transform.py 29.5 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
"""transformation.py - transformation functions for converting the
                       concrete into the abstract syntax tree

Copyright 2016  by Eckhart Arnold (arnold@badw.de)
                Bavarian Academy of Sciences an Humanities (badw.de)

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
implied.  See the License for the specific language governing
permissions and limitations under the License.
"""

import inspect
from functools import partial, reduce, singledispatch

23
from DHParser.syntaxtree import Node, WHITESPACE_PTYPE, TOKEN_PTYPE, MockParser
24

25
26
27
28
from DHParser.toolkit import expand_table, smart_list, re, typing

from typing import AbstractSet, Any, ByteString, Callable, cast, Container, Dict, \
    Iterator, List, NamedTuple, Sequence, Union, Text, Tuple
29

30
31
32
33
34
__all__ = ('TransformationDict',
           'TransformationProc',
           'ConditionFunc',
           'KeyFunc',
           'transformation_factory',
35
36
37
           'key_parser_name',
           'key_tag_name',
           'traverse',
38
           'is_named',
39
           'replace_by_single_child',
Eckhart Arnold's avatar
Eckhart Arnold committed
40
           'reduce_single_child',
41
           'replace_or_reduce',
42
43
           'replace_parser',
           'collapse',
44
           'merge_children',
45
46
           'replace_content',
           'apply_if',
47
           'traverse_locally',
48
           'is_anonymous',
49
50
51
52
           'is_whitespace',
           'is_empty',
           'is_expendable',
           'is_token',
53
           'is_one_of',
54
           'has_content',
55
56
57
58
59
60
61
62
           'lstrip',
           'rstrip',
           'strip',
           'keep_children',
           'keep_children_if',
           'keep_tokens',
           'keep_nodes',
           'keep_content',
63
           'remove_children_if',
eckhart's avatar
eckhart committed
64
           'remove_nodes',
65
66
67
68
69
           'remove_content',
           'remove_first',
           'remove_last',
           'remove_whitespace',
           'remove_empty',
di68kap's avatar
di68kap committed
70
           'remove_anonymous_empty',
71
72
           'remove_expendables',
           'remove_brackets',
73
74
           'remove_infix_operator',
           'remove_single_child',
75
76
77
78
           'remove_tokens',
           'flatten',
           'forbid',
           'require',
79
80
           'assert_content',
           'assert_condition',
eckhart's avatar
eckhart committed
81
           'assert_has_children')
82
83


84
TransformationProc = Callable[[List[Node]], None]
Eckhart Arnold's avatar
Eckhart Arnold committed
85
86
TransformationDict = Dict[str, Sequence[Callable]]
ProcessingTableType = Dict[str, Union[Sequence[Callable], TransformationDict]]
87
88
ConditionFunc = Callable  # Callable[[List[Node]], bool]
KeyFunc = Callable[[Node], str]
eckhart's avatar
eckhart committed
89
CriteriaType = Union[int, str, Callable]
90
91


92
def transformation_factory(t1=None, t2=None, t3=None, t4=None, t5=None):
93
    """Creates factory functions from transformation-functions that
94
    dispatch on the first parameter after the context parameter.
95
96
97
98

    Decorating a transformation-function that has more than merely the
    ``node``-parameter with ``transformation_factory`` creates a
    function with the same name, which returns a partial-function that
99
    takes just the context-parameter.
100
101
102
103
104
105
106
107
108
109

    Additionally, there is some some syntactic sugar for
    transformation-functions that receive a collection as their second
    parameter and do not have any further parameters. In this case a
    list of parameters passed to the factory function will be converted
    into a collection.

    Main benefit is readability of processing tables.

    Usage:
eckhart's avatar
eckhart committed
110
        @transformation_factory(AbstractSet[str])
111
        def remove_tokens(context, tokens):
112
113
114
            ...
      or, alternatively:
        @transformation_factory
115
        def remove_tokens(context, tokens: AbstractSet[str]):
116
117
118
119
120
121
            ...

