Merge pull request #1142 from coreos-inc/fixtorrentupload

Fix piece hash calculation

data/model/storage.py

@@ -1,6 +1,6 @@
 import logging
 
-from peewee import JOIN_LEFT_OUTER, fn, SQL
+from peewee import JOIN_LEFT_OUTER, fn, SQL, IntegrityError
 
 from data.model import (config, db_transaction, InvalidImageException, TorrentInfoDoesNotExist,
                         DataModelException, _basequery)
@@ -255,7 +255,11 @@ def get_storage_locations(uuid):
 
 
 def save_torrent_info(storage_object, piece_length, pieces):
-  TorrentInfo.create(storage=storage_object, piece_length=piece_length, pieces=pieces)
+  try:
+    TorrentInfo.create(storage=storage_object, piece_length=piece_length, pieces=pieces)
+  except IntegrityError:
+    # TorrentInfo already exists for this storage.
+    pass
 
 def get_torrent_info(blob):
   try:

endpoints/v2/blob.py

@@ -224,11 +224,17 @@ def _upload_chunk(namespace, repo_name, upload_uuid):
 
     input_fp = wrap_with_handler(input_fp, found.sha_state.update)
 
+    # Add a hasher for calculating SHA1s for torrents if this is the first chunk and/or we have
+    # already calculated hash data for the previous chunk(s).
     piece_hasher = None
-    # TODO remove this when all in-progress blob uploads reliably contain piece hashes
-    if start_offset == 0:
-      piece_hasher = PieceHasher(app.config['TORRENT_PIECE_SIZE'], start_offset, '',
-                                 resumablehashlib.sha1())
+    if found.chunk_count == 0 or found.piece_sha_state:
+      initial_sha1_value = found.piece_sha_state or resumablehashlib.sha1()
+      initial_sha1_pieces_value = found.piece_hashes or ''
+
+      piece_hasher = PieceHasher(app.config['TORRENT_PIECE_SIZE'], start_offset,
+                                 initial_sha1_pieces_value,
+                                 initial_sha1_value)
+
       input_fp = wrap_with_handler(input_fp, piece_hasher.update)
 
     # If this is the first chunk and we're starting at the 0 offset, add a handler to gunzip the