Update app.py

app.py

@@ -715,7 +715,90 @@ def _render_header_html(px: int) -> str:
 
 
 # ------------------------------
-# 11) UI (Blocks)
+# 11) Handlers (defined before UI)
+# ------------------------------
+def _split_chunks(files, csize: int):
+    files = files or []
+    c = max(1, int(csize))
+    return [files[i:i + c] for i in range(0, len(files), c)]
+
+
+def _tpms():
+    s = load_settings()
+    return s.get("temperature", 0.6), s.get("top_p", 0.9), s.get("max_tokens", 256)
+
+
+def _run_click(files, rows, instr, ms, mode, csize, budget_s, no_limit):
+    t, p, m = _tpms()
+    files = files or []
+    budget = None if no_limit else float(budget_s)
+
+    if mode == "Manual (step)" and files:
+        chunks = _split_chunks(files, int(csize))
+        batch = chunks[0]
+        remaining = sum(chunks[1:], [])
+        new_rows, gal, tbl, stamp, leftover_from_batch, done, total = run_batch(
+            batch, rows or [], instr, t, p, m, int(ms), budget
+        )
+        remaining = (leftover_from_batch or []) + remaining
+        panel_vis = gr.update(visible=bool(remaining))
+        msg = f"{len(remaining)} files remain. Process next chunk?"
+        prog = f"Batch progress: {done}/{total} processed in this step • Remaining overall: {len(remaining)}"
+        return new_rows, gal, tbl, stamp, remaining, panel_vis, gr.update(value=msg), gr.update(value=prog)
+
+    # Auto
+    new_rows, gal, tbl, stamp, leftover, done, total = run_batch(
+        files, rows or [], instr, t, p, m, int(ms), budget
+    )
+    panel_vis = gr.update(visible=bool(leftover))
+    msg = f"{len(leftover)} files remain. Process next chunk?" if leftover else ""
+    prog = f"Batch progress: {done}/{total} processed in this call • Remaining: {len(leftover)}"
+    return new_rows, gal, tbl, stamp, leftover, panel_vis, gr.update(value=msg), gr.update(value=prog)
+
+
+def _step_next(remain, rows, instr, ms, csize, budget_s, no_limit):
+    t, p, m = _tpms()
+    remain = remain or []
+    budget = None if no_limit else float(budget_s)
+
+    if not remain:
+        return (
+            rows,
+            gr.update(value="No files remaining."),
+            gr.update(visible=False),
+            [],
+            [],
+            [],
+            "Saved.",
+            gr.update(value="")
+        )
+
+    batch = remain[:int(csize)]
+    leftover = remain[int(csize):]
+    new_rows, gal, tbl, stamp, leftover_from_batch, done, total = run_batch(
+        batch, rows or [], instr, t, p, m, int(ms), budget
+    )
+    leftover = (leftover_from_batch or []) + leftover
+    panel_vis = gr.update(visible=bool(leftover))
+    msg = f"{len(leftover)} files remain. Process next chunk?" if leftover else "All done."
+    prog = f"Batch progress: {done}/{total} processed in this step • Remaining overall: {len(leftover)}"
+    return new_rows, msg, panel_vis, leftover, gal, tbl, stamp, gr.update(value=prog)
+
+
+def _step_finish():
+    return gr.update(visible=False), gr.update(value=""), []
+
+
+def sync_table_to_session(table_value: Any, session_rows: List[dict]) -> Tuple[List[dict], list, str]:
+    session_rows = _table_to_rows(table_value, session_rows or [])
+    save_session(session_rows)
+    gallery_pairs = [((r.get("thumb_path") or r.get("path")), r.get("caption", ""))
+                     for r in session_rows if (r.get("thumb_path") or r.get("path"))]
+    return session_rows, gallery_pairs, f"Saved • {time.strftime('%H:%M:%S')}"
+
+
+# ------------------------------
+# 12) UI (Blocks)
 # ------------------------------
 BASE_CSS = """
 :root{--galleryW:50%;--tableW:50%;}
@@ -782,11 +865,15 @@ with gr.Blocks(css=BASE_CSS, title="ForgeCaptions") as demo:
                             settings.get("caption_length", "long"),
                         ),
                     )
-                dataset_name = gr.Textbox(label="Dataset name (export title prefix)",
-                                          value=settings.get("dataset_name", "forgecaptions"))
-                max_side   = gr.Slider(256, 1024, settings.get("max_side", 896), step=32, label="Max side (resize)")
-                excel_thumb_px = gr.Slider(64, 256, value=settings.get("excel_thumb_px", 128),
-                                           step=8, label="Excel thumbnail size (px)")
+                dataset_name = gr.Textbox(
+                    label="Dataset name (export title prefix)",
+                    value=settings.get("dataset_name", "forgecaptions")
+                )
+                max_side = gr.Slider(256, 1024, settings.get("max_side", 896), step=32, label="Max side (resize)")
+                excel_thumb_px = gr.Slider(
+                    64, 256, value=settings.get("excel_thumb_px", 128),
+                    step=8, label="Excel thumbnail size (px)"
+                )
                 # Chunking
                 chunk_mode = gr.Radio(
                     choices=["Auto", "Manual (step)"],
@@ -824,6 +911,7 @@ with gr.Blocks(css=BASE_CSS, title="ForgeCaptions") as demo:
         cfg["dataset_name"] = sanitize_basename(name)
         save_settings(cfg)
         return gr.update()
+
     dataset_name.change(_save_dataset_name, inputs=[dataset_name], outputs=[])
 
