tools/findit/match_set.py

   1 # Copyright (c) 2014 The Chromium Authors. All rights reserved.
   2 # Use of this source code is governed by a BSD-style license that can be
   3 # found in the LICENSE file.
   4
   5 import re
   6
   7 from threading import Lock
   8
   9 import crash_utils
  10
  11
  12 REVIEW_URL_PATTERN = re.compile(r'Review URL:( *)(.*?)/(\d+)')
  13
  14
  15 class Match(object):
  16   """Represents a match entry.
  17
  18   A match is a CL that is suspected to have caused the crash. A match object
  19   contains information about files it changes, their authors, etc.
  20
  21   Attributes:
  22     is_revert: True if this CL is reverted by other CL.
  23     revert_of: If this CL is a revert of some other CL, a revision number/
  24                git hash of that CL.
  25     crashed_line_numbers: The list of lines that caused crash for this CL.
  26     function_list: The list of functions that caused the crash.
  27     min_distance: The minimum distance between the lines that CL changed and
  28                   lines that caused the crash.
  29     changed_files: The list of files that the CL changed.
  30     changed_file_urls: The list of URLs for the file.
  31     author: The author of the CL.
  32     component_name: The name of the component that this CL belongs to.
  33     stack_frame_indices: For files that caused crash, list of where in the
  34                          stackframe they occur.
  35     priorities: A list of priorities for each of the changed file. A priority
  36                 is 1 if the file changes a crashed line, and 2 if it changes
  37                 the file but not the crashed line.
  38     reivision_url: The revision URL of the CL.
  39     review_url: The codereview URL that reviews this CL.
  40     reviewers: The list of people that reviewed this CL.
  41     reason: The reason why this CL is suspected.
  42   """
  43   REVERT_PATTERN = re.compile(r'(revert\w*) r?(\d+)', re.I)
  44
  45   def __init__(self, revision, component_name):
  46     self.is_revert = False
  47     self.revert_of = None
  48     self.message = None
  49     self.crashed_line_numbers = []
  50     self.function_list = []
  51     self.min_distance = crash_utils.INFINITY
  52     self.min_distance_info = None
  53     self.changed_files = []
  54     self.changed_file_urls = []
  55     self.author = revision['author']
  56     self.component_name = component_name
  57     self.stack_frame_indices = []
  58     self.priorities = []
  59     self.revision_url = revision['url']
  60     self.review_url = ''
  61     self.reviewers = []
  62     self.reason = None
  63
  64   def ParseMessage(self, message, codereview_api_url):
  65     """Parses the message.
  66
  67     It checks the message to extract the code review website and list of
  68     reviewers, and it also checks if the CL is a revert of another CL.
  69
  70     Args:
  71       message: The message to parse.
  72       codereview_api_url: URL to retrieve codereview data from.
  73     """
  74     self.message = message
  75     for line in message.splitlines():
  76       line = line.strip()
  77       review_url_line_match = REVIEW_URL_PATTERN.match(line)
  78
  79       # Check if the line has the code review information.
  80       if review_url_line_match:
  81
  82         # Get review number for the code review site from the line.
  83         issue_number = review_url_line_match.group(3)
  84
  85         # Get JSON from the code review site, ignore the line if it fails.
  86         url = codereview_api_url % issue_number
  87         json_string = crash_utils.GetDataFromURL(url)
  88         if not json_string:
  89           continue
  90
  91         # Load the JSON from the string, and get the list of reviewers.
  92         code_review = crash_utils.LoadJSON(json_string)
  93         if code_review:
  94           self.reviewers = code_review['reviewers']
  95
  96       # Check if this CL is a revert of other CL.
  97       if line.lower().startswith('revert'):
  98         self.is_revert = True
  99
 100         # Check if the line says what CL this CL is a revert of.
 101         revert = self.REVERT_PATTERN.match(line)
 102         if revert:
 103           self.revert_of = revert.group(2)
 104         return
 105
 106
 107 class MatchSet(object):
 108   """Represents a set of matches.
 109
 110   Attributes:
 111     matches: A map from CL to a match object.
 112     cls_to_ignore: A set of CLs to ignore.
 113     matches_lock: A lock guarding matches dictionary.
 114   """
 115
 116   def __init__(self, codereview_api_url):
 117     self.codereview_api_url = codereview_api_url
 118     self.matches = {}
 119     self.cls_to_ignore = set()
 120     self.matches_lock = Lock()
 121
 122   def RemoveRevertedCLs(self):
 123     """Removes CLs that are revert."""
 124     for cl in self.matches:
 125       if cl in self.cls_to_ignore:
 126         del self.matches[cl]