Avoid unnecessary legacy hash computation

cary-rowen · cary-rowen · commit a8337ff11da4 · 2026-05-30T13:57:48.000+08:00
diff --git a/bookworm/gui/book_viewer/recents_manager.py b/bookworm/gui/book_viewer/recents_manager.py
@@ -1,4 +1,6 @@
 
+import sqlalchemy as sa
+
 from bookworm.database import PinnedDocument, RecentDocument
 from bookworm.document.hash_utils import get_persistent_content_hash
 from bookworm.logger import logger
@@ -42,6 +44,16 @@ def _get_legacy_documents_by_content_hash(model, content_hash, uri):
     return matching_documents
 
 
+def _has_legacy_documents(model, uri):
+    documents = model.query.filter(
+        sa.or_(
+            model.content_hash_version.is_(None),
+            model.content_hash_version != CURRENT_CONTENT_HASH_VERSION,
+        )
+    )
+    return any(doc.uri.format == uri.format for doc in documents)
+
+
 def _merge_matching_documents(uri_document, hash_matching_documents):
     matching_documents = []
     matching_ids = set()
@@ -90,19 +102,25 @@ def get_document_unique(model, document):
     )
     if content_hash is _CONTENT_HASH_UNSET:
         content_hash = document.get_content_hash()
-    legacy_content_hash = document.get_legacy_content_hash()
+    should_check_legacy_documents = content_hash is None or _has_legacy_documents(
+        model, uri
+    )
+    legacy_content_hash = (
+        document.get_legacy_content_hash() if should_check_legacy_documents else None
+    )
     stored_content_hash, stored_content_hash_version = get_persistent_content_hash(
         content_hash,
         legacy_content_hash,
     )
     hash_matching_documents = _get_documents_by_content_hash(model, content_hash, uri)
-    hash_matching_documents.extend(
-        _get_legacy_documents_by_content_hash(
-            model,
-            legacy_content_hash,
-            uri,
+    if should_check_legacy_documents:
+        hash_matching_documents.extend(
+            _get_legacy_documents_by_content_hash(
+                model,
+                legacy_content_hash,
+                uri,
+            )
         )
-    )
     matching_documents = _merge_matching_documents(doc, hash_matching_documents)
     doc = doc or (matching_documents[0] if matching_documents else None)
     if doc is not None:
diff --git a/bookworm/reader.py b/bookworm/reader.py
@@ -183,6 +183,15 @@ def _get_legacy_matching_document_records(self, model, *, uri_for_storage, legac
         )
         return [record for record in records if record.uri.format == uri_for_storage.format]
 
+    def _has_legacy_document_records(self, model, uri_for_storage):
+        records = model.query.filter(
+            sa.or_(
+                model.content_hash_version.is_(None),
+                model.content_hash_version != CURRENT_CONTENT_HASH_VERSION,
+            )
+        )
+        return any(record.uri.format == uri_for_storage.format for record in records)
+
     def _select_document_record(self, records, uri_for_storage):
         for record in records:
             if record.uri == uri_for_storage:
@@ -252,10 +261,12 @@ def _get_or_create_document_record(
             content_hash = uri_record.content_hash
         else:
             content_hash = content_hash_provider()
+        should_check_legacy_records = legacy_content_hash_provider is not None and (
+            content_hash is None
+            or self._has_legacy_document_records(model, uri_for_storage)
+        )
         legacy_content_hash = (
-            legacy_content_hash_provider()
-            if legacy_content_hash_provider is not None
-            else None
+            legacy_content_hash_provider() if should_check_legacy_records else None
         )
         stored_content_hash, stored_content_hash_version = get_persistent_content_hash(
             content_hash,
@@ -266,7 +277,7 @@ def _get_or_create_document_record(
             uri_for_storage=uri_for_storage,
             content_hash=content_hash,
         )
-        if legacy_content_hash_provider is not None:
+        if should_check_legacy_records:
             matching_records.extend(
                 self._get_legacy_matching_document_records(
                     model,
