Cache for reading annotation

copienator/commands/reading_grouped_annotations.py

@@ -27,6 +27,7 @@ from copienator.annotation_actions import apply_checkbox_actions, apply_score_ov
 from copienator.annotation_data import AnnotationData, RefaireList, load_annotation_data
 from copienator.annotation_documents import copy_documents, tablet_documents
 from copienator.annotation_layout import legacy_layout
+from copienator.annotation_read_cache import AnnotationReadCache
 from copienator.annotation_modifications import (
     BASELINE,
     baseline,
@@ -36,6 +37,7 @@ from copienator.annotation_modifications import (
 )
 from copienator.answer_info import build_answer_info
 from copienator.commands import annotating
+from copienator.commands.giving_names import update_named_return_answers
 from copienator.commands.reading_annotations import (
     concatenate,
     detect_checks_and_notes,
@@ -594,6 +596,18 @@ def run(
         and path.is_dir()
         and not path.name.startswith("Copie")
     ]
+    # Redo reads deliberately remain full reads: their selected labels and
+    # preserved baseline images have different dependencies.
+    cache = (AnnotationReadCache(workspace, annotation_dir, loaded.data, group_dirs)
+             if update_score and not refaire else None)
+    pending_ids = set(loaded.data)
+    if cache is not None:
+        pending_ids = {copy_id for copy_id in loaded.data if not cache.unchanged(copy_id)}
+        for copy_id in sorted(set(loaded.data) - pending_ids, key=utils.natural_key):
+            print(f"Skipping unchanged Copie{copy_id}")
+        if not pending_ids:
+            return ExitCode.PARTIAL if loaded.warnings else ExitCode.SUCCESS
+        only_ids = pending_ids
     # Each worker decodes a full-height returned group and its reference image.
     # Keep this stage deliberately narrow; answer regeneration below has its own
     # parallel executor and a much smaller per-task memory footprint.
@@ -657,6 +671,7 @@ def run(
         if refaire
         else sorted(loaded.data, key=utils.natural_key)
     )
+    regenerated: set[str] = set()
     with concurrent.futures.ThreadPoolExecutor(max_workers=4) as executor:
         futures = {
             executor.submit(
@@ -676,13 +691,19 @@ def run(
                 ),
             ): student_id
             for student_id in student_ids
-            if student_id in loaded.data and student_id not in skipped_students
+            if student_id in pending_ids and student_id not in skipped_students
         }
         for future in concurrent.futures.as_completed(futures):
             result, output = future.result()
             print(output)
             if result != ExitCode.SUCCESS:
                 status = ExitCode.PARTIAL
+            else:
+                regenerated.add(futures[future])
+    if update_named_return_answers(workspace, annotation_dir, copy_ids=regenerated) != ExitCode.SUCCESS:
+        status = ExitCode.PARTIAL
+    elif cache is not None:
+        cache.remember(regenerated)
     return status