Fix save-page scope dtype crash and prefill OCR selectors from ingest scope

smoke_signal_tab.py

@@ -76,6 +76,18 @@ MANIFEST_COLUMNS = [
     "story_pages_exclude",
     "safe_title",
 ]
+MANIFEST_TEXT_COLUMNS = {
+    "book_id",
+    "filename",
+    "sha256",
+    "rights_class",
+    "status",
+    "acquisition_date",
+    "notes",
+    "story_pages_include",
+    "story_pages_exclude",
+    "safe_title",
+}
 
 
 def _load_banner_image_css() -> str:
@@ -747,18 +759,29 @@ def sha256_file(path: Path) -> str:
 def load_manifest_df() -> pd.DataFrame:
     if not MANIFEST_CSV.exists():
         return pd.DataFrame(columns=MANIFEST_COLUMNS)
-    df = pd.read_csv(MANIFEST_CSV)
+    df = pd.read_csv(MANIFEST_CSV, dtype=str, keep_default_na=False)
     for col in MANIFEST_COLUMNS:
         if col not in df.columns:
             df[col] = ""
+    for col in MANIFEST_TEXT_COLUMNS:
+        if col in df.columns:
+            df[col] = df[col].fillna("").astype(str)
+    if "page_count" in df.columns:
+        df["page_count"] = df["page_count"].fillna("").astype(str)
     return df[MANIFEST_COLUMNS]
 
 
 def save_manifest_df(df: pd.DataFrame) -> None:
+    out = df.copy()
     for col in MANIFEST_COLUMNS:
-        if col not in df.columns:
-            df[col] = ""
-    df[MANIFEST_COLUMNS].to_csv(MANIFEST_CSV, index=False)
+        if col not in out.columns:
+            out[col] = ""
+    for col in MANIFEST_TEXT_COLUMNS:
+        if col in out.columns:
+            out[col] = out[col].fillna("").astype(str)
+    if "page_count" in out.columns:
+        out["page_count"] = out["page_count"].fillna("").astype(str)
+    out[MANIFEST_COLUMNS].to_csv(MANIFEST_CSV, index=False)
 
 
 def next_book_id(df: pd.DataFrame) -> str:
@@ -994,12 +1017,31 @@ def _normalize_saved_spec(value) -> str:
     return raw
 
 
+def _default_ocr_scope_values() -> tuple[str, str]:
+    """
+    Prefill OCR page selectors from saved ingest scope.
+    If multiple books exist, use the latest manifest row as default.
+    """
+    df = load_manifest_df()
+    if df.empty:
+        return "", ""
+    row = df.iloc[-1]
+    include_val = _normalize_saved_spec(row.get("story_pages_include", ""))
+    exclude_val = _normalize_saved_spec(row.get("story_pages_exclude", ""))
+    return include_val, exclude_val
+
+
 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 or not bid:
-        return _ingest_status_html("idle"), df, "Book ID is required."
-    if bid not in df["book_id"].values:
+    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."
 
     try:
@@ -2677,6 +2719,10 @@ def smoke_signal_tab():
                     outputs=[ocr_status, ocr_log],
                     show_progress="hidden",
                 )
+                ocr_tab.select(
+                    _default_ocr_scope_values,
+                    outputs=[ocr_page_selection, ocr_page_exclusion],
+                )
 
             # ── STEP 4: REVIEW ────────────────────────────────────────────────
             with gr.TabItem("④ Review", id="ss-review") as review_tab: