Upload app.py

app.py

@@ -262,6 +262,16 @@ def upload_document(file_obj) -> Tuple[bool, Optional[str], Optional[int], str]:
     return _upload_text(text_content, filename)
 
 
+def upload_pasted_text(text: str, filename: str = "pasted.txt") -> Tuple[bool, Optional[str], Optional[int], str]:
+    """Index user-pasted text directly."""
+    if not text or not text.strip():
+        return False, None, None, "⚠️ Please paste some text first."
+    filename = (filename or "pasted.txt").strip() or "pasted.txt"
+    if not filename.endswith(".txt"):
+        filename += ".txt"
+    return _upload_text(text, filename)
+
+
 def load_selected_samples(selected_labels: list) -> Tuple[bool, Optional[str], Optional[int], str]:
     """Combine and upload one or more sample annual reports."""
     if not selected_labels:
@@ -322,10 +332,10 @@ def query_backend(message: str) -> Generator[str, None, None]:
 def update_index_display(filename: Optional[str], is_indexed: bool, char_count: Optional[int] = None) -> str:
     if is_indexed and filename:
         detail = f" ({char_count:,} characters)" if char_count else ""
-        return f"📄 **Indexed Document:** {filename}{detail}"
+        return f"📄 **Indexed** {filename}{detail}"
     elif is_indexed:
-        return "📄 **Indexed Document:** ✅ Document indexed"
-    return "📄 **Indexed Document:** None"
+        return "📄 **Indexed** (document ready)"
+    return "📄 *No document indexed yet*"
 
 
 def load_index_state() -> Tuple[bool, Optional[str], Optional[int], str]:
@@ -422,117 +432,141 @@ def create_app() -> gr.Blocks:
         with gr.Column(elem_id="chat-page", visible=False) as chat_page:
             gr.Markdown(
                 """
-                <div style="text-align: center; margin-bottom: 1.5em;">
+                <div style="text-align: center; margin-bottom: 1em;">
                     <h1>📚 RAG Chat Interface</h1>
-                    <p style="color: #666;">Load sample reports or upload your own, then ask questions.</p>
                 </div>
                 """
             )
 
-            index_status = gr.Markdown("📄 **Indexed Document:** None")
-
-            # ── Sample report section ─────────────────────────────────────
-            gr.Markdown("### 📋 Sample Annual Reports")
-            sample_selector = gr.CheckboxGroup(
-                choices=list(SAMPLE_LABELS.keys()),
-                value=[],
-                label="Select one or more reports to index",
-                interactive=True,
-            )
+            index_status = gr.Markdown("📄 *No document indexed yet*")
 
-            load_samples_btn = gr.Button("🚀 Load Selected Reports", variant="primary")
-
-            sample_status = gr.Markdown("")
+            # Define respond up-front so any button below can reference it
+            def respond(message, history, is_indexed):
+                history = history or []
+                if not message or not message.strip():
+                    yield history, message
+                    return
+                user_msg = {"role": "user", "content": message}
+                if not is_indexed:
+                    yield history + [user_msg, {"role": "assistant", "content": "⚠️ Please index a document first (see section 1)."}], ""
+                    return
+                yield history + [user_msg, {"role": "assistant", "content": "🤔 Thinking..."}], ""
+                final_answer = "🤔 Thinking..."
+                for chunk in query_backend(message):
+                    final_answer = chunk
+                    yield history + [user_msg, {"role": "assistant", "content": final_answer}], ""
 
-            with gr.Accordion("📖 Preview Selected Documents", open=False):
+            # Returns (indexed?, filename, char_count, display_md, preview_text)
+            def _after_index(success, filename, char_count, source_text):
+                display = update_index_display(filename, success, char_count)
+                preview = source_text if success else ""
+                return success, filename, char_count, display, preview
+
+            # ── Section 1: Select files to index ──────────────────────────
+            with gr.Accordion("📥 1. Select files to index", open=True):
+                with gr.Tabs():
+                    # — Tab A: sample files —
+                    with gr.Tab("📋 Sample reports"):
+                        sample_selector = gr.CheckboxGroup(
+                            choices=list(SAMPLE_LABELS.keys()),
+                            value=[],
+                            label="Pick one or more sample annual reports",
+                            interactive=True,
+                        )
+                        load_samples_btn = gr.Button("🚀 Index Selected Reports", variant="primary")
+
+                    # — Tab B: paste text —
+                    with gr.Tab("✏️ Paste text"):
+                        paste_input = gr.Textbox(
+                            label="Paste text here",
+                            lines=10,
+                            max_lines=20,
+                            placeholder="Paste any text content you want to ask questions about…",
+                        )
+                        paste_filename = gr.Textbox(
+                            label="Filename (optional)",
+                            value="pasted.txt",
+                            lines=1,
+                        )
+                        load_paste_btn = gr.Button("🚀 Index Pasted Text", variant="primary")
+
+                    # — Tab C: upload file —
+                    with gr.Tab("📁 Upload file"):
+                        file_input = gr.File(
+                            label="Upload .txt file",
+                            file_types=[".txt"],
+                            file_count="single",
+                        )
+                        upload_btn = gr.Button("🚀 Index Uploaded File", variant="primary")
+
+            # ── Section 2: Preview indexed text ───────────────────────────
+            with gr.Accordion("📖 2. Preview indexed text", open=False):
                 preview_box = gr.Textbox(
                     value="",
                     lines=18,
                     max_lines=18,
                     interactive=False,
                     show_label=False,
-                    placeholder="Select reports above to preview their content here.",
+                    placeholder="Index a document in section 1 to see its content here.",
                 )
 
