supersize: Parse build.ninja concurrently

tools/binary_size/libsupersize/ninja_parser.py

Index: tools/binary_size/libsupersize/ninja_parser.py
diff --git a/tools/binary_size/libsupersize/ninja_parser.py b/tools/binary_size/libsupersize/ninja_parser.py
index 6486c5e435c95ac754e2ff0958127435c84b54a5..22d55198df1745aa91964c4fd90eba5640ff74e3 100644
--- a/tools/binary_size/libsupersize/ninja_parser.py
+++ b/tools/binary_size/libsupersize/ninja_parser.py
@@ -14,68 +14,21 @@ import re
 _REGEX = re.compile(r'build ([^:]+?\.[ao]): \w+ (.*?)(?: \||\n|$)')
 
 
-class SourceFileMapper(object):
-  def __init__(self, output_directory):
-    self._output_directory = output_directory
-    self._ninja_files_to_parse = ['build.ninja']
-    self._seen_ninja_files = set(('build.ninja',))
-    self._dep_map = {}
+class _SourceMapper(object):
+  def __init__(self, dep_map, parsed_file_count):
+    self._dep_map = dep_map
+    self.parsed_file_count = parsed_file_count
+    self._unmatched_paths = set()
 
-  def _ParseNinja(self, path):
-    with open(os.path.join(self._output_directory, path)) as obj:
-      self._ParseNinjaLines(obj)
-
-  def _ParseNinjaLines(self, lines):
-    dep_map = self._dep_map
-    sub_ninjas = []
-    for line in lines:
-      if line.startswith('subninja '):
-        subpath = line[9:-1]
-        assert subpath not in self._seen_ninja_files, (
-            'Double include of ' + subpath)
-        self._seen_ninja_files.add(subpath)
-        sub_ninjas.append(subpath)
-        continue
-      m = _REGEX.match(line)
-      if m:
-        output, srcs = m.groups()
-        output = output.replace('\\ ', ' ')
-        assert output not in dep_map, 'Duplicate output: ' + output
-        if output[-1] == 'o':
-          dep_map[output] = srcs.replace('\\ ', ' ')
-        else:
-          srcs = srcs.replace('\\ ', '\b')
-          obj_paths = (s.replace('\b', ' ') for s in srcs.split(' '))
-          dep_map[output] = {os.path.basename(p): p for p in obj_paths}
-
-    # Add reversed so that the first on encoundered is at the top of the stack.
-    self._ninja_files_to_parse.extend(reversed(sub_ninjas))
-
-  def _Lookup(self, path):
-    """Looks for |path| within self._dep_map.
-
-    If not found, continues to parse subninjas until it is found or there are no
-    more subninjas.
-    """
-    ret = self._dep_map.get(path)
-    while not ret and self._ninja_files_to_parse:
-      self._ParseNinja(self._ninja_files_to_parse.pop())
-      ret = self._dep_map.get(path)
-    return ret
-
-  def FindSourceForPath(self, path):
-    """Returns the source path for the given object path (or None if not found).
-
-    Paths for objects within archives should be in the format: foo/bar.a(baz.o)
-    """
+  def _FindSourceForPathInternal(self, path):
     if not path.endswith(')'):
-      return self._Lookup(path)
+      return self._dep_map.get(path)
 
     # foo/bar.a(baz.o)
     start_idx = path.index('(')
     lib_name = path[:start_idx]
     obj_name = path[start_idx + 1:-1]
-    by_basename = self._Lookup(lib_name)
+    by_basename = self._dep_map.get(lib_name)
     if not by_basename:
       return None
     obj_path = by_basename.get(obj_name)
@@ -83,7 +36,56 @@ class SourceFileMapper(object):
       # Found the library, but it doesn't list the .o file.
       logging.warning('no obj basename for %s', path)
       return None
-    return self._Lookup(obj_path)
+    return self._dep_map.get(obj_path)
+
+  def FindSourceForPath(self, path):
+    """Returns the source path for the given object path (or None if not found).
+
+    Paths for objects within archives should be in the format: foo/bar.a(baz.o)
+    """
+    ret = self._FindSourceForPathInternal(path)
+    if not ret and path not in self._unmatched_paths:
+      if self.unmatched_paths_count < 10:
+        logging.warning('Could not find source path for %s', path)
+      self._unmatched_paths.add(path)
+    return ret
+
+  @property
+  def unmatched_paths_count(self):
+    return len(self._unmatched_paths)
+
+
+def _ParseOneFile(lines, dep_map):
+  sub_ninjas = []
+  for line in lines:
+    if line.startswith('subninja '):
+      sub_ninjas.append(line[9:-1])
+      continue
+    m = _REGEX.match(line)
+    if m:
+      output, srcs = m.groups()
+      output = output.replace('\\ ', ' ')
+      assert output not in dep_map, 'Duplicate output: ' + output
+      if output[-1] == 'o':
+        dep_map[output] = srcs.replace('\\ ', ' ')
+      else:
+        srcs = srcs.replace('\\ ', '\b')
+        obj_paths = (s.replace('\b', ' ') for s in srcs.split(' '))
+        dep_map[output] = {os.path.basename(p): p for p in obj_paths}
+  return sub_ninjas
+
+
+def Parse(output_directory):
+  to_parse = ['build.ninja']
+  seen_paths = set(to_parse)
+  dep_map = {}
+  while to_parse:
+    path = os.path.join(output_directory, to_parse.pop())
+    with open(path) as obj:
+      sub_ninjas = _ParseOneFile(obj, dep_map)
+    for subpath in sub_ninjas:
+      assert subpath not in seen_paths, 'Double include of ' + subpath
+      seen_paths.add(subpath)
+    to_parse.extend(sub_ninjas)
 
-  def GetParsedFileCount(self):
-    return len(self._seen_ninja_files)
+  return _SourceMapper(dep_map, len(seen_paths))