transform.py 30.7 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
# transform.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.
18
19


20
21
22
"""
Module ``transform`` contains the functions for transforming the
concrete syntax tree (CST) into an abstract syntax tree (AST).
23

24
25
26
As these functions are very generic, they can in principle be
used for any kind of tree transformations, not necessarily only
for CST -> AST transformations.
27
28
"""

29

30
31
32
import inspect
from functools import partial, reduce, singledispatch

33
from DHParser.syntaxtree import Node, WHITESPACE_PTYPE, TOKEN_PTYPE, MockParser
34
35
36
37
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
38

39
40
41
42
43
__all__ = ('TransformationDict',
           'TransformationProc',
           'ConditionFunc',
           'KeyFunc',
           'transformation_factory',
44
45
46
           'key_parser_name',
           'key_tag_name',
           'traverse',
47
           'is_named',
48
           'replace_by_single_child',
Eckhart Arnold's avatar
Eckhart Arnold committed
49
           'reduce_single_child',
50
           'replace_or_reduce',
51
52
           'replace_parser',
           'collapse',
53
           'merge_children',
54
55
           'replace_content',
           'apply_if',
eckhart's avatar
eckhart committed
56
           'apply_unless',
57
           'traverse_locally',
58
           'is_anonymous',
59
60
61
62
           'is_whitespace',
           'is_empty',
           'is_expendable',
           'is_token',
63
           'is_one_of',
64
           'has_content',
di68kap's avatar
di68kap committed
65
           'has_parent',
66
67
68
69
70
71
72
73
           'lstrip',
           'rstrip',
           'strip',
           'keep_children',
           'keep_children_if',
           'keep_tokens',
           'keep_nodes',
           'keep_content',
74
           'remove_children_if',
eckhart's avatar
eckhart committed
75
           'remove_nodes',
76
77
78
79
80
           'remove_content',
           'remove_first',
           'remove_last',
           'remove_whitespace',
           'remove_empty',
di68kap's avatar
di68kap committed
81
           'remove_anonymous_empty',
82
83
           'remove_anonymous_expendables',
           'remove_anonymous_tokens',
84
85
           'remove_expendables',
           'remove_brackets',
86
87
           'remove_infix_operator',
           'remove_single_child',
88
89
90
91
           'remove_tokens',
           'flatten',
           'forbid',
           'require',
92
93
           'assert_content',
           'assert_condition',
eckhart's avatar
eckhart committed
94
           'assert_has_children')
95
96


97
TransformationProc = Callable[[List[Node]], None]
Eckhart Arnold's avatar
Eckhart Arnold committed
98
99
TransformationDict = Dict[str, Sequence[Callable]]
ProcessingTableType = Dict[str, Union[Sequence[Callable], TransformationDict]]
100
101
ConditionFunc = Callable  # Callable[[List[Node]], bool]
KeyFunc = Callable[[Node], str]
eckhart's avatar
eckhart committed
102
CriteriaType = Union[int, str, Callable]
103
104


105
def transformation_factory(t1=None, t2=None, t3=None, t4=None, t5=None):
106
    """Creates factory functions from transformation-functions that
107
    dispatch on the first parameter after the context parameter.
108
109
110
111

    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
112
    takes just the context-parameter.
113
114
115
116
117
118
119
120
121

    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.

122
123
    Usage::

eckhart's avatar
eckhart committed
124
        @transformation_factory(AbstractSet[str])
125
        def remove_tokens(context, tokens):
126
            ...
127
128
129

    or, alternatively::

130
        @transformation_factory
131
        def remove_tokens(context, tokens: AbstractSet[str]):
132
133
            ...

134
135
    Example::

136
        trans_table = { 'expression': remove_tokens('+', '-') }
137
138
139

    instead of::

140
        trans_table = { 'expression': partial(remove_tokens, tokens={'+', '-'}) }
141
142

    Parameters:
143
        t1:  type of the second argument of the transformation function,
144
145
            only necessary if the transformation functions' parameter list
            does not have type annotations.
146
147
148
149
150
151
152
    """

    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
153
        assert t1 or params[0].annotation != params[0].empty, \
154
            "No type information on second parameter found! Please, use type " \
eckhart's avatar
eckhart committed
155
            "annotation or provide the type information via transformer-decorator."
156
        p1type = t1 or params[0].annotation
157
        f = singledispatch(f)
eckhart's avatar
eckhart committed
158
159
160
161
162
163
164
165
166
167
168
169
        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
170
171
172
173
174
175

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

176
177
178
179
180
        for t in (p1type, t2, t3, t4, t5):
            if t:
                f.register(t, gen_partial)
            else:
                break
181
182
        return f

