fix: ocr unpack 3 vals, no duplicate rename, detected font in review, Reviewer label

smoke_signal_tab.py

@@ -1194,11 +1194,8 @@ def ingest_pdfs(files, rights_class: str, notes: str, book_code_hint: str = "",
             existing_book_id = df.loc[df["sha256"] == file_hash, "book_id"].values[0]
             existing_status = df.loc[df["sha256"] == file_hash, "status"].values[0]
             existing_rights = df.loc[df["sha256"] == file_hash, "rights_class"].values[0]
-            if desired_book_id and desired_book_id != existing_book_id and desired_book_id not in set(df["book_id"].tolist()):
-                _rename_book_id_references(existing_book_id, desired_book_id)
-                df.loc[df["sha256"] == file_hash, "book_id"] = desired_book_id
-                existing_book_id = desired_book_id
-                log.append(log_line(f"↻ Renamed duplicate book ID → {desired_book_id}"))
+            if False:  # disabled: do not rename on duplicate re-registration
+                pass
 
             if rights_class != "unknown" and existing_rights != rights_class:
                 df.loc[df["sha256"] == file_hash, "rights_class"] = rights_class
@@ -1950,7 +1947,7 @@ def run_full_pipeline(
 
     try:
         ocr_gen = run_ocr(scope_include, scope_exclude, replace_book_queue=True)
-        for ocr_status_html, ocr_log in ocr_gen:
+        for ocr_status_html, ocr_log, _ocr_done in ocr_gen:
             for line in str(ocr_log).splitlines():
                 if line.strip() and line not in log:
                     log.append(line)
@@ -2533,14 +2530,14 @@ def get_review_item(idx: int) -> tuple:
     q_df = load_queue_df()
     d_df = load_decisions_df()
     if q_df.empty:
-        return None, "", "", 0, 0
+        return None, "", "", 0, 0, ""
 
     decided_ids = set(d_df["region_id"].tolist()) if not d_df.empty else set()
     pending     = q_df[~q_df["region_id"].isin(decided_ids)]
     if not pending.empty and {"book_id", "page"}.issubset(pending.columns):
         pending = pending.sort_values(["book_id", "page"], ascending=[True, True], kind="stable")
     if pending.empty:
-        return None, "All items reviewed!", "", len(q_df), len(q_df)
+        return None, "All items reviewed!", "", len(q_df), len(q_df), ""
 
     idx  = idx % len(pending)
     item = pending.iloc[idx]
@@ -2581,7 +2578,8 @@ def get_review_item(idx: int) -> tuple:
             f"conf: <b style='color:{'var(--ss-red)' if float(item.get('confidence',0)) < 0.6 else 'var(--ss-gold)'}'>"
             f"{float(item.get('confidence',0)):.0%}</b></div>")
 
-    return img_path, item.get("raw_ocr",""), info, len(q_df) - len(pending), len(q_df)
+    font_name_val = str(item.get("font_name", "") or "").strip()
+    return img_path, item.get("raw_ocr",""), info, len(q_df) - len(pending), len(q_df), font_name_val
 
 
 def save_review_decision(idx: int, final_text: str, action: str, reviewer: str, reason: str, conf_override: bool = False) -> tuple:
@@ -3081,7 +3079,12 @@ def smoke_signal_tab():
                             lines=8,
                             interactive=True,
                         )
-                        reviewer_name = gr.Textbox(label="Your name", placeholder="e.g. jamal", scale=1)
+                        detected_font_display = gr.Textbox(
+                            label="Detected Font",
+                            interactive=False,
+                            scale=1,
+                        )
+                        reviewer_name = gr.Textbox(label="Reviewer", placeholder="e.g. jamal", scale=1)
                         reason_code   = gr.Dropdown(
                             label="Reason code",
                             choices=["","OCR_MISS","OCR_WRONG_WORD","DECORATIVE_FONT",
@@ -3108,14 +3111,14 @@ def smoke_signal_tab():
                 current_idx = gr.State(0)
 
