syntaxtree.py 30.7 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
"""syntaxtree.py - syntax tree classes and transformation functions for 
converting the concrete into the abstract syntax tree for DHParser

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.
"""

20
import copy
21
import inspect
22
import os
23
from functools import partial, singledispatch
24
25
26
27
try:
    import regex as re
except ImportError:
    import re
28
29
30
31
32
33
try:
    from typing import AbstractSet, Any, ByteString, Callable, cast, Container, Dict, \
        Iterator, List, NamedTuple, Sequence, Union, Text, Tuple
except ImportError:
    from .typing34 import AbstractSet, Any, ByteString, Callable, cast, Container, Dict, \
        Iterator, List, NamedTuple, Sequence, Union, Text, Tuple
34

35
from DHParser.toolkit import log_dir, repr_call, expand_table, line_col, smart_list
36

Eckhart Arnold's avatar
Eckhart Arnold committed
37

38
39
__all__ = ['WHITESPACE_PTYPE',
           'TOKEN_PTYPE',
Eckhart Arnold's avatar
Eckhart Arnold committed
40
           'ZOMBIE_PARSER',
Eckhart Arnold's avatar
Eckhart Arnold committed
41
           'ParserBase',
Eckhart Arnold's avatar
Eckhart Arnold committed
42
43
           'Error',
           'Node',
44
           'TransformationFunc',
Eckhart Arnold's avatar
Eckhart Arnold committed
45
           'transformation_factory',
46
47
           'key_parser_name',
           'key_tag_name',
48
           'traverse',
Eckhart Arnold's avatar
Eckhart Arnold committed
49
50
           'replace_by_single_child',
           'reduce_single_child',
51
           'reduce_children',
52
           'replace_parser',
53
54
           'collapse',
           'map_content',
Eckhart Arnold's avatar
Eckhart Arnold committed
55
56
57
58
           'is_whitespace',
           'is_empty',
           'is_expendable',
           'is_token',
59
60
           'has_name',
           'has_content',
Eckhart Arnold's avatar
Eckhart Arnold committed
61
           'remove_children_if',
62
63
64
65
           'remove_children',
           'remove_content',
           'remove_first',
           'remove_last',
Eckhart Arnold's avatar
Eckhart Arnold committed
66
           'remove_whitespace',
Eckhart Arnold's avatar
Eckhart Arnold committed
67
           'remove_empty',
Eckhart Arnold's avatar
Eckhart Arnold committed
68
69
70
           'remove_expendables',
           'remove_tokens',
           'flatten',
71
72
73
           'forbid',
           'require',
           'assert_content']
Eckhart Arnold's avatar
Eckhart Arnold committed
74
75


Eckhart Arnold's avatar
Eckhart Arnold committed
76
77
78
79
80
81
82
83
84
85
86
class ParserBase:
    """
    ParserBase is the base class for all real and mock parser classes.
    It is defined here, because Node objects require a parser object
    for instantiation.
    """
    def __init__(self, name=''):  # , pbases=frozenset()):
        self.name = name  # type: str
        self._ptype = ':' + self.__class__.__name__  # type: str

    def __str__(self):
87
        return self.name or self.ptype
Eckhart Arnold's avatar
Eckhart Arnold committed
88
89
90
91
92

    @property
    def ptype(self) -> str:
        return self._ptype

93
94
95
    @property
    def repr(self) -> str:
        return self.name if self.name else repr(self)
Eckhart Arnold's avatar
Eckhart Arnold committed
96
97

class MockParser(ParserBase):
98
99
    """
    MockParser objects can be used to reconstruct syntax trees from a
Eckhart Arnold's avatar
Eckhart Arnold committed
100
101
    serialized form like S-expressions or XML. Mock objects can mimic
    different parser types by assigning them a ptype on initialization.
102
103
104
105
106
    
    Mock objects should not be used for anything other than 
    syntax tree (re-)construction. In all other cases where a parser
    object substitute is needed, chose the singleton ZOMBIE_PARSER.
    """
Eckhart Arnold's avatar
Eckhart Arnold committed
107
    def __init__(self, name='', ptype=''):  # , pbases=frozenset()):
108
        assert not ptype or ptype[0] == ':'
Eckhart Arnold's avatar
Eckhart Arnold committed
109
        super(MockParser, self).__init__(name)
110
        self.name = name
Eckhart Arnold's avatar
Eckhart Arnold committed
111
        self._ptype = ptype or ':' + self.__class__.__name__
112
113
114


class ZombieParser(MockParser):
115
116
117
118
119
120
121
122
123
124
125
126
    """
    Serves as a substitute for a Parser instance.

