Merge pull request #1951 from lbryio/fix_invalid_blob_decode_startup

Fix invalid blob decode startup

lbrynet/blob/blob_file.py

@@ -151,6 +151,7 @@ class BlobFile:
             os.remove(self.file_path)
         self.verified.clear()
         self.finished_writing.clear()
+        self.length = None
 
     def decrypt(self, key: bytes, iv: bytes) -> bytes:
         """

lbrynet/stream/descriptor.py

@@ -131,7 +131,11 @@ class StreamDescriptor:
         assert os.path.isfile(blob.file_path)
         with open(blob.file_path, 'rb') as f:
             json_bytes = f.read()
-        decoded = json.loads(json_bytes.decode())
+        try:
+            decoded = json.loads(json_bytes.decode())
+        except json.JSONDecodeError:
+            blob.delete()
+            raise InvalidStreamDescriptorError("Does not decode as valid JSON")
         if decoded['blobs'][-1]['length'] != 0:
             raise InvalidStreamDescriptorError("Does not end with a zero-length blob.")
         if any([blob_info['length'] == 0 for blob_info in decoded['blobs'][:-1]]):

tests/integration/test_file_commands.py

@@ -133,6 +133,7 @@ class FileCommands(CommandTestCase):
         # restore blob
         os.rename(missing_blob.file_path + '__', missing_blob.file_path)
         self.server_blob_manager.blobs.clear()
+        missing_blob = self.server_blob_manager.get_blob(missing_blob_hash)
         await self.server_blob_manager.blob_completed(missing_blob)
         await asyncio.wait_for(self.wait_files_to_complete(), timeout=1)
 

tests/unit/stream/test_stream_descriptor.py

@@ -4,6 +4,7 @@ import tempfile
 import shutil
 import json
 
+from lbrynet.blob.blob_file import BlobFile
 from torba.testcase import AsyncioTestCase
 from lbrynet.conf import Config
 from lbrynet.error import InvalidStreamDescriptorError
@@ -107,3 +108,21 @@ class TestRecoverOldStreamDescriptors(AsyncioTestCase):
         self.assertEqual(stream_hash, descriptor.get_stream_hash())
         self.assertEqual(sd_hash, descriptor.calculate_old_sort_sd_hash())
         self.assertNotEqual(sd_hash, descriptor.calculate_sd_hash())
+
+    async def test_decode_corrupt_blob_raises_proper_exception_and_deletes_corrupt_file(self):
+        loop = asyncio.get_event_loop()
+        tmp_dir = tempfile.mkdtemp()
+        self.addCleanup(lambda: shutil.rmtree(tmp_dir))
+        sd_hash = '9313d1807551186126acc3662e74d9de29cede78d4f133349ace846273ef116b9bb86be86c54509eb84840e4b032f6b2'
+        with open(os.path.join(tmp_dir, sd_hash), 'wb') as handle:
+            handle.write(b'doesnt work')
+        blob = BlobFile(loop, tmp_dir, sd_hash)
+        self.assertTrue(blob.file_exists)
+        self.assertIsNotNone(blob.length)
+        with self.assertRaises(InvalidStreamDescriptorError):
+            await StreamDescriptor.from_stream_descriptor_blob(
+                loop, tmp_dir, blob
+            )
+        self.assertFalse(blob.file_exists)
+        # fixme: this is an emergency PR, plase move this to blob_file tests later
+        self.assertIsNone(blob.length)