dfd

cmdlet/merge_file.py

@@ -38,6 +38,7 @@ try:
     from metadata import (
         read_tags_from_file,
         dedup_tags_by_namespace,
+        merge_multiple_tag_lists,
     )
     HAS_METADATA_API = True
 except ImportError:
@@ -58,6 +59,17 @@ except ImportError:
     def dedup_tags_by_namespace(tags: List[str]) -> List[str]:
         return tags
 
+    def merge_multiple_tag_lists(sources: List[List[str]], strategy: str = 'first') -> List[str]:
+        out: List[str] = []
+        seen: set[str] = set()
+        for src in sources:
+            for t in (src or []):
+                s = str(t)
+                if s and s not in seen:
+                    out.append(s)
+                    seen.add(s)
+        return out
+
     def write_metadata(*_args: Any, **_kwargs: Any) -> None:
         return None
 
@@ -105,7 +117,8 @@ def _run(result: Any, args: Sequence[str], config: Dict[str, Any]) -> int:
     source_files: List[Path] = []
     source_hashes: List[str] = []
     source_url: List[str] = []
-    source_tags: List[str] = []  # NEW: collect tags from source files
+    source_tags: List[str] = []  # tags read from .tag sidecars
+    source_item_tag_lists: List[List[str]] = []  # tags carried in-memory on piped items
     for item in files_to_merge:
         raw_path = get_pipe_object_path(item)
         target_path = None
@@ -119,6 +132,16 @@ def _run(result: Any, args: Sequence[str], config: Dict[str, Any]) -> int:
         if target_path and target_path.exists():
             source_files.append(target_path)
 
+            # Track tags carried in the piped items (e.g. add-tag stage) so they survive merge.
+            try:
+                raw_tags = get_field(item, 'tag', [])
+                if isinstance(raw_tags, str) and raw_tags.strip():
+                    source_item_tag_lists.append([raw_tags.strip()])
+                elif isinstance(raw_tags, list):
+                    source_item_tag_lists.append([str(t) for t in raw_tags if t is not None and str(t).strip()])
+            except Exception:
+                pass
+
             # Track tags from the .tag sidecar for this source (if present)
             tags_file = target_path.with_suffix(target_path.suffix + '.tag')
             if tags_file.exists() and HAS_METADATA_API:
@@ -217,17 +240,46 @@ def _run(result: Any, args: Sequence[str], config: Dict[str, Any]) -> int:
         return 1
     
     log(f"Merged {len(source_files)} files into: {output_path}", file=sys.stderr)
-    
-    merged_tags: List[str] = [f"title:{output_path.stem}"]
 
-    # Merge tags from source files into the emitted PipeObject only.
-    # Sidecar files (.tag/.metadata) are written only during explicit filesystem export (add-file to a path).
-    if source_tags and HAS_METADATA_API:
-        merged_source_tags = dedup_tags_by_namespace(source_tags)
-        merged_tags.extend(merged_source_tags)
-        log(f"Merged {len(merged_source_tags)} unique tags from source files", file=sys.stderr)
-    elif source_tags:
-        merged_tags.extend(list(dict.fromkeys(source_tags)))  # Preserve order, remove duplicates
+    def _title_value_from_tags(tags: List[str]) -> Optional[str]:
+        for t in tags:
+            try:
+                s = str(t)
+            except Exception:
+                continue
+            if s.lower().startswith('title:'):
+                val = s.split(':', 1)[1].strip()
+                return val or None
+        return None
+
+    # Determine best title:
+    # - prefer a title tag shared across all inputs (typical when user did add-tag title:...)
+    # - otherwise fall back to first title tag encountered
+    shared_title: Optional[str] = None
+    try:
+        if source_item_tag_lists:
+            per_item_titles: List[Optional[str]] = [_title_value_from_tags(tl) for tl in source_item_tag_lists]
+            non_empty = [t for t in per_item_titles if t]
+            if non_empty:
+                candidate = non_empty[0]
+                if candidate and all((t == candidate) for t in non_empty):
+                    shared_title = candidate
+                else:
+                    shared_title = non_empty[0]
+    except Exception:
+        shared_title = None
+
+    merged_title = shared_title or output_path.stem
+
+    # Merge tags from:
+    # - in-memory PipeObject tags (from add-tag etc)
+    # - .tag sidecars (if present)
+    # Keep all unique plain tags, and keep the first value for namespaced tags.
+    merged_tags = merge_multiple_tag_lists(source_item_tag_lists + ([source_tags] if source_tags else []), strategy='combine')
+
+    # Ensure we always have a title tag (and make sure it's the chosen title)
+    merged_tags = [t for t in merged_tags if not str(t).lower().startswith('title:')]
+    merged_tags.insert(0, f"title:{merged_title}")
     
     # Emit a PipeObject-compatible dict so the merged file can be piped to next command
     try:
@@ -238,7 +290,7 @@ def _run(result: Any, args: Sequence[str], config: Dict[str, Any]) -> int:
             identifier=output_path.name,
             file_path=str(output_path),
             cmdlet_name="merge-file",
-            title=output_path.stem,
+            title=merged_title,
             hash_value=merged_hash,
             tag=merged_tags,
             url=source_url,