Update app.py

app.py

@@ -65,38 +65,6 @@ def process_file(file_obj):
             text = f.read()
     return text, name, get_empty_df()
 
-def build_transcript_html(text, df):
-    display_text = text
-    if df is not None and not df.empty:
-        for _, row in df.iterrows():
-            seg = row["Coded Segment"]
-            color = COLOR_MAP.get(row["Code"], "yellow")
-            display_text = display_text.replace(
-                seg, f"<span style='background-color:{color}'>{seg}</span>", 1
-            )
-    safe_text = display_text.replace("\n", "<br>")
-    html = f"""
-    <div id='transcript' style='white-space: pre-wrap; font-size:16px; line-height:1.5; max-height:600px; overflow:auto; border:1px solid #ccc; padding:5px;'>
-        {safe_text}
-    </div>
-    <script>
-    document.addEventListener('DOMContentLoaded', function() {{
-        const transcript = document.getElementById('transcript');
-        transcript.addEventListener('mouseup', function() {{
-            const sel = window.getSelection().toString();
-            if(sel.length>0){{
-                const state_input = document.getElementById('selected_segment_state');
-                if(state_input) {{
-                    state_input.value = sel;
-                    state_input.dispatchEvent(new Event("input", {{bubbles:true}}));
-                }}
-            }}
-        }});
-    }});
-    </script>
-    """
-    return html
-
 def apply_code(df, segment, code, file_id, *metadata_values):
     if not file_id:
         return df, "⚠️ Upload a file first"
@@ -121,13 +89,14 @@ def export_excel(df):
     df.to_excel(path, index=False)
     return path, "Excel ready"
 
+# ----------------------------
+# GRADIO APP
+# ----------------------------
 with gr.Blocks() as demo:
 
-    # States
     full_text = gr.State("")
     file_id = gr.State("")
     coded_df_state = gr.State(get_empty_df())
-    selected_segment_state = gr.State("")
     code_categories_state = gr.State(DEFAULT_CODES)
 
     # Metadata
@@ -136,22 +105,23 @@ with gr.Blocks() as demo:
         for k, lbl in METADATA_FIELDS.items():
             metadata_inputs.append(gr.Textbox(label=lbl))
 
-    # Main row: transcript left, coding right
+    # Main interface
     with gr.Row():
-        # Left: transcript
+        # Left: transcript selection
         with gr.Column(scale=3):
-            transcript_html = gr.HTML()
-            hidden_segment = gr.Textbox(
-                interactive=False, visible=False, elem_id="selected_segment_state"
+            transcript_text = gr.Textbox(
+                label="Transcript (Select text below to code)",
+                lines=25,
+                interactive=True,
+                placeholder="Select a segment of this text..."
             )
 
-        # Right: controls
+        # Right: coding
         with gr.Column(scale=2):
             gr.Markdown("## 🏷️ Code Category")
             code_dropdown = gr.Dropdown(label="Select code", choices=DEFAULT_CODES)
             code_input = gr.Textbox(label="Or type new code")
             add_code_btn = gr.Button("Add new code")
-
             apply_btn = gr.Button("Apply code to selected text")
 
             gr.Markdown("## 📊 Coded Segments")
@@ -160,48 +130,26 @@ with gr.Blocks() as demo:
             export_btn = gr.Button("Export XLSX")
             export_file = gr.File(visible=False)
 
-            file_input = gr.File(
-                label="Upload transcript", file_types=[".docx", ".vtt", ".txt"]
-            )
+            file_input = gr.File(label="Upload transcript", file_types=[".docx", ".vtt", ".txt"])
             status = gr.Textbox(label="Status", value="Ready")
 
     # Callbacks
-    file_input.change(
-        fn=process_file, inputs=file_input, outputs=[full_text, file_id, coded_df_state]
-    )
-
-    def update_transcript(text, df):
-        return build_transcript_html(text, df)
+    file_input.change(fn=process_file, inputs=file_input, outputs=[transcript_text, file_id, coded_df_state])
 
-    full_text.change(
-        update_transcript, inputs=[full_text, coded_df_state], outputs=transcript_html
-    )
-    coded_df_state.change(
-        update_transcript, inputs=[full_text, coded_df_state], outputs=transcript_html
-    )
-
-    add_code_btn.click(
-        add_new_code, inputs=[code_input, code_categories_state], outputs=[code_categories_state]
-    )
-    code_categories_state.change(
-        lambda codes: gr.update(choices=codes),
-        inputs=code_categories_state,
-        outputs=code_dropdown,
-    )
+    add_code_btn.click(add_new_code, inputs=[code_input, code_categories_state], outputs=[code_categories_state])
+    code_categories_state.change(lambda codes: gr.update(choices=codes), inputs=code_categories_state, outputs=code_dropdown)
 
-    # Apply button uses hidden textbox as input
     apply_btn.click(
         apply_code,
-        inputs=[coded_df_state, hidden_segment, code_dropdown, file_id] + metadata_inputs,
+        inputs=[coded_df_state, transcript_text, code_dropdown, file_id] + metadata_inputs,
         outputs=[coded_df_state, status],
     )
 
     coded_df_state.change(lambda df: df, inputs=coded_df_state, outputs=table)
-
     export_btn.click(export_excel, inputs=coded_df_state, outputs=[export_file, status]).then(
         lambda f: gr.update(visible=f is not None),
         inputs=export_file,
-        outputs=export_file,
+        outputs=export_file
     )
 
 demo.launch()