node.py

import re
import itertools
from typing import List, Iterable, Iterator, Optional, Union, Tuple, cast, Dict, Set, Sequence
try:
    from collections.abc import Sequence
except ImportError:
    from collections import Sequence

from .date import OrgDate, OrgDateClock, OrgDateRepeatedTask, parse_sdc
from .inline import to_plain_text
from .extra import to_rich_text, Rich
from .utils.py3compat import PY3, unicode


def lines_to_chunks(lines: Iterable[str]) -> Iterable[List[str]]:
    chunk: List[str] = []
    for l in lines:
        if RE_NODE_HEADER.search(l):
            yield chunk
            chunk = []
        chunk.append(l)
    yield chunk

RE_NODE_HEADER = re.compile(r"^\*+ ")


def parse_heading_level(heading):
    """
    Get star-stripped heading and its level

    >>> parse_heading_level('* Heading')
    ('Heading', 1)
    >>> parse_heading_level('******** Heading')
    ('Heading', 8)
    >>> parse_heading_level('*') # None since no space after star
    >>> parse_heading_level('*bold*') # None
    >>> parse_heading_level('not heading')  # None

    """
    match = RE_HEADING_STARS.search(heading)
    if match:
        return (match.group(2), len(match.group(1)))

RE_HEADING_STARS = re.compile(r'^(\*+)\s+(.*?)\s*$')


def parse_heading_tags(heading: str) -> Tuple[str, List[str]]:
    """
    Get first tags and heading without tags

    >>> parse_heading_tags('HEADING')
    ('HEADING', [])
    >>> parse_heading_tags('HEADING :TAG1:TAG2:')
    ('HEADING', ['TAG1', 'TAG2'])
    >>> parse_heading_tags('HEADING: this is still heading :TAG1:TAG2:')
    ('HEADING: this is still heading', ['TAG1', 'TAG2'])
    >>> parse_heading_tags('HEADING :@tag:_tag_:')
    ('HEADING', ['@tag', '_tag_'])

    Here is the spec of tags from Org Mode manual:

      Tags are normal words containing letters, numbers, ``_``, and
      ``@``.  Tags must be preceded and followed by a single colon,
      e.g., ``:work:``.

      -- (info "(org) Tags")

    """
    match = RE_HEADING_TAGS.search(heading)
    if match:
        heading = match.group(1)
        tagstr = match.group(2)
        tags = tagstr.split(':')
    else:
        tags = []
    return (heading, tags)

# Tags are normal words containing letters, numbers, '_', and '@'. https://orgmode.org/manual/Tags.html
RE_HEADING_TAGS = re.compile(r'(.*?)\s*:([\w@:]+):\s*$')


def parse_heading_todos(heading: str, todo_candidates: List[str]) -> Tuple[str, Optional[str]]:
    """
    Get TODO keyword and heading without TODO keyword.

    >>> todos = ['TODO', 'DONE']
    >>> parse_heading_todos('Normal heading', todos)
    ('Normal heading', None)
    >>> parse_heading_todos('TODO Heading', todos)
    ('Heading', 'TODO')

    """
    for todo in todo_candidates:
        if heading == todo:
            return ('', todo)
        if heading.startswith(todo + ' '):
            return (heading[len(todo) + 1:], todo)
    return (heading, None)


def parse_heading_priority(heading):
    """
    Get priority and heading without priority field.

    >>> parse_heading_priority('HEADING')
    ('HEADING', None)
    >>> parse_heading_priority('[#A] HEADING')
    ('HEADING', 'A')
    >>> parse_heading_priority('[#0] HEADING')
    ('HEADING', '0')
    >>> parse_heading_priority('[#A]')
    ('', 'A')

    """
    match = RE_HEADING_PRIORITY.search(heading)
    if match:
        return (match.group(2), match.group(1))
    else:
        return (heading, None)

RE_HEADING_PRIORITY = re.compile(r'^\s*\[#([A-Z0-9])\] ?(.*)$')

PropertyValue = Union[str, int]
def parse_property(line: str) -> Tuple[Optional[str], Optional[PropertyValue]]:
    """
    Get property from given string.

