syntaxtree.py 19.6 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 25 26 27
try:
    import regex as re
except ImportError:
    import re
28 29
try:
    from typing import AbstractSet, Any, ByteString, Callable, cast, Container, Dict, \
Eckhart Arnold's avatar
Eckhart Arnold committed
30
        Iterator, Iterable, List, NamedTuple, Sequence, Union, Text, Tuple, Hashable
31 32
except ImportError:
    from .typing34 import AbstractSet, Any, ByteString, Callable, cast, Container, Dict, \
Eckhart Arnold's avatar
Eckhart Arnold committed
33
        Iterator, Iterable, List, NamedTuple, Sequence, Union, Text, Tuple, Hashable
34

35 36
from DHParser.toolkit import is_logging, log_dir, identity, StringView
from DHParser.error import Error, linebreaks, line_col
37

38 39 40 41 42 43 44
__all__ = ('ParserBase',
           'WHITESPACE_PTYPE',
           'TOKEN_PTYPE',
           'MockParser',
           'ZombieParser',
           'ZOMBIE_PARSER',
           'Node',
45 46
           'mock_syntax_tree',
           'TransformationFunc')
Eckhart Arnold's avatar
Eckhart Arnold committed
47 48


49 50 51 52 53 54 55 56 57 58 59 60 61 62
#######################################################################
#
# 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()):
63
        self._name = name  # type: str
64 65 66 67 68 69 70 71
        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)

72 73 74 75
    @property
    def name(self):
        return self._name

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

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


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] == ':'
101
        super().__init__(name)
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143
        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
#
#######################################################################


144
ChildrenType = Tuple['Node', ...]
145 146
StrictResultType = Union[ChildrenType, StringView, str]
ResultType = Union[ChildrenType, 'Node', StringView, str, None]
147 148


Eckhart Arnold's avatar
Eckhart Arnold committed
149 150
def flatten_sxpr(sxpr: str) -> str:
    """Returns S-expression `sxpr` as a one-liner without unnecessary
Eckhart Arnold's avatar
Eckhart Arnold committed
151 152 153
    whitespace.

    Example:
Eckhart Arnold's avatar
Eckhart Arnold committed
154
    >>> flatten_sxpr('(a\\n    (b\\n        c\\n    )\\n)\\n')
Eckhart Arnold's avatar
Eckhart Arnold committed
155 156 157 158 159
    '(a (b c))'
    """
    return re.sub('\s(?=\))', '', re.sub('\s+', ' ', sxpr)).strip()


Eckhart Arnold's avatar
Eckhart Arnold committed
160
class Node(collections.abc.Sized):
161 162 163 164 165 166 167 168 169 170 171
    """
    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!
172 173 174 175 176 177
        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, ...)``.
178 179
        errors (list):  A list of parser- or compiler-errors:
            tuple(position, string) attached to this node
180 181 182
        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.
183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202
        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.
    """
203

204 205
    __slots__ = ['_result', 'children', '_errors', '_len', '_pos', 'parser', 'error_flag']

206

Eckhart Arnold's avatar
Eckhart Arnold committed
207
    def __init__(self, parser, result: ResultType) -> None:
208 209 210
        """Initializes the ``Node``-object with the ``Parser``-Instance
        that generated the node and the parser's result.
        """
Eckhart Arnold's avatar
Eckhart Arnold committed
211
        # self._result = ''  # type: StrictResultType
212
        # self.children = ()  # type: ChildrenType
213
        self.error_flag = 0   # type: bool
214
        self._errors = []  # type: List[Error]
215
        self.result = result
Eckhart Arnold's avatar
Eckhart Arnold committed
216
        self._len = len(self._result) if not self.children else \
217
            sum(child._len for child in self.children)  # type: int
Eckhart Arnold's avatar
Eckhart Arnold committed
218
        # self.pos: int  = 0  # continuous updating of pos values wastes a lot of time
219
        self._pos = -1  # type: int
220 221
        self.parser = parser or ZOMBIE_PARSER

Eckhart Arnold's avatar
Eckhart Arnold committed
222

223 224
    def __str__(self):
        if self.children:
225
            return "".join(str(child) for child in self.children)
226 227
        return str(self.result)

Eckhart Arnold's avatar
Eckhart Arnold committed
228

229 230 231 232 233 234
    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)
235

