feat: single-button pipeline — Register + OCR + Review in one click

smoke_signal_tab.py

@@ -1867,6 +1867,106 @@ def _run_tesseract_batch(
     return outputs
 
 
+def run_full_pipeline(
+    files,
+    rights_class: str,
+    notes: str,
+    book_code_hint: str,
+    publication_year: str,
+    scope_include: str,
+    scope_exclude: str,
+    safe_title: str,
+):
+    """
+    Single-button full pipeline:
+    1. Ingest + profile PDF
+    2. Update rights + save page scope
+    3. Run OCR
+    Yields log updates throughout so UI stays live.
+    """
+    log = []
+
+    def _line(msg):
+        from datetime import datetime
+        return f"[{datetime.now().strftime('%H:%M:%S')}] {msg}"
+
+    # Step 1: Ingest + profile
+    log.append(_line("① Registering PDF..."))
+    yield _ingest_status_html("idle"), "\n".join(log), 0
+
+    try:
+        status_html, manifest_df, ingest_log = ingest_pdfs(
+            files, rights_class, notes, book_code_hint, publication_year
+        )
+        for line in str(ingest_log).splitlines():
+            if line.strip():
+                log.append(line)
+    except Exception as e:
+        log.append(_line(f"✗ Ingest failed: {e}"))
+        yield _ingest_status_html("idle"), "\n".join(log), 0
+        return
+
+    yield status_html, "\n".join(log), 0
+
+    # Step 2: Save page scope + rights (resolve book_id from manifest)
+    df = load_manifest_df()
+    if df.empty:
+        log.append(_line("✗ No books in manifest after ingest"))
+        yield _ingest_status_html("idle"), "\n".join(log), 0
+        return
+
+    # Get the book we just registered/updated
+    latest_book_id = str(df.iloc[-1]["book_id"])
+
+    # Update rights if not unknown
+    if rights_class and rights_class != "unknown":
+        try:
+            _, _, rights_log = update_rights(latest_book_id, rights_class)
+            for line in str(rights_log).splitlines():
+                if line.strip() and "not found" not in line.lower():
+                    log.append(line)
+        except Exception as e:
+            log.append(_line(f"⚠ Rights update: {e}"))
+
+    # Save page scope
+    if scope_include or scope_exclude or safe_title:
+        log.append(_line(f"② Saving page scope for {latest_book_id}..."))
+        yield _ingest_status_html("done"), "\n".join(log), 0
+        try:
+            _, _, scope_log = save_book_scope(
+                latest_book_id, scope_include, scope_exclude, safe_title
+            )
+            for line in str(scope_log).splitlines():
+                if line.strip():
+                    log.append(line)
+        except Exception as e:
+            log.append(_line(f"⚠ Scope save: {e}"))
+
+    yield _ingest_status_html("done"), "\n".join(log), 0
+
+    # Step 3: Run OCR
+    log.append(_line("③ Starting OCR..."))
+    yield _ingest_status_html("done"), "\n".join(log), 0
+
+    try:
+        ocr_gen = run_ocr(scope_include, scope_exclude, replace_book_queue=True)
+        for ocr_status_html, ocr_log in ocr_gen:
+            for line in str(ocr_log).splitlines():
+                if line.strip() and line not in log:
+                    log.append(line)
+            yield _ingest_status_html("done"), "\n".join(log), 0
+    except Exception as e:
+        import traceback
+        log.append(_line(f"✗ OCR error: {e}"))
+        log.append(traceback.format_exc())
+        yield _ingest_status_html("done"), "\n".join(log), 0
+        return
+
+    log.append(_line("✓ Pipeline complete — go to Review tab"))
+    yield _ingest_status_html("done"), "\n".join(log), 1  # 1 = triggers review load
+
+
+
 def run_ocr(page_selection: str = "", page_exclusion: str = "", replace_book_queue: bool = True) -> tuple:
     """Run Surya OCR on all profiled PDFs."""
     df = load_manifest_df()
@@ -2714,14 +2814,7 @@ def smoke_signal_tab():
                 gr.HTML("""<div class="ss-panel-header" style="padding:20px 0 0 0">
                     <div class="ss-panel-icon">📥</div>
                     <div><p class="ss-panel-title">Source Registry</p>
-                    <p class="ss-panel-sub">Upload PDFs · Register · Auto-profile · Rights class</p></div>
-                </div>
-                <div style="background:#1a2535;border:1px solid #30363d;border-radius:8px;padding:14px 18px;margin:12px 0;font-size:13px;color:#8b949e;line-height:2.0">
-                    <b style="color:#e6edf3">&#9654; How to run OCR — follow this order:</b><br>
-                    <span style="color:#3fb950">1.</span> Drop your PDF &nbsp;·&nbsp; set <b style="color:#e6edf3">Rights Class</b> &nbsp;·&nbsp; enter <b style="color:#e6edf3">Book Code</b> + <b style="color:#e6edf3">Year</b> &nbsp;·&nbsp; click <b style="color:#f0883e">Register + Auto-Profile</b><br>
-                    <span style="color:#3fb950">2.</span> Enter <b style="color:#e6edf3">Story Pages Include</b> (e.g. 7-27) and <b style="color:#e6edf3">Story Pages Exclude</b> (e.g. 1,2,3) &nbsp;·&nbsp; click <b style="color:#f0883e">Save Book Pages</b><br>
-                    <span style="color:#3fb950">3.</span> Go to tab <b style="color:#e6edf3">③ OCR</b> &nbsp;·&nbsp; click <b style="color:#f0883e">Run OCR</b> &nbsp;·&nbsp; wait for the log to complete (may take 1-2 min)<br>
-                    <span style="color:#3fb950">4.</span> Go to tab <b style="color:#e6edf3">④ Review</b> &nbsp;·&nbsp; click <b style="color:#f0883e">↻ Load Review Queue</b> &nbsp;·&nbsp; accept / edit / reject pages
+                    <p class="ss-panel-sub">Drop PDF · Set pages · Click Run — everything else is automatic</p></div>
                 </div>""")
 