     # ---- Shape Aliases (with plural matching + persist)
@@ -847,7 +935,7 @@ with gr.Blocks(css=BASE_CSS, title="ForgeCaptions") as demo:
             headers=["shape (token or synonyms)", "name to insert"],
             value=init_rows,
             row_count=(max(1, len(init_rows)), "dynamic"),
-            datatype=["str","str"],
+            datatype=["str", "str"],
             type="array",
             interactive=True
         )
@@ -874,6 +962,7 @@ with gr.Blocks(css=BASE_CSS, title="ForgeCaptions") as demo:
             cfg["shape_aliases_persist"] = bool(v)
             save_settings(cfg)
             return gr.update()
+
         persist_aliases.change(_save_alias_persist_flag, inputs=[persist_aliases], outputs=[])
 
         save_btn.click(
@@ -905,7 +994,7 @@ with gr.Blocks(css=BASE_CSS, title="ForgeCaptions") as demo:
             run_button = gr.Button("Caption batch", variant="primary")
 
     # ---- Results area (gallery left / table right)
-    rows_state  = gr.State(load_session())
+    rows_state = gr.State(load_session())
     autosave_md = gr.Markdown("Ready.")
     progress_md = gr.Markdown("")
     remaining_state = gr.State([])
@@ -933,43 +1022,35 @@ with gr.Blocks(css=BASE_CSS, title="ForgeCaptions") as demo:
     # ---- Step panel
     step_panel = gr.Group(visible=False)
     with step_panel:
-        step_msg    = gr.Markdown("")
-        step_next   = gr.Button("Process next chunk")
+        step_msg = gr.Markdown("")
+        step_next = gr.Button("Process next chunk")
         step_finish = gr.Button("Finish")
 
     # ---- Exports
     with gr.Row():
         with gr.Column():
-            export_csv_btn  = gr.Button("Export CSV")
-            csv_file        = gr.File(label="CSV file", visible=False)
+            export_csv_btn = gr.Button("Export CSV")
+            csv_file = gr.File(label="CSV file", visible=False)
         with gr.Column():
             export_xlsx_btn = gr.Button("Export Excel (.xlsx) with thumbnails")
-            xlsx_file       = gr.File(label="Excel file", visible=False)
+            xlsx_file = gr.File(label="Excel file", visible=False)
         with gr.Column():
-            export_txt_btn  = gr.Button("Export captions as .txt (zip)")
-            txt_zip         = gr.File(label="TXT zip", visible=False)
+            export_txt_btn = gr.Button("Export captions as .txt (zip)")
+            txt_zip = gr.File(label="TXT zip", visible=False)
 