Eckhart Arnold's avatar
Eckhart Arnold committed
236

237 238 239 240 241 242 243 244 245
    def __len__(self):
        return self._len


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


246
    def __eq__(self, other):
247 248
        # return str(self.parser) == str(other.parser) and self.result == other.result
        return self.tag_name == other.tag_name and self.result == other.result
249

Eckhart Arnold's avatar
Eckhart Arnold committed
250

251
    def __hash__(self):
252
        return hash(self.tag_name)
253

Eckhart Arnold's avatar
Eckhart Arnold committed
254

255 256 257 258 259 260
    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
261

262
    @property   # this needs to be a (dynamic) property, in case sef.parser gets updated
263
    def tag_name(self) -> str:
264
        return self.parser.name or self.parser.ptype
265

Eckhart Arnold's avatar
Eckhart Arnold committed
266

267
    @property
Eckhart Arnold's avatar
Eckhart Arnold committed
268
    def result(self) -> StrictResultType:
269 270 271
        return self._result

    @result.setter
Eckhart Arnold's avatar
Eckhart Arnold committed
272
    def result(self, result: ResultType):
Eckhart Arnold's avatar
Eckhart Arnold committed
273
        # # made obsolete by static type checking with mypy
274 275 276
        # 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
277 278
        # Possible optimization: Do not allow single nodes as argument:
        # assert not isinstance(result, Node)
279
        self._result = (result,) if isinstance(result, Node) else str(result) \
280
            if isinstance(result, StringView) else result or ''  # type: StrictResultType
281
        self.children = cast(ChildrenType, self._result) \
Eckhart Arnold's avatar
Eckhart Arnold committed
282
            if isinstance(self._result, tuple) else cast(ChildrenType, ())  # type: ChildrenType
283 284 285
        if self.children:
            self.error_flag = max(self.error_flag,
                                  max(child.error_flag for child in self.children))  # type: bool
286 287

    @property
288
    def pos(self) -> int:
289 290 291 292
        assert self._pos >= 0, "position value not initialized!"
        return self._pos

    @pos.setter
293
    def pos(self, pos: int):
294 295
        self._pos = pos
        offset = 0
296
        # recursively adjust pos-values of all children
297 298
        for child in self.children:
            child.pos = pos + offset
299
            offset += len(child)
300 301 302
        # add pos-values to Error-objects
        for err in self._errors:
            err.pos = pos
303

Eckhart Arnold's avatar
Eckhart Arnold committed
304

305
    @property
306
    def errors(self) -> List[Error]:
307 308
        return self._errors.copy()

309
    def add_error(self, message: str, level: int = Error.ERROR, code: Hashable = 0) -> 'Node':
Eckhart Arnold's avatar
Eckhart Arnold committed
310
        self._errors.append(Error(message, level, code))
311
        self.error_flag = max(self.error_flag, self._errors[-1].level)
Eckhart Arnold's avatar
Eckhart Arnold committed
312 313
        return self

314 315
    def collect_errors(self, document: Union[StringView, str] = '', clear_errors=False) -> List[
        Error]:
Eckhart Arnold's avatar
Eckhart Arnold committed
316
        """
317
        Recursively adds line- and column-numbers to all error objects.
Eckhart Arnold's avatar
Eckhart Arnold committed
318 319 320 321
        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.
        """
322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343
        if self.error_flag:
            lbreaks = linebreaks(document) if document else []
            return self._collect_errors(lbreaks, clear_errors)
        else:
            return []

    def _collect_errors(self, lbreaks: List[int] = [], clear_errors=False) -> List[Error]:
        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 []
344

Eckhart Arnold's avatar
Eckhart Arnold committed
345

346
    def _tree_repr(self, tab, openF, closeF, dataF=identity, density=0) -> str:
347 348 349 350 351 352 353 354
        """
        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.

355
        Args:
356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371
            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:
372
            return head.rstrip() + tail.lstrip()
373

374
        D = None if density & 2 else ''
375 376 377

        if self.children:
            content = []
378
            for child in self.children:
379
                subtree = child._tree_repr(tab, openF, closeF, dataF, density).split('\n')
380
                content.append('\n'.join((tab + s) for s in subtree))
381
            return head + '\n'.join(content) + tail.lstrip(D)
382

383
        res = cast(str, self.result)  # safe, because if there are no children, result is a string
384 385 386 387
        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()
388
        else:
389
            return head + '\n'.join([tab + dataF(s) for s in res.split('\n')]) + tail.lstrip(D)
390

Eckhart Arnold's avatar
Eckhart Arnold committed
391

392
    def as_sxpr(self, src: str=None, compact: bool=False) -> str:
393 394 395
        """
        Returns content as S-expression, i.e. in lisp-like form.

