Add scripts to list or trace all test cases in a gtest executable.

tools/isolate/trace_inputs.py

 #!/usr/bin/env python
 # coding=utf-8
 # Copyright (c) 2012 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.
 
 """Runs strace or dtrace on a test and processes the logs to extract the
 dependencies from the source tree.
 
 Automatically extracts directories where all the files are used to make the
@@ skipping 183 matching lines @@
 
 
 def posix_relpath(path, root):
   """posix.relpath() that keeps trailing slash."""
   out = posixpath.relpath(path, root)
   if path.endswith('/'):
     out += '/'
   return out
 
 
+def cleanup_path(x):
+  """Cleans up a relative path. Converts any os.path.sep to '/' on Windows."""
+  if x:
+    x = x.rstrip(os.path.sep).replace(os.path.sep, '/')
+  if x == '.':
+    x = ''
+  if x:
+    x += '/'
+  return x
+
+
 class Strace(object):
   """strace implies linux."""
   IGNORED = (
     '/bin',
     '/dev',
     '/etc',
     '/lib',
     '/proc',
     '/sys',
     '/tmp',
@@ skipping 158 matching lines @@
             '_handle_file(%d, %s) -> %s' % (pid, old_filepath, filepath))
       else:
         logging.debug('_handle_file(%d, %s)' % (pid, filepath))
       if filepath not in self.files and filepath not in self.non_existent:
         if os.path.isfile(filepath):
           self.files.add(filepath)
         else:
           self.non_existent.add(filepath)
 
   @classmethod
-  def gen_trace(cls, cmd, cwd, logname):
+  def gen_trace(cls, cmd, cwd, logname, output):
     """Runs strace on an executable."""
-    logging.info('gen_trace(%s, %s, %s)' % (cmd, cwd, logname))
-    silent = not isEnabledFor(logging.INFO)
+    logging.info('gen_trace(%s, %s, %s, %s)' % (cmd, cwd, logname, output))
     stdout = stderr = None
-    if silent:
-      stdout = stderr = subprocess.PIPE
+    if output:
+      stdout = subprocess.PIPE
+      stderr = subprocess.STDOUT
     traces = ','.join(cls.Context.traces())
     trace_cmd = ['strace', '-f', '-e', 'trace=%s' % traces, '-o', logname]
     child = subprocess.Popen(
-        trace_cmd + cmd, cwd=cwd, stdout=stdout, stderr=stderr)
-    out, err = child.communicate()
+        trace_cmd + cmd,
+        cwd=cwd,
+        stdin=subprocess.PIPE,
+        stdout=stdout,
+        stderr=stderr)
+    out = child.communicate()[0]
     # Once it's done, inject a chdir() call to cwd to be able to reconstruct
     # the full paths.
     # TODO(maruel): cwd should be saved at each process creation, so forks needs
     # to be traced properly.
     if os.path.isfile(logname):
       with open(logname) as f:
         content = f.read()
       with open(logname, 'w') as f:
         pid = content.split(' ', 1)[0]
         f.write('%s chdir("%s") = 0\n' % (pid, cwd))
         f.write(content)
-
-    if child.returncode != 0:
-      print 'Failure: %d' % child.returncode
-      # pylint: disable=E1103
-      if out:
-        print ''.join(out.splitlines(True)[-100:])
-      if err:
-        print ''.join(err.splitlines(True)[-100:])
-    return child.returncode
+    return child.returncode, out
 
   @classmethod
   def parse_log(cls, filename, blacklist):
     """Processes a strace log and returns the files opened and the files that do
     not exist.
 
     It does not track directories.
 