    >>> parse_property(':Some_property: some value')
    ('Some_property', 'some value')
    >>> parse_property(':Effort: 1:10')
    ('Effort', 70)

    """
    prop_key = None
    prop_val: Optional[Union[str, int]] = None
    match = RE_PROP.search(line)
    if match:
        prop_key = match.group(1)
        prop_val = match.group(2)
        if prop_key == 'Effort':
            (h, m) = prop_val.split(":", 2)
            if h.isdigit() and m.isdigit():
                prop_val = int(h) * 60 + int(m)
    return (prop_key, prop_val)

RE_PROP = re.compile(r'^\s*:(.*?):\s*(.*?)\s*$')


def parse_comment(line: str): #  -> Optional[Tuple[str, Sequence[str]]]: # todo wtf?? it says 'ABCMeta isn't subscriptable??'
    """
    Parse special comment such as ``#+SEQ_TODO``

    >>> parse_comment('#+SEQ_TODO: TODO | DONE')
    ('SEQ_TODO', ['TODO | DONE'])
    >>> parse_comment('# not a special comment')  # None

    >>> parse_comment('#+FILETAGS: :tag1:tag2:')
    ('FILETAGS', ['tag1', 'tag2'])
    """
    match = re.match(r'\s*#\+', line)
    if match:
        end = match.end(0)
        comment = line[end:].split(':', maxsplit=1)
        if len(comment) >= 2:
            key   = comment[0]
            value = comment[1].strip()
            if key.upper() == 'FILETAGS':
                # just legacy behaviour; it seems like filetags is the only one that separated by ':'
                # see https://orgmode.org/org.html#In_002dbuffer-Settings
                return (key, [c.strip() for c in value.split(':') if len(c.strip()) > 0])
            else:
                return (key, [value])
    return None


def parse_seq_todo(line):
    """
    Parse value part of SEQ_TODO/TODO/TYP_TODO comment.

    >>> parse_seq_todo('TODO | DONE')
    (['TODO'], ['DONE'])
    >>> parse_seq_todo(' Fred  Sara   Lucy Mike  |  DONE  ')
    (['Fred', 'Sara', 'Lucy', 'Mike'], ['DONE'])
    >>> parse_seq_todo('| CANCELED')
    ([], ['CANCELED'])
    >>> parse_seq_todo('REPORT(r) BUG(b) KNOWNCAUSE(k) | FIXED(f)')
    (['REPORT', 'BUG', 'KNOWNCAUSE'], ['FIXED'])

    See also:

    * (info "(org) Per-file keywords")
    * (info "(org) Fast access to TODO states")

    """
    todo_done = line.split('|', 1)
    if len(todo_done) == 2:
        (todos, dones) = todo_done
    else:
        (todos, dones) = (line, '')
    strip_fast_access_key = lambda x: x.split('(', 1)[0]
    return (list(map(strip_fast_access_key, todos.split())),
            list(map(strip_fast_access_key, dones.split())))


class OrgEnv(object):

    """
    Information global to the file (e.g, TODO keywords).
    """

    def __init__(self, todos=['TODO'], dones=['DONE'],
                 filename='<undefined>'):
        self._todos = list(todos)
        self._dones = list(dones)
        self._todo_not_specified_in_comment = True
        self._filename = filename
        self._nodes = []

    @property
    def nodes(self):
        """
        A list of org nodes.

        >>> OrgEnv().nodes   # default is empty (of course)
        []

        >>> from orgparse import loads
        >>> loads('''
        ... * Heading 1
        ... ** Heading 2
        ... *** Heading 3
        ... ''').env.nodes      # doctest: +ELLIPSIS  +NORMALIZE_WHITESPACE
        [<orgparse.node.OrgRootNode object at 0x...>,
         <orgparse.node.OrgNode object at 0x...>,
         <orgparse.node.OrgNode object at 0x...>,
         <orgparse.node.OrgNode object at 0x...>]

        """
        return self._nodes

    def add_todo_keys(self, todos, dones):
        if self._todo_not_specified_in_comment:
            self._todos = []
            self._dones = []
            self._todo_not_specified_in_comment = False
        self._todos.extend(todos)
        self._dones.extend(dones)

    @property
    def todo_keys(self):
        """
        TODO keywords defined for this document (file).

        >>> env = OrgEnv()
        >>> env.todo_keys
        ['TODO']

        """
        return self._todos

    @property
    def done_keys(self):
        """
        DONE keywords defined for this document (file).

        >>> env = OrgEnv()
        >>> env.done_keys
        ['DONE']

        """
        return self._dones

    @property
    def all_todo_keys(self):
        """
        All TODO keywords (including DONEs).

        >>> env = OrgEnv()
        >>> env.all_todo_keys
        ['TODO', 'DONE']

        """
        return self._todos + self._dones

    @property
    def filename(self):
        """
        Return a path to the source file or similar information.

        If the org objects are not loaded from a file, this value
        will be a string of the form ``<SOME_TEXT>``.

        :rtype: str

        """
        return self._filename

    # parser

    def from_chunks(self, chunks):
        yield OrgRootNode.from_chunk(self, next(chunks))
        for chunk in chunks:
            yield OrgNode.from_chunk(self, chunk)


class OrgBaseNode(Sequence):

    """
    Base class for :class:`OrgRootNode` and :class:`OrgNode`

    .. attribute:: env

       An instance of :class:`OrgEnv`.
       All nodes in a same file shares same instance.