+    # ---- Scroll-sync JS injection (inside Blocks)
     gr.HTML("""
     <script>
     (() => {
-      // ---- Config: your wrapper IDs
-      const GAL_WRAP_SEL = "#cfGal";        // wrapper around the gallery
-      const TABLE_WRAP_SEL = "#cfTableWrap"; // wrapper around your table
-    
-      // ---- Helpers
+      const GAL_WRAP_SEL = "#cfGal";
+      const TABLE_WRAP_SEL = "#cfTableWrap";
       const clamp = (v, a, b) => Math.max(a, Math.min(b, v));
-    
       function findGalleryHost() {
         const wrap = document.querySelector(GAL_WRAP_SEL);
         if (!wrap) return null;
-        // Prefer explicit gallery node if present; otherwise use wrapper
         return wrap.querySelector('[data-testid="gallery"], [data-testid="image-gallery"]') || wrap;
       }
-    
-      function findTableHost() {
-        return document.querySelector(TABLE_WRAP_SEL);
-      }
-    
+      function findTableHost() { return document.querySelector(TABLE_WRAP_SEL); }
       function setMaxHeights(gal, tab) {
         const targetH = clamp(tab.clientHeight || 520, 360, 520);
         gal.style.maxHeight = targetH + "px";
@@ -977,202 +1058,74 @@ with gr.Blocks(css=BASE_CSS, title="ForgeCaptions") as demo:
         tab.style.maxHeight = targetH + "px";
         tab.style.overflowY = "auto";
       }
-    
       function attachScrollSync(a, b) {
         if (!a || !b) return () => {};
         let lock = false;
-    
-        // Use scroll ratio so different scrollHeights stay aligned
         const sync = (src, dst) => {
           const maxSrc = Math.max(1, src.scrollHeight - src.clientHeight);
           const r = src.scrollTop / maxSrc;
           const maxDst = Math.max(1, dst.scrollHeight - dst.clientHeight);
           const next = r * maxDst;
-          if (Math.abs(dst.scrollTop - next) > 1) {
-            dst.scrollTop = next;
-          }
+          if (Math.abs(dst.scrollTop - next) > 1) dst.scrollTop = next;
         };
-    
         const onA = () => { if (lock) return; lock = true; requestAnimationFrame(() => { sync(a, b); lock = false; }); };
         const onB = () => { if (lock) return; lock = true; requestAnimationFrame(() => { sync(b, a); lock = false; }); };
-    
         a.addEventListener("scroll", onA, { passive: true });
         b.addEventListener("scroll", onB, { passive: true });
-    
-        // Return cleanup
-        return () => {
-          a.removeEventListener("scroll", onA);
-          b.removeEventListener("scroll", onB);
-        };
+        return () => { a.removeEventListener("scroll", onA); b.removeEventListener("scroll", onB); };
       }
-    
-      let cleanupScroll = null;
-      let resizeObs = null;
-    
+      let cleanupScroll = null, resizeObs = null;
       function wireUp() {
-        const gal = findGalleryHost();
-        const tab = findTableHost();
+        const gal = findGalleryHost(), tab = findTableHost();
         if (!gal || !tab) return false;
-    
-        // (Re)apply heights & listeners
         setMaxHeights(gal, tab);
         if (cleanupScroll) cleanupScroll();
         cleanupScroll = attachScrollSync(gal, tab);
-    
-        // Keep heights in sync on resize/content changes
         if (resizeObs) resizeObs.disconnect();
         resizeObs = new ResizeObserver(() => setMaxHeights(gal, tab));
-        resizeObs.observe(tab);
-        resizeObs.observe(gal);
-    
+        resizeObs.observe(tab); resizeObs.observe(gal);
         return true;
       }
-    
-      // First attempt immediately (fast path)
       if (wireUp()) return;
-    
-      // Observe DOM for late mounts / Gradio re-renders
-      const mo = new MutationObserver(() => {
-        if (wireUp()) {
-          // Once wired, continue watching for *re-renders* to re-wire
-          // If you prefer, remove this `disconnect()` to keep watching forever
-          // but then you should also detect detach and clean up.
-          // Here we keep observing to handle re-renders:
-        }
-      });
-    
-      mo.observe(document.documentElement || document.body, {
-        childList: true,
-        subtree: true,
-      });
-    
-      // Optional: cleanup on page unload
+      const mo = new MutationObserver(() => { wireUp(); });
+      mo.observe(document.documentElement || document.body, { childList: true, subtree: true });
       window.addEventListener("beforeunload", () => {
-        mo.disconnect();
-        if (resizeObs) resizeObs.disconnect();
-        if (cleanupScroll) cleanupScroll();
+        mo.disconnect(); if (resizeObs) resizeObs.disconnect(); if (cleanupScroll) cleanupScroll();
       });
     })();
-    </script> """)
-
-
-# ---- Chunking logic
-def _split_chunks(files, csize: int):
-    files = files or []
-    c = max(1, int(csize))
-    return [files[i:i + c] for i in range(0, len(files), c)]
-
-
-def _tpms():
-    s = load_settings()
-    return s.get("temperature", 0.6), s.get("top_p", 0.9), s.get("max_tokens", 256)
-
-
-def _run_click(files, rows, instr, ms, mode, csize, budget_s, no_limit):
-    t, p, m = _tpms()
-    files = files or []
-    budget = None if no_limit else float(budget_s)
-
-    if mode == "Manual (step)" and files:
-        chunks = _split_chunks(files, int(csize))
-        batch = chunks[0]
-        remaining = sum(chunks[1:], [])
-        new_rows, gal, tbl, stamp, leftover_from_batch, done, total = run_batch(
-            batch, rows or [], instr, t, p, m, int(ms), budget
-        )
-        remaining = (leftover_from_batch or []) + remaining
-        panel_vis = gr.update(visible=bool(remaining))
-        msg = f"{len(remaining)} files remain. Process next chunk?"
-        prog = f"Batch progress: {done}/{total} processed in this step • Remaining overall: {len(remaining)}"
-        return new_rows, gal, tbl, stamp, remaining, panel_vis, gr.update(value=msg), gr.update(value=prog)
-
-    # Auto
-    new_rows, gal, tbl, stamp, leftover, done, total = run_batch(
-        files, rows or [], instr, t, p, m, int(ms), budget
+    </script>
+    """)
+
+    # ---- Event bindings (MUST be inside Blocks in Gradio v5)
+    run_button.click(
+        _run_click,
+        inputs=[input_files, rows_state, instruction_preview, max_side, chunk_mode, chunk_size, gpu_budget, no_time_limit],
+        outputs=[rows_state, gallery, table, autosave_md, remaining_state, step_panel, step_msg, progress_md]
     )
-    panel_vis = gr.update(visible=bool(leftover))
-    msg = f"{len(leftover)} files remain. Process next chunk?" if leftover else ""
-    prog = f"Batch progress: {done}/{total} processed in this call • Remaining: {len(leftover)}"
-    return new_rows, gal, tbl, stamp, leftover, panel_vis, gr.update(value=msg), gr.update(value=prog)
-
-
-run_button.click(
-    _run_click,
-    inputs=[input_files, rows_state, instruction_preview, max_side, chunk_mode, chunk_size, gpu_budget, no_time_limit],
-    outputs=[rows_state, gallery, table, autosave_md, remaining_state, step_panel, step_msg, progress_md]
-)
-
-
-def _step_next(remain, rows, instr, ms, csize, budget_s, no_limit):
-    t, p, m = _tpms()
-    remain = remain or []
-    budget = None if no_limit else float(budget_s)
 