     Most of the time, files that do not exist are temporary test files that
     should be put in /tmp instead. See http://crbug.com/116251
@@ skipping 262 matching lines @@
         if os.path.isfile(filepath):
           self.files.add(filepath)
         else:
           self.non_existent.add(filepath)
 
     @staticmethod
     def _handle_ignored(_ppid, pid, function, args, result):
       logging.debug('%d %s(%s) = %s' % (pid, function, args, result))
 
   @classmethod
-  def gen_trace(cls, cmd, cwd, logname):
+  def gen_trace(cls, cmd, cwd, logname, output):
     """Runs dtrace on an executable."""
-    logging.info('gen_trace(%s, %s, %s)' % (cmd, cwd, logname))
-    silent = not isEnabledFor(logging.INFO)
+    logging.info('gen_trace(%s, %s, %s, %s)' % (cmd, cwd, logname, output))
     logging.info('Running: %s' % cmd)
     signal = 'Go!'
     logging.debug('Our pid: %d' % os.getpid())
 
     # Part 1: start the child process.
     stdout = stderr = None
-    if silent:
-      stdout = stderr = subprocess.PIPE
+    if output:
+      stdout = subprocess.PIPE
+      stderr = subprocess.STDOUT
     child_cmd = [
       sys.executable, os.path.join(BASE_DIR, 'trace_child_process.py'),
     ]
     child = subprocess.Popen(
         child_cmd + cmd,
         stdin=subprocess.PIPE,
         stdout=stdout,
         stderr=stderr,
         cwd=cwd)
     logging.debug('Started child pid: %d' % child.pid)
@@ skipping 20 matching lines @@
     # ready.
     with open(logname, 'r') as logfile:
       while 'dtrace_BEGIN' not in logfile.readline():
         if dtrace.poll() is not None:
           break
 
     try:
       # Part 4: We can now tell our child to go.
       # TODO(maruel): Another pipe than stdin could be used instead. This would
       # be more consistent with the other tracing methods.
-      out, err = child.communicate(signal)
+      out = child.communicate(signal)[0]
 
       dtrace.wait()
       if dtrace.returncode != 0:
         print 'dtrace failure: %d' % dtrace.returncode
         with open(logname) as logfile:
           print ''.join(logfile.readlines()[-100:])
         # Find a better way.
         os.remove(logname)
       else:
         # Short the log right away to simplify our life. There isn't much
         # advantage in keeping it out of order.
         cls._sort_log(logname)
-      if child.returncode != 0:
-        print 'Failure: %d' % child.returncode
-        # pylint: disable=E1103
-        if out:
-          print ''.join(out.splitlines(True)[-100:])
-        if err:
-          print ''.join(err.splitlines(True)[-100:])
     except KeyboardInterrupt:
       # Still sort when testing.
       cls._sort_log(logname)
       raise
 
-    return dtrace.returncode or child.returncode
+    return dtrace.returncode or child.returncode, out
 
   @classmethod
   def parse_log(cls, filename, blacklist):
     """Processes a dtrace log and returns the files opened and the files that do
     not exist.
 
     It does not track directories.
 
     Most of the time, files that do not exist are temporary test files that
     should be put in /tmp instead. See http://crbug.com/116251
@@ skipping 223 matching lines @@
 
     # Also add their short path name equivalents.
     for i in list(self.IGNORED):
       self.IGNORED.add(GetShortPathName(i))
 
     # Add this one last since it has no short path name equivalent.
     self.IGNORED.add('\\systemroot')
     self.IGNORED = tuple(sorted(self.IGNORED))
 
   @classmethod
-  def gen_trace(cls, cmd, cwd, logname):
-    logging.info('gen_trace(%s, %s, %s)' % (cmd, cwd, logname))
+  def gen_trace(cls, cmd, cwd, logname, output):
+    logging.info('gen_trace(%s, %s, %s, %s)' % (cmd, cwd, logname, output))
     # Use "logman -?" for help.
 
     etl = logname + '.etl'
 
-    silent = not isEnabledFor(logging.INFO)
     stdout = stderr = None
-    if silent:
-      stdout = stderr = subprocess.PIPE
+    if output:
+      stdout = subprocess.PIPE
+      stderr = subprocess.STDOUT
 
