10.12., 9:00 - 11:00: Due to updates GitLab may be unavailable for some minutes between 09:00 and 11:00.

syntaxtree.py 17.7 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
#!/usr/bin/python3

"""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.
"""

22
import itertools
23 24
import os
from functools import partial
25 26 27 28
try:
    import regex as re
except ImportError:
    import re
29 30
from typing import NamedTuple

31
from .toolkit import IS_LOGGING, LOGS_DIR, expand_table, line_col, sequence
32

Eckhart Arnold's avatar
Eckhart Arnold committed
33 34 35 36 37 38

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


56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140
class ZombieParser:
    """
    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):
        assert not self.__class__.alive, "There can be only one!"
        assert self.__class__ == ZombieParser, "No derivatives, please!"
        self.name = "ZOMBIE"
        self.__class__.alive = True

    def __str__(self):
        return self.name

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


ZOMBIE_PARSER = ZombieParser()


class Error(NamedTuple):
    pos: int
    msg: str


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:
141
            return "".join(str(child) for child in self.result)
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194
        return str(self.result)

    @property
    def tag_name(self):
        return self.parser.name or self.parser.__class__.__name__

    @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):
        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.

195
        Args:
196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226
            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

227
    def as_sexpr(self, src=None, prettyprint=True):
228 229 230
        """
        Returns content as S-expression, i.e. in lisp-like form.

231
        Args:
232 233 234
            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.
235 236
            prettyprint(bool):  True (default), if pretty printing 
                of leaf nodes shall be applied for better readability.
237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253
        """

        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'\"')

254 255
        return self._tree_repr('    ', opening, lambda node: ')',
                               pretty if prettyprint else lambda s: s)
256 257 258 259 260

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

261
        Args:
262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305
            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.
        """
        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))
            return errors
        return []

    def log(self, log_file_name, ext):
306
        if IS_LOGGING():
307 308 309 310
            st_file_name = log_file_name + ext
            with open(os.path.join(LOGS_DIR(), st_file_name), "w", encoding="utf-8") as f:
                f.write(self.as_sexpr())

311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332
    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

333
    def navigate(self, path):
334 335 336 337 338
        """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)
339

340
        Args:
341 342
            path (str):  The path of the object, e.g. 'a/b/c'. The
                components of ``path`` can be regular expressions
343 344 345 346 347

        Returns:
            The object at the path, either a string or a Node or
            ``None``, if the path did not match.
        """
348 349 350 351
        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))
352
            else:
353 354
                return self.result,
        return nav(path.split('/'))
355 356 357 358 359 360 361 362 363


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


364 365 366 367
WHITESPACE_KEYWORD = 'WSP__'
TOKEN_KEYWORD = 'TOKEN__'


di68kap's avatar
di68kap committed
368
def traverse(root_node, processing_table):
369 370 371 372 373 374 375 376
    """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
377 378 379 380
        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 
381 382 383 384 385 386
            ``toolkit.expand_table`` or ``EBNFCompiler.EBNFTransTable``
            
    Example:
        table = { "term": [replace_by_single_child, flatten], 
            "factor, flowmarker, retrieveop": replace_by_single_child }
        traverse(node, table)
387
    """
di68kap's avatar
di68kap committed
388
    # normalize processing_table entries by turning single values into lists
389
    # with a single value
di68kap's avatar
di68kap committed
390
    table = {name: sequence(call) for name, call in list(processing_table.items())}
391 392
    table = expand_table(table)

di68kap's avatar
di68kap committed
393 394 395
    def traverse_recursive(node):
        if node.children:
            for child in node.result:
396
                traverse_recursive(child)
di68kap's avatar
di68kap committed
397 398
        sequence = table.get(node.parser.name,
                             table.get('~', [])) + table.get('*', [])
399
        for call in sequence:
di68kap's avatar
di68kap committed
400
            call(node)
401

di68kap's avatar
di68kap committed
402
    traverse_recursive(root_node)
403 404


di68kap's avatar
di68kap committed
405
def no_operation(node):
406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469
    pass


# ------------------------------------------------
#
# rearranging transformations:
#     - tree may be rearranged (flattened)
#     - 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


# ------------------------------------------------
#
# 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_scanner_token(node):
#     return isinstance(node.parser, ScannerToken)


def is_empty(node):
    return not node.result


def is_expendable(node):
    return is_empty(node) or is_whitespace(node)  # or is_scanner_token(node)


Eckhart Arnold's avatar
Eckhart Arnold committed
470
def is_token(node, token_set=frozenset()):
471 472 473 474 475 476 477 478 479 480 481 482 483 484 485
    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
486
def remove_tokens(node, tokens=frozenset()):
487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518
    """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)


519
def remove_enclosing_delimiters(node):
520 521 522 523 524 525 526 527
    """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]