update app.py

app.py

@@ -83,8 +83,10 @@ def get_flagged_groups():
     try:
         df = pd.read_csv(LABEL_FILE)
         if df.empty: return []
+        df['label'] = df['label'].astype(str).str.strip().str.lower()
         df['score'] = pd.to_numeric(df['score'], errors='coerce')
         df = df.drop_duplicates(subset=['url'], keep='last')
+        # Logic: Flag if score is 10 AND it's NOT living_room
         errors = df[(df['score'] == 10) & (df['label'] != 'living_room')]
         return errors['group_id'].unique().tolist()
     except: return []
@@ -100,10 +102,11 @@ def get_saved_values(gid, mode):
         df = df.drop_duplicates(subset=['url'], keep='last')
         rows = df[df['group_id'] == gid]
         for _, row in rows.iterrows():
+            lbl = str(row['label']).strip().lower() if mode in ["label", "fix"] else str(row['corrected_label']).strip().lower()
             if mode in ["label", "fix"]:
-                saved_data[row['url']] = {"score": row['score'], "label": row['label']}
+                saved_data[row['url']] = {"score": row['score'], "label": lbl}
             else:
-                saved_data[row['url']] = {"is_correct": row['is_correct'], "label": row['corrected_label'], "score": row['corrected_score']}
+                saved_data[row['url']] = {"is_correct": row['is_correct'], "label": lbl, "score": row['corrected_score']}
     except: pass
     return saved_data
 
@@ -114,13 +117,11 @@ def get_stats_text():
         df_l = pd.read_csv(LABEL_FILE).drop_duplicates(subset=['url'], keep='last')
         l = len(df_l['group_id'].unique())
     except: l = 0
-    err_msg = f" | ⚠️ **Fix:** {len(flagged)}" if flagged else ""
-    return f"**Total:** {len(all_gids)} | **Labeled:** {l}{err_msg}"
+    err_msg = f" | ⚠️ **To Fix:** {len(flagged)}" if flagged else ""
+    return f"**Total Properties:** {len(all_gids)} | **Labeled:** {l}{err_msg}"
 
 def render_workspace(mode, history, specific_index=None, move_back=False):
     all_ordered = get_ordered_groups()
-    flagged_pool = get_flagged_groups()
-    
     current_gid = history[-1] if history else None
     target_gid = None
 
@@ -130,6 +131,7 @@ def render_workspace(mode, history, specific_index=None, move_back=False):
         history.pop()
         target_gid = history[-1]
     else:
+        flagged_pool = get_flagged_groups()
         try:
             df_l = pd.read_csv(LABEL_FILE).drop_duplicates(subset=['url'], keep='last')
             l_done = set(df_l['group_id'].unique())
@@ -138,14 +140,17 @@ def render_workspace(mode, history, specific_index=None, move_back=False):
         except: l_done, v_done = set(), set()
         
         if mode == "fix":
+            # Deterministic selection for Fix mode to avoid jumping
             candidates = [g for g in flagged_pool if g != current_gid]
-            if not candidates and flagged_pool: candidates = flagged_pool 
+            if not candidates and flagged_pool: candidates = flagged_pool
+            if candidates: target_gid = candidates[0]
         else:
+            # Random selection for Label/Verify mode
             candidates = [g for g in all_ordered if (mode=="label" and g not in l_done) or (mode=="verify" and g in l_done and g not in v_done)]
+            if candidates: target_gid = random.choice(candidates)
         
-        if not candidates:
-             return {screen_menu: gr.update(visible=True), screen_work: gr.update(visible=False), log_box: "No more properties found in this mode."}
-        target_gid = candidates[0]
+        if not target_gid:
+             return {screen_menu: gr.update(visible=True), screen_work: gr.update(visible=False), log_box: "Mode Complete."}
 
     urls = get_group_urls(target_gid)
     if not history or history[-1] != target_gid: history.append(target_gid)
@@ -166,7 +171,7 @@ def render_workspace(mode, history, specific_index=None, move_back=False):
         screen_menu: gr.update(visible=False), screen_work: gr.update(visible=True),
         header_md: f"# {mode.upper()} Property #{target_idx + 1} ({target_gid})",
         state_urls: urls, state_hist: history, state_idx: target_idx,
-        top_stats: get_stats_text(), log_box: f"Viewing {target_gid}"
+        top_stats: get_stats_text(), log_box: f"Loaded {target_gid}"
     }
 