    Example:
        trans_table = { 'expression': remove_tokens('+', '-') }
      instead of:
        trans_table = { 'expression': partial(remove_tokens, tokens={'+', '-'}) }
122
123

    Parameters:
124
        t1:  type of the second argument of the transformation function,
125
126
            only necessary if the transformation functions' parameter list
            does not have type annotations.
127
128
129
130
131
132
133
    """

    def decorator(f):
        sig = inspect.signature(f)
        params = list(sig.parameters.values())[1:]
        if len(params) == 0:
            return f  # '@transformer' not needed w/o free parameters
134
        assert t1 or params[0].annotation != params[0].empty, \
135
136
            "No type information on second parameter found! Please, use type " \
            "annotation or provide the type information via transfomer-decorator."
137
        p1type = t1 or params[0].annotation
138
        f = singledispatch(f)
eckhart's avatar
eckhart committed
139
140
141
142
143
144
145
146
147
148
149
150
        try:
            if len(params) == 1 and issubclass(p1type, Container) \
                    and not issubclass(p1type, Text) and not issubclass(p1type, ByteString):
                def gen_special(*args):
                    c = set(args) if issubclass(p1type, AbstractSet) else \
                        list(args) if issubclass(p1type, Sequence) else args
                    d = {params[0].name: c}
                    return partial(f, **d)

                f.register(p1type.__args__[0], gen_special)
        except AttributeError:
            pass  # Union Type does not allow subclassing, but is not needed here
151
152
153
154
155
156

        def gen_partial(*args, **kwargs):
            d = {p.name: arg for p, arg in zip(params, args)}
            d.update(kwargs)
            return partial(f, **d)

157
158
159
160
161
        for t in (p1type, t2, t3, t4, t5):
            if t:
                f.register(t, gen_partial)
            else:
                break
162
163
        return f

164
    if isinstance(t1, type(lambda: 1)):
165
166
167
        # Provide for the case that transformation_factory has been
        # written as plain decorator and not as a function call that
        # returns the decorator proper.
168
169
        func = t1
        t1 = None
170
171
172
173
174
        return decorator(func)
    else:
        return decorator


175
def key_parser_name(node: Node) -> str:
176
177
178
    return node.parser.name


179
def key_tag_name(node: Node) -> str:
180
181
182
    return node.tag_name


183
def traverse(root_node: Node,
Eckhart Arnold's avatar
Eckhart Arnold committed
184
             processing_table: ProcessingTableType,
185
186
187
             key_func: KeyFunc=key_tag_name) -> None:
    """
    Traverses the snytax tree starting with the given ``node`` depth
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
    first and applies the sequences of callback-functions registered
    in the ``calltable``-dictionary.

    The most important use case is the transformation of a concrete
    syntax tree into an abstract tree (AST). But it is also imaginable
    to employ tree-traversal for the semantic analysis of the AST.

    In order to assign sequences of callback-functions to nodes, a
    dictionary ("processing table") is used. The keys usually represent
    tag names, but any other key function is possible. There exist
    three special keys:
        '+': always called (before any other processing function)
        '*': called for those nodes for which no (other) processing
             function appears in the table
        '~': always called (after any other processing function)

    Args:
        root_node (Node): The root-node of the syntax tree to be traversed
        processing_table (dict): node key -> sequence of functions that
            will be applied to matching nodes in order. This dictionary
208
209
            is interpreted as a `compact_table`. See
            `toolkit.expand_table` or ``EBNFCompiler.EBNFTransTable`
210
211
212
213
        key_func (function): A mapping key_func(node) -> keystr. The default
            key_func yields node.parser.name.

    Example:
214
215
        table = { "term": [replace_by_single_child, flatten],
            "factor, flowmarker, retrieveop": replace_by_single_child }
216
217
        traverse(node, table)
    """
218
219
220
221
222
223
    # Is this optimazation really needed?
    if '__cache__' in processing_table:
        # assume that processing table has already been expanded
        table = processing_table
        cache = processing_table['__cache__']
    else:
224
225
        # normalize processing_table entries by turning single values
        # into lists with a single value
226
227
        table = {name: cast(Sequence[Callable], smart_list(call))
                 for name, call in list(processing_table.items())}
228
        table = expand_table(table)
Eckhart Arnold's avatar
Eckhart Arnold committed
229
        cache = table.setdefault('__cache__', cast(TransformationDict, dict()))
230
231
        # change processing table in place, so its already expanded and cache filled next time
        processing_table.clear()
232
233
234
235
236
237
        processing_table.update(table)