                 ingest_status = gr.HTML(_ingest_status_html("idle"))
@@ -2729,14 +2822,14 @@ def smoke_signal_tab():
                 with gr.Row():
                     with gr.Column(scale=2):
                         pdf_upload = gr.File(
-                            label="Drop picture-book PDFs here",
+                            label="① Drop PDF here",
                             file_types=[".pdf"],
                             file_count="multiple",
                             type="filepath",
                         )
                     with gr.Column(scale=1):
                         rights_dd = gr.Dropdown(
-                            label="Rights Class",
+                            label="② Rights Class",
                             choices=["public-domain","licensed-owned","controlled-internal","unknown"],
                             value="unknown",
                         )
@@ -2751,47 +2844,45 @@ def smoke_signal_tab():
                             lines=1,
                         )
                         ingest_notes = gr.Textbox(label="Notes", placeholder="Source, edition, etc.", lines=2)
-                        ingest_btn = gr.Button("Register + Auto-Profile →", elem_classes=["ss-btn-run"])
+
+                with gr.Row():
+                    scope_include = gr.Textbox(
+                        label="③ Story Pages Include",
+                        placeholder="e.g. 7-27  or  7,8,9,11-27  or  all",
+                        scale=2,
+                    )
+                    scope_exclude = gr.Textbox(
+                        label="Story Pages Exclude",
+                        placeholder="e.g. 1,2,3,17,25",
+                        scale=2,
+                    )
+                    scope_title = gr.Textbox(
+                        label="Safe Book Title",
+                        placeholder="auto-filled from filename",
+                        scale=2,
+                    )
+
+                run_pipeline_btn = gr.Button(
+                    "④ Register + OCR + Send to Review →",
+                    elem_classes=["ss-btn-run"],
+                )
+                pipeline_done = gr.State(0)
+                ingest_log = gr.Textbox(label="Log", lines=10, interactive=False, elem_classes=["ss-log"])
+                ingest_copy_log_btn = gr.Button("Copy Log", size="sm")
 
