cultural_utils.py

"""Utility module for managing cultural context annotations."""
from typing import Dict, Tuple, List

# Initial database of idioms and their cultural context
# Format: {idiom: (literal_translation, cultural_explanation)}
ENGLISH_IDIOMS: Dict[str, Tuple[str, str]] = {
    "break the ice": ("شکستن یخ", "To initiate social interaction and reduce tension. In Persian culture, this concept is similar to 'گرم گرفتن' (warm taking) which emphasizes creating a warm, friendly atmosphere."),
    "costs an arm and a leg": ("به قیمت یک دست و پا", "Very expensive. In Persian, a similar expression is 'سر به فلک کشیدن' (reaching the sky) to describe extremely high prices."),
    "piece of cake": ("تکه کیک", "Something very easy to do. In Persian culture, the equivalent idiom is 'آب خوردن' (like drinking water) to describe a task that's very simple.")
}

PERSIAN_IDIOMS: Dict[str, Tuple[str, str]] = {
    "آب خوردن": ("drinking water", "Used to describe something very easy, similar to the English 'piece of cake'."),
    "دست و پنجه نرم کردن": ("softening hand and fingers", "To struggle or deal with something difficult, similar to 'wrestling with' in English."),
    "دیوار موش داره موش هم گوش داره": ("the wall has mice and mice have ears", "Be careful what you say as others might be listening, similar to 'walls have ears' in English.")
}

def detect_idioms(text: str, source_lang: str) -> List[Tuple[str, str, str]]:
    """
    Detect idioms in the input text and return their cultural context.
    
    Returns:
        List of tuples (idiom, literal_translation, cultural_explanation)
    """
    idioms_db = ENGLISH_IDIOMS if source_lang == "en" else PERSIAN_IDIOMS
    found_idioms = []
    
    for idiom in idioms_db:
        if idiom.lower() in text.lower():
            found_idioms.append((idiom, *idioms_db[idiom]))
            
    return found_idioms

def get_cultural_context(text: str, source_lang: str) -> Dict[str, List[Tuple[str, str, str]]]:
    """
    Get cultural context annotations for a given text.
    
    Returns:
        Dictionary with 'idioms' key containing list of detected idioms and their context
    """
    return {
        'idioms': detect_idioms(text, source_lang)
    }