Parse markdown string.

store = taskmark.loads(markdown)
store = taskmark.loads(markdown, minutes_per_day=360)

Postconditions:

• All valid tasks extracted
• Warnings collected for invalid syntax
• Inheritance resolved
• Subtasks sorted by priority

Parse file, following [text](file.md) links.

store = taskmark.load(path)
store = taskmark.load(path, minutes_per_day=360)

Postconditions:

• file and line set on each task
• All linked files parsed
• Circular links warned
• Missing links warned

Errors:

• FileNotFoundError if file missing

Raise if warnings exist.

store.check()  # raises ParseError if warnings

Update single task in-memory.

result = taskmark.update(store, title, project, changes)

Task lookup:

• Match by exact title
• Match by project_path (or None for no project)
• FAIL if no match or multiple matches

UpdateResult.status:

• SUCCESS - All changes applied
• PARTIAL - Some changes rejected (see rejected)
• FAIL - Task not found or ambiguous (see error)

Update multiple tasks.

result = taskmark.batch_update(store, updates)

Behavior:

• Each update independent
• Partial success allowed
• Returns results for each update

Write pending changes to files.

result = taskmark.write_updates(store)

Behavior:

• Process files independently
• Tasks ordered bottom-up (avoid line drift)
• Atomic writes (temp file + rename)

When a recurring task is marked done:

1. Original task is marked done (with done_date)
2. recurrence field is removed from completed task
3. New task is created with:
   • State: open
   • planned_date: next occurrence calculated from pattern
   • recurrence: pattern preserved
   • Inherited metadata: from headings (not explicit task metadata)

Base date priority:

1. planned_date (if set)
2. due_date (if set)
3. Today's date