                 manifest_table = gr.DataFrame(
                     label="Source Manifest",
                     interactive=False,
                     wrap=True,
                 )
-                ingest_log = gr.Textbox(label="Log", lines=6, interactive=False, elem_classes=["ss-log"])
-                ingest_copy_log_btn = gr.Button("Copy Log", size="sm")
                 refresh_btn = gr.Button("↻ Refresh Manifest", size="sm")
-                refresh_btn.click(lambda: load_manifest_df(), outputs=[manifest_table])
-                ingest_copy_log_btn.click(
-                    fn=None,
-                    inputs=[ingest_log],
-                    outputs=[],
-                    js="(logTxt) => { if (navigator && navigator.clipboard) { navigator.clipboard.writeText(logTxt || ''); } return []; }",
-                )
 
-                gr.Markdown("**Update rights class for existing book:**")
-                with gr.Row():
-                    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"],
-                        value="licensed-owned",
-                        scale=1,
-                    )
-                    update_rights_btn = gr.Button("Update Rights", size="sm", scale=1)
-                update_rights_btn.click(
-                    update_rights,
-                    inputs=[update_book_id, update_rights_dd],
-                    outputs=[ingest_status, manifest_table, ingest_log],
-                )
-
-                gr.Markdown("**Save Book Pages + Safe Title (persisted):**")
-                with gr.Row():
-                    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)
-                scope_save_btn = gr.Button("Save Book Pages →", size="sm")
+                # Keep hidden compat fields for wiring that references them later
+                update_book_id = gr.Textbox(visible=False, value="")
+                scope_book_id  = gr.Textbox(visible=False, value="")
+                update_rights_dd = gr.Dropdown(choices=["public-domain","licensed-owned","controlled-internal","unknown"], value="licensed-owned", visible=False)
+                scope_save_btn = gr.Button(visible=False)
+                ingest_btn     = gr.Button(visible=False)
 
                 # ── Auto-populate: book code + year → both book ID fields ──────
                 def _make_book_id(code, year):
@@ -2829,10 +2920,24 @@ def smoke_signal_tab():
                     outputs=[scope_title],
                 )
 
-                ingest_btn.click(
-                    ingest_pdfs,
-                    inputs=[pdf_upload, rights_dd, ingest_notes, book_code_input, pub_year_input],
-                    outputs=[ingest_status, manifest_table, ingest_log],
+                # Main pipeline button — does everything
+                run_pipeline_btn.click(
+                    run_full_pipeline,
+                    inputs=[pdf_upload, rights_dd, ingest_notes, book_code_input, pub_year_input,
+                            scope_include, scope_exclude, scope_title],
+                    outputs=[ingest_status, ingest_log, pipeline_done],
+                    show_progress="minimal",
+                    trigger_mode="once",
+                )
+                # Refresh manifest after pipeline completes
+                pipeline_done.change(lambda: load_manifest_df(), outputs=[manifest_table])
+
+                refresh_btn.click(lambda: load_manifest_df(), outputs=[manifest_table])
+                ingest_copy_log_btn.click(
+                    fn=None,
+                    inputs=[ingest_log],
+                    outputs=[],
+                    js="(logTxt) => { if (navigator && navigator.clipboard) { navigator.clipboard.writeText(logTxt || ''); } return []; }",
                 )
                 gr.HTML('<div style="height:16px"></div>')
                 # Auto-load removed — use Refresh button instead to avoid SSR hang
@@ -3037,6 +3142,7 @@ def smoke_signal_tab():
                 )
                 review_tab.select(load_review, outputs=review_load_outputs)
                 ocr_done.change(load_review, outputs=review_load_outputs)
+                pipeline_done.change(load_review, outputs=review_load_outputs)
 
                 accept_btn.click(do_accept, inputs=[current_idx, final_text_box, reviewer_name, reason_code], outputs=action_outputs)
                 edit_btn.click(do_accept, inputs=[current_idx, final_text_box, reviewer_name, reason_code], outputs=action_outputs)