utils/json_extractor.py

import json
import logging
from typing import List, Any

logger = logging.getLogger("json-extractor")

def find_balanced_closing_index(text: str, start_index: int) -> int:
    """
    Finds the matching closing bracket for the bracket at start_index.
    Ignores brackets inside strings and comments.
    """
    start_char = text[start_index]
    end_char = '}' if start_char == '{' else ']'
    
    depth = 0
    in_double_quote = False
    in_single_quote = False
    in_backtick = False
    in_line_comment = False
    in_block_comment = False
    is_escaped = False

    length = len(text)
    i = start_index

    while i < length:
        char = text[i]
        next_char = text[i+1] if i + 1 < length else ''
        
        # Handle Escaping
        if is_escaped:
            is_escaped = False
            i += 1
            continue
        if char == '\\' and not in_line_comment and not in_block_comment:
            is_escaped = True
            i += 1
            continue

        # Handle Comments
        if in_line_comment:
            if char == '\n': in_line_comment = False
            i += 1
            continue
        if in_block_comment:
            if char == '*' and next_char == '/':
                in_block_comment = False
                i += 2
                continue
            i += 1
            continue

        # Check comment starts
        if not in_double_quote and not in_single_quote and not in_backtick:
            if char == '/' and next_char == '/':
                in_line_comment = True
                i += 2
                continue
            if char == '/' and next_char == '*':
                in_block_comment = True
                i += 2
                continue

        # Handle Strings
        if in_double_quote:
            if char == '"': in_double_quote = False
            i += 1
            continue
        if in_single_quote:
            if char == "'": in_single_quote = False
            i += 1
            continue
        if in_backtick:
            if char == '`': in_backtick = False
            i += 1
            continue

        if char == '"': 
            in_double_quote = True
            i += 1
            continue
        if char == "'": 
            in_single_quote = True
            i += 1
            continue
        if char == '`': 
            in_backtick = True
            i += 1
            continue

        # Handle Bracket Counting
        if char == start_char:
            depth += 1
        elif char == end_char:
            depth -= 1
            if depth == 0:
                return i
        
        i += 1

    return -1

def extract_json_from_content(content: str) -> List[Any]:
    """
    Scans text for JSON objects/arrays using state machine logic.
    """
    if not content or not isinstance(content, str):
        return []

    found_blocks = []
    cursor = 0
    length = len(content)

    while cursor < length:
        if content[cursor] not in ['{', '[']:
            cursor += 1
            continue

        end_index = find_balanced_closing_index(content, cursor)

        if end_index != -1:
            raw_candidate = content[cursor : end_index + 1]
            try:
                parsed = json.loads(raw_candidate)
                found_blocks.append(parsed)
                cursor = end_index + 1
                continue
            except json.JSONDecodeError:
                pass
        
        cursor += 1

    return found_blocks