Intial checkin of layout test analyzer.

media/tools/layout_tests/layouttest_analyzer_helpers.py

Index: media/tools/layout_tests/layouttest_analyzer_helpers.py
diff --git a/media/tools/layout_tests/layouttest_analyzer_helpers.py b/media/tools/layout_tests/layouttest_analyzer_helpers.py
new file mode 100644
index 0000000000000000000000000000000000000000..f06e9feaff12a820f888c5865883e876981ebc7d
--- /dev/null
+++ b/media/tools/layout_tests/layouttest_analyzer_helpers.py
@@ -0,0 +1,437 @@
+#!/usr/bin/python
+# Copyright (c) 2011 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.
+
+"""Helper functions for the layout test analyzer."""
+
+import copy
+from datetime import datetime
+from email.mime.multipart import MIMEMultipart
+from email.mime.text import MIMEText
+import os
+import pickle
+import smtplib
+import socket
+import time
+import urllib
+
+from bug import Bug
+from test_expectations_history import TestExpectationsHistory
+
+
+class AnalyzerResultMap:
+  """A class to deal with joined result produed by the analyzer.
+
+  The join is done between layouttests and the test_expectations object
+  (based on the test expectation file). The instance variable |result_map|
+  contains the following keys: 'whole','skip','nonskip'. The value of 'whole'
+  contains information about all layouttests. The value of 'skip' contains
+  information about skipped layouttests where it has 'SKIP' in its entry in
+  the test expectation file. The value of 'nonskip' contains all information
+  about non skipped layout tests, which are in the test expectation file but
+  not skipped. The information is exactly same as the one parsed by the
+  analyzer.
+  """
+
+  def __init__(self, test_info_map):
+    """Initialize the AnalyzerResultMap based on test_info_map.
+
+    Test_info_map contains all layouttest information. The job here is to
+    classify them as 'whole', 'skip' or 'nonskip' based on that information.
+
+    Args:
+      test_info_map: the result map of |layouttests.JoinWithTestExpectation|.
+          The key of the map is test name such as 'media/media-foo.html'.
+          The value of the map is a map that contains the following keys:
+          'desc'(description), 'te_info' (test expectation information),
+          which is a list of test expectation information map. The key of the
+          test expectation information map is test expectation keywords such
+          as "SKIP" and other keywords (for full list of keywords, please
+          refer to |test_expectaions.ALL_TE_KEYWORDS|).
+    """
+    self.result_map = {}
+    self.result_map['whole'] = {}
+    self.result_map['skip'] = {}
+    self.result_map['nonskip'] = {}
+    if test_info_map:
+      for (k, v) in test_info_map.iteritems():
+        self.result_map['whole'][k] = v
+        if 'te_info' in v:
+          if any([True for x in v['te_info'] if 'SKIP' in x]):
+            self.result_map['skip'][k] = v
+          else:
+            self.result_map['nonskip'][k] = v
+
+  @staticmethod
+  def GetDiffString(diff_map_element, type_str):
+    """Get difference string out of diff map element.
+
+    The difference string shows difference between two analyzer results
+    (for example, a result for now and a result for sometime in the past)
+    in HTML format (with colors). This is used for generating email messages.
+
+    Args:
+      diff_map_element: An element of the compared map generated by
+         |CompareResultMaps()|. The element has two lists of test cases. One
+         is for test names that are in the current result but NOT in the
+         previous result. The other is for test names that are in the previous
+         results but NOT in the current result. Please refer to comments in
+         |CompareResultMaps()| for details.
+    type_str: a string indicating the test group to which |diff_map_element|
+        belongs; used for color determination.  Must be 'whole', 'skip', or
+        'nonskip'.
+
+    Returns:
+      a string in HTML format (with colors) to show difference between two
+          analyzer results.
+    """
+    diff = len(diff_map_element[0]) - len(diff_map_element[1])
+    if diff == 0:
+      return 'No Change'
+    color = ''
+    if diff > 0 and type_str != 'whole':
+      color = 'red'
+    else:
+      color = 'green'
+    diff_sign = ''
+    if diff > 0:
+      diff_sign = '+'
+    str = '<font color="%s">%s%d</font>' % (color, diff_sign, diff)
+    str1 = ''
+    for (name, v) in diff_map_element[0]:
+      str1 += name + ','