    # assert '__cache__' in processing_table
    # # Code without optimization
    # table = {name: smart_list(call) for name, call in list(processing_table.items())}
    # table = expand_table(table)
Eckhart Arnold's avatar
Eckhart Arnold committed
238
    # cache = {}  # type: Dict[str, List[Callable]]
239

240
241
    def traverse_recursive(context):
        node = context[-1]
242
243
        if node.children:
            for child in node.result:
244
245
                context.append(child)
                traverse_recursive(context)  # depth first
246
                node.error_flag = max(node.error_flag, child.error_flag)  # propagate error flag
247
                context.pop()
248
249

        key = key_func(node)
250
251
252
        try:
            sequence = cache[key]
        except KeyError:
253
254
255
256
257
258
259
260
261
262
            sequence = table.get('+', []) + \
                       table.get(key, table.get('*', [])) + \
                       table.get('~', [])
            # '+' always called (before any other processing function)
            # '*' called for those nodes for which no (other) processing function
            #     appears in the table
            # '~' always called (after any other processing function)
            cache[key] = sequence

        for call in sequence:
263
            call(context)
264

265
    traverse_recursive([root_node])
266
267
    # assert processing_table['__cache__']

268

269
#######################################################################
270
#
271
272
# meta transformations, i.e. transformations that call other
# transformations
273
#
274
#######################################################################
275
276


277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
@transformation_factory(Dict)
def traverse_locally(context: List[Node],
                     processing_table: Dict,            # actually: ProcessingTableType
                     key_func: Callable=key_tag_name):  # actually: KeyFunc
    """Transforms the syntax tree starting from the last node in the context
    according to the given processing table. The purpose of this function is
    to apply certain transformations locally, i.e. only for those nodes that
    have the last node in the context as their parent node.
    """
    traverse(context[-1], processing_table, key_func)


# @transformation_factory(List[Callable])
# def apply_to_child(context: List[Node], transformations: List[Callable], condition: Callable):
#     """Applies a list of transformations to those children that meet a specifc condition."""
#     node = context[-1]
#     for child in node.children:
#         context.append(child)
#         if condition(context):
#             for transform in transformations:
#                 transform(context)
#         context.pop()


@transformation_factory(Callable)
def apply_if(context: List[Node], transformation: Callable, condition: Callable):
    """Applies a transformation only if a certain condition is met."""
    if condition(context):
        transformation(context)


#######################################################################
#
# conditionals that determine whether the context (or the last node in
# the context for that matter) fulfill a specific condition.
# ---------------------------------------------------------------------
#
# The context of a node is understood as a list of all parent nodes
# leading up to and including the node itself. If represented as list,
# the last element of the list is the node itself.
#
#######################################################################


def is_single_child(context: List[Node]) -> bool:
    return len(context[-2].children) == 1


def is_named(context: List[Node]) -> bool:
    return bool(context[-1].parser.name)


def is_anonymous(context: List[Node]) -> bool:
    return not context[-1].parser.name


def is_whitespace(context: List[Node]) -> bool:
    """Removes whitespace and comments defined with the
    ``@comment``-directive."""
    return context[-1].parser.ptype == WHITESPACE_PTYPE


def is_empty(context: List[Node]) -> bool:
    return not context[-1].result


def is_expendable(context: List[Node]) -> bool:
    return is_empty(context) or is_whitespace(context)


@transformation_factory(AbstractSet[str])
def is_token(context: List[Node], tokens: AbstractSet[str] = frozenset()) -> bool:
    """Checks whether the last node in the context has `ptype == TOKEN_PTYPE`
    and it's content matches one of the given tokens. Leading and trailing
    whitespace-tokens will be ignored. In case an empty set of tokens is passed,
    any token is a match. If only ":" is given all anonymous tokens but no other
    tokens are a match.
    """
    def stripped(nd: Node) -> str:
        # assert node.parser.ptype == TOKEN_PTYPE
        if nd.children:
            i, k = 0, len(nd.children)
            while i < len(nd.children) and nd.children[i].parser.ptype == WHITESPACE_PTYPE:
                i += 1
            while k > 0 and nd.children[k-1].parser.ptype == WHITESPACE_PTYPE:
                k -= 1
            return "".join(child.content for child in node.children[i:k])
        return nd.content
    node = context[-1]
    return (node.parser.ptype == TOKEN_PTYPE
            and ((not tokens or stripped(node) in tokens)
                 or (not node.parser.name and len(tokens) == 1 and ":" in tokens)))