     for i in range(MAX_IMAGES):
@@ -176,14 +181,14 @@ def render_workspace(mode, history, specific_index=None, move_back=False):
             u = urls[i]
             updates[img_objs[i]] = gr.update(value=processed_images[i], visible=True)
             v_sc = int(float(saved_vals.get(u, {}).get('score', 5)))
-            v_lbl = saved_vals.get(u, {}).get('label', "living_room")
+            v_lbl = str(saved_vals.get(u, {}).get('label', "living_room")).strip().lower()
             is_err = (v_sc == 10 and v_lbl != "living_room")
             
             if mode in ["label", "fix"]:
                 updates[c_sld] = gr.update(visible=True, value=v_sc, interactive=True)
                 updates[c_drp] = gr.update(visible=True, value=v_lbl, interactive=True)
                 updates[c_chk] = gr.update(visible=False)
-                updates[c_lbl] = gr.update(visible=True if is_err else False, value="<span style='color:red'>⚠️ ERROR: Score 10 is Living Room ONLY</span>")
+                updates[c_lbl] = gr.update(visible=True if is_err else False, value="<span style='color:red'>⚠️ ERROR: Score 10 must be Living Room</span>")
             else:
                 p_lbl, p_sc = r1_vals.get(u, {}).get('label', "?"), r1_vals.get(u, {}).get('score', "?")
                 updates[c_sld] = gr.update(visible=True, value=v_sc if u in saved_vals else p_sc)
@@ -197,14 +202,12 @@ def render_workspace(mode, history, specific_index=None, move_back=False):
 
 def save_data(mode, history, urls, *args):
     if not history: return
-    gid = history[-1]
-    ts = datetime.now().isoformat()
-    rows = []
+    gid = history[-1]; ts = datetime.now().isoformat(); rows = []
     for i, u in enumerate(urls):
         sc, lbl, chk = args[i*4], args[i*4+1], args[i*4+2]
-        if mode in ["label", "fix"]: rows.append([ts, "user", gid, u, int(sc), lbl])
-        else: rows.append([ts, "user", gid, u, chk, lbl, int(sc)])
-    
+        clean_lbl = str(lbl).strip().lower()
+        if mode in ["label", "fix"]: rows.append([ts, "user", gid, u, int(sc), clean_lbl])
+        else: rows.append([ts, "user", gid, u, chk, clean_lbl, int(sc)])
     fname = LABEL_FILE if mode in ["label", "fix"] else VERIFY_FILE
     with FileLock(LOCK_FILE):
         with open(fname, "a", newline="") as f: csv.writer(f).writerows(rows)
@@ -220,13 +223,7 @@ def refresh_cat():
         df_v = pd.read_csv(VERIFY_FILE).drop_duplicates(subset=['url'], keep='last')
         v_set = set(df_v['group_id'].unique())
     except: l_set, v_set = set(), set()
-    data = []
-    for i, gid in enumerate(all_gids):
-        if gid in flagged: s = "⚠️ Fix Needed"
-        elif gid in v_set: s = "✅ Verified"
-        elif gid in l_set: s = "🔵 Labeled"
-        else: s = "⚪ Pending"
-        data.append([i+1, s, gid])
+    data = [[i+1, "⚠️ Fix Needed" if gid in flagged else "✅ Verified" if gid in v_set else "🔵 Labeled" if gid in l_set else "⚪ Pending", gid] for i, gid in enumerate(all_gids)]
     return pd.DataFrame(data, columns=["#", "Status", "ID"])
 
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
@@ -237,9 +234,9 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
     with gr.Tabs():
         with gr.Tab("Workspace"):
             with gr.Group() as screen_menu:
-                gr.Markdown("# Welcome")
+                gr.Markdown("# Property Labeler Pro")
                 with gr.Row():
-                    b_start_l, b_start_v, b_start_f = gr.Button("Label", variant="primary"), gr.Button("Verify"), gr.Button("🛠 Fix", variant="secondary")
+                    b_start_l, b_start_v, b_start_f = gr.Button("Start Labeling", variant="primary"), gr.Button("Start Verification"), gr.Button("🛠 Fix Errors", variant="secondary")
             with gr.Group(visible=False) as screen_work:
                 header_md = gr.Markdown()
                 img_objs, input_objs = [], []
@@ -257,7 +254,7 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
                 num_in = gr.Number(value=1, label="Prop #", precision=0)
                 b_go_l, b_go_v, b_go_f = gr.Button("Go Label"), gr.Button("Go Verify"), gr.Button("Go Fix")
             df_cat = gr.Dataframe(interactive=False)
-            b_ref_cat = gr.Button("Refresh")
+            b_ref_cat = gr.Button("Refresh Catalog")
 
     ALL_IO = [screen_menu, screen_work, header_md, state_urls, state_hist, state_idx, top_stats, log_box] + img_objs + input_objs
     b_start_l.click(lambda: "label", None, state_mode).then(render_workspace, [state_mode, state_hist], ALL_IO)