+    str1 = str1[:-1]
+    str2 = ''
+    for (name, v) in diff_map_element[1]:
+      str2 += name + ','
+    str2 = str2[:-1]
+    if str1 or str2:
+      str += ':'
+    if str1:
+      str += '<font color="%s">%s</font> ' % (color, str1)
+    if str2:
+      str += '<font color="%s">%s</font>' % (color, str2)
+    return str
+
+  def GetPassingRate(self):
+    """Get passing rate.
+
+    Returns:
+      layout test passing rate of this result in percent.
+
+    Raises:
+      ValueEror when the number of tests in test group "whole" is equal or less
+          than that of "skip".
+    """
+    d = len(self.result_map['whole'].keys()) - (
+        len(self.result_map['skip'].keys()))
+    if d <= 0:
+      raise ValueError('The number of tests in test group "whole" is equal or '
+                       'less than that of "skip"')
+    return 100 - len(self.result_map['nonskip'].keys()) * 100 / d
+
+  def ConvertToString(self, prev_time, diff_map, bug_anno_map):
+    """Convert this result to HTML display for email.
+
+    Args:
+      prev_time: the previous time string that are compared against.
+      diff_map: the compared map generated by |CompareResultMaps()|.
+      bug_anno_map: a annotation map where keys are bug names and values are
+          annotations for the bug.
+
+    Returns:
+      a analyzer result string in HTML format.
+    """
+
+    str = ('<b>Statistics (Diff Compared to %s):</b><ul>'
+           '<li>The number of tests: %d (%s)</li>'
+           '<li>The number of failing skipped tests: %d (%s)</li>'
+           '<li>The number of failing non-skipped tests: %d (%s)</li>'
+           '<li>Passing rate: %d %%</li></ul>') % (
+        prev_time,
+        len(self.result_map['whole'].keys()),
+        AnalyzerResultMap.GetDiffString(diff_map['whole'], 'whole'),
+        len(self.result_map['skip'].keys()),
+        AnalyzerResultMap.GetDiffString(diff_map['skip'], 'skip'),
+        len(self.result_map['nonskip'].keys()),
+        AnalyzerResultMap.GetDiffString(diff_map['nonskip'], 'nonskip'),
+        self.GetPassingRate())
+    str += '<b>Current issues about failing non-skipped tests:</b>'
+    for (bug_txt, test_info_list) in (
+        self.GetListOfBugsForNonSkippedTests().iteritems()):
+      if not bug_txt in bug_anno_map:
+        bug_anno_map[bug_txt] = '<font color="red">Needs investigation!</font>'
+      str += '<ul>%s (%s)' % (Bug(bug_txt), bug_anno_map[bug_txt])
+      for test_info in test_info_list:
+        (test_name, te_info) = test_info
+        gpu_link = ''
+        if 'GPU' in te_info:
+          gpu_link = 'group=%40ToT%20GPU%20Mesa%20-%20chromium.org&'
+        dashboard_link = ('http://test-results.appspot.com/dashboards/'
+                          'flakiness_dashboard.html#%stests=%s') % (
+                              gpu_link, test_name)
+        str += '<li><a href="%s">%s</a> (%s) </li>' % (
+            dashboard_link, test_name, ' '.join(te_info.keys()))
+      str += '</ul>\n'
+    return str
+
+  def CompareToOtherResultMap(self, other_result_map):
+    """Compare this result map with the other to see if there are any diff.
+
+    The comparison is done for layouttests which belong to 'whole', 'skip',
+    or 'nonskip'.
+
+    Args:
+      other_result_map: another result map to be compared against the result
+          map of the current object.
+
+    Returns:
+      a map that has 'whole', 'skip' and 'nonskip' as keys. The values of the
+          map are the result of |GetDiffBetweenMaps()|.
+          The element has two lists of test cases. One (with index 0) is for
+          test names that are in the current result but NOT in the previous
+          result. The other (with index 1) is for test names that are in the
+          previous results but NOT in the current result.
+           For example (test expectation information is omitted for
+           simplicity),
+             comp_result_map['whole'][0] = ['foo1.html']
+             comp_result_map['whole'][1] = ['foo2.html']
+           This means that current result has 'foo1.html' but NOT in the
+           previous result. This also means the previous result has 'foo2.html'
+           but it is NOT the current result.
+    """
+    comp_result_map = {}
+    for name in ['whole', 'skip', 'nonskip']:
+      if name == 'nonskip':
+        # Look into expectation to get diff only for non-skipped tests.
+        lookIntoTestExpectationInfo = True
+      else:
+        #  Otherwise, only test names are compared to get diff.
