syntaxtree.py 20.6 KB
Newer Older
1
"""syntaxtree.py - syntax tree classes for DHParser
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

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.
"""

Eckhart Arnold's avatar
Eckhart Arnold committed
19
import collections.abc
20
import copy
21
import os
22 23
from functools import partial

24
from DHParser.error import Error, linebreaks, line_col
25
from DHParser.stringview import StringView
di68kap's avatar
di68kap committed
26 27 28
from DHParser.toolkit import is_logging, log_dir, identity, re, typing
from typing import Any, Callable, cast, Iterator, List, Union, Tuple, Hashable, Optional

29

30 31 32 33 34 35 36
__all__ = ('ParserBase',
           'WHITESPACE_PTYPE',
           'TOKEN_PTYPE',
           'MockParser',
           'ZombieParser',
           'ZOMBIE_PARSER',
           'Node',
37 38
           'mock_syntax_tree',
           'TransformationFunc')
Eckhart Arnold's avatar
Eckhart Arnold committed
39 40


41 42 43 44 45 46 47 48 49 50 51 52 53 54
#######################################################################
#
# parser base and mock parsers
#
#######################################################################


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()):
55
        self._name = name  # type: str
56 57 58 59 60 61 62 63
        self._ptype = ':' + self.__class__.__name__  # type: str

    def __repr__(self):
        return self.name + self.ptype

    def __str__(self):
        return self.name + (' = ' if self.name else '') + repr(self)

64 65 66
    def __call__(self, text: StringView) -> Tuple[Optional['Node'], StringView]:
        return None, text

67 68 69 70
    @property
    def name(self):
        return self._name

71 72 73 74 75 76 77 78
    @property
    def ptype(self) -> str:
        return self._ptype

    @property
    def repr(self) -> str:
        return self.name if self.name else repr(self)

79 80 81 82 83 84 85 86 87
    def reset(self):
        pass

    def grammar(self) -> 'Grammar':
        return None

    def apply(self, func: Callable) -> bool:
        return False

88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104

WHITESPACE_PTYPE = ':Whitespace'
TOKEN_PTYPE = ':Token'


class MockParser(ParserBase):
    """
    MockParser objects can be used to reconstruct syntax trees from a
    serialized form like S-expressions or XML. Mock objects can mimic
    different parser types by assigning them a ptype on initialization.

    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.
    """
    def __init__(self, name='', ptype=''):  # , pbases=frozenset()):
        assert not ptype or ptype[0] == ':'
105
        super().__init__(name)
106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147
        self._ptype = ptype or ':' + self.__class__.__name__


class ZombieParser(MockParser):
    """
    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):
        super(ZombieParser, self).__init__("__ZOMBIE__")
        assert not self.__class__.alive, "There can be only one!"
        assert self.__class__ == ZombieParser, "No derivatives, please!"
        self.__class__.alive = True

    def __copy__(self):
        return self

    def __deepcopy__(self, memo):
        return self

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


ZOMBIE_PARSER = ZombieParser()


#######################################################################
#
# syntaxtree nodes
#
#######################################################################


148
ChildrenType = Tuple['Node', ...]
Eckhart Arnold's avatar
Eckhart Arnold committed
149
NoChildren = cast(ChildrenType, ())  # type: ChildrenType
150 151
StrictResultType = Union[ChildrenType, StringView, str]
ResultType = Union[ChildrenType, 'Node', StringView, str, None]
152 153


Eckhart Arnold's avatar
Eckhart Arnold committed
154 155
def flatten_sxpr(sxpr: str) -> str:
    """Returns S-expression `sxpr` as a one-liner without unnecessary
Eckhart Arnold's avatar
Eckhart Arnold committed
156 157 158
    whitespace.

