app/hadith_utils.py

"""Hadith utility functions — narrator extraction and highlighting."""

import re

# Arabic narrator markers
_AR_NARRATORS = [
    "حَدَّثَنَا", "أَخْبَرَنَا", "حَدَّثَنِي", "أَخْبَرَنِي", "عَنْ", "قَالَ",
    "سَمِعْتُ", "أَنْبَأَنَا", "ثنا", "أنا",
]

# English narrator markers
_EN_NARRATORS = [
    "Narrated", "told us", "informed us", "heard", "on the authority of",
    "from", "said", "reported",
]

def highlight_narrators(text: str, lang: str = "arabic") -> str:
    """Highlight common narrator markers in Hadith text using markdown bolding."""
    if not text:
        return text
        
    if lang == "arabic" or any(m in text for m in _AR_NARRATORS):
        # Highlight Arabic markers
        for marker in _AR_NARRATORS:
            # Match word with optional diacritics
            pattern = re.compile(f"({re.escape(marker)})", re.U)
            text = pattern.sub(r"**\1**", text)
    else:
        # Highlight English markers
        for marker in _EN_NARRATORS:
            pattern = re.compile(f"\\b({re.escape(marker)})\\b", re.I)
            text = pattern.sub(r"**\1**", text)
            
    return text