Fix ingest upload path resolution to prevent empty manifest

smoke_signal_tab.py

@@ -1034,6 +1034,60 @@ def _resolve_book_row(df: pd.DataFrame, selector: str) -> Optional[pd.Series]:
     return None
 
 
+def _resolve_uploaded_file_path(file_obj) -> tuple[Optional[Path], str]:
+    """
+    Robustly resolve an uploaded file path across Gradio runtime object shapes.
+    Returns (path, debug_hint).
+    """
+    candidates: list[str] = []
+    if file_obj is None:
+        return None, "upload item is None"
+
+    # Candidate 1: direct string form (often full temp path in Gradio)
+    try:
+        s = str(file_obj).strip()
+        if s:
+            candidates.append(s)
+    except Exception:
+        pass
+
+    # Candidate 2: .name attribute (file-like wrappers)
+    try:
+        n = getattr(file_obj, "name", "")
+        n = str(n).strip()
+        if n:
+            candidates.append(n)
+    except Exception:
+        pass
+
+    # Candidate 3: explicit path attr used by some wrappers
+    try:
+        p = getattr(file_obj, "path", "")
+        p = str(p).strip()
+        if p:
+            candidates.append(p)
+    except Exception:
+        pass
+
+    # Deduplicate in order
+    seen = set()
+    uniq = []
+    for c in candidates:
+        if c not in seen:
+            uniq.append(c)
+            seen.add(c)
+
+    for c in uniq:
+        try:
+            path = Path(c)
+            if path.exists():
+                return path, f"resolved from '{c}'"
+        except Exception:
+            continue
+
+    return None, f"no existing path in candidates={uniq!r}"
+
+
 def _clean_page_spec(spec: str) -> str:
     raw = (spec or "").strip().lower()
     if raw in ("", "all", "*", "none", "-"):
@@ -1120,9 +1174,9 @@ def ingest_pdfs(files, rights_class: str, notes: str, book_code_hint: str = "",
     auto_profile_ids = []
 
     for file in files:
-        path = Path(file.name) if hasattr(file, "name") else Path(file)
-        if not path.exists():
-            log.append(log_line(f"⚠ File not found: {path.name}"))
+        path, resolve_hint = _resolve_uploaded_file_path(file)
+        if path is None:
+            log.append(log_line(f"⚠ Upload path unresolved: {resolve_hint}"))
             continue
 
         file_hash = sha256_file(path)
@@ -1160,7 +1214,12 @@ def ingest_pdfs(files, rights_class: str, notes: str, book_code_hint: str = "",
         # Copy to source_pdfs
         dest = SOURCE_DIR / path.name
         import shutil
-        shutil.copy2(path, dest)
+        try:
+            if path.resolve() != dest.resolve():
+                shutil.copy2(path, dest)
+        except Exception as e:
+            log.append(log_line(f"✗ Failed to copy {path.name}: {e}"))
+            continue
 
         # Page count
         page_count = None
@@ -1189,7 +1248,7 @@ def ingest_pdfs(files, rights_class: str, notes: str, book_code_hint: str = "",
             "safe_title": _safe_title_slug(Path(path.name).stem),
         }])
         df = pd.concat([df, new_row], ignore_index=True)
-        log.append(log_line(f"✓ Registered {book_id} — {path.name} ({page_count or '?'} pages)"))
+        log.append(log_line(f"✓ Registered {book_id} — {path.name} ({page_count or '?'} pages) [{resolve_hint}]"))
         new_count += 1
         auto_profile_ids.append(book_id)