     # 1. Start the log collection. Requires administrative access. logman.exe is
     # synchronous so no need for a "warmup" call.
     # 'Windows Kernel Trace' is *localized* so use its GUID instead.
     # The GUID constant name is SystemTraceControlGuid. Lovely.
     cmd_start = [
       'logman.exe',
       'start',
       'NT Kernel Logger',
       '-p', '{9e814aad-3204-11d2-9a82-006008a86939}',
       '(process,img,file,fileio)',
       '-o', etl,
       '-ets',  # Send directly to kernel
     ]
     logging.debug('Running: %s' % cmd_start)
-    subprocess.check_call(cmd_start, stdout=stdout, stderr=stderr)
+    subprocess.check_call(
+        cmd_start,
+        stdin=subprocess.PIPE,
+        stdout=subprocess.PIPE,
+        stderr=subprocess.STDOUT)
 
     # 2. Run the child process.
     logging.debug('Running: %s' % cmd)
     try:
-      child = subprocess.Popen(cmd, cwd=cwd, stdout=stdout, stderr=stderr)
-      out, err = child.communicate()
+      child = subprocess.Popen(
+          cmd, cwd=cwd, stdin=subprocess.PIPE, stdout=stdout, stderr=stderr)
+      out = child.communicate()[0]
     finally:
       # 3. Stop the log collection.
       cmd_stop = [
         'logman.exe',
         'stop',
         'NT Kernel Logger',
         '-ets',  # Send directly to kernel
       ]
       logging.debug('Running: %s' % cmd_stop)
-      subprocess.check_call(cmd_stop, stdout=stdout, stderr=stderr)
+      subprocess.check_call(
+          cmd_stop,
+          stdin=subprocess.PIPE,
+          stdout=subprocess.PIPE,
+          stderr=subprocess.STDOUT)
 
     # 4. Convert the traces to text representation.
     # Use "tracerpt -?" for help.
     LOCALE_INVARIANT = 0x7F
     windll.kernel32.SetThreadLocale(LOCALE_INVARIANT)
     cmd_convert = [
       'tracerpt.exe',
       '-l', etl,
       '-o', logname,
       '-gmt',  # Use UTC
@@ skipping 10 matching lines @@
       cmd_convert.extend(['-of', 'CSV'])
     elif logformat == 'csv_utf16':
       # This causes it to use UTF-16, which doubles the log size but ensures the
       # log is readable for non-ASCII characters.
       cmd_convert.extend(['-of', 'CSV', '-en', 'Unicode'])
     elif logformat == 'xml':
       cmd_convert.extend(['-of', 'XML'])
     else:
       assert False, logformat
     logging.debug('Running: %s' % cmd_convert)
-    subprocess.check_call(cmd_convert, stdout=stdout, stderr=stderr)
+    subprocess.check_call(
+        cmd_convert, stdin=subprocess.PIPE, stdout=stdout, stderr=stderr)
 
-    if child.returncode != 0:
-      print 'Failure: %d' % child.returncode
-      # pylint: disable=E1103
-      if out:
-        print ''.join(out.splitlines(True)[-100:])
-      if err:
-        print ''.join(err.splitlines(True)[-100:])
-    return child.returncode
+    return child.returncode, out
 
   @classmethod
   def parse_log(cls, filename, blacklist):
     logging.info('parse_log(%s, %s)' % (filename, blacklist))
 
     # Auto-detect the log format
     with open(filename, 'rb') as f:
       hdr = f.read(2)
       assert len(hdr) == 2
       if hdr == '<E':
@@ skipping 149 matching lines @@
       elif item in (True, False, None):
         stdout.write('%s\n' % item)
       else:
         assert False, item
 
   stdout.write('{\n')
   loop_dict('  ', variables)
   stdout.write('}\n')
 
