summaryrefslogtreecommitdiff
path: root/giscanner/annotationparser.py
blob: f70bfa452e00a4184ff5d3fa6bd0ee007219cd8a (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
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
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
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
195
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
227
228
229
230
231
232
233
234
235
236
237
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
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
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
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
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
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
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
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
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
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
# -*- Mode: Python -*-
# GObject-Introspection - a framework for introspecting GObject libraries
# Copyright (C) 2008  Johan Dahlin
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
# 02110-1301, USA.
#

# AnnotationParser - parses gtk-doc annotations

import re
import sys

from .ast import (Array, Bitfield, Callback, Class, Enum, Field, Function,
                  Interface, List, Map, Parameter, Record, Return, Type, Union,
                  Varargs,
                  default_array_types,
                  BASIC_GIR_TYPES,
                  PARAM_DIRECTION_INOUT,
                  PARAM_DIRECTION_IN,
                  PARAM_DIRECTION_OUT,
                  PARAM_TRANSFER_NONE,
                  PARAM_TRANSFER_CONTAINER,
                  PARAM_TRANSFER_FULL,
                  TYPE_ANY, TYPE_NONE)
from .odict import odict
from .glibast import GLibBoxed

# All gtk-doc comments needs to start with this:
_COMMENT_HEADER = '*\n '

# Tags - annotations applyed to comment blocks
TAG_VFUNC = 'virtual'
TAG_SINCE = 'since'
TAG_DEPRECATED = 'deprecated'
TAG_RETURNS = 'returns'
TAG_RETURNS_ALT = 'return value'
TAG_ATTRIBUTES = 'attributes'
TAG_RENAME_TO = 'rename to'

# Options - annotations for parameters and return values
OPT_ALLOW_NONE = 'allow-none'
OPT_ARRAY = 'array'
OPT_ELEMENT_TYPE = 'element-type'
OPT_IN = 'in'
OPT_INOUT = 'inout'
OPT_INOUT_ALT = 'in-out'
OPT_OUT = 'out'
OPT_SCOPE = 'scope'
OPT_TRANSFER = 'transfer'
OPT_TYPE = 'type'
OPT_CLOSURE = 'closure'
OPT_DESTROY = 'destroy'
OPT_SKIP = 'skip'

# Specific option values
OPT_VAL_BITFIELD = 'bitfield'

# Array options - array specific annotations
OPT_ARRAY_FIXED_SIZE = 'fixed-size'
OPT_ARRAY_LENGTH = 'length'
OPT_ARRAY_ZERO_TERMINATED = 'zero-terminated'

OPT_SCOPE_ASYNC = 'async'
OPT_SCOPE_CALL = 'call'
OPT_SCOPE_NOTIFIED = 'notified'

class InvalidAnnotationError(Exception):
    pass


class DocBlock(object):

    def __init__(self, name, options):
        self.name = name
        self.options = options
        self.value = None
        self.tags = odict()
        self.comment = None

    def __repr__(self):
        return '<DocBlock %r %r>' % (self.name, self.options)

    def get(self, name):
        if name == TAG_RETURNS:
            value = self.tags.get(name)
            if value is None:
                return self.tags.get(TAG_RETURNS_ALT)
            else:
                return value
        else:
            return self.tags.get(name)


class DocTag(object):

    def __init__(self, name):
        self.name = name
        self.options = {}
        self.comment = None

    def __repr__(self):
        return '<DocTag %r %r>' % (self.name, self.options)

class Option(object):

    def __init__(self, option):
        self._array = []
        self._dict = {}
        for p in option.split(' '):
            if '=' in p:
                name, value = p.split('=', 1)
            else:
                name = p
                value = None
            self._dict[name] = value
            if value is None:
                self._array.append(name)
            else:
                self._array.append((name, value))

    def __repr__(self):
        return '<Option %r>' % (self._array, )

    def one(self):
        assert len(self._array) == 1
        return self._array[0]

    def flat(self):
        return self._array

    def all(self):
        return self._dict


class AnnotationParser(object):

    def __init__(self, namespace, source_scanner, transformer):
        self._blocks = {}
        self._namespace = namespace
        self._transformer = transformer
        for comment in source_scanner.get_comments():
            self._parse_comment(comment)

    def parse(self):
        aa = AnnotationApplier(self._blocks, self._transformer)
        aa.parse(self._namespace)