    Example:
Eckhart Arnold's avatar
Eckhart Arnold committed
159
    >>> flatten_sxpr('(a\\n    (b\\n        c\\n    )\\n)\\n')
Eckhart Arnold's avatar
Eckhart Arnold committed
160 161 162 163 164
    '(a (b c))'
    """
    return re.sub('\s(?=\))', '', re.sub('\s+', ' ', sxpr)).strip()


Eckhart Arnold's avatar
Eckhart Arnold committed
165
class Node(collections.abc.Sized):
166 167 168
    """
    Represents a node in the concrete or abstract syntax tree.

169
    Attributes and Properties:
170 171 172 173 174 175 176
        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!
Eckhart Arnold's avatar
Eckhart Arnold committed
177
        parser (Parser):  The parser which generated this node.
178 179
            WARNING: In case you use mock syntax trees for testing or
            parser replacement during the AST-transformation: DO NOT
Eckhart Arnold's avatar
Eckhart Arnold committed
180 181
            rely on this being a real parser object in any phase after
            parsing (i.e. AST-transformation and compiling), for
182
            example by calling ``isinstance(node.parer, ...)``.
183 184
        errors (list):  A list of parser- or compiler-errors:
            tuple(position, string) attached to this node
185 186 187
        error_flag (int):  0 if no error occurred in either the node
            itself or any of its descendants. Otherwise contains the
            highest warning or error level or all errors that occurred.
188 189 190 191 192 193 194
        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.

Eckhart Arnold's avatar
Eckhart Arnold committed
195
            The value of ``pos`` is -1 meaning invalid by default.
196
            Setting this value will set the positions of all child
Eckhart Arnold's avatar
Eckhart Arnold committed
197
            nodes relative to this value.
198 199

            To set the pos values of all nodes in a syntax tree, the
Eckhart Arnold's avatar
Eckhart Arnold committed
200
            pos value of the root node should be set to 0 right
201 202
            after parsing.

Eckhart Arnold's avatar
Eckhart Arnold committed
203
            Other than that, this value should be considered READ ONLY.
204 205 206 207
            At any rate, it should only be reassigned only during
            parsing stage and never during or after the
            AST-transformation.
    """
208

209 210
    __slots__ = ['_result', 'children', '_errors', '_len', '_pos', 'parser', 'error_flag']

211

212
    def __init__(self, parser, result: ResultType, leafhint: bool=False) -> None:
213 214 215
        """Initializes the ``Node``-object with the ``Parser``-Instance
        that generated the node and the parser's result.
        """
216 217
        self.error_flag = 0             # type: int
        self._errors = []               # type: List[Error]
di68kap's avatar
di68kap committed
218 219
        # Assignment to self.result initializes the attributes _result, children and _len
        # The following if-clause is merely an optimization, i.e. a fast-path for leaf-Nodes
220 221 222
        if leafhint:
            self._result = result       # type: StrictResultType
            self.children = NoChildren  # type: ChildrenType
223
            self._len = -1              # type: int  # lazy evaluation
224 225
        else:
            self.result = result
Eckhart Arnold's avatar
Eckhart Arnold committed
226
        # self.pos: int  = 0  # continuous updating of pos values wastes a lot of time
227
        self._pos = -1  # type: int
228 229
        self.parser = parser or ZOMBIE_PARSER

Eckhart Arnold's avatar
Eckhart Arnold committed
230

231 232
    def __str__(self):
        if self.children:
233
            return "".join(str(child) for child in self.children)
234 235 236
        elif isinstance(self.result, StringView):
            self.result = str(self.result)
        return self.result
237

Eckhart Arnold's avatar
Eckhart Arnold committed
238

239 240 241 242 243 244
    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)
245

Eckhart Arnold's avatar
Eckhart Arnold committed
246

247
    def __len__(self):
248 249 250
        if self._len < 0:
            self._len = sum(child._len for child in self.children) \
                if self.children else len(self._result)
251 252 253
        return self._len


254

255 256 257 258 259
    def __bool__(self):
        # A node that is not None is always True, even if it's empty
        return True


260
    def __eq__(self, other):
261 262
        # return str(self.parser) == str(other.parser) and self.result == other.result
        return self.tag_name == other.tag_name and self.result == other.result
263