 
+def get_api():
+  flavor = get_flavor()
+  if flavor == 'linux':
+    return Strace()
+  elif flavor == 'mac':
+    return Dtrace()
+  elif sys.platform == 'win32':
+    return LogmanTrace()
+  else:
+    print >> sys.stderr, 'Unsupported platform %s' % sys.platform
+    sys.exit(1)
+
+
+def get_blacklist(api):
+  git_path = os.path.sep + '.git' + os.path.sep
+  svn_path = os.path.sep + '.svn' + os.path.sep
+  return lambda f: (
+      f.startswith(api.IGNORED) or
+      f.endswith('.pyc') or
+      git_path in f or
+      svn_path in f)
+
+
+def generate_dict(files, cwd_dir, product_dir):
+  """Converts the list of files into a .isolate dictionary.
+
+  Both cwd_dir and product_dir are relative directories to root_dir.
+  """
+  cwd_dir = cleanup_path(cwd_dir)
+  product_dir = cleanup_path(product_dir)
+
+  def fix(f):
+    """Bases the file on the most restrictive variable."""
+    logging.debug('fix(%s)' % f)
+    # Important, GYP stores the files with / and not \.
+    f = f.replace(os.path.sep, '/')
+    if product_dir and f.startswith(product_dir):
+      return '<(PRODUCT_DIR)/%s' % f[len(product_dir):]
+    else:
+      # cwd_dir is usually the directory containing the gyp file. It may be
+      # empty if the whole directory containing the gyp file is needed.
+      return posix_relpath(f, cwd_dir) or './'
+
+  corrected = [fix(f) for f in files]
+  tracked = [f for f in corrected if not f.endswith('/') and ' ' not in f]
+  untracked = [f for f in corrected if f.endswith('/') or ' ' in f]
+  variables = {}
+  if tracked:
+    variables[KEY_TRACKED] = tracked
+  if untracked:
+    variables[KEY_UNTRACKED] = untracked
+  return variables
+
+
+def trace(logfile, cmd, cwd, api, output):
+  """Traces an executable. Returns returncode, output."""
+  cmd = fix_python_path(cmd)
+  assert os.path.isabs(cmd[0]), cmd[0]
+  if os.path.isfile(logfile):
+    os.remove(logfile)
+  return api.gen_trace(cmd, cwd, logfile, output)
+
+
+def load_trace(logfile, root_dir, api):

[Roger Tawa OOO till Jul 10th, 2012/05/11 14:17:07]

Please add docs for this function, and arg docs for the new functions above too.

+  files, non_existent = api.parse_log(logfile, get_blacklist(api))
+  expected, unexpected = relevant_files(
+      files, root_dir.rstrip(os.path.sep) + os.path.sep)
+  # In case the file system is case insensitive.
+  expected = sorted(set(get_native_path_case(root_dir, f) for f in expected))
+  simplified = extract_directories(expected, root_dir)
+  return files, expected, unexpected, non_existent, simplified
+
+
 def trace_inputs(logfile, cmd, root_dir, cwd_dir, product_dir, force_trace):
   """Tries to load the logs if available. If not, trace the test.
 
   Symlinks are not processed at all.
 
   Arguments:
   - logfile:     Absolute path to the OS-specific trace.
   - cmd:         Command list to run.
   - root_dir:    Base directory where the files we care about live.
   - cwd_dir:     Cwd to use to start the process, relative to the root_dir
                  directory.
   - product_dir: Directory containing the executables built by the build
                  process, relative to the root_dir directory. It is used to
                  properly replace paths with <(PRODUCT_DIR) for gyp output.
   - force_trace: Will force to trace unconditionally even if a trace already
                  exist.
   """
   logging.debug(
       'trace_inputs(%s, %s, %s, %s, %s, %s)' % (
         logfile, cmd, root_dir, cwd_dir, product_dir, force_trace))
 
+  def print_if(txt):
+    if cwd_dir is None:
+      print txt
+
   # It is important to have unambiguous path.
   assert os.path.isabs(root_dir), root_dir
   assert os.path.isabs(logfile), logfile
   assert not cwd_dir or not os.path.isabs(cwd_dir), cwd_dir
   assert not product_dir or not os.path.isabs(product_dir), product_dir
 
