w3m

app.py

@@ -5,7 +5,7 @@ A unified HTML parser optimized for extracting mathematical content.
 """
 
 import gradio as gr
-from ultradata_math_parser.parsers.unified_parser import UnifiedParser
+from ultradata_math_parser import GeneralParser
 
 
 def parse_html(
@@ -14,9 +14,10 @@ def parse_html(
     process_math: bool = True,
     include_tables: bool = True,
     enable_forum_assembly: bool = True,
+    html_type: str = "unified",
 ) -> dict:
     """
-    Parse HTML content using UnifiedParser.
+    Parse HTML content using GeneralParser.
     
     Args:
         html_content: Raw HTML string to parse
@@ -24,6 +25,7 @@ def parse_html(
         process_math: Whether to process and convert math expressions
         include_tables: Whether to preserve table elements
         enable_forum_assembly: Whether to enable forum post assembly
+        html_type: Parser type (unified/article/forum)
     
     Returns:
         Dictionary containing parsed results
@@ -32,6 +34,7 @@ def parse_html(
         return {
             "title": "",
             "html": "",
+            "text": "",
             "text_length": 0,
             "xp_num": "",
             "fallback_strategy": "",
@@ -39,7 +42,7 @@ def parse_html(
             "error": "Please provide HTML content to parse.",
         }
     
-    parser = UnifiedParser()
+    parser = GeneralParser()
     
     try:
         result = parser.extract(
@@ -48,11 +51,13 @@ def parse_html(
             process_math=process_math,
             include_tables=include_tables,
             enable_forum_assembly=enable_forum_assembly,
+            html_type=html_type,
         )
         
         return {
             "title": result.get("title", ""),
             "html": result.get("html", ""),
+            "text": result.get("text", ""),
             "text_length": result.get("text_length", 0),
             "xp_num": result.get("xp_num", ""),
             "fallback_strategy": result.get("fallback_strategy", ""),
@@ -63,6 +68,7 @@ def parse_html(
         return {
             "title": "",
             "html": "",
+            "text": "",
             "text_length": 0,
             "xp_num": "",
             "fallback_strategy": "",
@@ -79,6 +85,7 @@ def format_output(result: dict) -> tuple:
             "",
             "",
             "",
+            "",
         )
     
     # Build metadata string
@@ -94,11 +101,12 @@ def format_output(result: dict) -> tuple:
         metadata,
         result.get("title", ""),
         result.get("html", ""),
+        result.get("text", ""),
         result.get("html", ""),  # For HTML preview
     )
 
 
-def process_input(html_content, base_url, process_math, include_tables, enable_forum):
+def process_input(html_content, base_url, process_math, include_tables, enable_forum, html_type):
     """Main processing function for Gradio interface."""
     result = parse_html(
         html_content=html_content,
@@ -106,6 +114,7 @@ def process_input(html_content, base_url, process_math, include_tables, enable_f
         process_math=process_math,
         include_tables=include_tables,
         enable_forum_assembly=enable_forum,
+        html_type=html_type,
     )
     return format_output(result)
 
@@ -298,6 +307,12 @@ with gr.Blocks(css=custom_css, title="UltraData Math Parser") as demo:
             )
             
             with gr.Accordion("⚙️ Advanced Options", open=False):
+                html_type = gr.Radio(
+                    choices=["unified", "article", "forum"],
+                    value="unified",
+                    label="Parser Type",
+                    info="Select the parsing strategy",
+                )
                 process_math = gr.Checkbox(
                     label="Process Math Expressions",
                     value=True,
@@ -340,6 +355,13 @@ with gr.Blocks(css=custom_css, title="UltraData Math Parser") as demo:
                         max_lines=20,
                         interactive=False,
                     )
+                with gr.TabItem("📄 Plain Text"):
+                    text_output = gr.Textbox(
+                        label="Plain Text (w3m rendered)",
+                        lines=12,
+                        max_lines=20,
+                        interactive=False,
+                    )
                 with gr.TabItem("👁️ Preview"):
                     preview_output = gr.HTML(
                         label="HTML Preview",
@@ -348,16 +370,16 @@ with gr.Blocks(css=custom_css, title="UltraData Math Parser") as demo:
     # Event handlers
     parse_btn.click(
         fn=process_input,
-        inputs=[html_input, base_url_input, process_math, include_tables, enable_forum],
-        outputs=[metadata_output, title_output, html_output, preview_output],
+        inputs=[html_input, base_url_input, process_math, include_tables, enable_forum, html_type],
+        outputs=[metadata_output, title_output, html_output, text_output, preview_output],
     )
     
     def clear_all():
-        return "", "", "", "", "", ""
+        return "", "", "", "", "", "", ""
     
     clear_btn.click(
         fn=clear_all,
-        outputs=[html_input, base_url_input, metadata_output, title_output, html_output, preview_output],
+        outputs=[html_input, base_url_input, metadata_output, title_output, html_output, text_output, preview_output],
     )
     
     # Footer info

packages.txt

@@ -0,0 +1 @@
+w3m