project.json

[
  {
    "children": [], 
    "name": "a task @done", 
    "tags": {
      "done": null
    }
  }, 
  {
    "children": [
      {
        "children": [], 
        "name": "a child task", 
        "tags": {}
      }, 
      {
        "children": [], 
        "name": "another child task", 
        "tags": {}
      }
    ], 
    "name": "another task @delay(2014-01-01)", 
    "tags": {
      "delay": "2014-01-01"
    }
  }
]

project.md

Test Project

Some prose about the goals of the project, whatever.

Run this project like so:

PRETTY=1 python task_parser.py project.md

• a task @done
• another task @delay(2014-01-01)
  • a child task
  • another child task

task_parser.py

from itertools import chain
import json
import re
import yaml

task_re = re.compile(' +\+ .+')
tag_re = re.compile('@(?P<tag>[\w\-_]+)(\((?P<value>.+?)\))?')

def tasks_for_file(fname):
    with open(fname, 'r') as f:
        text = f.read()

    tasks = task_re.findall(text)

    sanitized = [
        task.replace('+', '-', 1) + ':'
        for task in tasks
    ]

    tasks = yaml.safe_load('\n'.join(sanitized))
    tasks = sanitize_tasks(tasks)

    return tasks

def sanitize_tasks(tasks):
    for task in tasks:
        name = task.keys()[0]
        children = list(sanitize_tasks(task.values()[0] or []))

        tags = {
            tag.groupdict()['tag']: tag.groupdict()['value']
            for tag in tag_re.finditer(name)
        }

        yield {
            'name': name,
            'tags': tags,
            'children': children,
        }

def main(files, indent=2):
    tasks = chain.from_iterable(
        tasks_for_file(fname)
        for fname in files
    )

    return json.dumps(list(tasks), indent=indent)

if __name__ == '__main__':
    import os, sys
    sys.stdout.write(main(
        sys.argv[1:],
        2 if os.environ.get('PRETTY', '0') == '1' else None
    ) + '\n')