Bnot : split large copies into pieces.

copienator/annotation_documents.py

@@ -0,0 +1,27 @@
+"""Locate tablet documents, including the parts of a large checked copy."""
+from pathlib import Path
+
+from copienator.json_io import read_json
+
+MAX_HEIGHT_PX = 32000
+
+
+def copy_documents(directory: Path) -> list[Path]:
+    try:
+        metadata = read_json(directory / "bnote.json", default={})
+    except (OSError, ValueError):
+        # Ordinary documents need no manifest. Let the reader report bad metadata.
+        return [directory]
+    parts = metadata.get("parts", []) if isinstance(metadata, dict) else []
+    return [directory / name for name in parts] if parts else [directory]
+
+
+def tablet_documents(root: Path) -> list[tuple[str, Path]]:
+    documents = []
+    for directory in sorted(root.iterdir()):
+        if not directory.is_dir():
+            continue
+        for part in copy_documents(directory):
+            name = directory.name if part == directory else f"{directory.name} {part.name}"
+            documents.append((name, part))
+    return documents