6203: Use faster =~ instead of match.

[arvados.git] / sdk / python / arvados / collection.py

diff --git a/sdk/python/arvados/collection.py b/sdk/python/arvados/collection.py
index 497bf7fcd5672d45e5ba2d9ba6507f0bfc7b88cd..30828732d8d4908ca0922bc780c11d2a6943578e 100644 (file)
--- a/sdk/python/arvados/collection.py
+++ b/sdk/python/arvados/collection.py
@@ -822,12 +822,48 @@ class RichCollectionBase(CollectionBase):
 
         target_dir.add(source_obj, target_name, overwrite)
 
-    @synchronized
+    def portable_manifest_text(self, stream_name="."):
+        """Get the manifest text for this collection, sub collections and files.
+
+        This method does not flush outstanding blocks to Keep.  It will return
+        a normalized manifest with access tokens stripped.
+
+        :stream_name:
+          Name to use for this stream (directory)
+
+        """
+        return self._get_manifest_text(stream_name, True, True)
+
     def manifest_text(self, stream_name=".", strip=False, normalize=False):
         """Get the manifest text for this collection, sub collections and files.
 
+        This method will flush outstanding blocks to Keep.  By default, it will
+        not normalize an unmodified manifest or strip access tokens.
+
+        :stream_name:
+          Name to use for this stream (directory)
+
+        :strip:
+          If True, remove signing tokens from block locators if present.
+          If False (default), block locators are left unchanged.
+
+        :normalize:
+          If True, always export the manifest text in normalized form
+          even if the Collection is not modified.  If False (default) and the collection
+          is not modified, return the original manifest text even if it is not
+          in normalized form.
+
+        """
+
+        self._my_block_manager().commit_all()
+        return self._get_manifest_text(stream_name, strip, normalize)
+
+    @synchronized
+    def _get_manifest_text(self, stream_name, strip, normalize):
+        """Get the manifest text for this collection, sub collections and files.
+
         :stream_name:
-          Name of the stream (directory)
+          Name to use for this stream (directory)
 
         :strip:
           If True, remove signing tokens from block locators if present.
@@ -861,7 +897,7 @@ class RichCollectionBase(CollectionBase):
             if stream:
                 buf.append(" ".join(normalize_stream(stream_name, stream)) + "\n")
             for dirname in [s for s in sorted_keys if isinstance(self[s], RichCollectionBase)]:
-                buf.append(self[dirname].manifest_text(stream_name=os.path.join(stream_name, dirname), strip=strip))
+                buf.append(self[dirname].manifest_text(stream_name=os.path.join(stream_name, dirname), strip=strip, normalize=True))
             return "".join(buf)
         else:
             if strip:
@@ -941,7 +977,7 @@ class RichCollectionBase(CollectionBase):
 
     def portable_data_hash(self):
         """Get the portable data hash for this collection's manifest."""
-        stripped = self.manifest_text(strip=True)
+        stripped = self.portable_manifest_text()
         return hashlib.md5(stripped).hexdigest() + '+' + str(len(stripped))
 
     @synchronized
@@ -1390,7 +1426,7 @@ class Collection(RichCollectionBase):
                 block_locator = re.match(r'[0-9a-f]{32}\+(\d+)(\+\S+)*', tok)
                 if block_locator:
                     blocksize = long(block_locator.group(1))
-                    blocks.append(Range(tok, streamoffset, blocksize))
+                    blocks.append(Range(tok, streamoffset, blocksize, 0))
                     streamoffset += blocksize
                 else:
                     state = SEGMENTS