syntaxtree.py 22.8 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 itertools
22
23
import os
from functools import partial
24
25
26
27
try:
    import regex as re
except ImportError:
    import re
28
29
from typing import NamedTuple

Eckhart Arnold's avatar
Eckhart Arnold committed
30
from .toolkit import is_logging, log_dir, expand_table, line_col, smart_list
31

Eckhart Arnold's avatar
Eckhart Arnold committed
32
33
34
35
36
37

__all__ = ['WHITESPACE_KEYWORD',
           'TOKEN_KEYWORD',
           'ZOMBIE_PARSER',
           'Error',
           'Node',
38
           'mock_syntax_tree',
39
           'traverse',
di68kap's avatar
di68kap committed
40
           'no_operation',
Eckhart Arnold's avatar
Eckhart Arnold committed
41
42
           'replace_by_single_child',
           'reduce_single_child',
di68kap's avatar
di68kap committed
43
           'change_parser',
Eckhart Arnold's avatar
Eckhart Arnold committed
44
45
46
47
48
49
50
51
52
           'is_whitespace',
           'is_empty',
           'is_expendable',
           'is_token',
           'remove_children_if',
           'remove_whitespace',
           'remove_expendables',
           'remove_tokens',
           'flatten',
53
54
55
56
           'remove_enclosing_delimiters',
           'forbid',
           'require',
           'assert_content']
Eckhart Arnold's avatar
Eckhart Arnold committed
57
58


59
class MockParser:
60
61
62
63
64
65
66
67
68
69
    """
    MockParser objects can be used to reconstruct syntax trees from a
    serialized form like S-expressions or XML. Mock objects are needed,
    because Node objects require a parser object for instantiation.
    Mock objects have just enough properties to serve that purpose. 
    
    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.
    """
70
    def __init__(self, name='', class_name=''):
71
        self.name = name
72
        self.class_name = class_name or self.__class__.__name__
73
74

    def __str__(self):
75
        return self.name or self.class_name
76
77
78


class ZombieParser(MockParser):
79
80
81
82
83
84
85
86
87
88
89
90
    """
    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):
91
        super(ZombieParser, self).__init__("ZOMBIE")
92
93
94
95
        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
96
97
98
99
100
101
    def __copy__(self):
        return self

    def __deepcopy__(self, memo):
        return self

102
103
104
105
    def __call__(self, text):
        """Better call Saul ;-)"""
        return None, text

106
107
108
109

ZOMBIE_PARSER = ZombieParser()


110
111
112
113
114
# # Python 3.6:
# class Error(NamedTuple):
#     pos: int
#     msg: str
Error = NamedTuple('Error', [('pos', int), ('msg', str)])
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
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!
        parser (Parser):  The parser which generated this node.
        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.
    """

    def __init__(self, parser, result):
        """Initializes the ``Node``-object with the ``Parser``-Instance
        that generated the node and the parser's result.
        """
        self.result = result
        self.parser = parser or ZOMBIE_PARSER
        self._errors = []
        self.error_flag = any(r.error_flag for r in self.result) if self.children else False
        self._len = len(self.result) if not self.children else \
            sum(child._len for child in self.children)
        # self.pos = 0  # coninuous updating of pos values
        self._pos = -1

    def __str__(self):
        if self.children:
168
            return "".join(str(child) for child in self.result)
169
170
        return str(self.result)

171
    def __eq__(self, other):
172
173
        # return str(self.parser) == str(other.parser) and self.result == other.result
        return self.tag_name == other.tag_name and self.result == other.result
174

175
    def __hash__(self):
176
        return hash(str(self.parser))
177
178
179
180
181
182
183

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

184
185
    @property
    def tag_name(self):
186
        return str(self.parser)
187
        # ONLY FOR DEBUGGING: return self.parser.name + ':' + self.parser.__class__.__name__
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216

    @property
    def result(self):
        return self._result

    @result.setter
    def result(self, result):
        assert ((isinstance(result, tuple) and all(isinstance(child, Node) for child in result))
                or isinstance(result, Node)
                or isinstance(result, str)), str(result)
        self._result = (result,) if isinstance(result, Node) else result or ''
        self._children = self._result if isinstance(self._result, tuple) else ()

    @property
    def children(self):
        return self._children

    @property
    def len(self):
        # DEBUGGING:  print(str(self.parser), str(self.pos), str(self._len), str(self)[:10].replace('\n','.'))
        return self._len

    @property
    def pos(self):
        assert self._pos >= 0, "position value not initialized!"
        return self._pos

    @pos.setter
    def pos(self, pos):
217
        assert isinstance(pos, int)
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
        self._pos = pos
        offset = 0
        for child in self.children:
            child.pos = pos + offset
            offset += child.len

    @property
    def errors(self):
        return [Error(self.pos, err) for err in self._errors]

    def _tree_repr(self, tab, openF, closeF, dataF=lambda s: s):
        """
        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.