    :class:`OrgBaseNode` is an iterable object:

    >>> from orgparse import loads
    >>> root = loads('''
    ... * Heading 1
    ... ** Heading 2
    ... *** Heading 3
    ... * Heading 4
    ... ''')
    >>> for node in root:
    ...     print(node)
    <BLANKLINE>
    * Heading 1
    ** Heading 2
    *** Heading 3
    * Heading 4

    Note that the first blank line is due to the root node, as
    iteration contains the object itself.  To skip that, use
    slice access ``[1:]``:

    >>> for node in root[1:]:
    ...     print(node)
    * Heading 1
    ** Heading 2
    *** Heading 3
    * Heading 4

    It also supports sequence protocol.

    >>> print(root[1])
    * Heading 1
    >>> root[0] is root  # index 0 means itself
    True
    >>> len(root)   # remember, sequence contains itself
    5

    Note the difference between ``root[1:]`` and ``root[1]``:

    >>> for node in root[1]:
    ...     print(node)
    * Heading 1
    ** Heading 2
    *** Heading 3

    Nodes remember the line number information (1-indexed):

    >>> print(root.children[1].linenumber)
    5
    """

    _body_lines: List[str] # set by the child classes

    def __init__(self, env, index=None) -> None:
        """
        Create an :class:`OrgBaseNode` object.

        :type env: :class:`OrgEnv`
        :arg  env: This will be set to the :attr:`env` attribute.

        """
        self.env = env

        self.linenumber = cast(int, None) # set in parse_lines

        # content
        self._lines: List[str] = []

        # FIXME: use `index` argument to set index.  (Currently it is
        # done externally in `parse_lines`.)
        if index is not None:
            self._index = index
            """
            Index of `self` in `self.env.nodes`.

            It must satisfy an equality::

                self.env.nodes[self._index] is self

            This value is used for quick access for iterator and
            tree-like traversing.

            """

    def __iter__(self):
        yield self
        level = self.level
        for node in self.env._nodes[self._index + 1:]:
            if node.level > level:
                yield node
            else:
                break

    def __len__(self):
        return sum(1 for _ in self)

    def __nonzero__(self):
        # As self.__len__ returns non-zero value always this is not
        # needed.  This function is only for performance.
        return True

    __bool__ = __nonzero__  # PY3

    def __getitem__(self, key):
        if isinstance(key, slice):
            return itertools.islice(self, key.start, key.stop, key.step)
        elif isinstance(key, int):
            if key < 0:
                key += len(self)
            for (i, node) in enumerate(self):
                if i == key:
                    return node
            raise IndexError("Out of range {0}".format(key))
        else:
            raise TypeError("Inappropriate type {0} for {1}"
                            .format(type(key), type(self)))

    # tree structure

    def _find_same_level(self, iterable):
        for node in iterable:
            if node.level < self.level:
                return
            if node.level == self.level:
                return node

    @property
    def previous_same_level(self):
        """
        Return previous node if exists or None otherwise.

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node 1
        ... * Node 2
        ... ** Node 3
        ... ''')
        >>> (n1, n2, n3) = list(root[1:])
        >>> n1.previous_same_level is None
        True
        >>> n2.previous_same_level is n1
        True
        >>> n3.previous_same_level is None  # n2 is not at the same level
        True

        """
        return self._find_same_level(reversed(self.env._nodes[:self._index]))

    @property
    def next_same_level(self):
        """
        Return next node if exists or None otherwise.

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node 1
        ... * Node 2
        ... ** Node 3
        ... ''')
        >>> (n1, n2, n3) = list(root[1:])
        >>> n1.next_same_level is n2
        True
        >>> n2.next_same_level is None  # n3 is not at the same level
        True
        >>> n3.next_same_level is None
        True

        """
        return self._find_same_level(self.env._nodes[self._index + 1:])

    # FIXME: cache parent node
    def _find_parent(self):
        for node in reversed(self.env._nodes[:self._index]):
            if node.level < self.level:
                return node

    def get_parent(self, max_level=None):
        """
        Return a parent node.

        :arg int max_level:
            In the normally structured org file, it is a level
            of the ancestor node to return.  For example,
            ``get_parent(max_level=0)`` returns a root node.

            In the general case, it specify a maximum level of the
            desired ancestor node.  If there is no ancestor node
            whose level is equal to ``max_level``, this function
            try to find an ancestor node which level is smaller
            than ``max_level``.

