@@ -283,6 +283,22 @@ class ParsedTree(_Base, _Tree):
283283
284284 List format
285285 Not implemented.
286+
287+ Penn Treebank format
288+ Used for :meth:`from_penn` and :meth:`to_penn`.
289+
290+ .. code-block:: python
291+
292+ [
293+ 'S',
294+ [ 'Head:Nab', '中文字', ],
295+ [ 'particle:Td', '耶', ],
296+ ]
297+
298+ .. note::
299+
300+ One may use :meth:`to_penn` together with `SvgLing <https://pypi.org/project/svgling/>`_ to generate SVG tree graphs.
301+
286302 """
287303
288304 node_class = ParsedNode
@@ -303,6 +319,8 @@ def normalize_text(tree_text):
303319 def __str__ (self ):
304320 self .to_text ()
305321
322+ ########################################################################################################################
323+
306324 @classmethod
307325 def from_text (cls , data , * , normalize = True ):
308326 """Construct an instance from text format.
@@ -319,33 +337,33 @@ def from_text(cls, data, *, normalize=True):
319337
320338 tree = cls ()
321339 node_id = 0
322- node_queue = [None ]
340+ node_stack = [None ]
323341 text = ''
324342 ending = True
325343
326344 for char in data :
327345 if char == '(' :
328346 node_data = cls .node_class .data_class .from_text (text )
329- tree .create_node (tag = text , identifier = node_id , parent = node_queue [- 1 ], data = node_data )
347+ tree .create_node (tag = text , identifier = node_id , parent = node_stack [- 1 ], data = node_data )
330348
331- node_queue .append (node_id )
349+ node_stack .append (node_id )
332350 node_id += 1
333351 text = ''
334352
335353 elif char == ')' :
336354 if not ending :
337355 node_data = cls .node_class .data_class .from_text (text )
338- tree .create_node (tag = text , identifier = node_id , parent = node_queue [- 1 ], data = node_data )
356+ tree .create_node (tag = text , identifier = node_id , parent = node_stack [- 1 ], data = node_data )
339357 node_id += 1
340358
341- node_queue .pop ()
359+ node_stack .pop ()
342360 text = ''
343361 ending = True
344362
345363 elif char == '|' :
346364 if not ending :
347365 node_data = cls .node_class .data_class .from_text (text )
348- tree .create_node (tag = text , identifier = node_id , parent = node_queue [- 1 ], data = node_data )
366+ tree .create_node (tag = text , identifier = node_id , parent = node_stack [- 1 ], data = node_data )
349367 node_id += 1
350368
351369 text = ''
@@ -383,7 +401,7 @@ def to_text(self, node_id=None):
383401
384402 @classmethod
385403 def from_dict (cls , data ):
386- """Construct an instance a from python built-in containers.
404+ """Construct an instance from python built-in containers.
387405
388406 Parameters
389407 ----------
@@ -392,22 +410,22 @@ def from_dict(cls, data):
392410 """
393411 tree = cls ()
394412
395- queue = _deque ()
396- queue .append ((data , None ,))
413+ node_queue = _deque ()
414+ node_queue .append ((data , None ,))
397415
398- while queue :
399- node_dict , parent_id = queue .popleft ()
416+ while node_queue :
417+ node_dict , parent_id = node_queue .popleft ()
400418 node_id = node_dict ['id' ]
401419 node_data = cls .node_class .data_class .from_dict (node_dict ['data' ])
402420 tree .create_node (tag = node_data .to_text (), identifier = node_id , parent = parent_id , data = node_data )
403421
404422 for child in node_dict ['children' ]:
405- queue .append ((child , node_id ,))
423+ node_queue .append ((child , node_id ,))
406424
407425 return tree
408426
409427 def to_dict (self , node_id = None ):
410- """Construct an instance a from python built-in containers.
428+ """Transform to python built-in containers.
411429
412430 Parameters
413431 ----------
@@ -429,6 +447,70 @@ def to_dict(self, node_id=None):
429447
430448 return tree_dict
431449
450+ @classmethod
451+ def from_penn (cls , data ):
452+ """Construct an instance from Penn Treebank format."""
453+ tree = cls ()
454+
455+ node_stack = _deque ()
456+ node_stack .append ((data , None ,))
457+
458+ node_id = 0
459+
460+ while node_stack :
461+ penn_data , parent_id = node_stack .pop ()
462+
463+ if not penn_data :
464+ raise SyntaxError (f'Empty node #{ node_id } )
465+
466+ if not isinstance (penn_data [0 ], str ):
467+ raise SyntaxError (f'First element of a node must be string, got { type (penn_data [0 ])} )
468+
469+ if len (penn_data ) == 2 and isinstance (penn_data [- 1 ], str ):
470+ penn_data = (':' .join (penn_data ),)
471+
472+ node_data = cls .node_class .data_class .from_text (penn_data [0 ])
473+ tree .create_node (tag = node_data .to_text (), identifier = node_id , parent = parent_id , data = node_data )
474+
475+ for child in penn_data [- 1 :0 :- 1 ]:
476+ node_stack .append ((child , node_id ,))
477+ node_id += 1
478+
479+ return tree
480+
481+ def to_penn (self , node_id = None , * , with_role = True , with_word = True , sep = ':' ):
482+ """Transform to Penn Treebank format.
483+
484+ Parameters
485+ ----------
486+ node_id : int
487+ Output the plain text format for the subtree under **node_id**.
488+ with_role : bool
489+ Contains role-tag or not.
490+ with_word : bool
491+ Contains word or not.
492+ sep : str
493+ The seperator between role and POS-tag.
494+
495+ Returns
496+ -------
497+ list
498+ """
499+ if node_id is None :
500+ node_id = self .root
501+ node = self [node_id ]
502+
503+ penn_data = [f'{ node .data .role } { sep } { node .data .pos } if with_role and node .data .role else node .data .pos ,]
504+ if with_word and node .data .word :
505+ penn_data .append (node .data .word )
506+
507+ for child in self .children (node_id ):
508+ penn_data .append (self .to_penn (child .identifier , with_role = with_role , with_word = with_word , sep = sep ))
509+
510+ return penn_data
511+
512+ ########################################################################################################################
513+
432514 def show (self , * ,
433515 key = lambda node : node .identifier ,
434516 idhidden = False ,
@@ -541,10 +623,14 @@ def get_relations(self, root_id=None, *, semantic=True):
541623 for tail in children :
542624 if tail .data .role != 'Head' and tail not in head_children :
543625 if tail .is_leaf ():
544- yield ParsedRelation (head = head_node , tail = tail , relation = tail ) # pylint: disable=no-value-for-parameter
626+ yield ParsedRelation ( # pylint: disable=no-value-for-parameter
627+ head = head_node , tail = tail , relation = tail ,
628+ )
545629 else :
546630 for node in self .get_heads (tail .identifier , semantic = semantic ):
547- yield ParsedRelation (head = head_node , tail = node , relation = tail ) # pylint: disable=no-value-for-parameter
631+ yield ParsedRelation ( # pylint: disable=no-value-for-parameter
632+ head = head_node , tail = node , relation = tail ,
633+ )
548634
549635 # Recursion
550636 for child in children :
0 commit comments