label_parser.py

"""
Project: Images Dataset Collector
File: label_parser.py
Description: Utilities for parsing the input keywords JSON file. Extracts labels, categories,
             and thresholds for the collection process.
"""
import json
from pathlib import Path
from typing import Optional, List, Dict, Any

from config import KEYWORDS_JSON


def get_all_labels(json_path: Optional[Path] = None) -> List[Dict[str, Any]]:
    """
    Parses the configuration JSON to extract all available labels.
    
    Args:
        json_path: Path to the forensic_keywords.json file using 'categories' and 'keywords' schema.
                   Defaults to config.KEYWORDS_JSON.
                   
    Returns:
        List of dictionaries containing 'text', 'category', and 'threshold' for each label.
    """
    if json_path is None:
        json_path = KEYWORDS_JSON
    
    try:
        with open(json_path, 'r', encoding='utf-8') as f:
            data = json.load(f)
    except FileNotFoundError:
        print(f"Error: Keywords file not found at {json_path}")
        return []
    
    default_threshold = data.get("default_threshold", 0.5)
    labels = []
    
    # Iterate through categories to flatten the label list
    for category in data.get("categories", []):
        category_name = category.get("name", "unknown")
        for keyword in category.get("keywords", []):
            labels.append({
                "text": keyword.get("text", ""),
                "category": category_name,
                "threshold": keyword.get("threshold", default_threshold)
            })
    
    return labels


def get_labels_by_category(category_name: str, json_path: Optional[Path] = None) -> List[Dict[str, Any]]:
    """Retrieves all labels belonging to a specific category (case-insensitive)."""
    all_labels = get_all_labels(json_path)
    return [l for l in all_labels if l["category"].lower() == category_name.lower()]


def get_category_names(json_path: Optional[Path] = None) -> List[str]:
    """Retrieves a list of all unique category names defined in the configuration."""
    if json_path is None:
        json_path = KEYWORDS_JSON
        
    try:
        with open(json_path, 'r', encoding='utf-8') as f:
            data = json.load(f)
        return [cat.get("name", "") for cat in data.get("categories", [])]
    except FileNotFoundError:
        return []