markdown2unicode

gistfile1.txt

import re

# Bold: Mathematical Sans-Serif Bold
def char_to_bold(c):
    if 'a' <= c <= 'z':
        return chr(0x1D5EE + (ord(c) - ord('a')))
    elif 'A' <= c <= 'Z':
        return chr(0x1D5D4 + (ord(c) - ord('A')))
    elif '0' <= c <= '9':
        return chr(0x1D7CE + (ord(c) - ord('0')))
    return c

# Italic: Mathematical Italic
def char_to_italic(c):
    if 'a' <= c <= 'z':
        return chr(0x1D44E + (ord(c) - ord('a')))
    elif 'A' <= c <= 'Z':
        return chr(0x1D434 + (ord(c) - ord('A')))
    return c

# Bold-Italic: Mathematical Sans-Serif Bold Italic
def char_to_bold_italic(c):
    if 'a' <= c <= 'z':
        return chr(0x1D656 + (ord(c) - ord('a')))
    elif 'A' <= c <= 'Z':
        return chr(0x1D63C + (ord(c) - ord('A')))
    return c

# Bold-Italic: Mathematical Sans-Serif Bold Italic
def char_to_bold_italic(c):
    if 'a' <= c <= 'z':
        return chr(0x1D656 + (ord(c) - ord('a')))
    elif 'A' <= c <= 'Z':
        return chr(0x1D63C + (ord(c) - ord('A')))
    return c

# Underline: Add combining underline (U+0332)
def underline_repl(match):
    text = match.group(1)
    return ''.join([c + '\u0332' for c in text])

def process_bold_section(content):
    # Convert italic (_..._ or *...*) within bold to bold-italic
    content = re.sub(
        r'([_*])(?=\S)(.*?)(?<=\S)\1',
        lambda m: ''.join(char_to_bold_italic(c) for c in m.group(2)),
        content,
        flags=re.DOTALL
    )
    # Convert remaining text to bold
    return ''.join(char_to_bold(c) for c in content)

def convert_markdown_to_unicode(text):
    # Process bold-italic (***...***)
    text = re.sub(r'\*\*\*(?=\S)(.*?)(?<=\S)\*\*\*', 
                 lambda m: ''.join(char_to_bold_italic(c) for c in m.group(1)), 
                 text, flags=re.DOTALL)
    
    # Process bold (**...**) with nested italic
    text = re.sub(r'\*\*(?=\S)(.*?)(?<=\S)\*\*', 
                 lambda m: process_bold_section(m.group(1)), 
                 text, flags=re.DOTALL)
    
    # Process remaining italic (_..._ or *...*)
    text = re.sub(r'([_*])(?=\S)(.*?)(?<=\S)\1', 
                 lambda m: ''.join(char_to_italic(c) for c in m.group(2)), 
                 text, flags=re.DOTALL)
    
    # Process underline (__...__)
    text = re.sub(r'__(?=\S)(.*?)(?<=\S)__', underline_repl, text, flags=re.DOTALL)
    
    return text

if __name__ == '__main__':
    # Example usage
    markdown_text = "**Is _This_ the Future of AEC?** *Regular italic* __Underline__"
    converted_text = convert_markdown_to_unicode(markdown_text)
    print(converted_text)