results.py
diff --git a/appengine/findit/crash/results.py b/appengine/findit/crash/results.py
new file mode 100644
index 0000000000000000000000000000000000000000..abe6c258cd127fe5f8a57ac9fea8f4f6a450d82e
--- /dev/null
+++ b/appengine/findit/crash/results.py
@@ -0,0 +1,137 @@
+# 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.
+
+_INFINITY = 1000
+
+
+class Result(object):
+
+  def __init__(self, changelog, dep_path, component,
+               confidence=None, reason=None):
+    self.changelog = changelog
+    self.dep_path = dep_path
+    self.component = component
+    self.confidence = confidence
+    self.reason = reason
+
+    self.file_to_stack_infos = {}
+
+  def ToDict(self):
+    changelog_json = self.changelog.ToDict()

[stgao, 2016/04/15 18:35:19]

This seems unnecessary as we only refer to several fields. Why not refer to them
directly?

[Sharu, 2016/04/15 22:59:47]

Done.

+    return {
+        'url': changelog_json['commit_url'],
+        'revision': changelog_json['revision'],
+        'dep_path': self.dep_path,
+        'component': self.component,
+        'author': changelog_json['author_email'],
+        'time': changelog_json['author_time'],
+        'reason': self.reason,
+        'confidence': self.confidence,
+    }
+
+  def ToString(self):
+    if not self.file_to_stack_infos:
+      return ''
+
+    lines = []
+    for file_path, stack_infos in self.file_to_stack_infos.iteritems():
+      line_parts = []
+      for frame, _ in stack_infos:
+        line_parts.append('%s (#%d)' % (frame.function, frame.index))
+
+      lines.append('Changed file %s which crashed in %s' % (
+          file_path, ', '.join(line_parts)))
+
+    return '\n'.join(lines)
+
+  def __str__(self):
+    return self.ToString()
+
+  def __repr__(self):
+    return self.ToString()
+
+
+class MatchResult(Result):
+
+  def __init__(self, changelog, dep_path, component,
+               confidence=None, reason=None):
+    super(MatchResult, self).__init__(
+        changelog, dep_path, component, confidence, reason)
+
+    self.min_distance = _INFINITY
+
+  def Update(self, file_path, stack_infos, blame):
+    """Updates match result once a file_path is found both shown in
+    stacktrace and touched by the changelog of this result.

[stgao, 2016/04/15 18:35:19]

changelog here might not be that clear enough. It seems we meant to refer to a
commit or revision.

[Sharu, 2016/04/15 22:59:47]

Done.

+
+    Inserts the file path and its stack infos, and updates the min distance
+    if less distance is found between touched lines of this result and
+    crashed lines in the file path.
+
+    Args:
+      file_path (str): File path of the crashed file.
+      stack_infos (list): List of (StackFrame, callstack priority) tuples,
+        represents frames of this file and the callstack priorities of those
+        frames.
+      blame (Blame): Blame oject of this file.
+    """
+    self.file_to_stack_infos[file_path] = stack_infos
+
+    for region in blame:
+      if region.revision != self.changelog.revision:
+        continue
+
+      region_lines = range(region.start, region.start + region.count)
+
+      for frame, _ in stack_infos:
+        self.min_distance = min(self.min_distance, self._DistanceOfTwoRegions(
+            frame.crashed_line_numbers, region_lines))
+
+  def _DistanceOfTwoRegions(self, region1, region2):
+
+    if set(region1).intersection(set(region2)):
+      return 0
+
+    if region1[-1] < region2[0]:
+      return region2[0] - region1[-1]
+
+    return region1[0] - region2[-1]
+
+
+class MatchResults(dict):
+
+  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 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 (StackFrame, callstack priority) tuples,
+        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:
+        # TODO(katesonia): Enable component classifier later. Get it from
+        # file_path and dep_path.
+        component = ''
+        self[changelog.revision] = MatchResult(changelog, dep_path, component)
+
+      match_result = self[changelog.revision]
+
+      match_result.Update(file_path, stack_infos, blame)