Eckhart Arnold's avatar
Eckhart Arnold committed
264

265
    def __hash__(self):
266
        return hash(self.tag_name)
267

Eckhart Arnold's avatar
Eckhart Arnold committed
268

269 270 271 272 273 274
    def __deepcopy__(self, memodict={}):
        result = copy.deepcopy(self.result)
        other = Node(self.parser, result)
        other._pos = self._pos
        return other

Eckhart Arnold's avatar
Eckhart Arnold committed
275

276
    @property   # this needs to be a (dynamic) property, in case sef.parser gets updated
277
    def tag_name(self) -> str:
278
        return self.parser.name or self.parser.ptype
279

Eckhart Arnold's avatar
Eckhart Arnold committed
280

281
    @property
Eckhart Arnold's avatar
Eckhart Arnold committed
282
    def result(self) -> StrictResultType:
283 284 285
        return self._result

    @result.setter
Eckhart Arnold's avatar
Eckhart Arnold committed
286
    def result(self, result: ResultType):
Eckhart Arnold's avatar
Eckhart Arnold committed
287
        # # made obsolete by static type checking with mypy
288 289 290
        # assert ((isinstance(result, tuple) and all(isinstance(child, Node) for child in result))
        #         or isinstance(result, Node)
        #         or isinstance(result, str)), str(result)
Eckhart Arnold's avatar
Eckhart Arnold committed
291 292
        # Possible optimization: Do not allow single nodes as argument:
        # assert not isinstance(result, Node)
293
        self._len = -1  # lazy evaluation
294 295 296 297 298 299 300 301
        if isinstance(result, Node):
            self.children = (result,)
            self._result = self.children
            self.error_flag = result.error_flag
        else:
            if isinstance(result, tuple):
                self.children = result
                self._result = result or ''
302 303 304
                if result:
                    if self.error_flag == 0:
                        self.error_flag = max(child.error_flag for child in self.children)
305 306 307 308 309 310 311 312 313 314 315
            else:
                self.children = NoChildren
                self._result = result
        # # shorter but slower:
        # self._result = (result,) if isinstance(result, Node) else result or ''  # type: StrictResultType
        # self.children = cast(ChildrenType, self._result) \
        #     if isinstance(self._result, tuple) else NoChildren  # type: ChildrenType
        # if self.children:
        #     self.error_flag = max(self.error_flag,
        #                           max(child.error_flag for child in self.children))  # type: bool

316
    @property
317
    def pos(self) -> int:
318 319 320 321
        assert self._pos >= 0, "position value not initialized!"
        return self._pos

    @pos.setter
322
    def pos(self, pos: int):
323 324
        self._pos = pos
        offset = 0
325
        # recursively adjust pos-values of all children
326 327
        for child in self.children:
            child.pos = pos + offset
328
            offset += len(child)
329 330 331
        # add pos-values to Error-objects
        for err in self._errors:
            err.pos = pos
332

Eckhart Arnold's avatar
Eckhart Arnold committed
333

334
    @property
335
    def errors(self) -> List[Error]:
336 337
        return self._errors.copy()

338

339
    def add_error(self, message: str, level: int = Error.ERROR, code: Hashable = 0) -> 'Node':
Eckhart Arnold's avatar
Eckhart Arnold committed
340
        self._errors.append(Error(message, level, code))
341
        self.error_flag = max(self.error_flag, self._errors[-1].level)
Eckhart Arnold's avatar
Eckhart Arnold committed
342 343
        return self

344

345 346
    def collect_errors(self, document: Union[StringView, str] = '', clear_errors=False) \
            -> List[Error]:
Eckhart Arnold's avatar
Eckhart Arnold committed
347
        """
348
        Recursively adds line- and column-numbers to all error objects.
Eckhart Arnold's avatar
Eckhart Arnold committed
349 350 351 352
        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.
        """
353 354 355 356 357 358
        if self.error_flag:
            lbreaks = linebreaks(document) if document else []
            return self._collect_errors(lbreaks, clear_errors)
        else:
            return []