    ``ZombieParser`` is the class of the singelton object
    ``ZOMBIE_PARSER``. The  ``ZOMBIE_PARSER`` has a name and can be
    called, but it never matches. It serves as a substitute where only
    these (or one of these properties) is needed, but no real Parser-
    object is instantiated.
    """
    alive = False

    def __init__(self):
127
        super(ZombieParser, self).__init__("__ZOMBIE__")
128
129
130
131
        assert not self.__class__.alive, "There can be only one!"
        assert self.__class__ == ZombieParser, "No derivatives, please!"
        self.__class__.alive = True

Eckhart Arnold's avatar
Eckhart Arnold committed
132
133
134
135
136
137
    def __copy__(self):
        return self

    def __deepcopy__(self, memo):
        return self

138
139
140
141
    def __call__(self, text):
        """Better call Saul ;-)"""
        return None, text

142
143
144
145

ZOMBIE_PARSER = ZombieParser()


146
147
148
149
150
# # Python 3.6:
# class Error(NamedTuple):
#     pos: int
#     msg: str
Error = NamedTuple('Error', [('pos', int), ('msg', str)])
151

Eckhart Arnold's avatar
Eckhart Arnold committed
152
153
StrictResultType = Union[Tuple['Node', ...], str]
ResultType = Union[Tuple['Node', ...], 'Node', str, None]
154
155


156
157
158
159
160
161
162
163
164
165
166
167
class Node:
    """
    Represents a node in the concrete or abstract syntax tree.

    Attributes:
        tag_name (str):  The name of the node, which is either its
            parser's name or, if that is empty, the parser's class name
        result (str or tuple):  The result of the parser which
            generated this node, which can be either a string or a
            tuple of child nodes.
        children (tuple):  The tuple of child nodes or an empty tuple
            if there are no child nodes. READ ONLY!
168
169
170
171
172
173
        parser (Parser):  The parser which generated this node. 
            WARNING: In case you use mock syntax trees for testing or
            parser replacement during the AST-transformation: DO NOT
            rely on this being a real parser object in any phase after 
            parsing (i.e. AST-transformation and compiling), for 
            example by calling ``isinstance(node.parer, ...)``.
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
        errors (list):  A list of parser- or compiler-errors:
            tuple(position, string) attached to this node
        len (int):  The full length of the node's string result if the
            node is a leaf node or, otherwise, the concatenated string
            result's of its descendants. The figure always represents
            the length before AST-transformation ans will never change
            through AST-transformation. READ ONLY!
        pos (int):  the position of the node within the parsed text.

            The value of ``pos`` is -1 meaning invalid by default. 
            Setting this value will set the positions of all child
            nodes relative to this value.  

            To set the pos values of all nodes in a syntax tree, the
            pos value of the root node should be set to 0 right 
            after parsing.

            Other than that, this value should be considered READ ONLY. 
            At any rate, it should only be reassigned only during
            parsing stage and never during or after the
            AST-transformation.
    """
196

Eckhart Arnold's avatar
Eckhart Arnold committed
197
    def __init__(self, parser, result: ResultType) -> None:
198
199
200
        """Initializes the ``Node``-object with the ``Parser``-Instance
        that generated the node and the parser's result.
        """
Eckhart Arnold's avatar
Eckhart Arnold committed
201
        self._result = ''  # type: StrictResultType
202
        self._errors = []  # type: List[str]
203
        self._children = ()  # type: Tuple['Node', ...]
204
        self.result = result
205
206
207
208
        self._len = len(self.result) if not self.children else \
            sum(child._len for child in self.children)  # type: int
        # self.pos: int  = 0  # continuous updating of pos values
        self._pos = -1  # type: int
209
        self.parser = parser or ZOMBIE_PARSER
210
211
        self.error_flag = any(r.error_flag for r in self.children) \
            if self.children else False  # type: bool
212
213
214

    def __str__(self):
        if self.children:
215
            return "".join(str(child) for child in self.children)
216
217
218
219
220
221
222
223
        return str(self.result)

    def __repr__(self):
        mpargs = {'name': self.parser.name, 'ptype': self.parser.ptype}
        parg = "MockParser({name}, {ptype})".format(**mpargs)
        rarg = str(self) if not self.children else \
               "(" + ", ".join(repr(child) for child in self.children) + ")"
        return "Node(%s, %s)" % (parg, rarg)
224