183
    if isinstance(t1, type(lambda: 1)):
184
185
186
        # Provide for the case that transformation_factory has been
        # written as plain decorator and not as a function call that
        # returns the decorator proper.
187
188
        func = t1
        t1 = None
189
190
191
192
193
        return decorator(func)
    else:
        return decorator


194
def key_parser_name(node: Node) -> str:
195
196
197
    return node.parser.name


198
def key_tag_name(node: Node) -> str:
199
200
201
    return node.tag_name


202
def traverse(root_node: Node,
Eckhart Arnold's avatar
Eckhart Arnold committed
203
             processing_table: ProcessingTableType,
204
205
206
             key_func: KeyFunc=key_tag_name) -> None:
    """
    Traverses the snytax tree starting with the given ``node`` depth
207
    first and applies the sequences of callback-functions registered
208
    in the ``processing_table``-dictionary.
209
210
211
212
213
214
215
216
217

    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:
218
219
220
221
222

    - '+': 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)
223
224
225
226
227

    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
228
229
            is interpreted as a ``compact_table``. See
            :func:`expand_table` or :func:`EBNFCompiler.EBNFTransTable`
230
231
232
        key_func (function): A mapping key_func(node) -> keystr. The default
            key_func yields node.parser.name.

233
234
    Example::

235
        table = { "term": [replace_by_single_child, flatten],
236
                  "factor, flowmarker, retrieveop": replace_by_single_child }
237
        traverse(node, table)
238

239
    """
240
241
242
243
244
245
    # 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:
246
247
        # normalize processing_table entries by turning single values
        # into lists with a single value
248
249
        table = {name: cast(Sequence[Callable], smart_list(call))
                 for name, call in list(processing_table.items())}
250
        table = expand_table(table)
Eckhart Arnold's avatar
Eckhart Arnold committed
251
        cache = table.setdefault('__cache__', cast(TransformationDict, dict()))
252
253
        # change processing table in place, so its already expanded and cache filled next time
        processing_table.clear()
254
255
256
257
258
259
        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
260
    # cache = {}  # type: Dict[str, List[Callable]]
261

262
263
    def traverse_recursive(context):
        node = context[-1]
264
265
        if node.children:
            for child in node.result:
266
267
                context.append(child)
                traverse_recursive(context)  # depth first
268
                node.error_flag = max(node.error_flag, child.error_flag)  # propagate error flag
269
                context.pop()
270
271

        key = key_func(node)
272
273
274
        try:
            sequence = cache[key]
        except KeyError:
275
276
277
            sequence = table.get('+', []) \
                + table.get(key, table.get('*', [])) \
                + table.get('~', [])
278
279
280
281
282
283
284
            # '+' 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:
285
            call(context)
286

287
    traverse_recursive([root_node])
288
289
    # assert processing_table['__cache__']

290

291
#######################################################################
292
#
293
294
# meta transformations, i.e. transformations that call other
# transformations
295
#
296
#######################################################################
297
298


eckhart's avatar
eckhart committed
299
@transformation_factory(dict)
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
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)


eckhart's avatar
eckhart committed
330
331
332
333
334
335
336
@transformation_factory(Callable)
def apply_unless(context: List[Node], transformation: Callable, condition: Callable):
    """Applies a transformation if a certain condition is *not* met."""
    if not condition(context):
        transformation(context)


337
338
339
340
341
342
343
344
345
346
347
348
349
350
#######################################################################
#
# 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:
eckhart's avatar
eckhart committed
351
    """Returns ``True`` if the current node does not have any siblings."""
352
353
354
355
    return len(context[-2].children) == 1


def is_named(context: List[Node]) -> bool:
eckhart's avatar
eckhart committed
356
    """Returns ``True`` if the current node's parser is a named parser."""
357
358
359
360
    return bool(context[-1].parser.name)


def is_anonymous(context: List[Node]) -> bool:
eckhart's avatar
eckhart committed
361
    """Returns ``True`` if the current node's parser is an anonymous parser."""
362
363
364
365
366
367
368
369
370
371
    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:
eckhart's avatar
eckhart committed
372
    """Returns ``True`` if the current node's content is empty."""
373
374
375
376
    return not context[-1].result


def is_expendable(context: List[Node]) -> bool:
eckhart's avatar
eckhart committed
377
378
    """Returns ``True`` if the current node either is a node containing
    whitespace or an empty node."""
379
380
381
    return is_empty(context) or is_whitespace(context)


