1#!/usr/bin/env python
2#
3# Copyright 2013 The Chromium Authors. All rights reserved.
4# Use of this source code is governed by a BSD-style license that can be
5# found in the LICENSE file.
6
7"""Symbolizes stack traces generated by Chromium for Android.
8
9Sample usage:
10  adb logcat chromium:V | symbolize.py
11"""
12
13import os
14import re
15import sys
16
17from pylib import constants
18
19# Uses symbol.py from third_party/android_platform, not python's.
20sys.path.insert(0,
21                os.path.join(constants.DIR_SOURCE_ROOT,
22                            'third_party/android_platform/development/scripts'))
23import symbol
24
25# Sample output from base/debug/stack_trace_android.cc
26#00 0x693cd34f /path/to/some/libfoo.so+0x0007434f
27TRACE_LINE = re.compile('(?P<frame>\#[0-9]+ 0x[0-9a-f]{8,8}) '
28                        '(?P<lib>[^+]+)\+0x(?P<addr>[0-9a-f]{8,8})')
29
30class Symbolizer(object):
31  def __init__(self, output):
32    self._output = output
33
34  def write(self, data):
35    while True:
36      match = re.search(TRACE_LINE, data)
37      if not match:
38        self._output.write(data)
39        break
40
41      frame = match.group('frame')
42      lib = match.group('lib')
43      addr = match.group('addr')
44
45      # TODO(scherkus): Doing a single lookup per line is pretty slow,
46      # especially with larger libraries. Consider caching strategies such as:
47      # 1) Have Python load the libraries and do symbol lookups instead of
48      #    calling out to addr2line each time.
49      # 2) Have Python keep multiple addr2line instances open as subprocesses,
50      #    piping addresses and reading back symbols as we find them
51      # 3) Read ahead the entire stack trace until we find no more, then batch
52      #    the symbol lookups.
53      #
54      # TODO(scherkus): These results are memoized, which could result in
55      # incorrect lookups when running this script on long-lived instances
56      # (e.g., adb logcat) when doing incremental development. Consider clearing
57      # the cache when modification timestamp of libraries change.
58      sym = symbol.SymbolInformation(lib, addr, False)[0][0]
59
60      if not sym:
61        post = match.end('addr')
62        self._output.write(data[:post])
63        data = data[post:]
64        continue
65
66      pre = match.start('frame')
67      post = match.end('addr')
68
69      self._output.write(data[:pre])
70      self._output.write(frame)
71      self._output.write(' ')
72      self._output.write(sym)
73
74      data = data[post:]
75
76  def flush(self):
77    self._output.flush()
78
79
80def main():
81  symbolizer = Symbolizer(sys.stdout)
82  for line in sys.stdin:
83    symbolizer.write(line)
84  symbolizer.flush()
85
86
87if __name__ == '__main__':
88  main()
89