-            sample_selector.change(
-                fn=lambda labels: build_preview_text(labels)[0],
-                inputs=[sample_selector],
-                outputs=[preview_box],
-            )
+            # ── Section 3: Chat ────────────────────────────────────────────
+            with gr.Accordion("💬 3. Ask questions", open=True):
+                gr.Markdown("<small style='color:#888'>Quick questions:</small>")
+                with gr.Row():
+                    q_btns = [gr.Button(q, size="sm") for q in SAMPLE_QUESTIONS]
+
+                chatbot = gr.Chatbot(height=420, show_label=False, value=[])
+
+                with gr.Row():
+                    question_input = gr.Textbox(
+                        placeholder="Type your question and press Enter…",
+                        lines=1,
+                        scale=9,
+                        show_label=False,
+                        container=False,
+                    )
+                    send_btn = gr.Button("Send ▶", scale=1, variant="primary")
 
-            def _after_load(success, filename, char_count, message):
-                display = update_index_display(filename, success, char_count)
-                return success, filename, char_count, message, display
+            # ── Wiring for Section 1 actions ──────────────────────────────
+            def _samples_action(selected_labels):
+                success, filename, char_count, _msg = load_selected_samples(selected_labels)
+                source_text = build_preview_text(selected_labels)[0] if success else ""
+                return _after_index(success, filename, char_count, source_text)
 
             load_samples_btn.click(
-                fn=load_selected_samples,
+                fn=_samples_action,
                 inputs=[sample_selector],
-                outputs=[document_indexed, indexed_filename, indexed_char_count, sample_status],
-            ).then(
-                fn=_after_load,
-                inputs=[document_indexed, indexed_filename, indexed_char_count, sample_status],
-                outputs=[document_indexed, indexed_filename, indexed_char_count, sample_status, index_status],
+                outputs=[document_indexed, indexed_filename, indexed_char_count, index_status, preview_box],
             )
 
-            # ── Own file upload ────────────────────────────────────────────
-            gr.Markdown("### 📁 Or Upload Your Own Document")
-            with gr.Row():
-                with gr.Column(scale=2):
-                    file_input = gr.File(
-                        label="Upload .txt file",
-                        file_types=[".txt"],
-                        file_count="single",
-                    )
-                with gr.Column(scale=1):
-                    upload_btn = gr.Button("📤 Build Index", variant="primary", size="lg")
+            def _paste_action(text, filename):
+                success, fn_used, char_count, _msg = upload_pasted_text(text, filename)
+                return _after_index(success, fn_used, char_count, text if success else "")
 
-            upload_status = gr.Markdown("**Status:** Waiting for document...")
+            load_paste_btn.click(
+                fn=_paste_action,
+                inputs=[paste_input, paste_filename],
+                outputs=[document_indexed, indexed_filename, indexed_char_count, index_status, preview_box],
+            )
+
+            def _upload_action(file_obj):
+                success, fn_used, char_count, _msg = upload_document(file_obj)
+                source_text = ""
+                if success and file_obj is not None and os.path.exists(file_obj.name):
+                    try:
+                        with open(file_obj.name, "r", encoding="utf-8") as f:
+                            source_text = f.read()
+                    except Exception:
+                        source_text = ""
+                return _after_index(success, fn_used, char_count, source_text)
 
             upload_btn.click(
-                fn=upload_document,
+                fn=_upload_action,
                 inputs=[file_input],
-                outputs=[document_indexed, indexed_filename, indexed_char_count, upload_status],
-            ).then(
-                fn=_after_load,
-                inputs=[document_indexed, indexed_filename, indexed_char_count, upload_status],
-                outputs=[document_indexed, indexed_filename, indexed_char_count, upload_status, index_status],
+                outputs=[document_indexed, indexed_filename, indexed_char_count, index_status, preview_box],
             )
 
-            # ── Chat section ───────────────────────────────────────────────
-            gr.Markdown("### 💬 Ask Questions")
-
-            # Define respond BEFORE wiring any buttons that reference it
-            def respond(message, history, is_indexed):
-                history = history or []
-                if not message or not message.strip():
-                    yield history, message
-                    return
-                user_msg = {"role": "user", "content": message}
-                if not is_indexed:
-                    yield history + [user_msg, {"role": "assistant", "content": "⚠️ Please load a sample report or upload a document first."}], ""
-                    return
-                yield history + [user_msg, {"role": "assistant", "content": "🤔 Thinking..."}], ""
-                final_answer = "🤔 Thinking..."
-                for chunk in query_backend(message):
-                    final_answer = chunk
-                    yield history + [user_msg, {"role": "assistant", "content": final_answer}], ""
-
-            gr.Markdown("<small style='color:#888'>Quick questions:</small>")
-            with gr.Row():
-                q_btns = [gr.Button(q, size="sm") for q in SAMPLE_QUESTIONS]
-
-            chatbot = gr.Chatbot(height=420, show_label=False, value=[])
-
-            with gr.Row():
-                question_input = gr.Textbox(
-                    placeholder="Type your question and press Enter…",
-                    lines=1,
-                    scale=9,
-                    show_label=False,
-                    container=False,
-                )
-                send_btn = gr.Button("Send ▶", scale=1, variant="primary")
-
             # Quick-question buttons: fill the textbox, then auto-submit
             for btn, q in zip(q_btns, SAMPLE_QUESTIONS):
                 btn.click(