359
    def _collect_errors(self, lbreaks: List[int]=[], clear_errors=False) -> List[Error]:
360 361 362 363 364 365 366 367 368 369 370 371 372 373 374
        if self.error_flag:
            errors = self.errors
            if lbreaks:
                for err in errors:
                    err.pos = self.pos
                    err.line, err.column = line_col(lbreaks, err.pos)
            if clear_errors:
                self._errors = []
                self.error_flag = 0
            if self.children:
                for child in self.children:
                    errors.extend(child._collect_errors(lbreaks, clear_errors))
            return errors
        else:
            return []
375

Eckhart Arnold's avatar
Eckhart Arnold committed
376

377
    def _tree_repr(self, tab, openF, closeF, dataF=identity, density=0) -> str:
378 379 380 381 382 383 384 385
        """
        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.

386
        Args:
387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402
            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:
403
            return head.rstrip() + tail.lstrip()
404

405
        D = None if density & 2 else ''
406 407 408

        if self.children:
            content = []
409
            for child in self.children:
410
                subtree = child._tree_repr(tab, openF, closeF, dataF, density).split('\n')
411
                content.append('\n'.join((tab + s) for s in subtree))
412
            return head + '\n'.join(content) + tail.lstrip(D)
413

414
        res = cast(str, self.result)  # safe, because if there are no children, result is a string
415 416 417 418
        if density & 1 and res.find('\n') < 0:  # and head[0] == "<":
            # except for XML, add a gap between opening statement and content
            gap = ' ' if head.rstrip()[-1] != '>' else ''
            return head.rstrip() + gap + dataF(self.result) + tail.lstrip()
419
        else:
420
            return head + '\n'.join([tab + dataF(s) for s in res.split('\n')]) + tail.lstrip(D)
421

Eckhart Arnold's avatar
Eckhart Arnold committed
422

423
    def as_sxpr(self, src: str=None, compact: bool=False) -> str:
424 425 426
        """
        Returns content as S-expression, i.e. in lisp-like form.

427
        Args:
428 429 430
            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.
Eckhart Arnold's avatar
Eckhart Arnold committed
431 432 433
            compact:  If True a compact representation is returned where
                brackets are ommited and only the indentation indicates the
                tree structure.
434 435
        """

436 437
        lB, rB, D = ('', '', 1) if compact else ('(', '\n)', 0)

438
        def opening(node) -> str:
439
            s = lB + node.tag_name
440 441
            # s += " '(pos %i)" % node.pos
            if src:
442
                s += " '(pos %i " % node.pos  # + " %i %i)" % line_col(src, node.pos)
443 444 445
            if node.errors:
                s += " '(err '(%s))" % ' '.join(str(err).replace('"', r'\"')
                                                for err in node.errors)
446
            return s + '\n'
447 448 449 450 451 452

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

453
        return self._tree_repr('    ', opening, lambda node: rB, pretty, density=D)
454

Eckhart Arnold's avatar
Eckhart Arnold committed
455

456
    def as_xml(self, src: str=None) -> str:
457 458 459
        """
        Returns content as XML-tree.

460
        Args:
461 462 463 464 465
            src:  The source text or `None`. In case the source text is
                given the position will also be reported as line and
                column.
        """

466
        def opening(node) -> str:
467 468 469 470 471 472
            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)
473
            return s + ">\n"
474 475

        def closing(node):
476
            return '\n</' + node.tag_name + '>'
477

478
        return self._tree_repr('    ', opening, closing, density=1)
479

Eckhart Arnold's avatar
Eckhart Arnold committed
480

Eckhart Arnold's avatar
Eckhart Arnold committed
481 482 483
    def structure(self) -> str:
        """Return structure (and content) as S-expression on a single line
        without any line breaks."""
Eckhart Arnold's avatar
Eckhart Arnold committed
484 485
        return flatten_sxpr(self.as_sxpr())

