mllm_task_file.py

from dataclasses import dataclass
from typing import List, Optional
import re

@dataclass
class Item:
    indent: int
    task_id: Optional[str]
    content: List[str]

def parse_text(text: str) -> List[Item]:
    """
    Parse the custom task file format into a list of Items.
    """
    lines = text.splitlines()
    items: List[Item] = []

    # Holds the item being built as we parse; None if we aren't in an item yet.
    current_item = None

    # Regex to detect a new item line:
    # Capture leading spaces in group(1), optional "[task_id]" in group(2), and the remainder in group(3).
    # Example match: "   - [12] wash stuff"
    # group(1) -> "   "
    # group(2) -> "12"
    # group(3) -> "wash stuff"
    item_pattern = re.compile(
        r'^(\s*)-\s*(?:\[([0-9A-Fa-f]+)\]\s*)?(.*)$'
    )

    for line in lines:
        match = item_pattern.match(line)
        if match:
            # We have a new item starting on this line.
            # First, commit the previous item if there is one
            if current_item is not None:
                items.append(current_item)

            indent_str, task_id, rest = match.groups()

            # Prepare a new current_item
            current_item = Item(
                indent=len(indent_str),
                task_id=task_id,  # may be None if not present
                content=[rest] if rest else []
            )
        else:
            # This line is part of the current item's content (if any).
            if current_item is None:
                # If we hit a line that doesn't match new-item syntax 
                # but we don't have a current item, we can either:
                # - ignore it,
                # - or create an item with no dash,
                #   but that doesn't really fit the grammar, so let's append it
                #   to a "dummy" item or just start a new item with no dash.
                # For simplicity, let's create a "dummy" item with indent=0.
                current_item = Item(indent=0, task_id=None, content=[])
            
            current_item.content.append(line)

    # End of file: commit the last item
    if current_item is not None:
        items.append(current_item)

    return items

def unparse_items(items: List[Item]) -> str:
    """
    Convert a list of Items back into the custom-format text.
    """
    lines = []
    for item in items:
        # Start with indentation + '-'
        prefix = ' ' * item.indent + '-'
        
        # If we have a task_id, include it
        if item.task_id is not None:
            prefix += f' [{item.task_id}]'

        if item.content:
            # Put the first content line on the same line, separated by a space
            line0 = prefix
            # If the first content line isn't empty, add a space first
            if item.content[0]:
                line0 += " " + item.content[0]
            lines.append(line0)

            # The rest of the content lines each go on its own line as-is
            for more_line in item.content[1:]:
                lines.append(more_line)
        else:
            # If there's no content, just append prefix
            lines.append(prefix)

    return "\n".join(lines)