Update app.py

app.py

@@ -75,16 +75,34 @@ def register_local_fonts():
     return sorted(text_font_names), emoji_font_name
 
 def apply_emoji_font(text: str, emoji_font_name: str) -> str:
-    """Wraps emoji characters in a <font> tag to use the dedicated emoji font."""
+    """
+    Intelligently wraps emoji characters in a <font> tag, preserving existing HTML-like tags.
+    This prevents invalid nested tags like <b><font>...</font></b> which ReportLab handles poorly.
+    """
     if not emoji_font_name:
         return text
 
+    # Regex to find emojis
     emoji_pattern = re.compile(f"([{re.escape(''.join(map(chr, range(0x1f600, 0x1f650))))}"
                                f"{re.escape(''.join(map(chr, range(0x1f300, 0x1f5ff))))}"
                                f"{re.escape(''.join(map(chr, range(0x1f900, 0x1f9ff))))}"
                                f"{re.escape(''.join(map(chr, range(0x2600, 0x26ff))))}"
                                f"{re.escape(''.join(map(chr, range(0x2700, 0x27bf))))}]+)")
-    return emoji_pattern.sub(fr'<font name="{emoji_font_name}">\1</font>', text)
+
+    # Regex to split the string by existing tags (<b>, <i>)
+    tag_pattern = re.compile(r"(<[^>]+>)")
+    parts = tag_pattern.split(text)
+    
+    result = []
+    for part in parts:
+        if tag_pattern.match(part):
+            # It's a tag, append it as is
+            result.append(part)
+        else:
+            # It's text, apply emoji font to any emojis within this segment
+            result.append(emoji_pattern.sub(fr'<font name="{emoji_font_name}">\1</font>', part))
+            
+    return "".join(result)
 
 
 # --- PDF Generation & Handling ---
@@ -92,7 +110,6 @@ def apply_emoji_font(text: str, emoji_font_name: str) -> str:
 def markdown_to_story(markdown_text: str, font_name: str, emoji_font: str):
     """Converts markdown to a ReportLab story, with enhanced styling and page breaks."""
     styles = getSampleStyleSheet()
-    # Define styles for various markdown elements
     style_normal = ParagraphStyle('BodyText', fontName=font_name, spaceAfter=6, leading=14, fontSize=10)
     style_h1 = ParagraphStyle('h1', parent=styles['h1'], fontName=font_name, spaceBefore=12, fontSize=24, leading=28, textColor=colors.darkblue)
     style_h2 = ParagraphStyle('h2', parent=styles['h2'], fontName=font_name, fontSize=18, leading=22, spaceBefore=10)
@@ -135,18 +152,21 @@ def markdown_to_story(markdown_text: str, font_name: str, emoji_font: str):
                 story.append(table); story.append(Spacer(1, 0.2 * inch))
                 table_data = []
         