@transformation_factory(AbstractSet[str])
def is_one_of(context: List[Node], tag_name_set: AbstractSet[str]) -> bool:
    """Returns true, if the node's tag_name is one of the given tag names."""
    return context[-1].tag_name in tag_name_set


@transformation_factory(str)
def has_content(context: List[Node], regexp: str) -> bool:
    """Checks a node's content against a regular expression."""
    return bool(re.match(regexp, context[-1].content))


@transformation_factory(AbstractSet[str])
def has_parent(context: List[Node], tag_name_set: AbstractSet[str]) -> bool:
    """Checks whether a node with one of the given tag names appears somewhere
     in the context before the last node in the context."""
    for i in range(2, len(context)):
        if context[-i].tag_name in tag_name_set:
            return True
    return False


#######################################################################
#
# utility functions (private)
#
#######################################################################


def _replace_by(node: Node, child: Node):
401
402
403
404
405
406
407
408
    if not child.parser.name:
        child.parser = MockParser(node.parser.name, child.parser.ptype)
        # parser names must not be overwritten, else: child.parser.name = node.parser.name
    node.parser = child.parser
    node._errors.extend(child._errors)
    node.result = child.result


409
def _reduce_child(node: Node, child: Node):
410
411
412
413
    node._errors.extend(child._errors)
    node.result = child.result


414
def _pick_child(context: List[Node], criteria: CriteriaType):
eckhart's avatar
eckhart committed
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
    """Returns the first child that meets the criteria."""
    if isinstance(criteria, int):
        try:
            return context[-1].children[criteria]
        except IndexError:
            return None
    elif isinstance(criteria, str):
        for child in context[-1].children:
            if child.tag_name == criteria:
                return child
        return None
    else:  # assume criteria has type ConditionFunc
        for child in context[-1].children:
            context.append(child)
            evaluation = criteria(context)
            context.pop()
            if evaluation:
                return child
        return None
434
435


436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
#######################################################################
#
# rearranging transformations
#
# - tree may be rearranged (e.g.flattened)
# - nodes that are not leaves may be dropped
# - order is preserved
# - leave content is preserved (though not necessarily the leaves
#   themselves)
#
#######################################################################


# @transformation_factory(int, str, Callable)
# def replace_by_child(context: List[Node], criteria: CriteriaType=is_single_child):
#     """
#     Replaces a node by the first of its immediate descendants
#     that meets the `criteria`. The criteria can either be the
#     index of the child (counting from zero), or the tag name or
#     a boolean-valued function on the context of the child.
#     If no child matching the criteria is found, the node will
#     not be replaced.
#     With the default value for `criteria` the same semantics is
#     the same that of `replace_by_single_child`.
#     """
#     child = _pick_child(context, criteria)
#     if child:
#         _replace_by(context[-1], child)
#
#
# @transformation_factory(int, str, Callable)
# def content_from_child(context: List[Node], criteria: CriteriaType = is_single_child):
#     """
#     Reduces a node, by transferring the result of the first of its
#     immediate descendants that meets the `criteria` to this node,
#     but keeping this node's parser entry. The criteria can either
#     be the index of the child (counting from zero), or the tag
#     name or a boolean-valued function on the context of the child.
#     If no child matching the criteria is found, the node will
#     not be replaced.
#     With the default value for `criteria` this has the same semantics
#     as `content_from_single_child`.
#     """
#     child = _pick_child(context, criteria)
#     if child:
#         _reduce_child(context[-1], child)
482
483


