dep update

README.md

@@ -4,7 +4,7 @@ emoji: 🐨
 colorFrom: red
 colorTo: yellow
 sdk: gradio
-sdk_version: 5.6.0
+sdk_version: 5.9.1
 app_file: app.py
 pinned: false
 license: mit
@@ -14,9 +14,24 @@ license: mit
 
 ## Usage
 
+Run the Gradio app
 ```bash
-# Run the Gradio app 
-gradio app.py  # auto reload
-python app.py  # static
+# auto reload
+gradio app.py
 
+# manual reload
+python app.py
+```
+
+Install with virtual conda env:
+```bash
+conda create -n seed-crawl-annotator python=3.10
+conda activate seed-crawl-annotator
+pip install -r requirements.txt
+```
+
+
+Build Docker image
+```bash
+docker build -t seed-crawl-annotator  .
 ```

app.py

@@ -35,7 +35,7 @@ def fetch_screenshot_and_text_from_url(url):
     screen_width = 1080
     height = 350
     text = ""
-    
+
     if OFFLINE:
         screenshot = Image.new('RGB', (350, height))
         text = f"Some dummy text for {url} (offline mode enabled)"
@@ -122,9 +122,15 @@ with gr.Blocks(fill_height=True) as demo:
             load_btn = gr.Button("Annotate URL", variant="primary", interactive=True)
 
         with gr.Row():
-            extracted_text = gr.Textbox(label="Extracted text", max_lines=15, lines=15, visible=False, placeholder="Click on `Load URL` to fetch Web page's text content.")
+            extracted_text = gr.Textbox(
+                label="Extracted text", 
+                max_lines=15, 
+                lines=15, 
+                visible=True, 
+                placeholder="Click on `Load URL` to fetch Web page's text content."
+            )
             
-            screenshot_scrollable = gr.HTML(visible=False)
+            screenshot_scrollable = gr.HTML("", visible=False)
 
         with gr.Column(visible=False) as output_col:
             with gr.Row():
@@ -186,7 +192,7 @@ with gr.Blocks(fill_height=True) as demo:
             return {
                 url_field: "",
                 output_col: gr.update(visible=False),
-                extracted_text: gr.update(value=None, visible=False),
+                extracted_text: gr.update(value=None, visible=True),
                 screenshot_scrollable: gr.update(value="", visible=False),
             }
 

requirements.txt

@@ -1,5 +1,5 @@
 selenium >=4.0.0, < 5.0.0
-gradio>=3.40.1
+gradio==5.9.1
 Pillow>=8.3.1,<9.0
-trafilatura
+trafilatura==2.0.0
 gradio[oauth]