-        line_with_emoji = apply_emoji_font(line, emoji_font)
-        formatted_line = re.sub(r'_(.*?)_', r'<i>\1</i>', re.sub(r'\*\*(.*?)\*\*', r'<b>\1</b>', line_with_emoji))
-        
+        # Apply base formatting first
+        formatted_line = re.sub(r'_(.*?)_', r'<i>\1</i>', re.sub(r'\*\*(.*?)\*\*', r'<b>\1</b>', line))
+        # Then apply emoji font to the already formatted line
+        line_with_emoji = apply_emoji_font(formatted_line, emoji_font)
+
         if line.startswith("# "):
             if not first_heading: story.append(PageBreak())
-            story.append(Paragraph(formatted_line[2:], style_h1)); first_heading = False
-        elif line.startswith("## "): story.append(Paragraph(formatted_line[3:], style_h2))
-        elif line.startswith("### "): story.append(Paragraph(formatted_line[4:], style_h3))
-        elif line.strip().startswith(("* ", "- ")): story.append(Paragraph(formatted_line.strip()[2:], style_normal, bulletText='•'))
-        elif re.match(r'^\d+\.\s', line.strip()): story.append(Paragraph(formatted_line.strip(), style_normal))
+            # Strip markdown characters from the original line before applying style
+            story.append(Paragraph(apply_emoji_font(re.sub(r'\*\*(.*?)\*\*', r'<b>\1</b>', line[2:]), emoji_font), style_h1)); first_heading = False
+        elif line.startswith("## "): story.append(Paragraph(apply_emoji_font(re.sub(r'\*\*(.*?)\*\*', r'<b>\1</b>', line[3:]), emoji_font), style_h2))
+        elif line.startswith("### "): story.append(Paragraph(apply_emoji_font(re.sub(r'\*\*(.*?)\*\*', r'<b>\1</b>', line[4:]), emoji_font), style_h3))
+        elif line.strip().startswith(("- ", "* ")): story.append(Paragraph(line_with_emoji.strip()[2:], style_normal, bulletText='•'))
+        elif re.match(r'^\d+\.\s', line.strip()): story.append(Paragraph(line_with_emoji.strip(), style_normal))
         elif line.strip() == "": story.append(Spacer(1, 0.1 * inch))
-        else: story.append(Paragraph(formatted_line, style_normal))
+        else: story.append(Paragraph(line_with_emoji, style_normal))
 
     return story
 
@@ -222,8 +242,34 @@ def generate_pdfs_api(files, layouts, fonts, num_columns, page_w_mult, page_h_mu
 
 # --- Gradio UI Definition ---
 AVAILABLE_FONTS, EMOJI_FONT_NAME = register_local_fonts()
-SAMPLE_MARKDOWN = "# Document Title 🚀\n\nThis is the first section.\n\n## Subsection 1.1\n\nHere is some text with **bold** and _italic_ elements. And an emoji: 😊\n\n# Chapter Two\n\nThis new chapter starts on a new page automatically.\n\n### A Table of Data\n| Item | Category | Status |\n|---|---|---|\n| Chocolate 🍫| Food | Delicious |\n| Computer 💻| Tech | Necessary |\n\n```python\n# This is a code block.\ndef main():\n    print(\"Hello, PDF World!\")\n```"
-with open(CWD / "sample.md", "w") as f: f.write(SAMPLE_MARKDOWN)
+SAMPLE_MARKDOWN = """# Deities Guide: Mythology and Moral Lessons 
+
+1. 📜 **Introduction**
+   - **Purpose**: Explore deities, spirits, saints, and beings with their epic stories and morals!
+   - **Usage**: A guide for learning and storytelling across traditions. ️
+   - **Themes**: Justice ️, faith , hubris , redemption , cosmic order .
+
+2. 🛠️ **Core Concepts of Divinity**
+   - **Powers**: Creation , omniscience ️‍️, shapeshifting  across entities.
+   - **Life Cycle**: Mortality , immortality , transitions like saints and avatars .
+   - **Communication**: Omens ️, visions ️, miracles  from gods and spirits.
+
+# ⚔️ Arthurian Legends
+ - **Merlin, Morgan le Fay, Arthur**: Mentor , rival ‍️, son .
+ - **Relation**: Family tests loyalty .
+ - **Lesson**: Honor ️ vs. betrayal ️.
+
+# 🏛️ Greek Mythology
+ - **Zeus, Hera, Athena**: Father , mother , daughter .
+ - **Relation**: Family rules with tension ️.
+ - **Lesson**: Hubris  meets wisdom .
+
+# 🕉️ Hindu Trimurti
+ - **Brahma, Vishnu, Shiva**: Creator , preserver ️, destroyer .
+ - **Relation**: Divine trio cycles existence .
+ - **Lesson**: Balance ️ sustains life .
+"""
+with open(CWD / "sample.md", "w", encoding="utf-8") as f: f.write(SAMPLE_MARKDOWN)
 
 with gr.Blocks(theme=gr.themes.Soft(), title="Advanced PDF Generator") as demo:
     gr.Markdown("# 📄 Advanced PDF Layout Engine")