Updating flickr html parsing, fixing #2688

[redakcja.git] / lib / vstorage / __init__.py

diff --git a/lib/vstorage/__init__.py b/lib/vstorage/__init__.py
index 8d22e10..2708ed7 100644 (file)
--- a/lib/vstorage/__init__.py
+++ b/lib/vstorage/__init__.py
@@ -147,14 +147,22 @@ class VersionedStorage(object):
         self.repo = mercurial.hg.repository(self.ui, self.repo_path)
 
     def _file_path(self, title, type='.xml'):
-        return os.path.join(self.path, urlquote(title, safe='')) + type
+        """ Return plain version if exists in repo, add extension otherwise. """
+        path = os.path.join(self.path, urlquote(title, safe=''))
+        if type and self._title_to_file(title, '') not in self.repo['tip']:
+            path += type
+        return path
 
     def _title_to_file(self, title, type=".xml"):
-        return os.path.join(self.repo_prefix, urlquote(title, safe='')) + type
+        """ Return plain version if exists in repo, add extension otherwise. """
+        path = os.path.join(self.repo_prefix, urlquote(title, safe=''))
+        if type and path not in self.repo['tip']:
+            path += type
+        return path
 
     def _file_to_title(self, filename):
         assert filename.startswith(self.repo_prefix)
-        name = filename[len(self.repo_prefix):].strip('/').split('.', 1)[0]
+        name = filename[len(self.repo_prefix):].strip('/').rsplit('.', 1)[0]
         return urlunquote(name)
 
     def __contains__(self, title):
@@ -211,7 +219,7 @@ class VersionedStorage(object):
             filectx_tip = changectx[repo_file]
             current_page_rev = filectx_tip.filerev()
         except mercurial.revlog.LookupError:
-            self.repo.add([repo_file])
+            self.repo[None].add([repo_file])
             current_page_rev = -1
 
         if parent is not None and current_page_rev != parent:
@@ -263,7 +271,7 @@ class VersionedStorage(object):
             os.unlink(file_path)
         except OSError:
             pass
-        self.repo.remove([repo_file])
+        self.repo[None].remove([repo_file])
         self._commit([repo_file], text, user)
 
     def page_text(self, title, revision=None):
@@ -325,42 +333,28 @@ class VersionedStorage(object):
         return guess_mime(self._file_path(title))
 
     def _find_filectx(self, title, rev=None):
-        """Find the last revision in which the file existed."""
-        tip = self._changectx()  # start with tip
-
-        def tree_search(tip, repo_file):
-            logging.info("Searching for %r", repo_file)
-            current = tip
-            visited = set()
-
-            stack = [current]
-            visited.add(current)
-
-            while repo_file not in current:
-                if not stack:
-                    raise LookupError
-
-                current = stack.pop()
-                for parent in current.parents():
-                    if parent not in visited:
-                        stack.append(parent)
-                        visited.add(parent)
-
-            fctx = current[repo_file]
-            if rev is not None:
-                fctx = fctx.filectx(rev)
-                fctx.filerev()
-            return fctx
-
-        try:
-            return tree_search(tip, self._title_to_file(title))
-        except (IndexError, LookupError) as e:
-            logging.info("XML file not found, trying plain")
-            try:
-                return tree_search(tip, self._title_to_file(title, type=''))
-            except (IndexError, LookupError) as e:
+        """
+        Find the revision of the file in repo.
+        Only look for files still existing in repo's tip.
+        """
+        tip = self._changectx()
+        file = self._title_to_file(title)
+        logging.info('Looking for %s', file)
+        if file in tip:
+            fctx = tip[file]
+        else:
+            file = self._title_to_file(title, type='')
+            logging.info('.xml not found, trying plain')
+            if file in tip:
+                fctx = tip[file]
+            else:
                 raise DocumentNotFound(title)
 
+        if rev is not None:
+            fctx = fctx.filectx(rev)
+            fctx.filerev()
+        return fctx
+
     def page_history(self, title):
         """Iterate over the page's history."""