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node 1
        ... ** Node 2
        ... ** Node 3
        ... ''')
        >>> (n1, n2, n3) = list(root[1:])
        >>> n1.get_parent() is root
        True
        >>> n2.get_parent() is n1
        True
        >>> n3.get_parent() is n1
        True

        For simplicity, accessing :attr:`parent` is alias of calling
        :meth:`get_parent` without argument.

        >>> n1.get_parent() is n1.parent
        True
        >>> root.parent is None
        True

        This is a little bit pathological situation -- but works.

        >>> root = loads('''
        ... * Node 1
        ... *** Node 2
        ... ** Node 3
        ... ''')
        >>> (n1, n2, n3) = list(root[1:])
        >>> n1.get_parent() is root
        True
        >>> n2.get_parent() is n1
        True
        >>> n3.get_parent() is n1
        True

        Now let's play with `max_level`.

        >>> root = loads('''
        ... * Node 1 (level 1)
        ... ** Node 2 (level 2)
        ... *** Node 3 (level 3)
        ... ''')
        >>> (n1, n2, n3) = list(root[1:])
        >>> n3.get_parent() is n2
        True
        >>> n3.get_parent(max_level=2) is n2  # same as default
        True
        >>> n3.get_parent(max_level=1) is n1
        True
        >>> n3.get_parent(max_level=0) is root
        True

        """
        if max_level is None:
            max_level = self.level - 1
        parent = self._find_parent()
        while parent.level > max_level:
            parent = parent.get_parent()
        return parent

    @property
    def parent(self):
        """
        Alias of :meth:`get_parent()` (calling without argument).
        """
        return self.get_parent()

    # FIXME: cache children nodes
    def _find_children(self):
        nodeiter = iter(self.env._nodes[self._index + 1:])
        try:
            node = next(nodeiter)
        except StopIteration:
            return
        if node.level <= self.level:
            return
        yield node
        last_child_level = node.level
        for node in nodeiter:
            if node.level <= self.level:
                return
            if node.level <= last_child_level:
                yield node
                last_child_level = node.level

    @property
    def children(self):
        """
        A list of child nodes.

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node 1
        ... ** Node 2
        ... *** Node 3
        ... ** Node 4
        ... ''')
        >>> (n1, n2, n3, n4) = list(root[1:])
        >>> (c1, c2) = n1.children
        >>> c1 is n2
        True
        >>> c2 is n4
        True

        Note the difference to ``n1[1:]``, which returns the Node 3 also:

        >>> (m1, m2, m3) = list(n1[1:])
        >>> m2 is n3
        True

        """
        return list(self._find_children())

    @property
    def root(self):
        """
        The root node.

        >>> from orgparse import loads
        >>> root = loads('* Node 1')
        >>> n1 = root[1]
        >>> n1.root is root
        True

        """
        root = self
        while True:
            parent = root.get_parent()
            if not parent:
                return root
            root = parent

    # parser

    @classmethod
    def from_chunk(cls, env, lines):
        self = cls(env)
        self._lines = lines
        self._parse_comments()
        return self

    def _parse_comments(self):
        special_comments: Dict[str, List[str]] = {}
        for line in self._lines:
            parsed = parse_comment(line)
            if parsed:
                (key, vals) = parsed
                key = key.upper() # case insensitive, so keep as uppercase
                special_comments.setdefault(key, []).extend(vals)
        self._special_comments = special_comments
        # parse TODO keys and store in OrgEnv
        for todokey in ['TODO', 'SEQ_TODO', 'TYP_TODO']:
            for val in special_comments.get(todokey, []):
                self.env.add_todo_keys(*parse_seq_todo(val))

    # misc

    @property
    def level(self):
        """
        Level of this node.

        :rtype: int

        """
        raise NotImplemented

    def _get_tags(self, inher=False) -> Set[str]:
        """
        Return tags

        :arg bool inher:
            Mix with tags of all ancestor nodes if ``True``.

        :rtype: set

        """
        return set()

    @property
    def tags(self) -> Set[str]:
        """
        Tags of this and parent's node.

        >>> from orgparse import loads
        >>> n2 = loads('''
        ... * Node 1    :TAG1:
        ... ** Node 2   :TAG2:
        ... ''')[2]
        >>> n2.tags == set(['TAG1', 'TAG2'])
        True

        """
        return self._get_tags(inher=True)

    @property
    def shallow_tags(self) -> Set[str]:
        """
        Tags defined for this node (don't look-up parent nodes).

        >>> from orgparse import loads
        >>> n2 = loads('''
        ... * Node 1    :TAG1:
        ... ** Node 2   :TAG2:
        ... ''')[2]
        >>> n2.shallow_tags == set(['TAG2'])
        True

        """
        return self._get_tags(inher=False)

    @staticmethod
    def _get_text(text, format='plain'):
        if format == 'plain':
            return to_plain_text(text)
        elif format == 'raw':
            return text
        elif format == 'rich':
            return to_rich_text(text)
        else:
            raise ValueError('format={0} is not supported.'.format(format))

    def get_body(self, format='plain') -> str:
        """
        Return a string of body text.