225
    def __eq__(self, other):
226
227
        # return str(self.parser) == str(other.parser) and self.result == other.result
        return self.tag_name == other.tag_name and self.result == other.result
228

229
    def __hash__(self):
230
        return hash(self.tag_name)
231
232
233
234
235
236
237

    def __deepcopy__(self, memodict={}):
        result = copy.deepcopy(self.result)
        other = Node(self.parser, result)
        other._pos = self._pos
        return other

238
    @property   # this needs to be a (dynamic) property, in case sef.parser gets updated
239
    def tag_name(self) -> str:
240
        return self.parser.name or self.parser.ptype
241
242

    @property
Eckhart Arnold's avatar
Eckhart Arnold committed
243
    def result(self) -> StrictResultType:
244
245
246
        return self._result

    @result.setter
Eckhart Arnold's avatar
Eckhart Arnold committed
247
    def result(self, result: ResultType):
248
249
250
251
        # # made obsolete by static type checking with mypy is done
        # assert ((isinstance(result, tuple) and all(isinstance(child, Node) for child in result))
        #         or isinstance(result, Node)
        #         or isinstance(result, str)), str(result)
252
        self._result = (result,) if isinstance(result, Node) else result or ''
253
254
        self._children = cast(Tuple['Node', ...], self._result) \
            if isinstance(self._result, tuple) else cast(Tuple['Node', ...], ())
255
256

    @property
257
    def children(self) -> Tuple['Node', ...]:
258
259
260
        return self._children

    @property
261
    def len(self) -> int:
262
        # DEBUGGING:  print(self.tag_name, str(self.pos), str(self._len), str(self)[:10].replace('\n','.'))
263
264
265
        return self._len

    @property
266
    def pos(self) -> int:
267
268
269
270
        assert self._pos >= 0, "position value not initialized!"
        return self._pos

    @pos.setter
271
272
    def pos(self, pos: int):
        # assert isinstance(pos, int)
273
274
275
276
277
278
279
        self._pos = pos
        offset = 0
        for child in self.children:
            child.pos = pos + offset
            offset += child.len

    @property
280
    def errors(self) -> List[Error]:
281
282
        return [Error(self.pos, err) for err in self._errors]

283
284
    def show(self) -> str:
        """Returns content as string, inserting error messages where
285
        errors occurred.
286
        """
287
        s = "".join(child.show() for child in self.children) if self.children \
288
289
290
            else str(self.result)
        return (' <<< Error on "%s" | %s >>> ' % (s, '; '.join(self._errors))) if self._errors else s

291
    def _tree_repr(self, tab, openF, closeF, dataF=lambda s: s) -> str:
292
293
294
295
296
297
298
299
        """
        Generates a tree representation of this node and its children
        in string from.

        The kind ot tree-representation that is determined by several
        function parameters. This could be an XML-representation or a
        lisp-like S-expression.

300
        Args:
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
            tab (str):  The indentation string, e.g. '\t' or '    '
            openF:  (Node->str) A function that returns an opening
                string (e.g. an XML-tag_name) for a given node
            closeF:  (Node->str) A function that returns a closeF
                string (e.g. an XML-tag_name) for a given node.
            dataF:  (str->str) A function that filters the data string
                before printing, e.g. to add quotation marks

        Returns (str):
            A string that contains a (serialized) tree representation
            of the node and its children.
        """
        head = openF(self)
        tail = closeF(self)

        if not self.result:
            return head + tail

        head = head + '\n'  # place the head, tail and content
        tail = '\n' + tail  # of the node on different lines

        if self.children:
            content = []
324
            for child in self.children:
325
326
327
328
                subtree = child._tree_repr(tab, openF, closeF, dataF).split('\n')
                content.append('\n'.join((tab + s) for s in subtree))
            return head + '\n'.join(content) + tail

329
330
        res = cast(str, self.result)  # safe, because if there are no children, result is a string
        if head[0] == "<" and res.find('\n') < 0:
331
332
333
            # for XML: place tags for leaf-nodes on one line if possible
            return head[:-1] + self.result + tail[1:]
        else:
334
            return head + '\n'.join([tab + dataF(s) for s in res.split('\n')]) + tail
335

336
    def as_sexpr(self, src: str=None) -> str:
337
338
339
        """
        Returns content as S-expression, i.e. in lisp-like form.