484
485
def replace_by_single_child(context: List[Node]):
    """
486
487
488
    Removes single branch node, replacing it by its immediate descendant.
    Replacement only takes place, if the last node in the context has
    exactly one child.
489
490
491
    """
    node = context[-1]
    if len(node.children) == 1:
492
        _replace_by(node, node.children[0])
493
494


Eckhart Arnold's avatar
Eckhart Arnold committed
495
def reduce_single_child(context: List[Node]):
496
    """
497
    Reduces a single branch node by transferring the result of its
498
    immediate descendant to this node, but keeping this node's parser entry.
499
500
    Reduction only takes place if the last node in the context has
    exactly one child.
501
502
503
    """
    node = context[-1]
    if len(node.children) == 1:
504
        _reduce_child(node, node.children[0])
505
506
507


@transformation_factory(Callable)
508
509
510
def replace_or_reduce(context: List[Node], condition: Callable=is_named):
    """
    Replaces node by a single child, if condition is met on child,
511
512
    otherwise (i.e. if the child is anonymous) reduces the child.
    """
513
    node = context[-1]
514
    if len(node.children) == 1:
di68kap's avatar
di68kap committed
515
        child = node.children[0]
516
        if condition(context):
517
            _replace_by(node, child)
518
        else:
519
            _reduce_child(node, child)
520
521
522


@transformation_factory
523
524
525
def replace_parser(context: List[Node], name: str):
    """
    Replaces the parser of a Node with a mock parser with the given
526
527
528
    name.

    Parameters:
529
530
        context: the context where the parser shall be replaced
        name: "NAME:PTYPE" of the surogate. The ptype is optional
531
    """
532
    node = context[-1]
533
534
535
536
537
    name, ptype = (name.split(':') + [''])[:2]
    node.parser = MockParser(name, ptype)


@transformation_factory(Callable)
538
539
540
def flatten(context: List[Node], condition: Callable=is_anonymous, recursive: bool=True):
    """
    Flattens all children, that fulfil the given `condition`
541
542
543
544
545
546
547
548
549
550
551
552
    (default: all unnamed children). Flattening means that wherever a
    node has child nodes, the child nodes are inserted in place of the
    node.
     If the parameter `recursive` is `True` the same will recursively be
    done with the child-nodes, first. In other words, all leaves of
    this node and its child nodes are collected in-order as direct
    children of this node.
     Applying flatten recursively will result in these kinds of
    structural transformation:
        (1 (+ 2) (+ 3)     ->   (1 + 2 + 3)
        (1 (+ (2 + (3))))  ->   (1 + 2 + 3)
    """
553
    node = context[-1]
554
    if node.children:
Eckhart Arnold's avatar
Eckhart Arnold committed
555
        new_result = []     # type: List[Node]
556
        for child in node.children:
557
558
            context.append(child)
            if child.children and condition(context):
559
                if recursive:
560
                    flatten(context, condition, recursive)
561
562
563
                new_result.extend(child.children)
            else:
                new_result.append(child)
564
            context.pop()
565
566
567
        node.result = tuple(new_result)


568
569
570
def collapse(context: List[Node]):
    """
    Collapses all sub-nodes of a node by replacing them with the
571
    string representation of the node.
572
    """
573
    node = context[-1]
574
    node.result = node.content
575
576
577


@transformation_factory
578
579
def merge_children(context: List[Node], tag_names: List[str]):
    """
580
581
582
    Joins all children next to each other and with particular tag-names
    into a single child node with a mock-parser with the name of the
    first tag-name in the list.
583
    """
Eckhart Arnold's avatar
Eckhart Arnold committed
584
    node = context[-1]
585
    result = []
586
    name, ptype = ('', tag_names[0]) if tag_names[0][:1] == ':' else (tag_names[0], '')
587
    if node.children:
588
        i = 0
589
590
591
592
593
594
595
596
597
598
599
        L = len(node.children)
        while i < L:
            while i < L and not node.children[i].tag_name in tag_names:
                result.append(node.children[i])
                i += 1
            k = i + 1
            while (k < L and node.children[k].tag_name in tag_names
                   and bool(node.children[i].children) == bool(node.children[k].children)):
                k += 1
            if i < L:
                result.append(Node(MockParser(name, ptype),
Eckhart Arnold's avatar
Eckhart Arnold committed
600
601
                                   reduce(lambda a, b: a + b,
                                          (node.children for node in node.children[i:k]))))
602
603
604
605
606
            i = k
        node.result = tuple(result)