        See also: :meth:`get_heading`.

        """
        return self._get_text(
            '\n'.join(self._body_lines), format) if self._lines else ''

    @property
    def body(self) -> str:
        """Alias of ``.get_body(format='plain')``."""
        return self.get_body()

    @property
    def body_rich(self) -> Iterator[Rich]:
        r = self.get_body(format='rich')
        return cast(Iterator[Rich], r) # meh..

    @property
    def heading(self) -> str:
        raise NotImplementedError

    def is_root(self):
        """
        Return ``True`` when it is a root node.

        >>> from orgparse import loads
        >>> root = loads('* Node 1')
        >>> root.is_root()
        True
        >>> n1 = root[1]
        >>> n1.is_root()
        False

        """
        return False

    def __unicode__(self):
        return unicode("\n").join(self._lines)

    if PY3:
        __str__ = __unicode__
    else:
        def __str__(self):
            return unicode(self).encode('utf-8')

    # todo hmm, not sure if it really belongs here and not to OrgRootNode?
    def get_file_property_list(self, property):
        """
        Return a list of the selected property
        """
        vals = self._special_comments.get(property.upper(), None)
        return [] if vals is None else vals

    def get_file_property(self, property):
        """
        Return a single element of the selected property or None if it doesn't exist
        """
        vals = self._special_comments.get(property.upper(), None)
        if vals is None:
            return None
        elif len(vals) == 1:
            return vals[0]
        else:
            raise RuntimeError('Multiple values for property {}: {}'.format(property, vals))


class OrgRootNode(OrgBaseNode):

    """
    Node to represent a file

    See :class:`OrgBaseNode` for other available functions.

    """

    @property
    def _body_lines(self) -> List[str]: # type: ignore[override]
        # todo hacky..
        # for root node, the body is whatever is before the first node
        return self._lines

    @property
    def heading(self) -> str:
        return ''

    def _get_tags(self, inher=False) -> Set[str]:
        filetags = self.get_file_property_list('FILETAGS')
        return set(filetags)

    @property
    def level(self):
        return 0

    def get_parent(self, max_level=None):
        return None

    def is_root(self):
        return True


class OrgNode(OrgBaseNode):

    """
    Node to represent normal org node

    See :class:`OrgBaseNode` for other available functions.

    """

    def __init__(self, *args, **kwds) -> None:
        super(OrgNode, self).__init__(*args, **kwds)
        # fixme instead of casts, should organize code in such a way that they aren't necessary
        self._heading = cast(str, None)
        self._level = None
        self._tags = cast(List[str], None)
        self._todo: Optional[str] = None
        self._priority = None
        self._properties: Dict[str, PropertyValue] = {}
        self._scheduled = OrgDate(None)
        self._deadline = OrgDate(None)
        self._closed = OrgDate(None)
        self._timestamps: List[OrgDate] = []
        self._clocklist: List[OrgDateClock] = []
        self._body_lines: List[str] = []
        self._repeated_tasks: List[OrgDateRepeatedTask] = []

    # parser

    def _parse_pre(self):
        """Call parsers which must be called before tree structuring"""
        self._parse_heading()
        # FIXME: make the following parsers "lazy"
        ilines: Iterator[str] = iter(self._lines)
        try:
            next(ilines)            # skip heading
        except StopIteration:
            return
        ilines = self._iparse_sdc(ilines)
        ilines = self._iparse_clock(ilines)
        ilines = self._iparse_properties(ilines)
        ilines = self._iparse_repeated_tasks(ilines)
        ilines = self._iparse_timestamps(ilines)
        self._body_lines = list(ilines)

    def _parse_heading(self) -> None:
        heading = self._lines[0]
        (heading, self._level) = parse_heading_level(heading)
        (heading, self._tags) = parse_heading_tags(heading)
        (heading, self._todo) = parse_heading_todos(
            heading, self.env.all_todo_keys)
        (heading, self._priority) = parse_heading_priority(heading)
        self._heading = heading

    # The following ``_iparse_*`` methods are simple generator based
    # parser.  See ``_parse_pre`` for how it is used.  The principle
    # is simple: these methods get an iterator and returns an iterator.
    # If the item returned by the input iterator must be dedicated to
    # the parser, do not yield the item or yield it as-is otherwise.

    def _iparse_sdc(self, ilines: Iterator[str]) -> Iterator[str]:
        """
        Parse SCHEDULED, DEADLINE and CLOSED time tamps.