eckhart's avatar
eckhart committed
382
@transformation_factory  # (AbstractSet[str])
383
384
385
386
387
388
389
390
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:
391
        """Removes leading and trailing whitespace-nodes from content."""
392
393
394
395
396
        # 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
397
            while k > 0 and nd.children[k - 1].parser.ptype == WHITESPACE_PTYPE:
398
399
400
401
                k -= 1
            return "".join(child.content for child in node.children[i:k])
        return nd.content
    node = context[-1]
402
    return node.parser.ptype == TOKEN_PTYPE and (not tokens or stripped(node) in tokens)
403
404


eckhart's avatar
eckhart committed
405
@transformation_factory
406
407
408
409
410
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


eckhart's avatar
eckhart committed
411
@transformation_factory
412
413
414
415
416
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))


eckhart's avatar
eckhart committed
417
@transformation_factory
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
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):
435
436
437
438
439
440
441
442
    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


443
def _reduce_child(node: Node, child: Node):
444
445
446
447
    node._errors.extend(child._errors)
    node.result = child.result


448
def _pick_child(context: List[Node], criteria: CriteriaType):
eckhart's avatar
eckhart committed
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
    """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
468
469


470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
#######################################################################
#
# 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)
516
517


518
519
def replace_by_single_child(context: List[Node]):
    """
520
521
522
    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.
523
524
525
    """
    node = context[-1]
    if len(node.children) == 1:
526
        _replace_by(node, node.children[0])
527
528


Eckhart Arnold's avatar
Eckhart Arnold committed
529
def reduce_single_child(context: List[Node]):
530
    """
531
    Reduces a single branch node by transferring the result of its
532
    immediate descendant to this node, but keeping this node's parser entry.
533
534
    Reduction only takes place if the last node in the context has
    exactly one child.
535
536
537
    """
    node = context[-1]
    if len(node.children) == 1:
538
        _reduce_child(node, node.children[0])
539
540
541


@transformation_factory(Callable)
542
543
544
def replace_or_reduce(context: List[Node], condition: Callable=is_named):
    """
    Replaces node by a single child, if condition is met on child,
545
546
    otherwise (i.e. if the child is anonymous) reduces the child.
    """
547
    node = context[-1]
548
    if len(node.children) == 1:
di68kap's avatar
di68kap committed
549
        child = node.children[0]
550
        if condition(context):
551
            _replace_by(node, child)
552
        else:
553
            _reduce_child(node, child)
554
555
556


@transformation_factory
557
558
559
def replace_parser(context: List[Node], name: str):
    """
    Replaces the parser of a Node with a mock parser with the given
560
561
562
    name.

    Parameters:
563
564
        context: the context where the parser shall be replaced
        name: "NAME:PTYPE" of the surogate. The ptype is optional
565
    """
566
    node = context[-1]
567
568
569
570
571
    name, ptype = (name.split(':') + [''])[:2]
    node.parser = MockParser(name, ptype)


@transformation_factory(Callable)
572
573
def flatten(context: List[Node], condition: Callable=is_anonymous, recursive: bool=True):
    """
574
    Flattens all children, that fulfil the given ``condition``
575
576
577
    (default: all unnamed children). Flattening means that wherever a
    node has child nodes, the child nodes are inserted in place of the
    node.
578
579

    If the parameter ``recursive`` is ``True`` the same will recursively be
580
581
582
    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.
583
584
585
586

    Applying flatten recursively will result in these kinds of
    structural transformation::

587
588
589
        (1 (+ 2) (+ 3)     ->   (1 + 2 + 3)
        (1 (+ (2 + (3))))  ->   (1 + 2 + 3)
    """
590
    node = context[-1]
591
    if node.children:
Eckhart Arnold's avatar
Eckhart Arnold committed
592
        new_result = []     # type: List[Node]
593
        for child in node.children:
594
595
            context.append(child)
            if child.children and condition(context):
596
                if recursive:
597
                    flatten(context, condition, recursive)
598
599
600
                new_result.extend(child.children)
            else:
                new_result.append(child)
601
            context.pop()
602
603
604
        node.result = tuple(new_result)


605
606
607
def collapse(context: List[Node]):
    """
    Collapses all sub-nodes of a node by replacing them with the
608
    string representation of the node.
609
    """
610
    node = context[-1]
611
    node.result = node.content
612
613
614