@transformation_factory
607
def replace_content(context: List[Node], func: Callable):  # Callable[[Node], ResultType]
608
609
610
    """Replaces the content of the node. ``func`` takes the node
    as an argument an returns the mapped result.
    """
611
    node = context[-1]
612
613
614
    node.result = func(node.result)


615
616
617
618
619
620
621
622
623
#######################################################################
#
# destructive transformations:
#
# - leaves may be dropped (e.g. if deemed irrelevant)
# - errors of dropped leaves will be lost
# - no promise that order will be preserved
#
#######################################################################
624
625


626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
@transformation_factory(Callable)
def lstrip(context: List[Node], condition: Callable = is_expendable):
    """Recursively removes all leading child-nodes that fulfill a given condition."""
    node = context[-1]
    i = 1
    while i > 0 and node.children:
        lstrip(context + [node.children[0]], condition)
        i, L = 0, len(node.children)
        while i < L and condition(context + [node.children[i]]):
            i += 1
        if i > 0:
            node.result = node.children[i:]


@transformation_factory(Callable)
def rstrip(context: List[Node], condition: Callable = is_expendable):
    """Recursively removes all leading nodes that fulfill a given condition."""
    node = context[-1]
    i, L = 0, len(node.children)
    while i < L and node.children:
        rstrip(context + [node.children[-1]], condition)
        L = len(node.children)
        i = L
        while i > 0 and condition(context + [node.children[i-1]]):
            i -= 1
        if i < L:
            node.result = node.children[:i]


@transformation_factory(Callable)
eckhart's avatar
eckhart committed
656
def strip(context: List[Node], condition: Callable = is_expendable):
657
658
659
660
661
    """Removes leading and trailing child-nodes that fulfill a given condition."""
    lstrip(context, condition)
    rstrip(context, condition)


662
@transformation_factory(slice)
663
def keep_children(context: List[Node], section: slice = slice(None)):
664
    """Keeps only child-nodes which fall into a slice of the result field."""
665
    node = context[-1]
666
    if node.children:
667
        node.result = node.children[section]
668
669
670


@transformation_factory(Callable)
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
def keep_children_if(context: List[Node], condition: Callable):
    """Removes all children for which `condition()` returns `True`."""
    node = context[-1]
    if node.children:
        node.result = tuple(c for c in node.children if condition(context + [c]))


@transformation_factory
def keep_tokens(context: List[Node], tokens: AbstractSet[str] = frozenset()):
    """Removes any among a particular set of tokens from the immediate
    descendants of a node. If ``tokens`` is the empty set, all tokens
    are removed."""
    keep_children_if(context, partial(is_token, tokens=tokens))


@transformation_factory
def keep_nodes(context: List[Node], tag_names: AbstractSet[str]):
    """Removes children by tag name."""
    keep_children_if(context, partial(is_one_of, tag_name_set=tag_names))


@transformation_factory
def keep_content(context: List[Node], regexp: str):
    """Removes children depending on their string value."""
    keep_children_if(context, partial(has_content, regexp=regexp))


@transformation_factory(Callable)
def remove_children_if(context: List[Node], condition: Callable):
700
701
702
703
704
705
    """Removes all children for which `condition()` returns `True`."""
    node = context[-1]
    if node.children:
        node.result = tuple(c for c in node.children if not condition(context + [c]))


eckhart's avatar
eckhart committed
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
# @transformation_factory(Callable)
# def remove_children(context: List[Node],
#                     condition: Callable = TRUE_CONDITION,
#                     section: slice = slice(None)):
#     """Removes all nodes from a slice of the result field if the function
#     `condition(child_node)` evaluates to `True`."""
#     node = context[-1]
#     if node.children:
#         c = node.children
#         N = len(c)
#         rng = range(*section.indices(N))
#         node.result = tuple(c[i] for i in range(N)
#                             if i not in rng or not condition(context + [c[i]]))
#         # selection = []
#         # for i in range(N):
#         #     context.append(c[i])
#         #     if not i in rng or not condition(context):
#         #         selection.append(c[i])
#         #     context.pop()
#         # if len(selection) != c:
#         #     node.result = tuple(selection)
727
728
729
730