486

Eckhart Arnold's avatar
Eckhart Arnold committed
487
    def content(self) -> str:
488
        """
Eckhart Arnold's avatar
Eckhart Arnold committed
489 490
        Returns content as string, inserting error messages where
        errors occurred.
491
        """
Eckhart Arnold's avatar
Eckhart Arnold committed
492 493
        s = "".join(child.content() for child in self.children) if self.children \
            else str(self.result)
494 495
        return (' <<< Error on "%s" | %s >>> '
                % (s, '; '.join(e.message for e in self._errors))) if self._errors else s
496

Eckhart Arnold's avatar
Eckhart Arnold committed
497

498
    def find(self, match_function: Callable) -> Iterator['Node']:
499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518
        """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

Eckhart Arnold's avatar
Eckhart Arnold committed
519

520 521 522 523 524
    def tree_size(self) -> int:
        """Recursively counts the number of nodes in the tree including the root node."""
        return sum(child.tree_size() for child in self.children) + 1


Eckhart Arnold's avatar
Eckhart Arnold committed
525
    def log(self, log_file_name):
526
        if is_logging():
Eckhart Arnold's avatar
Eckhart Arnold committed
527 528 529 530
            path = os.path.join(log_dir(), log_file_name)
            if os.path.exists(path):
                print('WARNING: Log-file "%s" already exists and will be overwritten!' % path)
            with open(path, "w", encoding="utf-8") as f:
531
                f.write(self.as_sxpr())
Eckhart Arnold's avatar
Eckhart Arnold committed
532

533

534
def mock_syntax_tree(sxpr):
535 536 537 538 539 540 541 542 543 544 545 546 547 548
    """
    Generates a tree of nodes from an S-expression.

    Example:
    >>> mock_syntax_tree("(a (b c))").as_sxpr()
    '(a\\n    (b\\n        "c"\\n    )\\n)'
    """

    def next_block(s):
        s = s.strip()
        while s[0] != ')':
            if s[0] != '(': raise ValueError('"(" expected, not ' + s[:10])
            # assert s[0] == '(', s
            level = 1
549
            k = 1
550
            while level > 0:
551
                if s[k] == '(':
552
                    level += 1
553
                elif s[k] == ')':
554
                    level -= 1
555 556 557
                k += 1
            yield s[:k]
            s = s[k:].strip()
558

559 560 561 562 563 564 565 566 567
    sxpr = sxpr.strip()
    if sxpr[0] != '(': raise ValueError('"(" expected, not ' + sxpr[:10])
    # assert sxpr[0] == '(', sxpr
    sxpr = sxpr[1:].strip()
    m = re.match('[\w:]+', sxpr)
    name, class_name = (sxpr[:m.end()].split(':') + [''])[:2]
    sxpr = sxpr[m.end():].strip()
    if sxpr[0] == '(':
        result = tuple(mock_syntax_tree(block) for block in next_block(sxpr))
568 569
    else:
        lines = []
570
        while sxpr and sxpr[0] != ')':
571
            for qm in ['"""', "'''", '"', "'"]:
572
                m = re.match(qm + r'.*?' + qm, sxpr, re.DOTALL)
573 574
                if m:
                    i = len(qm)
575 576
                    lines.append(sxpr[i:m.end() - i])
                    sxpr = sxpr[m.end():].strip()
577 578
                    break
            else:
579 580 581
                m = re.match(r'(?:(?!\)).)*', sxpr, re.DOTALL)
                lines.append(sxpr[:m.end()])
                sxpr = sxpr[m.end():]
582 583 584 585
        result = "\n".join(lines)
    return Node(MockParser(name, ':' + class_name), result)


586
TransformationFunc = Union[Callable[[Node], Any], partial]
587

588 589 590 591 592

# if __name__ == "__main__":
#     st = mock_syntax_tree("(alpha (beta (gamma i\nj\nk) (delta y)) (epsilon z))")
#     print(st.as_sxpr())
#     print(st.as_xml())