-    if not remain:
-        return (
-            rows,
-            gr.update(value="No files remaining."),
-            gr.update(visible=False),
-            [],
-            [],
-            [],
-            "Saved.",
-            gr.update(value="")
-        )
-
-    batch = remain[:int(csize)]
-    leftover = remain[int(csize):]
-    new_rows, gal, tbl, stamp, leftover_from_batch, done, total = run_batch(
-        batch, rows or [], instr, t, p, m, int(ms), budget
+    step_next.click(
+        _step_next,
+        inputs=[remaining_state, rows_state, instruction_preview, max_side, chunk_size, gpu_budget, no_time_limit],
+        outputs=[rows_state, step_msg, step_panel, remaining_state, gallery, table, autosave_md, progress_md]
     )
-    leftover = (leftover_from_batch or []) + leftover
-    panel_vis = gr.update(visible=bool(leftover))
-    msg = f"{len(leftover)} files remain. Process next chunk?" if leftover else "All done."
-    prog = f"Batch progress: {done}/{total} processed in this step • Remaining overall: {len(leftover)}"
-    return new_rows, msg, panel_vis, leftover, gal, tbl, stamp, gr.update(value=prog)
-
-
-step_next.click(
-    _step_next,
-    inputs=[remaining_state, rows_state, instruction_preview, max_side, chunk_size, gpu_budget, no_time_limit],
-    outputs=[rows_state, step_msg, step_panel, remaining_state, gallery, table, autosave_md, progress_md]
-)
-
-
-def _step_finish():
-    return gr.update(visible=False), gr.update(value=""), []
-
 