remove_whitespace = remove_children_if(is_whitespace)  # partial(remove_children_if, condition=is_whitespace)
remove_empty = remove_children_if(is_empty)
di68kap's avatar
di68kap committed
731
remove_anonymous_empty = remove_children_if(lambda ctx: is_empty(ctx) and is_anonymous(ctx))
732
remove_expendables = remove_children_if(is_expendable)  # partial(remove_children_if, condition=is_expendable)
733
remove_anonymous_expendables = remove_children_if(lambda ctx: is_anonymous(ctx) and is_expendable(ctx))
734
735
736
remove_first = apply_if(keep_children(slice(1, None)), lambda ctx: len(ctx[-1].children) > 1)
remove_last = apply_if(keep_children(slice(None, -1)), lambda ctx: len(ctx[-1].children) > 1)
remove_brackets = apply_if(keep_children(slice(1, -1)), lambda ctx: len(ctx[-1].children) >= 2)
737
remove_infix_operator = keep_children(slice(0, None, 2))
738
remove_single_child = apply_if(keep_children(slice(0)), lambda ctx: len(ctx[-1].children) == 1)
739
740
741


@transformation_factory
742
def remove_tokens(context: List[Node], tokens: AbstractSet[str] = frozenset()):
743
    """Removes any among a particular set of tokens from the immediate
744
745
    descendants of a node. If ``tokens`` is the empty set, all tokens
    are removed."""
746
    remove_children_if(context, partial(is_token, tokens=tokens))
747
748
749


@transformation_factory
eckhart's avatar
eckhart committed
750
def remove_nodes(context: List[Node], tag_names: AbstractSet[str]):
Eckhart Arnold's avatar
Eckhart Arnold committed
751
    """Removes children by tag name."""
752
    remove_children_if(context, partial(is_one_of, tag_name_set=tag_names))
753
754
755


@transformation_factory
756
def remove_content(context: List[Node], regexp: str):
757
    """Removes children depending on their string value."""
758
    remove_children_if(context, partial(has_content, regexp=regexp))
759
760
761
762


########################################################################
#
763
# AST semantic validation functions (EXPERIMENTAL!!!)
764
765
766
#
########################################################################

767
@transformation_factory(Callable)
Eckhart Arnold's avatar
Eckhart Arnold committed
768
def assert_condition(context: List[Node], condition: Callable, error_msg: str = ''):
769
    """Checks for `condition`; adds an error message if condition is not met."""
770
    node = context[-1]
771
    if not condition(context):
772
773
774
775
776
777
778
779
780
781
782
783
784
        if error_msg:
            node.add_error(error_msg % node.tag_name if error_msg.find("%s") > 0 else error_msg)
        else:
            cond_name = condition.__name__ if hasattr(condition, '__name__') \
                        else condition.__class__.__name__ if hasattr(condition, '__class__') \
                        else '<unknown>'
            node.add_error("transform.assert_condition: Failed to meet condition " + cond_name)


assert_has_children = assert_condition(lambda nd: nd.children, 'Element "%s" has no children')


@transformation_factory
785
def assert_content(context: List[Node], regexp: str):
786
    node = context[-1]
787
    if not has_content(context, regexp):
788
        node.add_error('Element "%s" violates %s on %s' %
789
                       (node.parser.name, str(regexp), node.content))
790

791
792

@transformation_factory
793
def require(context: List[Node], child_tags: AbstractSet[str]):
794
    node = context[-1]
795
796
797
798
799
800
801
    for child in node.children:
        if child.tag_name not in child_tags:
            node.add_error('Element "%s" is not allowed inside "%s".' %
                           (child.parser.name, node.parser.name))


@transformation_factory
802
def forbid(context: List[Node], child_tags: AbstractSet[str]):
803
    node = context[-1]
804
805
806
807
    for child in node.children:
        if child.tag_name in child_tags:
            node.add_error('Element "%s" cannot be nested inside "%s".' %
                           (child.parser.name, node.parser.name))