@transformation_factory
615
616
def merge_children(context: List[Node], tag_names: List[str]):
    """
617
618
619
    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.
620
    """
Eckhart Arnold's avatar
Eckhart Arnold committed
621
    node = context[-1]
622
    result = []
623
    name, ptype = ('', tag_names[0]) if tag_names[0][:1] == ':' else (tag_names[0], '')
624
    if node.children:
625
        i = 0
626
627
628
629
630
631
632
633
634
635
636
        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
637
638
                                   reduce(lambda a, b: a + b,
                                          (node.children for node in node.children[i:k]))))
639
640
641
642
            i = k
        node.result = tuple(result)


eckhart's avatar
eckhart committed
643
@transformation_factory(Callable)
644
def replace_content(context: List[Node], func: Callable):  # Callable[[Node], ResultType]
di68kap's avatar
di68kap committed
645
    """Replaces the content of the node. ``func`` takes the node's result
646
647
    as an argument an returns the mapped result.
    """
648
    node = context[-1]
649
650
651
    node.result = func(node.result)


652
653
654
655
656
657
658
659
660
#######################################################################
#
# 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
#
#######################################################################
661
662


663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
@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
693
def strip(context: List[Node], condition: Callable = is_expendable):
694
695
696
697
698
    """Removes leading and trailing child-nodes that fulfill a given condition."""
    lstrip(context, condition)
    rstrip(context, condition)


699
@transformation_factory(slice)
700
def keep_children(context: List[Node], section: slice = slice(None)):
701
    """Keeps only child-nodes which fall into a slice of the result field."""
702
    node = context[-1]
703
    if node.children:
704
        node.result = node.children[section]
705
706
707


@transformation_factory(Callable)
708
709
710
711
712
713
714
715
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
eckhart's avatar
eckhart committed
716
def keep_tokens(context: List[Node], tokens: AbstractSet[str]=frozenset()):
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
    """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):
737
738
739
740
741
742
    """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
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
# @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)
764
765


766
767
remove_whitespace = remove_children_if(is_whitespace)
# partial(remove_children_if, condition=is_whitespace)
768
remove_empty = remove_children_if(is_empty)
di68kap's avatar
di68kap committed
769
remove_anonymous_empty = remove_children_if(lambda ctx: is_empty(ctx) and is_anonymous(ctx))
770
771
772
773
remove_expendables = remove_children_if(is_expendable)
# partial(remove_children_if, condition=is_expendable)
remove_anonymous_expendables = remove_children_if(lambda ctx: is_anonymous(ctx)
                                                  and is_expendable(ctx))
774
remove_anonymous_tokens = remove_children_if(lambda ctx: is_token(ctx) and is_anonymous(ctx))
775
776
777
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)
778
remove_infix_operator = keep_children(slice(0, None, 2))
779
remove_single_child = apply_if(keep_children(slice(0)), lambda ctx: len(ctx[-1].children) == 1)
780
781


eckhart's avatar
eckhart committed
782
783
@transformation_factory(AbstractSet[str])
def remove_tokens(context: List[Node], tokens: AbstractSet[str]=frozenset()):
784
    """Removes any among a particular set of tokens from the immediate
785
786
    descendants of a node. If ``tokens`` is the empty set, all tokens
    are removed."""
787
    remove_children_if(context, partial(is_token, tokens=tokens))
788
789
790


@transformation_factory
eckhart's avatar
eckhart committed
791
def remove_nodes(context: List[Node], tag_names: AbstractSet[str]):
Eckhart Arnold's avatar
Eckhart Arnold committed
792
    """Removes children by tag name."""
793
    remove_children_if(context, partial(is_one_of, tag_name_set=tag_names))
794
795
796


@transformation_factory
797
def remove_content(context: List[Node], regexp: str):
798
    """Removes children depending on their string value."""
799
    remove_children_if(context, partial(has_content, regexp=regexp))
800
801
802
803


########################################################################
#
804
# AST semantic validation functions (EXPERIMENTAL!!!)
805
806
807
#
########################################################################

808
@transformation_factory(Callable)
Eckhart Arnold's avatar
Eckhart Arnold committed
809
def assert_condition(context: List[Node], condition: Callable, error_msg: str = ''):
810
    """Checks for `condition`; adds an error message if condition is not met."""
811
    node = context[-1]
812
    if not condition(context):
813
814
815
816
817
818
819
820
821
822
823
824
825
        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
826
def assert_content(context: List[Node], regexp: str):
827
    node = context[-1]
828
    if not has_content(context, regexp):
829
        node.add_error('Element "%s" violates %s on %s' %
830
                       (node.parser.name, str(regexp), node.content))
831

832
833

@transformation_factory
834
def require(context: List[Node], child_tags: AbstractSet[str]):
835
    node = context[-1]
836
837
838
839
840
841
842
    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
843
def forbid(context: List[Node], child_tags: AbstractSet[str]):
844
    node = context[-1]
845
846
847
848
    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))