237
        Args:
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
            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 = []
            for child in self.result:
                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

        return head + '\n'.join([tab + dataF(s)
                                 for s in str(self.result).split('\n')]) + tail

269
    def as_sexpr(self, src=None):
270
271
272
        """
        Returns content as S-expression, i.e. in lisp-like form.

273
        Args:
274
275
276
            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.
277
278
            prettyprint(bool):  True (default), if pretty printing 
                of leaf nodes shall be applied for better readability.
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
        """

        def opening(node):
            s = '(' + node.tag_name
            # s += " '(pos %i)" % node.pos
            if src:
                s += " '(pos %i  %i %i)" % (node.pos, *line_col(src, node.pos))
            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'\"')

296
        return self._tree_repr('    ', opening, lambda node: ')', pretty)  # pretty if prettyprint else lambda s: s)
297
298
299
300
301

    def as_xml(self, src=None):
        """
        Returns content as XML-tree.

302
        Args:
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
            src:  The source text or `None`. In case the source text is
                given the position will also be reported as line and
                column.
        """

        def opening(node):
            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)

    def add_error(self, error_str):
        self._errors.append(error_str)
        self.error_flag = True
        return self

    def collect_errors(self, clear_errors=False):
        """
        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.
        """
335
        errors = []
336
337
338
339
340
341
342
343
        if self.error_flag:
            errors = self.errors
            if clear_errors:
                self._errors = []
                self.error_flag = False
            if self.children:
                for child in self.result:
                    errors.extend(child.collect_errors(clear_errors))
344
        return errors
345

Eckhart Arnold's avatar
Eckhart Arnold committed
346
347
348
349
    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())
350

351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
    def find(self, match_function):
        """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

372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
    # 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
    #     """


389
    def navigate(self, path):
390
391
392
393
394
        """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)
395

396
        Args:
397
398
            path (str):  The path of the object, e.g. 'a/b/c'. The
                components of ``path`` can be regular expressions
399
400
401
402
403

        Returns:
            The object at the path, either a string or a Node or
            ``None``, if the path did not match.
        """
404
405
406
407
        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))
408
            else:
409
410
                return self.result,
        return nav(path.split('/'))
411
412


413
414
415
416
417
418
419
def mock_syntax_tree(sexpr):
    """Generates a tree of nodes from an S-expression.

    Example: 
    >>> mock_syntax_tree("(a (b c))").as_sexpr()
    (a 
        (b 
420
            "c" 
421
422
423
424
425
426
        )
    )
    """
    def next_block(s):
        s = s.strip()
        while s[0] != ')':
427
428
            if s[0] != '(': raise ValueError('"(" expected, not ' + s[:10])
            # assert s[0] == '(', s
429
430
431
432
433
434
435
436
437
438
439
440
            level = 1;
            i = 1
            while level > 0:
                if s[i] == '(':
                    level += 1
                elif s[i] == ')':
                    level -= 1
                i += 1
            yield s[:i]
            s = s[i:].strip()

    sexpr = sexpr.strip()
441
442
    if sexpr[0] != '(': raise ValueError('"(" expected, not ' + sexpr[:10])
    # assert sexpr[0] == '(', sexpr
443
    sexpr = sexpr[1:].strip()
444
445
    m = re.match('[\w:]+', sexpr)
    name, class_name = (sexpr[:m.end()].split(':') + [''])[:2]
446
447
448
449
    sexpr = sexpr[m.end():].strip()
    if sexpr[0] == '(':
        result = tuple(mock_syntax_tree(block) for block in next_block(sexpr))
    else:
450
451
452
453
454
455
456
457
458
459
460
461
462
463
        lines = []
        while sexpr and sexpr[0] != ')':
            for qm in ['"""', "'''", '"', "'"]:
                m = re.match(qm + r'.*?' + qm, sexpr)
                if m:
                    i = len(qm)
                    lines.append(sexpr[i:m.end() - i])
                    sexpr = sexpr[m.end():].strip()
                    break
            else:
                m = re.match(r'(?:(?!\)).)*', sexpr)
                lines.append(sexpr[:m.end()])
                sexpr = sexpr[m.end():]
        result = "\n".join(lines)
464
    return Node(MockParser(name, class_name), result)
465
466


467
468
469
470
471
472
473
########################################################################
#
# syntax tree transformation functions
#
########################################################################


474
475
476
477
WHITESPACE_KEYWORD = 'WSP__'
TOKEN_KEYWORD = 'TOKEN__'


di68kap's avatar
di68kap committed
478
def traverse(root_node, processing_table):
479
480
481
482
483
484
485
486
    """Traverses the snytax tree starting with the given ``node`` depth
    first and applies the sequences of callback functions registered
    in the ``calltable``-dictionary.
    