                 def load_review():
-                    img, raw, info, done, total = get_review_item(0)
+                    img, raw, info, done, total, font = get_review_item(0)
                     status = _review_status_html()
-                    return 0, status, img, raw, raw, info
+                    return 0, status, img, raw, raw, info, font
 
                 def next_item(idx):
                     new_idx = idx + 1
-                    img, raw, info, done, total = get_review_item(new_idx)
-                    return new_idx, img, raw, raw, info
+                    img, raw, info, done, total, font = get_review_item(new_idx)
+                    return new_idx, img, raw, raw, info, font
 
                 def do_accept(idx, final, reviewer, reason, conf_ov=False):
                     action = "edited" if final.strip() != "" else "accepted"
@@ -3124,27 +3127,27 @@ def smoke_signal_tab():
                     )
                     status = _review_status_html()
                     new_idx = idx + 1
-                    img2, raw2, info2, _, _ = get_review_item(new_idx)
-                    return fb, status, new_idx, img2, raw2, raw2, info2
+                    img2, raw2, info2, _, _, font2 = get_review_item(new_idx)
+                    return fb, status, new_idx, img2, raw2, raw2, info2, font2
 
                 def do_reject(idx, final, reviewer, reason):
                     fb, img, raw, info, _, _ = save_review_decision(idx, final, "rejected", reviewer, reason)
                     status = _review_status_html()
                     new_idx = idx + 1
-                    img2, raw2, info2, _, _ = get_review_item(new_idx)
-                    return fb, status, new_idx, img2, raw2, raw2, info2
+                    img2, raw2, info2, _, _, font2 = get_review_item(new_idx)
+                    return fb, status, new_idx, img2, raw2, raw2, info2, font2
 
                 def do_quarantine(idx, final, reviewer, reason):
                     fb, img, raw, info, _, _ = save_review_decision(idx, final, "quarantined", reviewer, reason)
                     status = _review_status_html()
                     new_idx = idx + 1
-                    img2, raw2, info2, _, _ = get_review_item(new_idx)
-                    return fb, status, new_idx, img2, raw2, raw2, info2
+                    img2, raw2, info2, _, _, font2 = get_review_item(new_idx)
+                    return fb, status, new_idx, img2, raw2, raw2, info2, font2
 
                 action_outputs = [training_feedback, review_status, current_idx,
-                                  review_image, raw_text_box, final_text_box, item_info]
+                                  review_image, raw_text_box, final_text_box, item_info, detected_font_display]
 
-                review_load_outputs = [current_idx, review_status, review_image, raw_text_box, final_text_box, item_info]
+                review_load_outputs = [current_idx, review_status, review_image, raw_text_box, final_text_box, item_info, detected_font_display]
 
                 load_review_btn.click(
                     load_review,
@@ -3161,18 +3164,18 @@ def smoke_signal_tab():
 
                 def _next(idx):
                     new_idx = idx + 1
-                    img, raw, info, _, _ = get_review_item(new_idx)
-                    return new_idx, img, raw, raw, info
+                    img, raw, info, _, _, font = get_review_item(new_idx)
+                    return new_idx, img, raw, raw, info, font
 
                 def _prev(idx):
                     new_idx = max(0, idx - 1)
-                    img, raw, info, _, _ = get_review_item(new_idx)
-                    return new_idx, img, raw, raw, info
+                    img, raw, info, _, _, font = get_review_item(new_idx)
+                    return new_idx, img, raw, raw, info, font
 
                 prev_btn.click(_prev, inputs=[current_idx],
-                               outputs=[current_idx, review_image, raw_text_box, final_text_box, item_info])
+                               outputs=[current_idx, review_image, raw_text_box, final_text_box, item_info, detected_font_display])
                 next_btn.click(_next, inputs=[current_idx],
-                               outputs=[current_idx, review_image, raw_text_box, final_text_box, item_info])
+                               outputs=[current_idx, review_image, raw_text_box, final_text_box, item_info, detected_font_display])
 
                 # Review now auto-loads on tab select and after OCR run completion.