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
  20 class PathCanonicalizer(object):
  21   '''Transforms paths into their canonical forms. Since the docserver has had
  22   many incarnations - e.g. there didn't use to be apps/ - there may be old
  23   paths lying around the webs. We try to redirect those to where they are now.
  24   '''
  25   def __init__(self,
  26                file_system,
  27                object_store_creator,
  28                strip_extensions):
  29     # |strip_extensions| is a list of file extensions (e.g. .html) that should
  30     # be stripped for a path's canonical form.
  31     self._cache = object_store_creator.Create(
  32         PathCanonicalizer, category=file_system.GetIdentity())
  33     self._file_system = file_system
  34     self._strip_extensions = strip_extensions
  35
  36   def _LoadCache(self):
  37     cached_future = self._cache.GetMulti(('canonical_paths',
  38                                           'simplified_paths_map'))
  39
  40     def resolve():
  41       # |canonical_paths| is the pre-calculated set of canonical paths.
  42       # |simplified_paths_map| is a lazily populated mapping of simplified file
  43       # names to a list of full paths that contain them. For example,
  44       #  - browseraction: [extensions/browserAction.html]
  45       #  - storage: [apps/storage.html, extensions/storage.html]
  46       cached = cached_future.Get()
  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
  77     return Future(callback=resolve)
  78
  79   def Canonicalize(self, path):
  80     '''Returns the canonical path for |path|.
  81     '''
  82     canonical_paths, simplified_paths_map = self._LoadCache().Get()
  83
  84     # Path may already be the canonical path.
  85     if path in canonical_paths:
  86       return path
  87
  88     # Path not found. Our single heuristic: find |base| in the directory
  89     # structure with the longest common prefix of |path|.
  90     _, base = SplitParent(path)
  91
  92     # Paths with a non-extension dot separator lose information in
  93     # _SimplifyFileName, so we try paths both with and without the dot to
  94     # maximize the possibility of finding the right path.
  95     potential_paths = (
  96         simplified_paths_map.get(_Normalize(base), []) +
  97         simplified_paths_map.get(_Normalize(base, splittext=True), []))
  98
  99     if potential_paths == []:
 100       # There is no file with anything close to that name.
 101       return path
 102
 103     # The most likely canonical file is the one with the longest common prefix
 104     # with |path|. This is slightly weaker than it could be; |path| is
 105     # compared without symbols, not the simplified form of |path|,
 106     # which may matter.
 107     max_prefix = potential_paths[0]
 108     max_prefix_length = len(posixpath.commonprefix((max_prefix, path)))
 109     for path_for_file in potential_paths[1:]:
 110       prefix_length = len(posixpath.commonprefix((path_for_file,
 111           _Normalize(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 Cron(self):
 118     return self._LoadCache()