        They are assumed be in the first line.

        """
        try:
            line = next(ilines)
        except StopIteration:
            return
        (self._scheduled, self._deadline, self._closed) = parse_sdc(line)

        if not (self._scheduled or
                self._deadline or
                self._closed):
            yield line  # when none of them were found

        for line in ilines:
            yield line

    def _iparse_clock(self, ilines: Iterator[str]) -> Iterator[str]:
        self._clocklist = []
        for line in ilines:
            cl = OrgDateClock.from_str(line)
            if cl:
                self._clocklist.append(cl)
            else:
                yield line

    def _iparse_timestamps(self, ilines: Iterator[str]) -> Iterator[str]:
        self._timestamps = []
        self._timestamps.extend(OrgDate.list_from_str(self._heading))
        for l in ilines:
            self._timestamps.extend(OrgDate.list_from_str(l))
            yield l

    def _iparse_properties(self, ilines: Iterator[str]) -> Iterator[str]:
        self._properties = {}
        in_property_field = False
        for line in ilines:
            if in_property_field:
                if line.find(":END:") >= 0:
                    break
                else:
                    (key, val) = parse_property(line)
                    if key is not None and val is not None:
                        self._properties.update({key: val})
            elif line.find(":PROPERTIES:") >= 0:
                in_property_field = True
            else:
                yield line
        for line in ilines:
            yield line

    def _iparse_repeated_tasks(self, ilines: Iterator[str]) -> Iterator[str]:
        self._repeated_tasks = []
        for line in ilines:
            match = self._repeated_tasks_re.search(line)
            if match:
                # FIXME: move this parsing to OrgDateRepeatedTask.from_str
                mdict = match.groupdict()
                done_state = mdict['done']
                todo_state = mdict['todo']
                date = OrgDate.from_str(mdict['date'])
                self._repeated_tasks.append(
                    OrgDateRepeatedTask(date.start, todo_state, done_state))
            else:
                yield line

    _repeated_tasks_re = re.compile(
        r'''
        \s+ - \s+
        State \s+ "(?P<done> [^"]+)" \s+
        from  \s+ "(?P<todo> [^"]+)" \s+
        \[ (?P<date> [^\]]+) \]''',
        re.VERBOSE)

    def get_heading(self, format='plain'):
        """
        Return a string of head text without tags and TODO keywords.

        >>> from orgparse import loads
        >>> node = loads('* TODO Node 1').children[0]
        >>> node.get_heading()
        'Node 1'

        It strips off inline markup by default (``format='plain'``).
        You can get the original raw string by specifying
        ``format='raw'``.

        >>> node = loads('* [[link][Node 1]]').children[0]
        >>> node.get_heading()
        'Node 1'
        >>> node.get_heading(format='raw')
        '[[link][Node 1]]'

        """
        return self._get_text(self._heading, format)

    @property
    def heading(self) -> str:
        """Alias of ``.get_heading(format='plain')``."""
        return self.get_heading()

    @property
    def level(self):
        return self._level
        """
        Level attribute of this node.  Top level node is level 1.

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node 1
        ... ** Node 2
        ... ''')
        >>> (n1, n2) = root.children
        >>> root.level
        0
        >>> n1.level
        1
        >>> n2.level
        2

        """

    @property
    def priority(self):
        """
        Priority attribute of this node.  It is None if undefined.

        >>> from orgparse import loads
        >>> (n1, n2) = loads('''
        ... * [#A] Node 1
        ... * Node 2
        ... ''').children
        >>> n1.priority
        'A'
        >>> n2.priority is None
        True

        """
        return self._priority

    def _get_tags(self, inher=False) -> Set[str]:
        tags = set(self._tags)
        if inher:
            parent = self.get_parent()
            if parent:
                return tags | parent._get_tags(inher=True)
        return tags

    @property
    def todo(self) -> Optional[str]:
        """
        A TODO keyword of this node if exists or None otherwise.

        >>> from orgparse import loads
        >>> root = loads('* TODO Node 1')
        >>> root.children[0].todo
        'TODO'

        """
        return self._todo

    def get_property(self, key, val=None) -> Optional[PropertyValue]:
        """
        Return property named ``key`` if exists or ``val`` otherwise.

        :arg str key:
            Key of property.

        :arg val:
            Default value to return.

        """
        return self._properties.get(key, val)

    @property
    def properties(self) -> Dict[str, PropertyValue]:
        """
        Node properties as a dictionary.

