adding adding source fetcher

app.py

@@ -5,6 +5,8 @@ from markdown.extensions.fenced_code import FencedCodeExtension
 from markdown.extensions.toc import TocExtension
 from markdown.extensions.attr_list import AttrListExtension
 from markdown.extensions.codehilite import CodeHiliteExtension
+import requests
+from bs4 import BeautifulSoup
 
 # For ReaderLM-2 functionality
 from transformers import pipeline
@@ -36,7 +38,23 @@ def render_markdown(md_text):
     )
 
 ######################################
-# 2) READERLM-2 FUNCTIONALITY        #
+# 2) VIEW SOURCE INFO FUNCTIONALITY  #
+######################################
+def view_source_info(url):
+    """
+    Fetch the HTML source of the given URL.
+    - Supports `view-source:` prefix or plain URLs.
+    """
+    if url.startswith("view-source:"):
+        url = url.replace("view-source:", "").strip()
+    if not url.startswith(("http://", "https://")):
+        url = "https://" + url
+    print(f"Fetching source for URL: {url}...")  # Debug log
+    response = requests.get(url)
+    return response.text
+
+######################################
+# 3) READERLM-2 FUNCTIONALITY        #
 ######################################
 # Load the JinaAI ReaderLM-v2 model
 model_name = "jinaai/ReaderLM-v2"
@@ -67,7 +85,7 @@ def convert_html(html_input, output_format, custom_prompt=None):
     return converted_output
 
 ######################################
-# 3) READERLM-1 FUNCTIONALITY        #
+# 4) READERLM-1 FUNCTIONALITY        #
 ######################################
 # Prepare models and tokenizers for ReaderLM-1
 print("Loading ReaderLM-1 models and tokenizers...")  # Debug log
@@ -178,7 +196,29 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme", css=css) as demo:
         md_input.change(fn=render_markdown, inputs=md_input, outputs=md_output)
 
     ########################################################
-    # TAB 2: ReaderLM-2 Converter (HTML → Markdown/JSON)
+    # TAB 2: View Source Info
+    ########################################################
+    with gr.Tab("View Source Info"):
+        gr.Markdown("## View HTML Source Code")  # Tab description
+
+        source_input = gr.Textbox(
+            label="Enter URL (with or without 'view-source:')",
+            placeholder="e.g., https://example.com or view-source:example.com"
+        )
+        source_output = gr.Textbox(
+            label="HTML Source Code",
+            lines=20
+        )
+        source_button = gr.Button("Fetch Source")  # Button to fetch source
+
+        source_button.click(
+            fn=view_source_info,
+            inputs=source_input,
+            outputs=source_output
+        )
+
+    ########################################################
+    # TAB 3: ReaderLM-2 Converter (HTML → Markdown/JSON)
     ########################################################
     with gr.Tab("ReaderLM-2 Converter"):
         gr.Markdown("## HTML to Markdown/JSON Converter (ReaderLM-v2)")  # Tab description
@@ -231,7 +271,7 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme", css=css) as demo:
         )
 
     ########################################################
-    # TAB 3: ReaderLM-1 HTML-to-Markdown
+    # TAB 4: ReaderLM-1 HTML-to-Markdown
     ########################################################
     with gr.Tab("ReaderLM-1 Converter"):
         gr.Markdown("""