+        lookIntoTestExpectationInfo = False
+      comp_result_map[name] = GetDiffBetweenMaps(
+         self.result_map[name], other_result_map.result_map[name],
+         lookIntoTestExpectationInfo)
+    return comp_result_map
+
+  @staticmethod
+  def Load(file_path):
+    """Load the object from |file_path| using pickle library.
+
+    Args:
+      file_path: the string path to the file from which to read the result.
+
+    Returns:
+       a AnalyzerResultMap object read from |file_path|.
+    """
+    file_object = open(file_path)
+    analyzer_result_map = pickle.load(file_object)
+    file_object.close()
+    return analyzer_result_map
+
+  def Save(self, file_path):
+    """Save the object to |file_path| using pickle library.
+
+    Args:
+       file_path: the string path to the file in which to store the result.
+    """
+    file_object = open(file_path, 'wb')
+    pickle.dump(self, file_object)
+    file_object.close()
+
+  def GetListOfBugsForNonSkippedTests(self):
+    """Get a list of bugs for non-skipped layout tests.
+
+    This is used for generating email content.
+
+    Returns:
+      a mapping from bug modifier text (e.g., BUGCR1111) to a test name and
+          main test information string which excludes comments and bugs. This
+          is used for grouping test names by bug.
+    """
+    bug_map = {}
+    for (name, v) in self.result_map['nonskip'].iteritems():
+      for te_info in v['te_info']:
+        main_te_info = {}
+        for k in te_info.keys():
+          if k != 'Comments' and k != 'Bugs':
+            main_te_info[k] = True
+        if 'Bugs' in te_info:
+          for bug in te_info['Bugs']:
+            if bug not in bug_map:
+              bug_map[bug] = []
+            bug_map[bug].append((name, main_te_info))
+    return bug_map
+
+
+def SendStatusEmail(prev_time, analyzer_result_map, prev_analyzer_result_map,
+                    bug_anno_map, receiver_email_address):
+  """Send status email.
+
+  Args:
+    prev_time: the date string such as '2011-10-09-11'. This format has been
+        used in this analyzer.
+    analyzer_result_map: current analyzer result.
+    prev_analyzer_result_map: previous analyzer result, which is read from
+         a file.
+    bug_anno_map: bug annotation map where bug name and annotations are
+         stored.
+    receiver_email_address: receiver's email address.
+  """
+  diff_map = analyzer_result_map.CompareToOtherResultMap(
+      prev_analyzer_result_map)
+  str = analyzer_result_map.ConvertToString(prev_time, diff_map, bug_anno_map)
+  # Add diff info about skipped/non-skipped test.
+  prev_time = datetime.strptime(prev_time, '%Y-%m-%d-%H')
+  prev_time = time.mktime(prev_time.timetuple())
+  testname_map = {}
+  for type in ['skip', 'nonskip']:
+    for i in range(2):
+      for (k, _) in diff_map[type][i]:
+        testname_map[k] = True
+  now = time.time()
+
+  rev_infos = TestExpectationsHistory.GetDiffBetweenTimes(now, prev_time,
+                                                          testname_map.keys())
+  if rev_infos:
+    str += '<br><b>Revision Information:</b>'
+    for rev_info in rev_infos:
+      (old_rev, new_rev, author, date, message, target_lines) = rev_info
+      link = urllib.unquote('http://trac.webkit.org/changeset?new=%d%40trunk'
+                            '%2FLayoutTests%2Fplatform%2Fchromium%2F'
+                            'test_expectations.txt&old=%d%40trunk%2F'
+                            'LayoutTests%2Fplatform%2Fchromium%2F'
+                            'test_expectations.txt') % (new_rev, old_rev)
+      str += '<ul><a href="%s">%s->%s</a>\n' % (link, old_rev, new_rev)
+      str += '<li>%s</li>\n' % author
+      str += '<li>%s</li>\n<ul>' % date
+      for line in target_lines:
+        str += '<li>%s</li>\n' % line
+      str += '</ul></ul>'
+  localtime = time.asctime(time.localtime(time.time()))
+  # TODO(imasaki): remove my name from here.
+  SendEmail('imasaki@chromium.org', 'Kenji Imasaki',
+            [receiver_email_address], ['Layout Test Analyzer Result'],
+            'Layout Test Analyzer Result : ' + localtime, str)
+
+
+def SendEmail(sender_email_address, sender_name, receivers_email_addresses,
+              receivers_names, subject, message):
+    """Send email using localhost's mail server.
+
+    Args:
+        sender_email_address: sender's email address.
+        sender_name: sender's name.
+        receivers_email_addresses: receiver's email addresses.
+        receivers_names:  receiver's names.