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node
        ...   :PROPERTIES:
        ...   :SomeProperty: value
        ...   :END:
        ... ''')
        >>> root.children[0].properties['SomeProperty']
        'value'

        """
        return self._properties

    @property
    def scheduled(self):
        """
        Return scheduled timestamp

        :rtype: a subclass of :class:`orgparse.date.OrgDate`

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node
        ...   SCHEDULED: <2012-02-26 Sun>
        ... ''')
        >>> root.children[0].scheduled
        OrgDateScheduled((2012, 2, 26))

        """
        return self._scheduled

    @property
    def deadline(self):
        """
        Return deadline timestamp.

        :rtype: a subclass of :class:`orgparse.date.OrgDate`

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node
        ...   DEADLINE: <2012-02-26 Sun>
        ... ''')
        >>> root.children[0].deadline
        OrgDateDeadline((2012, 2, 26))

        """
        return self._deadline

    @property
    def closed(self):
        """
        Return timestamp of closed time.

        :rtype: a subclass of :class:`orgparse.date.OrgDate`

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node
        ...   CLOSED: [2012-02-26 Sun 21:15]
        ... ''')
        >>> root.children[0].closed
        OrgDateClosed((2012, 2, 26, 21, 15, 0))

        """
        return self._closed

    @property
    def clock(self):
        """
        Return a list of clocked timestamps

        :rtype: a list of a subclass of :class:`orgparse.date.OrgDate`

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node
        ...   CLOCK: [2012-02-26 Sun 21:10]--[2012-02-26 Sun 21:15] =>  0:05
        ... ''')
        >>> root.children[0].clock
        [OrgDateClock((2012, 2, 26, 21, 10, 0), (2012, 2, 26, 21, 15, 0))]

        """
        return self._clocklist

    def get_timestamps(self, active=False, inactive=False,
                       range=False, point=False):
        """
        Return a list of timestamps in the body text.

        :type   active: bool
        :arg    active: Include active type timestamps.
        :type inactive: bool
        :arg  inactive: Include inactive type timestamps.
        :type    range: bool
        :arg     range: Include timestamps which has end date.
        :type    point: bool
        :arg     point: Include timestamps which has no end date.

        :rtype: list of :class:`orgparse.date.OrgDate` subclasses


        Consider the following org node:

        >>> from orgparse import loads
        >>> node = loads('''
        ... * Node
        ...   CLOSED: [2012-02-26 Sun 21:15] SCHEDULED: <2012-02-26 Sun>
        ...   CLOCK: [2012-02-26 Sun 21:10]--[2012-02-26 Sun 21:15] =>  0:05
        ...   Some inactive timestamp [2012-02-23 Thu] in body text.
        ...   Some active timestamp <2012-02-24 Fri> in body text.
        ...   Some inactive time range [2012-02-25 Sat]--[2012-02-27 Mon].
        ...   Some active time range <2012-02-26 Sun>--<2012-02-28 Tue>.
        ... ''').children[0]

        The default flags are all off, so it does not return anything.

        >>> node.get_timestamps()
        []

        You can fetch appropriate timestamps using keyword arguments.

        >>> node.get_timestamps(inactive=True, point=True)
        [OrgDate((2012, 2, 23), None, False)]
        >>> node.get_timestamps(active=True, point=True)
        [OrgDate((2012, 2, 24))]
        >>> node.get_timestamps(inactive=True, range=True)
        [OrgDate((2012, 2, 25), (2012, 2, 27), False)]
        >>> node.get_timestamps(active=True, range=True)
        [OrgDate((2012, 2, 26), (2012, 2, 28))]

        This is more complex example.  Only active timestamps,
        regardless of range/point type.

        >>> node.get_timestamps(active=True, point=True, range=True)
        [OrgDate((2012, 2, 24)), OrgDate((2012, 2, 26), (2012, 2, 28))]

        """
        return [
            ts for ts in self._timestamps if
            (((active and ts.is_active()) or
              (inactive and not ts.is_active())) and
             ((range and ts.has_end()) or
              (point and not ts.has_end())))]

    @property
    def datelist(self):
        """
        Alias of ``.get_timestamps(active=True, inactive=True, point=True)``.

        :rtype: list of :class:`orgparse.date.OrgDate` subclasses

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node with point dates <2012-02-25 Sat>
        ...   CLOSED: [2012-02-25 Sat 21:15]
        ...   Some inactive timestamp [2012-02-26 Sun] in body text.
        ...   Some active timestamp <2012-02-27 Mon> in body text.
        ... ''')
        >>> root.children[0].datelist      # doctest: +NORMALIZE_WHITESPACE
        [OrgDate((2012, 2, 25)),
         OrgDate((2012, 2, 26), None, False),
         OrgDate((2012, 2, 27))]

        """
        return self.get_timestamps(active=True, inactive=True, point=True)

    @property
    def rangelist(self):
        """
        Alias of ``.get_timestamps(active=True, inactive=True, range=True)``.