340
        Args:
341
342
343
344
345
            src:  The source text or `None`. In case the source text is
                given the position of the element in the text will be
                reported as line and column.
        """

346
        def opening(node) -> str:
347
348
349
            s = '(' + node.tag_name
            # s += " '(pos %i)" % node.pos
            if src:
Eckhart Arnold's avatar
Eckhart Arnold committed
350
                s += " '(pos %i " % node.pos + " %i %i)" % line_col(src, node.pos)
351
352
353
354
355
356
357
358
359
360
            if node.errors:
                s += " '(err '(%s))" % ' '.join(str(err).replace('"', r'\"')
                                                for err in node.errors)
            return s

        def pretty(s):
            return '"%s"' % s if s.find('"') < 0 \
                else "'%s'" % s if s.find("'") < 0 \
                else '"%s"' % s.replace('"', r'\"')

361
        return self._tree_repr('    ', opening, lambda node: ')', pretty)
362

363
    def as_xml(self, src: str=None) -> str:
364
365
366
        """
        Returns content as XML-tree.

367
        Args:
368
369
370
371
372
            src:  The source text or `None`. In case the source text is
                given the position will also be reported as line and
                column.
        """

373
        def opening(node) -> str:
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
            s = '<' + node.tag_name
            # s += ' pos="%i"' % node.pos
            if src:
                s += ' line="%i" col="%i"' % line_col(src, node.pos)
            if node.errors:
                s += ' err="%s"' % ''.join(str(err).replace('"', r'\"') for err in node.errors)
            s += ">"
            return s

        def closing(node):
            s = '</' + node.tag_name + '>'
            return s

        return self._tree_repr('    ', opening, closing)

389
    def add_error(self, error_str) -> 'Node':
390
391
392
393
        self._errors.append(error_str)
        self.error_flag = True
        return self

394
395
    def propagate_error_flags(self) -> None:
        """Recursively propagates error flags set on child nodes to its
396
397
398
399
400
        parents. This can be used if errors are added to descendant 
        nodes after syntaxtree construction, i.e. in the compile phase.
        """
        for child in self.children:
            child.propagate_error_flags()
401
            self.error_flag = self.error_flag or child.error_flag
402

403
    def collect_errors(self, clear_errors=False) -> List[Error]:
404
405
406
407
408
        """
        Returns all errors of this node or any child node in the form
        of a set of tuples (position, error_message), where position
        is always relative to this node.
        """
409
410
411
412
413
        errors = self.errors
        if clear_errors:
            self._errors = []
            self.error_flag = False
        if self.children:
414
            for child in self.children:
415
                errors.extend(child.collect_errors(clear_errors))
416
        return errors
417

Eckhart Arnold's avatar
Eckhart Arnold committed
418
419
420
421
    def log(self, log_file_name):
        st_file_name = log_file_name
        with open(os.path.join(log_dir(), st_file_name), "w", encoding="utf-8") as f:
            f.write(self.as_sexpr())
422

423
    def find(self, match_function) -> Iterator['Node']:
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
        """Finds nodes in the tree that match a specific criterion.
        
        ``find`` is a generator that yields all nodes for which the
        given ``match_function`` evaluates to True. The tree is 
        traversed pre-order.
        
        Args:
            match_function (function): A function  that takes as Node
                object as argument and returns True or False
        Yields:
            Node: all nodes of the tree for which 
            ``match_function(node)`` returns True
        """
        if match_function(self):
            yield self
        else:
            for child in self.children:
                for nd in child.find(match_function):
                    yield nd

444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
    # def range(self, match_first, match_last):
    #     """Iterates over the range of nodes, starting from the first
    #     node for which ``match_first`` becomes True until the first node
    #     after this one for which ``match_last`` becomes true or until
    #     the end if it never does.
    #
    #     Args:
    #         match_first (function): A function  that takes as Node
    #             object as argument and returns True or False
    #         match_last (function): A function  that takes as Node
    #             object as argument and returns True or False
    #     Yields:
    #         Node: all nodes of the tree for which
    #         ``match_function(node)`` returns True
    #     """


