Make rights/page-scope updates resilient and auto-resolve latest book

smoke_signal_tab.py

@@ -998,6 +998,42 @@ def _safe_title_slug(title: str) -> str:
     return slug[:120]
 
 
+def _resolve_book_row(df: pd.DataFrame, selector: str) -> Optional[pd.Series]:
+    if df.empty:
+        return None
+    raw = str(selector or "").strip()
+    if not raw:
+        return df.iloc[-1]
+
+    # Exact book_id
+    exact = df[df["book_id"].astype(str) == raw]
+    if not exact.empty:
+        return exact.iloc[0]
+
+    low = raw.lower()
+
+    # Case-insensitive book_id
+    bid_match = df[df["book_id"].astype(str).str.lower() == low]
+    if not bid_match.empty:
+        return bid_match.iloc[0]
+
+    # Safe title match
+    if "safe_title" in df.columns:
+        st_match = df[df["safe_title"].astype(str).str.lower() == low]
+        if not st_match.empty:
+            return st_match.iloc[0]
+
+    # Filename / stem match
+    name_match = df[df["filename"].astype(str).str.lower() == low]
+    if not name_match.empty:
+        return name_match.iloc[0]
+    stem_match = df[df["filename"].astype(str).str.lower().str.replace(".pdf", "", regex=False) == low]
+    if not stem_match.empty:
+        return stem_match.iloc[0]
+
+    return None
+
+
 def _clean_page_spec(spec: str) -> str:
     raw = (spec or "").strip().lower()
     if raw in ("", "all", "*", "none", "-"):
@@ -1033,16 +1069,12 @@ def _default_ocr_scope_values() -> tuple[str, str]:
 
 def save_book_scope(book_id: str, include_spec: str, exclude_spec: str, safe_title: str) -> tuple:
     df = load_manifest_df()
-    bid = (book_id or "").strip()
     if df.empty:
         return _ingest_status_html("idle"), df, "No books in manifest yet."
-    if not bid:
-        if len(df) == 1:
-            bid = str(df.iloc[0]["book_id"])
-        else:
-            bid = str(df.iloc[-1]["book_id"])
-    if bid not in df["book_id"].astype(str).values:
-        return _ingest_status_html("idle"), df, f"Book ID {bid} not found."
+    row = _resolve_book_row(df, book_id)
+    if row is None:
+        return _ingest_status_html("idle"), df, f"Book ID or title '{book_id}' not found."
+    bid = str(row["book_id"])
 
     try:
         include_clean = _clean_page_spec(include_spec)
@@ -1058,7 +1090,7 @@ def save_book_scope(book_id: str, include_spec: str, exclude_spec: str, safe_tit
     if title_raw:
         safe = _safe_title_slug(title_raw)
     else:
-        filename = str(df.loc[df["book_id"] == bid, "filename"].values[0] or "")
+        filename = str(row.get("filename", "") or "")
         stem = Path(filename).stem if filename else bid
         safe = _safe_title_slug(stem)
 
@@ -1201,14 +1233,19 @@ def _ingest_status_html(state: str, new=0, dups=0) -> str:
 def update_rights(book_id: str, new_rights: str) -> tuple:
     """Update rights class for an existing book."""
     df = load_manifest_df()
-    if df.empty or book_id not in df["book_id"].values:
-        return _ingest_status_html("idle"), df, f"Book ID {book_id} not found."
-    prev_status = df.loc[df["book_id"] == book_id, "status"].values[0]
-    df.loc[df["book_id"] == book_id, "rights_class"] = new_rights
+    if df.empty:
+        return _ingest_status_html("idle"), df, "No books in manifest yet."
+    row = _resolve_book_row(df, book_id)
+    if row is None:
+        return _ingest_status_html("idle"), df, f"Book ID or title '{book_id}' not found."
+    resolved_book_id = str(row["book_id"])
+
+    prev_status = df.loc[df["book_id"] == resolved_book_id, "status"].values[0]
+    df.loc[df["book_id"] == resolved_book_id, "rights_class"] = new_rights
     save_manifest_df(df)
-    msg = f"[{datetime.utcnow().strftime('%H:%M:%S')}] ✓ Updated {book_id} rights → {new_rights}"
+    msg = f"[{datetime.utcnow().strftime('%H:%M:%S')}] ✓ Updated {resolved_book_id} rights → {new_rights}"
     if prev_status == "pending" and new_rights not in ("unknown", "excluded"):
-        _, profile_log = run_profile(book_ids=[book_id])
+        _, profile_log = run_profile(book_ids=[resolved_book_id])
         if profile_log:
             msg = msg + "\n" + str(profile_log)
     return _ingest_status_html("done"), load_manifest_df(), msg
@@ -1737,6 +1774,16 @@ def run_ocr(page_selection: str = "", page_exclusion: str = "", replace_book_que
     debug = f"[DEBUG] SS_ROOT={SS_ROOT}\nMANIFEST_CSV={MANIFEST_CSV}\nCSV exists={MANIFEST_CSV.exists()}\n"
     if not df.empty:
         debug += f"Manifest rows={len(df)}\nStatuses={df['status'].value_counts().to_dict()}\n"
+        try:
+            row_summaries = []
+            for _, r in df.iterrows():
+                row_summaries.append(
+                    f"{r.get('book_id','?')} rights={r.get('rights_class','?')} status={r.get('status','?')}"
+                )
+            if row_summaries:
+                debug += "Rows:\n- " + "\n- ".join(row_summaries) + "\n"
+        except Exception:
+            pass
     else:
         debug += "Manifest is EMPTY\n"
 
@@ -2614,7 +2661,7 @@ def smoke_signal_tab():
 
                 gr.Markdown("**Update rights class for existing book:**")
                 with gr.Row():
-                    update_book_id = gr.Textbox(label="Book ID", placeholder="grf1999", scale=1)
+                    update_book_id = gr.Textbox(label="Book ID (optional)", placeholder="grf1999 or leave blank to use latest", scale=1)
                     update_rights_dd = gr.Dropdown(
                         label="New Rights Class",
                         choices=["public-domain","licensed-owned","controlled-internal","unknown"],
@@ -2630,7 +2677,7 @@ def smoke_signal_tab():
 
                 gr.Markdown("**Save Book Pages + Safe Title (persisted):**")
                 with gr.Row():
-                    scope_book_id = gr.Textbox(label="Book ID", placeholder="grf1999", scale=1)
+                    scope_book_id = gr.Textbox(label="Book ID (optional)", placeholder="grf1999 or leave blank to use latest", scale=1)
                     scope_include = gr.Textbox(label="Story Pages Include", placeholder="all or 7-27 or 7,8,9,11-27", scale=1)
                     scope_exclude = gr.Textbox(label="Story Pages Exclude", placeholder="e.g. 1,2,3,17,25", scale=1)
                     scope_title = gr.Textbox(label="Safe Book Title", placeholder="e.g. the-gruffalo", scale=1)