Docserver: Make the .html extension unnecessary for content pages, for example,

chrome/common/extensions/docs/server2/content_provider.py

 # Copyright 2013 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.
 
 import mimetypes
 import posixpath
 
 from compiled_file_system import SingleFile
 from directory_zipper import DirectoryZipper
 from docs_server_utils import ToUnicode
 from file_system import FileNotFoundError
 from future import Gettable, Future
 from path_canonicalizer import PathCanonicalizer
 from path_util import AssertIsValid, ToDirectory
+from special_paths import WEBMASTER_PROOF
 from third_party.handlebar import Handlebar
 from third_party.markdown import markdown
 
 
 _MIMETYPE_OVERRIDES = {
   # SVG is not supported by mimetypes.guess_type on AppEngine.
   '.svg': 'image/svg+xml',
 }
 
 
 class ContentAndType(object):
   '''Return value from ContentProvider.GetContentAndType.
   '''
 
   def __init__(self, content, content_type):
     self.content = content
     self.content_type = content_type
 
 
 class ContentProvider(object):
   '''Returns file contents correctly typed for their content-types (in the HTTP
   sense). Content-type is determined from Python's mimetype library which
   guesses based on the file extension.
 
   Typically the file contents will be either str (for binary content) or
   unicode (for text content). However, HTML files *may* be returned as
-  Handlebar templates (if supports_templates is True on construction), in which
-  case the caller will presumably want to Render them.
+  Handlebar templates (if |supports_templates| is True on construction), in
+  which case the caller will presumably want to Render them.
+
+  Zip file are automatically created and returned for .zip file extensions if
+  |supports_zip| is True.
+
+  |default_extensions| is a list of file extensions which are queried when no
+  file extension is given to GetCanonicalPath/GetContentAndType.  Typically
+  this will include .html.
   '''
 
   def __init__(self,
                name,
                compiled_fs_factory,
                file_system,
+               object_store_creator,
+               default_extensions=(),
                supports_templates=False,
                supports_zip=False):
     # Public.
     self.name = name
     self.file_system = file_system
     # Private.
     self._content_cache = compiled_fs_factory.Create(file_system,
                                                      self._CompileContent,
                                                      ContentProvider)
-    self._path_canonicalizer = PathCanonicalizer(compiled_fs_factory,
-                                                 file_system)
+    self._path_canonicalizer = PathCanonicalizer(file_system,
+                                                 object_store_creator,
+                                                 default_extensions)
+    self._default_extensions = default_extensions
     self._supports_templates = supports_templates
     if supports_zip:
       self._directory_zipper = DirectoryZipper(compiled_fs_factory, file_system)
     else:
       self._directory_zipper = None
 
   @SingleFile
   def _CompileContent(self, path, text):
     assert text is not None, path
     _, ext = posixpath.splitext(path)
@@ skipping 13 matching lines @@
       content = ToUnicode(text)
       if self._supports_templates:
         content = Handlebar(content, name=path)
     elif (mimetype.startswith('text/') or
           mimetype in ('application/javascript', 'application/json')):
       content = ToUnicode(text)
     else:
       content = text
     return ContentAndType(content, mimetype)
 
-  def _MaybeMarkdown(self, path):
-    base, ext = posixpath.splitext(path)
-    if ext != '.html':
-      return path
-    if self.file_system.Exists(path).Get():
-      return path
-    as_md = base + '.md'
-    if self.file_system.Exists(as_md).Get():
-      return as_md
-    return path
-
   def GetCanonicalPath(self, path):
     '''Gets the canonical location of |path|. This class is tolerant of
     spelling errors and missing files that are in other directories, and this
     returns the correct/canonical path for those.
 
     For example, the canonical path of "browseraction" is probably
     "extensions/browserAction.html".
+
+    Note that the canonical path is relative to this content provider i.e.
+    given relative to |path|. It does not add the "serveFrom" prefix which
+    would have been pulled out in ContentProviders, callers must do that
+    themselves.
     '''
     AssertIsValid(path)
-    _, ext = posixpath.splitext(path)
-    if not ext or ext == '.html':
-      return self._path_canonicalizer.Canonicalize(path)
-    return path
+    base, ext = posixpath.splitext(path)
+    if self._directory_zipper and ext == '.zip':
+      # The canonical location of zip files is the canonical location of the
+      # directory with that zip suffix.

[Yoyo Zhou, 2014/02/12 22:14:05]

This "with" is confusing.

[not at google - send to devlin, 2014/02/13 03:34:34]

Done.

+      return self._path_canonicalizer.Canonicalize(base + '/').rstrip('/') + ext
+    return self._path_canonicalizer.Canonicalize(path)
 
   def GetContentAndType(self, path):
     '''Returns the ContentAndType of the file at |path|.
     '''
     AssertIsValid(path)
     base, ext = posixpath.splitext(path)
 
     # Check for a zip file first, if zip is enabled.
     if self._directory_zipper and ext == '.zip':
       zip_future = self._directory_zipper.Zip(ToDirectory(base))
       return Future(delegate=Gettable(
           lambda: ContentAndType(zip_future.Get(), 'application/zip')))
 
-    return self._content_cache.GetFromFile(self._MaybeMarkdown(path))
+    # If there is no file extension, look for a file with one of the default
+    # extensions.
+    #
+    # Note that it would make sense to guard this on Exists(path), since a file
+    # without an extension may actually exist, but it's such an uncommon case
+    # it hardly seems worth it.

[Yoyo Zhou, 2014/02/12 22:14:05]

Could be worked around by having the empty string in default_extensions.

[not at google - send to devlin, 2014/02/13 03:34:34]

True, though more that Exists isn't necessarily free. It's probably free. I'll
update comment.

+    if not ext:
+      for default_ext in self._default_extensions:
+        if self.file_system.Exists(path + default_ext).Get():
+          path += default_ext
+          break
+
+    return self._content_cache.GetFromFile(path)
 
   def Cron(self):
-    # Running Refresh() on the file system is enough to pull GitHub content,
-    # which is all we need for now while the full render-every-page cron step
-    # is in effect.
     futures = []
     for root, _, files in self.file_system.Walk(''):
-      futures += [self.GetContentAndType(posixpath.join(root, filename))
-                  for filename in files]
+      for f in files:
+        futures.append(self.GetContentAndType(posixpath.join(root, f)))
+        # Also cache the extension-less version of the file if needed.
+        base, ext = posixpath.splitext(f)
+        if f != WEBMASTER_PROOF and ext in self._default_extensions:

[Yoyo Zhou, 2014/02/12 22:14:05]

I'm surprised you need the special case for WEBMASTER_PROOF here.

[not at google - send to devlin, 2014/02/13 03:34:34]

I think otherwise we'll try to fetch the WEBMASTER_PROOF file without its
extension, which doesn't really make sense? I mean, it probably wouldn't hurt. I
might just leave it in here also to mark this spot, if/when this logic needs to
get less hacky.

+          futures.append(self.GetContentAndType(posixpath.join(root, base)))
+      # TODO(kalman): Cache .zip files for each directory (if supported).
     return Future(delegate=Gettable(lambda: [f.Get() for f in futures]))
 
   def __repr__(self):
     return 'ContentProvider of <%s>' % repr(self.file_system)