461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
    # def navigate(self, path):
    #     """Yields the results of all descendant elements matched by
    #     ``path``, e.g.
    #     'd/s' yields 'l' from (d (s l)(e (r x1) (r x2))
    #     'e/r' yields 'x1', then 'x2'
    #     'e'   yields (r x1)(r x2)
    #
    #     Args:
    #         path (str):  The path of the object, e.g. 'a/b/c'. The
    #             components of ``path`` can be regular expressions
    #
    #     Returns:
    #         The object at the path, either a string or a Node or
    #         ``None``, if the path did not match.
    #     """
    #     def nav(node, pl):
    #         if pl:
    #             return itertools.chain(nav(child, pl[1:]) for child in node.children
    #                                    if re.match(pl[0], child.tag_name))
    #         else:
    #             return self.result,
    #     return nav(path.split('/'))
483
484
485
486
487
488
489
490
491


########################################################################
#
# syntax tree transformation functions
#
########################################################################


492
493
494
495
TransformationFunc = Union[Callable[[Node], Any], partial]


def transformation_factory(t=None):
Eckhart Arnold's avatar
Eckhart Arnold committed
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
    """Creates factory functions from transformation-functions that
    dispatch on the first parameter after the node parameter.

    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
    takes just the node-parameter.

    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.
511
512
513
514
515

    Usage:
        @transformation_factory(AbtractSet[str])
        def remove_tokens(node, tokens):
            ...
Eckhart Arnold's avatar
Eckhart Arnold committed
516
      or, alternatively:
517
518
519
        @transformation_factory
        def remove_tokens(node, tokens: AbstractSet[str]):
            ...
Eckhart Arnold's avatar
Eckhart Arnold committed
520
521
522
523
524

    Example:
        trans_table = { 'expression': remove_tokens('+', '-') }
      instead of:
        trans_table = { 'expression': partial(remove_tokens, tokens={'+', '-'}) }
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
    """

    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
        assert t or params[0].annotation != params[0].empty, \
            "No type information on second parameter found! Please, use type " \
            "annotation or provide the type information via transfomer-decorator."
        p1type = t or params[0].annotation
        f = singledispatch(f)
        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)

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

        f.register(p1type, gen_partial)
        return f

    if isinstance(t, type(lambda: 1)):
Eckhart Arnold's avatar
Eckhart Arnold committed
556
557
558
559
        # Provide for the case that transformation_factory has been
        # written as plain decorator and not as a function call that
        # returns the decorator proper.
        func = t;  t = None
560
561
562
        return decorator(func)
    else:
        return decorator
563
564


565
566
WHITESPACE_PTYPE = ':Whitespace'
TOKEN_PTYPE = ':Token'
567
568


569
def key_parser_name(node) -> str:
570
571
572
    return node.parser.name


573
def key_tag_name(node) -> str:
574
575
576
    return node.tag_name


577
def traverse(root_node, processing_table, key_func=key_tag_name) -> None:
578
    """Traverses the snytax tree starting with the given ``node`` depth
579
    first and applies the sequences of callback-functions registered
580
581
    in the ``calltable``-dictionary.
    
582
583
    The most important use case is the transformation of a concrete
    syntax tree into an abstract tree (AST). But it is also imaginable
584
    to employ tree-traversal for the semantic analysis of the AST.
585
586
587
588
589
590
591
592
593
594

    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)

595
    Args:
di68kap's avatar
di68kap committed
596
        root_node (Node): The root-node of the syntax tree to be traversed 
597
        processing_table (dict): node key -> sequence of functions that
di68kap's avatar
di68kap committed
598
599
            will be applied to matching nodes in order. This dictionary
            is interpreted as a ``compact_table``. See 
600
            ``toolkit.expand_table`` or ``EBNFCompiler.EBNFTransTable``
601
602
        key_func (function): A mapping key_func(node) -> keystr. The default
            key_func yields node.parser.name.
603
604
605
606
607
            
