fix: Avoid collecting unnecessary snapshots

When the logs are collected in relaxed mode it does not need to compute
the current snapshot for every single entry.

This fix speeds up large registers by ~2x.

Signed-off-by: Arnau Siches <arnau.siches@digital.cabinet-office.gov.uk>

registers/log.py

@@ -149,43 +149,10 @@ def collect(commands: List[Command], # pylint: disable=too-many-branches
  data = log or Log()
  metadata = metalog or Log()
  blobs = {**data.blobs, **metadata.blobs}
- errors = []
+ errors: List[ValidationError] = []
 
  for command in commands:
- if command.action == Action.AssertRootHash:
- digest = cast(Hash, command.value)
-
- if digest != data.digest():
- raise InconsistentLog(digest, data.digest(), data.size)
-
- elif command.action == Action.AddItem:
- blobs[command.value.digest()] = cast(Blob, command.value)
-
- elif command.action == Action.AppendEntry:
- entry = cast(Entry, command.value)
- blob = blobs.get(entry.blob_hash)
-
- if blob is None:
- raise OrphanEntry(entry)
-
- if entry.scope == Scope.System:
- metadata.insert(blob)
- record = metadata.snapshot().get(entry.key)
- (_, err) = _collect_entry(entry, record)
-
- if err and not relaxed:
- errors.append(err)
- else:
- metadata.insert(entry)
- else:
- data.insert(blob)
- record = data.snapshot().get(entry.key)
- (_, err) = _collect_entry(entry, record)
-
- if err and not relaxed:
- errors.append(err)
- else:
- data.insert(entry)
+ _collect_command(command, data, metadata, blobs, errors, relaxed)
 
  return {"data": data, "metadata": metadata, "errors": errors}
 
@@ -207,6 +174,49 @@ def slice(log: Log, start_position: int) -> List[Command]:
 Result = Tuple[Optional[Entry], Optional[ValidationError]]
 
 
+def _collect_command(command: Command,
+ data: Log,
+ metadata: Log,
+ blobs: Dict[Hash, Blob],
+ errors: List[ValidationError],
+ relaxed: bool):
+ if command.action == Action.AssertRootHash:
+ digest = cast(Hash, command.value)
+
+ if digest != data.digest():
+ raise InconsistentLog(digest, data.digest(), data.size)
+
+ elif command.action == Action.AddItem:
+ blobs[command.value.digest()] = cast(Blob, command.value)
+
+ elif command.action == Action.AppendEntry:
+ entry = cast(Entry, command.value)
+ blob = blobs.get(entry.blob_hash)
+
+ if blob is None:
+ raise OrphanEntry(entry)
+
+ if entry.scope == Scope.System:
+ _collect_pair(entry, blob, metadata, errors, relaxed)
+
+ else:
+ _collect_pair(entry, blob, data, errors, relaxed)
+
+
+def _collect_pair(entry: Entry, blob: Blob, log: Log,
+ errors: List[ValidationError], relaxed: bool):
+ log.insert(blob)
+
+ if not relaxed:
+ record = log.snapshot().get(entry.key)
+ (_, err) = _collect_entry(entry, record)
+
+ if err:
+ errors.append(err)
+
+ log.insert(entry)
+
+
 def _collect_entry(entry: Entry, record: Optional[Record]) -> Result:
  if record and record.blob.digest() == entry.blob_hash:
  return (None, DuplicatedEntry(entry.key, record.blob))