Update app.py

app.py

@@ -108,15 +108,19 @@ def apply_emoji_font(text: str, emoji_font_name: str) -> str:
 # --- PDF Generation & Handling ---
 
 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."""
+    """
+    Converts markdown to a ReportLab story, with enhanced styling and page breaks.
+    This version correctly separates structural parsing from content formatting.
+    """
     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)
     style_h3 = ParagraphStyle('h3', parent=styles['h3'], fontName=font_name, fontSize=14, leading=18, spaceBefore=8)
     style_code = ParagraphStyle('Code', fontName='Courier', backColor=colors.whitesmoke, textColor=colors.darkred, borderWidth=1, borderColor=colors.lightgrey, padding=8, leading=12, fontSize=9)
     style_table_header = ParagraphStyle('TableHeader', parent=style_normal, fontName=font_name + "-Bold" if font_name != 'Helvetica' else 'Helvetica-Bold')
-
+    
     story = []
     lines = markdown_text.split('\n')
     
@@ -125,51 +129,70 @@ def markdown_to_story(markdown_text: str, font_name: str, emoji_font: str):
     first_heading = True
 
     for line in lines:
-        if line.strip().startswith("```"):
+        stripped_line = line.strip()
+
+        if stripped_line.startswith("```"):
             if in_code_block:
                 story.append(Paragraph(code_block_text.replace('\n', '<br/>'), style_code)); story.append(Spacer(1, 0.1 * inch))
                 in_code_block = False; code_block_text = ""
-            else:
-                in_code_block = True
+            else: in_code_block = True
             continue
         if in_code_block:
             code_block_text += line.replace('&', '&amp;').replace('<', '&lt;').replace('>', '&gt;') + '\n'
             continue
         
-        if line.strip().startswith('|'):
+        if stripped_line.startswith('|'):
             if not in_table: in_table = True
-            if all(c in '-|: ' for c in line.strip()): continue
-            cells = [apply_emoji_font(cell.strip(), emoji_font) for cell in line.strip().strip('|').split('|')]
+            if all(c in '-|: ' for c in stripped_line): continue
+            cells = [cell.strip() for cell in stripped_line.strip('|').split('|')]
             table_data.append(cells)
             continue
         if in_table:
             in_table = False
             if table_data:
-                header = [Paragraph(cell, style_table_header) for cell in table_data[0]]
-                formatted_rows = [[Paragraph(cell, style_normal) for cell in row] for row in table_data[1:]]
+                header_content = [apply_emoji_font(re.sub(r'\*\*(.*?)\*\*', r'<b>\1</b>', cell), emoji_font) for cell in table_data[0]]
+                header = [Paragraph(cell, style_table_header) for cell in header_content]
+                
+                formatted_rows = []
+                for row in table_data[1:]:
+                    formatted_cells = [apply_emoji_font(re.sub(r'\*\*(.*?)\*\*', r'<b>\1</b>', cell), emoji_font) for cell in row]
+                    formatted_rows.append([Paragraph(cell, style_normal) for cell in formatted_cells])
+                
                 table = Table([header] + formatted_rows, hAlign='LEFT', repeatRows=1)
                 table.setStyle(TableStyle([('BACKGROUND', (0, 0), (-1, 0), colors.lightgrey), ('GRID', (0, 0), (-1, -1), 1, colors.darkgrey), ('VALIGN', (0,0), (-1,-1), 'MIDDLE')]))
                 story.append(table); story.append(Spacer(1, 0.2 * inch))
-                table_data = []
-        
-        # 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)
+            table_data = []
 
-        if line.startswith("# "):
-            if not first_heading: story.append(PageBreak())
-            # 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(line_with_emoji, style_normal))
+        if not stripped_line:
+            story.append(Spacer(1, 0.1 * inch))
+            continue
 
+        # Default content is the whole stripped line
+        content = stripped_line
+        style = style_normal
+        extra_args = {}
+        
+        # Detect structural elements and extract the raw content
+        if stripped_line.startswith("# "):
+            if not first_heading: story.append(PageBreak())
+            content = stripped_line.lstrip('# '); style = style_h1; first_heading = False
+        elif stripped_line.startswith("## "):
+            content = stripped_line.lstrip('## '); style = style_h2
+        elif stripped_line.startswith("### "):
+            content = stripped_line.lstrip('### '); style = style_h3
+        elif stripped_line.startswith(("- ", "* ")):
+            content = stripped_line[2:]; extra_args['bulletText'] = '•'
+        
+        # Now, format the extracted content
+        # Apply markdown formatting for bold/italic
+        formatted_content = re.sub(r'_(.*?)_', r'<i>\1</i>', re.sub(r'\*\*(.*?)\*\*', r'<b>\1</b>', content))
+        # Then apply emoji font to the already formatted line
+        final_content = apply_emoji_font(formatted_content, emoji_font)
+        
+        story.append(Paragraph(final_content, style, **extra_args))
     return story
 
+
 def create_pdf_preview(pdf_path: Path):
     preview_path = PREVIEW_DIR / f"{pdf_path.stem}.png"
     try:
@@ -247,27 +270,27 @@ 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 .
+   - **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.
+   - **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 ️.
+ - **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 .
+ - **Zeus, Hera, Athena**: Father ⚡, mother 👑, daughter 🦉.
+ - **Relation**: Family rules with tension 🌩️.
+ - **Lesson**: Hubris  अहंकार meets wisdom 🧠.
 
 # 🕉️ Hindu Trimurti
- - **Brahma, Vishnu, Shiva**: Creator , preserver ️, destroyer .
- - **Relation**: Divine trio cycles existence .
- - **Lesson**: Balance ️ sustains life .
+ - **Brahma, Vishnu, Shiva**: Creator  Brahma, preserver  Vishnu, destroyer  Shiva.
+ - **Relation**: Divine trio cycles existence 🔄.
+ - **Lesson**: Balance ⚖️ sustains life 💖.
 """
 with open(CWD / "sample.md", "w", encoding="utf-8") as f: f.write(SAMPLE_MARKDOWN)