doocus improvemnts

doocus/tree.py

@@ -31,6 +31,11 @@ EXTRACT_EXTS = {"docx", "pdf", "pptx", "xlsx"}
 # Videos are meetings → meetus territory.
 MEETING_EXTS = {"mp4", "mkv", "mov", "m4v", "webm", "avi", "wmv"}
 
+# Sidecar suffix meetus writes a meeting's output into. Must match the batch
+# invocation `make batch ... OUT_FORMAT="{name}.meetus"` so the index pointer set
+# here (deterministically, at index time) resolves once meetus has run.
+MEETING_SIDECAR_SUFFIX = ".meetus"
+
 
 def classify(ext: str) -> str:
     if ext in EXTRACT_EXTS:
@@ -60,6 +65,7 @@ def build_index(source_root, output_dir, options=None, dry_run=False) -> dict:
         mode = classify(ext)
         counts[mode] += 1
         stat = p.stat()
+        birth = getattr(stat, "st_birthtime", None)  # not on every filesystem
         node = {
             "path": rel,
             "name": p.name,
@@ -68,11 +74,17 @@ def build_index(source_root, output_dir, options=None, dry_run=False) -> dict:
             "mode": mode,
             "bytes": stat.st_size,
             "modified": datetime.fromtimestamp(stat.st_mtime).isoformat(),
+            "created": datetime.fromtimestamp(birth or stat.st_mtime).isoformat(),
             "url": None,  # Drive link — filled later from a URL source
         }
 
         if mode == "extracted" and not dry_run:
             _extract_sidecar(p, rel, out, options, node, counts)
+        elif mode == "meeting":
+            # Deterministic pointer to meetus's output sidecar (produced later by
+            # `make batch ... OUT_FORMAT="{name}.meetus"`). No relink needed.
+            node["out"] = rel + MEETING_SIDECAR_SUFFIX
+            node["transcript"] = f"{rel}{MEETING_SIDECAR_SUFFIX}/{p.stem}_enhanced.txt"
 
         files.append(node)