    Example:
        table = { "term": [replace_by_single_child, flatten], 
            "factor, flowmarker, retrieveop": replace_by_single_child }
        traverse(node, table)
608
    """
609
610
611
612
    # commented, because this approach is too error prone!
    # def funclist(call):
    #     return [as_partial(func) for func in smart_list(call)]

di68kap's avatar
di68kap committed
613
    # normalize processing_table entries by turning single values into lists
614
    # with a single value
615
    table = {name: smart_list(call) for name, call in list(processing_table.items())}
616
    table = expand_table(table)
617
    cache = {}  # type: Dict[str, List[Callable]]
618

di68kap's avatar
di68kap committed
619
620
621
    def traverse_recursive(node):
        if node.children:
            for child in node.result:
622
                traverse_recursive(child)            # depth first
623
                node.error_flag |= child.error_flag  # propagate error flag
624
625
626
627
628
629
630
631
632
633
634
635
636

        key = key_func(node)
        sequence = cache.get(key, None)
        if sequence is None:
            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

637
        for call in sequence:
di68kap's avatar
di68kap committed
638
            call(node)
639

di68kap's avatar
di68kap committed
640
    traverse_recursive(root_node)
641
642
643
644
645


# ------------------------------------------------
#
# rearranging transformations:
di68kap's avatar
di68kap committed
646
647
#     - tree may be rearranged (e.g.flattened)
#     - nodes that are not leaves may be dropped
648
649
650
651
652
653
#     - order is preserved
#     - all leaves are kept
#
# ------------------------------------------------


654
655
656
657
658
659
660
661
662
663
664
665
666
@transformation_factory
def replace_parser(node, name: str):
    """Replaces the parser of a Node with a mock parser with the given
    name.

    Parameters:
        name(str): "NAME:PTYPE" of the surogate. The ptype is optional
        node(Node): The node where the parser shall be replaced
    """
    name, ptype = (name.split(':') + [''])[:2]
    node.parser = MockParser(name, ptype)


667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
def replace_by_single_child(node):
    """Remove single branch node, replacing it by its immediate descendant.
    (In case the descendant's name is empty (i.e. anonymous) the
    name of this node's parser is kept.)
    """
    if node.children and len(node.result) == 1:
        if not node.result[0].parser.name:
            node.result[0].parser.name = node.parser.name
        node.parser = node.result[0].parser
        node._errors.extend(node.result[0].errors)
        node.result = node.result[0].result


def reduce_single_child(node):
    """Reduce a single branch node, by transferring the result of its
    immediate descendant to this node, but keeping this node's parser entry.
    """
    if node.children and len(node.result) == 1:
        node._errors.extend(node.result[0].errors)
        node.result = node.result[0].result


689
690
691
692
693
694
@transformation_factory(Callable)
def reduce_children(node, condition=lambda node: not node.name):
    """Replaces those children of node that have children themselves
    ans fulfil the given condition (default unnamed nodes).
    In contrast to ``flatten`` (see below) this transformation does not
    operate recursively.
di68kap's avatar
di68kap committed
695
    """
696
697
698
699
700
701
702
703
    if node.children:
        new_result = []
        for child in node.children:
            if child.children and condition(child):
                new_result.extend(child.children)
            else:
                new_result.append(child)
        node.result = tuple(new_result)
di68kap's avatar
di68kap committed
704
705


706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
def flatten(node):
    """Recursively flattens all unnamed sub-nodes, in case there is more
    than one sub-node present. Flattening means that
    wherever a node has child nodes, the child nodes are inserted in place
    of the node. In other words, all leaves of this node and its child nodes
    are collected in-order as direct children of this node.
    This is meant to achieve these kinds of structural transformation:
        (1 (+ 2) (+ 3)     ->   (1 + 2 + 3)
        (1 (+ (2 + (3))))  ->   (1 + 2 + 3)

    Warning: Use with care. Du tue its recursive nature, flattening can
    have unexpected side-effects.
    """
    if node.children:
        new_result = []
        for child in node.children:
            if not child.parser.name and child.children:
                flatten(child)
724
                new_result.extend(child.children)
725
726
727
728
729
730
731
732
733
734
735
736
            else:
                new_result.append(child)
        node.result = tuple(new_result)


def collapse(node):
    """Collapses all sub-nodes by replacing the node's result with it's
    string representation.
    """
    node.result = str(node)