        :rtype: list of :class:`orgparse.date.OrgDate` subclasses

        >>> from orgparse import loads
        >>> root = loads('''
        ... * Node with range dates <2012-02-25 Sat>--<2012-02-28 Tue>
        ...   CLOCK: [2012-02-26 Sun 21:10]--[2012-02-26 Sun 21:15] => 0:05
        ...   Some inactive time range [2012-02-25 Sat]--[2012-02-27 Mon].
        ...   Some active time range <2012-02-26 Sun>--<2012-02-28 Tue>.
        ...   Some time interval <2012-02-27 Mon 11:23-12:10>.
        ... ''')
        >>> root.children[0].rangelist     # doctest: +NORMALIZE_WHITESPACE
        [OrgDate((2012, 2, 25), (2012, 2, 28)),
         OrgDate((2012, 2, 25), (2012, 2, 27), False),
         OrgDate((2012, 2, 26), (2012, 2, 28)),
         OrgDate((2012, 2, 27, 11, 23, 0), (2012, 2, 27, 12, 10, 0))]

        """
        return self.get_timestamps(active=True, inactive=True, range=True)

    def has_date(self):
        """
        Return ``True`` if it has any kind of timestamp
        """
        return (self.scheduled or
                self.deadline or
                self.datelist or
                self.rangelist)

    @property
    def repeated_tasks(self):
        """
        Get repeated tasks marked DONE in an entry having repeater.

        :rtype: list of :class:`orgparse.date.OrgDateRepeatedTask`

        >>> from orgparse import loads
        >>> node = loads('''
        ... * TODO Pay the rent
        ...   DEADLINE: <2005-10-01 Sat +1m>
        ...   - State "DONE"  from "TODO"  [2005-09-01 Thu 16:10]
        ...   - State "DONE"  from "TODO"  [2005-08-01 Mon 19:44]
        ...   - State "DONE"  from "TODO"  [2005-07-01 Fri 17:27]
        ... ''').children[0]
        >>> node.repeated_tasks            # doctest: +NORMALIZE_WHITESPACE
        [OrgDateRepeatedTask((2005, 9, 1, 16, 10, 0), 'TODO', 'DONE'),
         OrgDateRepeatedTask((2005, 8, 1, 19, 44, 0), 'TODO', 'DONE'),
         OrgDateRepeatedTask((2005, 7, 1, 17, 27, 0), 'TODO', 'DONE')]
        >>> node.repeated_tasks[0].before
        'TODO'
        >>> node.repeated_tasks[0].after
        'DONE'

        Repeated tasks in ``:LOGBOOK:`` can be fetched by the same code.

        >>> node = loads('''
        ... * TODO Pay the rent
        ...   DEADLINE: <2005-10-01 Sat +1m>
        ...   :LOGBOOK:
        ...   - State "DONE"  from "TODO"  [2005-09-01 Thu 16:10]
        ...   - State "DONE"  from "TODO"  [2005-08-01 Mon 19:44]
        ...   - State "DONE"  from "TODO"  [2005-07-01 Fri 17:27]
        ...   :END:
        ... ''').children[0]
        >>> node.repeated_tasks            # doctest: +NORMALIZE_WHITESPACE
        [OrgDateRepeatedTask((2005, 9, 1, 16, 10, 0), 'TODO', 'DONE'),
         OrgDateRepeatedTask((2005, 8, 1, 19, 44, 0), 'TODO', 'DONE'),
         OrgDateRepeatedTask((2005, 7, 1, 17, 27, 0), 'TODO', 'DONE')]

        See: `(info "(org) Repeated tasks")
        <http://orgmode.org/manual/Repeated-tasks.html>`_

        """
        return self._repeated_tasks


def parse_lines(lines: Iterable[str], filename, env=None) -> OrgNode:
    if not env:
        env = OrgEnv(filename=filename)
    elif env.filename != filename:
        raise ValueError('If env is specified, filename must match')

    # parse into node of list (environment will be parsed)
    ch1, ch2 = itertools.tee(lines_to_chunks(lines))
    linenos = itertools.accumulate(itertools.chain([0], (len(c) for c in ch1)))
    nodes = env.from_chunks(ch2)
    nodelist = []
    for lineno, node in zip(linenos, nodes):
        lineno += 1 # in text editors lines are 1-indexed
        node.linenumber = lineno
        nodelist.append(node)
    # parse headings (level, TODO, TAGs, and heading)
    nodelist[0]._index = 0
    for (i, node) in enumerate(nodelist[1:], 1):   # nodes except root node
        node._index = i
        node._parse_pre()
    env._nodes = nodelist
    return nodelist[0]  # root