-step_finish.click(_step_finish, inputs=None, outputs=[step_panel, step_msg, remaining_state])
+    step_finish.click(_step_finish, inputs=None, outputs=[step_panel, step_msg, remaining_state])
 
+    table.change(sync_table_to_session, inputs=[table, rows_state], outputs=[rows_state, gallery, autosave_md])
 
-# ---- Table edits → persist + refresh gallery
-def sync_table_to_session(table_value: Any, session_rows: List[dict]) -> Tuple[List[dict], list, str]:
-    session_rows = _table_to_rows(table_value, session_rows or [])
-    save_session(session_rows)
-    gallery_pairs = [((r.get("thumb_path") or r.get("path")), r.get("caption", ""))
-                     for r in session_rows if (r.get("thumb_path") or r.get("path"))]
-    return session_rows, gallery_pairs, f"Saved • {time.strftime('%H:%M:%S')}"
-
-
-table.change(sync_table_to_session, inputs=[table, rows_state], outputs=[rows_state, gallery, autosave_md])
-
-
-# ---- Exports
-export_csv_btn.click(
-    lambda tbl, ds: (export_csv_from_table(tbl, ds), gr.update(visible=True)),
-    inputs=[table, dataset_name], outputs=[csv_file, csv_file]
-)
-
-export_xlsx_btn.click(
-    lambda tbl, rows, px, ds: (export_excel_with_thumbs(tbl, rows or [], int(px), ds), gr.update(visible=True)),
-    inputs=[table, rows_state, excel_thumb_px, dataset_name], outputs=[xlsx_file, xlsx_file]
-)
+    export_csv_btn.click(
+        lambda tbl, ds: (export_csv_from_table(tbl, ds), gr.update(visible=True)),
+        inputs=[table, dataset_name], outputs=[csv_file, csv_file]
+    )
+    export_xlsx_btn.click(
+        lambda tbl, rows, px, ds: (export_excel_with_thumbs(tbl, rows or [], int(px), ds), gr.update(visible=True)),
+        inputs=[table, rows_state, excel_thumb_px, dataset_name], outputs=[xlsx_file, xlsx_file]
+    )
+    export_txt_btn.click(
+        lambda tbl, ds: (export_txt_zip(tbl, ds), gr.update(visible=True)),
+        inputs=[table, dataset_name], outputs=[txt_zip, txt_zip]
+    )
 
-export_txt_btn.click(
-    lambda tbl, ds: (export_txt_zip(tbl, ds), gr.update(visible=True)),
-    inputs=[table, dataset_name], outputs=[txt_zip, txt_zip]
-)