src/chrome/common/extensions/docs/server2/path_canonicalizer.py

   1 # Copyright 2013 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 from collections import defaultdict
   6 import posixpath
   7
   8 from future import Future
   9 from path_util import SplitParent
  10 from special_paths import SITE_VERIFICATION_FILE
  11
  12 def _Normalize(file_name, splittext=False):
  13   normalized = file_name
  14   if splittext:
  15     normalized = posixpath.splitext(file_name)[0]
  16   normalized = normalized.replace('.', '').replace('-', '').replace('_', '')
  17   return normalized.lower()
  18
  19 def _CommonNormalizedPrefix(first_file, second_file):
  20   return posixpath.commonprefix((_Normalize(first_file),
  21                                  _Normalize(second_file)))
  22
  23
  24 class PathCanonicalizer(object):
  25   '''Transforms paths into their canonical forms. Since the docserver has had
  26   many incarnations - e.g. there didn't use to be apps/ - there may be old
  27   paths lying around the webs. We try to redirect those to where they are now.
  28   '''
  29   def __init__(self,
  30                file_system,
  31                object_store_creator,
  32                strip_extensions):
  33     # |strip_extensions| is a list of file extensions (e.g. .html) that should
  34     # be stripped for a path's canonical form.
  35     self._cache = object_store_creator.Create(
  36         PathCanonicalizer, category=file_system.GetIdentity())
  37     self._file_system = file_system
  38     self._strip_extensions = strip_extensions
  39
  40   def _LoadCache(self):
  41     def load(cached):
  42       # |canonical_paths| is the pre-calculated set of canonical paths.
  43       # |simplified_paths_map| is a lazily populated mapping of simplified file
  44       # names to a list of full paths that contain them. For example,
  45       #  - browseraction: [extensions/browserAction.html]
  46       #  - storage: [apps/storage.html, extensions/storage.html]
  47       canonical_paths, simplified_paths_map = (
  48           cached.get('canonical_paths'), cached.get('simplified_paths_map'))
  49
  50       if canonical_paths is None:
  51         assert simplified_paths_map is None
  52         canonical_paths = set()
  53         simplified_paths_map = defaultdict(list)
  54         for base, dirs, files in self._file_system.Walk(''):
  55           for path in dirs + files:
  56             path_without_ext, ext = posixpath.splitext(path)
  57             canonical_path = posixpath.join(base, path_without_ext)
  58             if (ext not in self._strip_extensions or
  59                 path == SITE_VERIFICATION_FILE):
  60               canonical_path += ext
  61             canonical_paths.add(canonical_path)
  62             simplified_paths_map[_Normalize(path, splittext=True)].append(
  63                 canonical_path)
  64         # Store |simplified_paths_map| sorted. Ties in length are broken by
  65         # taking the shortest, lexicographically smallest path.
  66         for path_list in simplified_paths_map.itervalues():
  67           path_list.sort(key=lambda p: (len(p), p))
  68         self._cache.SetMulti({
  69           'canonical_paths': canonical_paths,
  70           'simplified_paths_map': simplified_paths_map,
  71         })
  72       else:
  73         assert simplified_paths_map is not None
  74
  75       return canonical_paths, simplified_paths_map
  76     return self._cache.GetMulti(('canonical_paths',
  77                                  'simplified_paths_map')).Then(load)
  78
  79
  80   def Canonicalize(self, path):
  81     '''Returns the canonical path for |path|.
  82     '''
  83     canonical_paths, simplified_paths_map = self._LoadCache().Get()
  84
  85     # Path may already be the canonical path.
  86     if path in canonical_paths:
  87       return path
  88
  89     # Path not found. Our single heuristic: find |base| in the directory
  90     # structure with the longest common prefix of |path|.
  91     _, base = SplitParent(path)
  92
  93     # Paths with a non-extension dot separator lose information in
  94     # _SimplifyFileName, so we try paths both with and without the dot to
  95     # maximize the possibility of finding the right path.
  96     potential_paths = (
  97         simplified_paths_map.get(_Normalize(base), []) +
  98         simplified_paths_map.get(_Normalize(base, splittext=True), []))
  99
 100     if potential_paths == []:
 101       # There is no file with anything close to that name.
 102       return path
 103
 104     # The most likely canonical file is the one with the longest common prefix
 105     # with |path|. This is slightly weaker than it could be; |path| is
 106     # compared without symbols, not the simplified form of |path|,
 107     # which may matter.
 108     max_prefix = potential_paths[0]
 109     max_prefix_length = len(_CommonNormalizedPrefix(max_prefix, path))
 110     for path_for_file in potential_paths[1:]:
 111       prefix_length = len(_CommonNormalizedPrefix(path_for_file, path))
 112       if prefix_length > max_prefix_length:
 113         max_prefix, max_prefix_length = path_for_file, prefix_length
 114
 115     return max_prefix
 116
 117   def Refresh(self):
 118     return self._LoadCache()