+        subject: subject string.
+        message: email message.
+    """
+    whole_message = ''.join([
+        'From: %s<%s>\n' % (sender_name, sender_email_address),
+        'To: %s<%s>\n' % (receivers_names[0],
+                          receivers_email_addresses[0]),
+        'Subject: %s\n' % subject, message])
+
+    try:
+      html_top = """
+        <html>
+        <head></head>
+        <body>
+      """
+      html_bot = """
+        </body>
+        </html>
+      """
+      html = html_top + message + html_bot
+      msg = MIMEMultipart('alternative')
+      msg['Subject'] = subject
+      msg['From'] = sender_email_address
+      msg['To'] = receivers_email_addresses[0]
+      part1 = MIMEText(html, 'html')
+      smtpObj = smtplib.SMTP('localhost')
+      msg.attach(part1)
+      smtpObj.sendmail(sender_email_address, receivers_email_addresses,
+                       msg.as_string())
+      print 'Successfully sent email'
+    except smtplib.SMTPException, e:
+      print 'Authentication failed:', e
+      print 'Error: unable to send email'
+    except (socket.gaierror, socket.error, socket.herror), e:
+      print e
+      print 'Error: unable to send email'
+
+
+def FindLatestTime(time_list):
+  """Find latest time from |time_list|.
+
+  The current status is compared to the status of the latest file in
+  |RESULT_DIR|.
+
+  Args:
+    time_list: a list of time string in the form of '2011-10-23-23'.
+
+  Returns:
+     a string representing latest time among the time_list or None if
+         |time_list| is empty.
+  """
+  if not time_list:
+    return None
+  latest_date = None
+  for t in time_list:
+    item_date = datetime.strptime(t, '%Y-%m-%d-%H')
+    if latest_date == None or latest_date < item_date:
+      latest_date = item_date
+  return latest_date.strftime('%Y-%m-%d-%H')
+
+
+def FindLatestResult(result_dir):
+  """Find the latest result in |result_dir| and read and return them.
+
+  This is used for comparison of analyzer result between current analyzer
+  and most known latest result.
+
+  Args:
+    result_dir: the result directory.
+
+  Returns:
+    a tuple of filename (latest_time) of the and the latest analyzer result.
+  """
+  dirList = os.listdir(result_dir)
+  file_name = FindLatestTime(dirList)
+  file_path = os.path.join(result_dir, file_name)
+  return (file_name, AnalyzerResultMap.Load(file_path))
+
+
+def GetDiffBetweenMaps(map1, map2, lookIntoTestExpectationInfo=False):
+  """Get difference between maps.
+
+  Args:
+    map1: analyzer result map to be compared.
+    map2: analyzer result map to be compared.
+    lookIntoTestExpectationInfo: a boolean to indicate whether to compare
+        test expectation information in addition to just the test case names.
+
+  Returns:
+    a tuple of |name1_list| and |name2_list|. |Name1_list| contains all test
+        name and the test expectation information in |map1| but not in |map2|.
+        |Name2_list| contains all test name and the test expectation
+        information in |map2| but not in |map1|.
+  """
+
+  def GetDiffBetweenMapsHelper(map1, map2, lookIntoTestExpectationInfo):
+    name_list = []
+    for (name, v1) in map1.iteritems():
+      if name in map2:
+        if lookIntoTestExpectationInfo and 'te_info' in v1:
+          list1 = v1['te_info']
+          list2 = map2[name]['te_info']
+          te_diff = [item for item in list1 if not item in list2]
+          if te_diff:
+              name_list.append((name, te_diff))
+      else:
+        name_list.append((name, v1))
+    return name_list
+
+  return (GetDiffBetweenMapsHelper(map1, map2, lookIntoTestExpectationInfo),

[dennis_jeffrey, 2011/08/27 00:04:50]

I think there's a code structure problem here.  This return is actually outside
of any function.  The function defined at line 406 above seems to be empty. 
Could you double-check lines 406-437 and make sure it's correct?

[imasaki1, 2011/08/29 21:32:49]

I thought this is return value of GetDiffBetweenMaps.

It is called in unit tests function and has been tested on my desktop.. Are
there anything I miss here.

[dennis_jeffrey, 2011/08/29 23:52:26]

Yes, I mis-read the code here.  I see now that the function defined at line 422
is an inner function, inside the function defined at line 406.  And the return
at line 436 is the implementation of the outer function (which just calls the
inner function).  So it does look fine as is.  Sorry about that!

+          GetDiffBetweenMapsHelper(map2, map1, lookIntoTestExpectationInfo))