    Possible use cases are the transformation of a concrete syntax tree
    into an abstract tree (AST) or the semantic analysis of the AST.
    
    Args:
di68kap's avatar
di68kap committed
487
488
489
490
        root_node (Node): The root-node of the syntax tree to be traversed 
        processing_table (dict): parser.name -> sequence of functions that
            will be applied to matching nodes in order. This dictionary
            is interpreted as a ``compact_table``. See 
491
492
493
494
495
496
            ``toolkit.expand_table`` or ``EBNFCompiler.EBNFTransTable``
            
    Example:
        table = { "term": [replace_by_single_child, flatten], 
            "factor, flowmarker, retrieveop": replace_by_single_child }
        traverse(node, table)
497
    """
di68kap's avatar
di68kap committed
498
    # normalize processing_table entries by turning single values into lists
499
    # with a single value
500
    table = {name: smart_list(call) for name, call in list(processing_table.items())}
501
502
    table = expand_table(table)

di68kap's avatar
di68kap committed
503
504
505
    def traverse_recursive(node):
        if node.children:
            for child in node.result:
506
                traverse_recursive(child)
507
                node.error_flag |= child.error_flag  # propagate error flag
508
509
510
511
512
513
        sequence = table.get('*', []) + \
                   table.get(node.parser.name, table.get('?', [])) + \
                   table.get('~', [])
        # '*' always called (before any other processing function)
        # '?' called for those nodes for which no (other) processing functions is in the table
        # '~' always called (after any other processing function)
514
        for call in sequence:
di68kap's avatar
di68kap committed
515
            call(node)
516

di68kap's avatar
di68kap committed
517
    traverse_recursive(root_node)
518
519


di68kap's avatar
di68kap committed
520
def no_operation(node):
521
522
523
524
525
526
    pass


# ------------------------------------------------
#
# rearranging transformations:
di68kap's avatar
di68kap committed
527
528
#     - tree may be rearranged (e.g.flattened)
#     - nodes that are not leaves may be dropped
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
556
#     - order is preserved
#     - all leaves are kept
#
# ------------------------------------------------


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


di68kap's avatar
di68kap committed
557
558
559
560
561
562
563
def change_parser(node, new_parser_name):
    """Changes the parser of a Node to a mock parser with the given 
    name.
    """
    node.parser = MockParser(new_parser_name)


564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
# ------------------------------------------------
#
# 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."""
    return node.parser.name == WHITESPACE_KEYWORD


def is_empty(node):
    return not node.result


def is_expendable(node):
586
    return is_empty(node) or is_whitespace(node)
587
588


Eckhart Arnold's avatar
Eckhart Arnold committed
589
def is_token(node, token_set=frozenset()):
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
    return node.parser.name == TOKEN_KEYWORD and (not token_set or node.result in token_set)


def remove_children_if(node, condition):
    """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))


remove_whitespace = partial(remove_children_if, condition=is_whitespace)
# remove_scanner_tokens = partial(remove_children_if, condition=is_scanner_token)
remove_expendables = partial(remove_children_if, condition=is_expendable)


Eckhart Arnold's avatar
Eckhart Arnold committed
605
def remove_tokens(node, tokens=frozenset()):
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
    """Reomoves any among a particular set of tokens from the immediate
    descendants of a node. If ``tokens`` is the empty set, all tokens
    are removed.
    """
    remove_children_if(node, partial(is_token, token_set=tokens))


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:
                assert child.children, node.as_sexpr()
                flatten(child)
                new_result.extend(child.result)
            else:
                new_result.append(child)
        node.result = tuple(new_result)


638
def remove_enclosing_delimiters(node):
639
640
641
642
643
644
645
646
    """Removes any enclosing delimiters from a structure (e.g. quotation marks
    from a literal or braces from a group).
    """
    if len(node.children) >= 3:
        assert not node.children[0].children and not node.children[-1].children, node.as_sexpr()
        node.result = node.result[1:-1]


647
648
649
########################################################################
#
# syntax tree validation functions
di68kap's avatar
di68kap committed
650
# EXPERIMENTAL!
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
#
########################################################################


def require(node, child_tags):
    for child in node.children:
        if child.tag_name not in child_tags:
            node.add_error('Element "%s" is not allowed inside "%s".' %
                           (child.tag_name, node.tag_name))


def forbid(node, child_tags):
    for child in node.children:
        if child.tag_name in child_tags:
            node.add_error('Element "%s" cannot be nested inside "%s".' %
                           (child.tag_name, node.tag_name))


def assert_content(node, regex):
    content = str(node)
    if not re.match(regex, content):
        node.add_error('Element "%s" violates %s on %s' %
                       (node.tag_name, str(regex), content))