-  cmd = fix_python_path(cmd)
-  assert (
-      (os.path.isfile(logfile) and not force_trace) or os.path.isabs(cmd[0])
-      ), cmd[0]
-
+  api = get_api()
   # Resolve any symlink
   root_dir = os.path.realpath(root_dir)
-
-  def print_if(txt):
-    if cwd_dir is None:
-      print(txt)
-
-  flavor = get_flavor()
-  if flavor == 'linux':
-    api = Strace()
-  elif flavor == 'mac':
-    api = Dtrace()
-  elif sys.platform == 'win32':
-    api = LogmanTrace()
-  else:
-    print >> sys.stderr, 'Unsupported platform %s' % sys.platform
-    return 1
-
   if not os.path.isfile(logfile) or force_trace:
-    if os.path.isfile(logfile):
-      os.remove(logfile)
     print_if('Tracing... %s' % cmd)
-    cwd = root_dir
     # Use the proper relative directory.
-    if cwd_dir:
-      cwd = os.path.join(cwd, cwd_dir)
-    returncode = api.gen_trace(cmd, cwd, logfile)
+    cwd = root_dir if not cwd_dir else os.path.join(root_dir, cwd_dir)
+    silent = not isEnabledFor(logging.WARNING)
+    returncode, _ = trace(logfile, cmd, cwd, api, silent)
     if returncode and not force_trace:
       return returncode
 
-  git_path = os.path.sep + '.git' + os.path.sep
-  svn_path = os.path.sep + '.svn' + os.path.sep
-  def blacklist(f):
-    """Strips ignored paths."""
-    return (
-        f.startswith(api.IGNORED) or
-        f.endswith('.pyc') or
-        git_path in f or
-        svn_path in f)
-
   print_if('Loading traces... %s' % logfile)
-  files, non_existent = api.parse_log(logfile, blacklist)
+  files, expected, unexpected, non_existent, simplified = load_trace(
+      logfile, root_dir, api)
 
   print_if('Total: %d' % len(files))
   print_if('Non existent: %d' % len(non_existent))
   for f in non_existent:
     print_if('  %s' % f)
-
-  expected, unexpected = relevant_files(
-      files, root_dir.rstrip(os.path.sep) + os.path.sep)
   if unexpected:
     print_if('Unexpected: %d' % len(unexpected))
     for f in unexpected:
       print_if('  %s' % f)
-
-  # In case the file system is case insensitive.
-  expected = sorted(set(get_native_path_case(root_dir, f) for f in expected))
-
-  simplified = extract_directories(expected, root_dir)
   print_if('Interesting: %d reduced to %d' % (len(expected), len(simplified)))
   for f in simplified:
     print_if('  %s' % f)
 
   if cwd_dir is not None:
-    def cleanuppath(x):
-      """Cleans up a relative path. Converts any os.path.sep to '/' on Windows.
-      """
-      if x:
-        x = x.rstrip(os.path.sep).replace(os.path.sep, '/')
-      if x == '.':
-        x = ''
-      if x:
-        x += '/'
-      return x
-
-    # Both are relative directories to root_dir.
-    cwd_dir = cleanuppath(cwd_dir)
-    product_dir = cleanuppath(product_dir)
-
-    def fix(f):
-      """Bases the file on the most restrictive variable."""
-      logging.debug('fix(%s)' % f)
-      # Important, GYP stores the files with / and not \.
-      f = f.replace(os.path.sep, '/')
-
-      if product_dir and f.startswith(product_dir):
-        return '<(PRODUCT_DIR)/%s' % f[len(product_dir):]
-      else:
-        # cwd_dir is usually the directory containing the gyp file. It may be
-        # empty if the whole directory containing the gyp file is needed.
-        return posix_relpath(f, cwd_dir) or './'
-
-    corrected = [fix(f) for f in simplified]
-    tracked = [f for f in corrected if not f.endswith('/') and ' ' not in f]
-    untracked = [f for f in corrected if f.endswith('/') or ' ' in f]
-    variables = {}
-    if tracked:
-      variables[KEY_TRACKED] = tracked
-    if untracked:
-      variables[KEY_UNTRACKED] = untracked
     value = {
       'conditions': [
-        ['OS=="%s"' % flavor, {
-          'variables': variables,
+        ['OS=="%s"' % get_flavor(), {
+          'variables': generate_dict(simplified, cwd_dir, product_dir),
         }],
       ],
     }
     pretty_print(value, sys.stdout)
   return 0
 
 
 def main():
   parser = optparse.OptionParser(
       usage='%prog <options> [cmd line...]')
@@ skipping 40 matching lines @@
       os.path.abspath(options.log),
       args,
       options.root_dir,
       options.cwd,
       options.product_dir,
       options.force)
 
 
 if __name__ == '__main__':
   sys.exit(main())