737
738
739
740
741
742
743
744
745
746
747
748
749
750
# ------------------------------------------------
#
# destructive transformations:
#     - tree may be rearranged (flattened),
#     - order is preserved
#     - but (irrelevant) leaves may be dropped
#     - errors of dropped leaves will be lost
#
# ------------------------------------------------


def is_whitespace(node):
    """Removes whitespace and comments defined with the
    ``@comment``-directive."""
751
    return node.parser.ptype == WHITESPACE_PTYPE
752
753
754
755
756

def is_empty(node):
    return not node.result

def is_expendable(node):
757
    return is_empty(node) or is_whitespace(node)
758

759
def is_token(node, tokens: AbstractSet[str] = frozenset()) -> bool:
760
    return node.parser.ptype == TOKEN_PTYPE and (not tokens or node.result in tokens)
761
762


763
@transformation_factory
764
def has_name(node, tag_names: AbstractSet[str]) -> bool:
765
766
    """Checks if node has any of a given set of `tag names`.
    See property `Node.tagname`."""
767
768
769
    return node.tag_name in tag_names


770
@transformation_factory
771
def has_content(node, contents: AbstractSet[str]) -> bool:
772
773
    """Checks if the node's content (i.e. `str(node)`) matches any of
    a given set of strings."""
774
775
776
    return str(node) in contents


777
@transformation_factory(Callable)  # @singledispatch
778
def remove_children_if(node, condition):
779
780
781
782
783
784
    """Removes all nodes from the result field if the function 
    ``condition(child_node)`` evaluates to ``True``."""
    if node.children:
        node.result = tuple(c for c in node.children if not condition(c))


785
remove_whitespace = remove_children_if(is_whitespace)  # partial(remove_children_if, condition=is_whitespace)
Eckhart Arnold's avatar
Eckhart Arnold committed
786
remove_empty = remove_children_if(is_empty)
787
remove_expendables = remove_children_if(is_expendable)  # partial(remove_children_if, condition=is_expendable)
788
789


790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
@transformation_factory(Callable)
def remove_first(node, condition=lambda node: True):
    """Removes the first child if the condition is met.
    Otherwise does nothing."""
    if node.children:
        if condition(node.children[0]):
            node.result = node.result[1:]


@transformation_factory(Callable)
def remove_last(node, condition=lambda node: True):
    """Removes the last child if the condition is met.
    Otherwise does nothing."""
    if node.children:
        if condition(node.children[-1]):
            node.result = node.result[:-1]


808
809
@transformation_factory
def remove_tokens(node, tokens: AbstractSet[str] = frozenset()):
810
811
    """Reomoves any among a particular set of tokens from the immediate
    descendants of a node. If ``tokens`` is the empty set, all tokens
812
    are removed."""
813
    remove_children_if(node, partial(is_token, tokens=tokens))
814
815


816
@transformation_factory
817
def remove_children(node, tag_names: AbstractSet[str]):
818
819
820
821
822
823
824
825
    """Removes children by 'tag name'."""
    remove_children_if(node, partial(has_name, tag_names=tag_names))


@transformation_factory
def remove_content(node, contents: AbstractSet[str]):
    """Removes children depending on their string value."""
    remove_children_if(node, partial(has_content, contents=contents))
826
827


Eckhart Arnold's avatar
Eckhart Arnold committed
828
829
@transformation_factory
def map_content(node, func: Callable):      # Callable[[Node], ResultType]
830
831
832
833
834
835
    """Replaces the content of the node. ``func`` takes the node
    as an argument an returns the mapped result.
    """
    node.result = func(node.result)


836
837
########################################################################
#
838
# AST semantic validation functions
di68kap's avatar
di68kap committed
839
# EXPERIMENTAL!
840
841
842
843
#
########################################################################


844
845
@transformation_factory
def require(node, child_tags: AbstractSet[str]):
846
    for child in node.children:
847
        if child.tag_name not in child_tags:
848
            node.add_error('Element "%s" is not allowed inside "%s".' %
849
                           (child.parser.name, node.parser.name))
850
851


852
853
@transformation_factory
def forbid(node, child_tags: AbstractSet[str]):
854
    for child in node.children:
855
        if child.tag_name in child_tags:
856
            node.add_error('Element "%s" cannot be nested inside "%s".' %
857
                           (child.parser.name, node.parser.name))
858
859


860
861
@transformation_factory
def assert_content(node, regex: str):
862
863
864
    content = str(node)
    if not re.match(regex, content):
        node.add_error('Element "%s" violates %s on %s' %
865
                       (node.parser.name, str(regex), content))
866