396
        Args:
397 398 399
            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
400 401 402
            compact:  If True a compact representation is returned where
                brackets are ommited and only the indentation indicates the
                tree structure.
403 404
        """

405 406
        lB, rB, D = ('', '', 1) if compact else ('(', '\n)', 0)

407
        def opening(node) -> str:
408
            s = lB + node.tag_name
409 410
            # s += " '(pos %i)" % node.pos
            if src:
411
                s += " '(pos %i " % node.pos  # + " %i %i)" % line_col(src, node.pos)
412 413 414
            if node.errors:
                s += " '(err '(%s))" % ' '.join(str(err).replace('"', r'\"')
                                                for err in node.errors)
415
            return s + '\n'
416 417 418 419 420 421

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

422
        return self._tree_repr('    ', opening, lambda node: rB, pretty, density=D)
423

Eckhart Arnold's avatar
Eckhart Arnold committed
424

425
    def as_xml(self, src: str=None) -> str:
426 427 428
        """
        Returns content as XML-tree.

429
        Args:
430 431 432 433 434
            src:  The source text or `None`. In case the source text is
                given the position will also be reported as line and
                column.
        """

435
        def opening(node) -> str:
436 437 438 439 440 441
            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)
442
            return s + ">\n"
443 444

        def closing(node):
445
            return '\n</' + node.tag_name + '>'
446

447
        return self._tree_repr('    ', opening, closing, density=1)
448

Eckhart Arnold's avatar
Eckhart Arnold committed
449

Eckhart Arnold's avatar
Eckhart Arnold committed
450 451 452
    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
453 454
        return flatten_sxpr(self.as_sxpr())

455

Eckhart Arnold's avatar
Eckhart Arnold committed
456
    def content(self) -> str:
457
        """
Eckhart Arnold's avatar
Eckhart Arnold committed
458 459
        Returns content as string, inserting error messages where
        errors occurred.
460
        """
Eckhart Arnold's avatar
Eckhart Arnold committed
461 462
        s = "".join(child.content() for child in self.children) if self.children \
            else str(self.result)
463 464
        return (' <<< Error on "%s" | %s >>> '
                % (s, '; '.join(e.message for e in self._errors))) if self._errors else s
465

Eckhart Arnold's avatar
Eckhart Arnold committed
466

467
    def find(self, match_function: Callable) -> Iterator['Node']:
468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487
        """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
488

489 490 491 492 493
    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
494
    def log(self, log_file_name):
495
        if is_logging():
Eckhart Arnold's avatar
Eckhart Arnold committed
496 497 498 499
            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:
500
                f.write(self.as_sxpr())
Eckhart Arnold's avatar
Eckhart Arnold committed
501

502

503
def mock_syntax_tree(sxpr):
504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527
    """
    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
            i = 1
            while level > 0:
                if s[i] == '(':
                    level += 1
                elif s[i] == ')':
                    level -= 1
                i += 1
            yield s[:i]
            s = s[i:].strip()

528 529 530 531 532 533 534 535 536
    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))
537 538
    else:
        lines = []
539
        while sxpr and sxpr[0] != ')':
540
            for qm in ['"""', "'''", '"', "'"]:
541
                m = re.match(qm + r'.*?' + qm, sxpr, re.DOTALL)
542 543
                if m:
                    i = len(qm)
544 545
                    lines.append(sxpr[i:m.end() - i])
                    sxpr = sxpr[m.end():].strip()
546 547
                    break
            else:
548 549 550
                m = re.match(r'(?:(?!\)).)*', sxpr, re.DOTALL)
                lines.append(sxpr[:m.end()])
                sxpr = sxpr[m.end():]
551 552 553 554
        result = "\n".join(lines)
    return Node(MockParser(name, ':' + class_name), result)


555
TransformationFunc = Union[Callable[[Node], Any], partial]
556

557 558 559 560 561

# 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())