    def _parse_comment(self, comment):
        # We're looking for gtk-doc comments here, they look like this:
        # /**
        #   * symbol:
        #
        # Or, alternatively, with options:
        # /**
        #   * symbol: (name value) ...
        #
        # symbol is currently one of:
        #  - function: gtk_widget_show
        #  - signal:   GtkWidget::destroy
        #  - property: GtkWidget:visible
        #
        comment = comment.lstrip()
        if not comment.startswith(_COMMENT_HEADER):
            return
        comment = comment[len(_COMMENT_HEADER):]
        comment = comment.strip()
        if not comment.startswith('* '):
            return
        comment = comment[2:]

        pos = comment.find('\n ')
        if pos == -1:
            return
        block_header = comment[:pos]
        block_header = block_header.strip()
        cpos = block_header.find(': ')
        if cpos:
            block_name = block_header[:cpos]
            block_options, rest = self._parse_options(block_header[cpos+2:])
            if rest:
                return
        else:
            block_name, block_options = block_header, {}
        block = DocBlock(block_name, block_options)
        comment_lines = []
        for line in comment[pos+1:].split('\n'):
            line = line.lstrip()
            line = line[2:].strip() # Skip ' *'
            if not line:
                continue
            if line.startswith('@'):
                line = line[1:]
            elif not ': ' in line:
                comment_lines.append(line)
                continue
            tag_name, value = self._split_tag_namevalue(line)
            canon_name = tag_name.lower()
            if canon_name in block.tags:
                print >> sys.stderr, (
                    "Symbol %s has multiple definition of tag %r" % (
                    block_name, canon_name, ))
            block.tags[canon_name] = self._create_tag(canon_name, value)
        block.comment = '\n'.join(comment_lines)
        self._blocks[block.name] = block

    def _split_tag_namevalue(self, raw):
        """Split a line into tag name and value"""
        parts = raw.split(': ', 1)
        if len(parts) == 1:
            tag_name = parts[0]
            value = ''
            if tag_name.endswith(':'):
                tag_name = tag_name[:-1]
        else:
            tag_name, value = parts
        return (tag_name, value)

    def _create_tag(self, tag_name, value):
        # Tag: bar
        # Tag: bar opt1 opt2
        tag = DocTag(tag_name)
        tag.value = value
        options, rest = self._parse_options(tag.value)
        tag.options = options
        tag.comment = rest
        return tag

    def _parse_options(self, value):
        # (foo)
        # (bar opt1 opt2...)
        opened = -1
        options = {}
        last = None
        for i, c in enumerate(value):
            if c == '(' and opened == -1:
                opened = i+1
            if c == ')' and opened != -1:
                segment = value[opened:i]
                parts = segment.split(' ', 1)
                if len(parts) == 2:
                    name, option = parts
                elif len(parts) == 1:
                    name = parts[0]
                    option = None
                else:
                    raise AssertionError
                if option is not None:
                    option = Option(option)
                options[name] = option
                last = i + 2
                opened = -1

        if last is not None:
            rest = value[last:].strip()
        else:
            rest = None
        return options, rest


class AnnotationApplier(object):

    def __init__(self, blocks, transformer):
        self._blocks = blocks
        self._transformer = transformer

    def _get_tag(self, block, tag_name):
        if block is None:
            return None

        return block.get(tag_name)

    def parse(self, namespace):
        self._namespace = namespace
        for node in namespace.nodes[:]:
            self._parse_node(node)
        del self._namespace

    # Boring parsing boilerplate.

    def _parse_node(self, node):
        if isinstance(node, Function):
            self._parse_function(node)
        elif isinstance(node, Enum):
            self._parse_enum(node)
        elif isinstance(node, Bitfield):
            self._parse_bitfield(node)
        elif isinstance(node, Class):
            self._parse_class(node)
        elif isinstance(node, Interface):
            self._parse_interface(node)
        elif isinstance(node, Callback):
            self._parse_callback(node)
        elif isinstance(node, Record):
            self._parse_record(node)
        elif isinstance(node, Union):
            self._parse_union(node)
        elif isinstance(node, GLibBoxed):
            self._parse_boxed(node)

    def _parse_class(self, class_):
        block = self._blocks.get(class_.type_name)
        self._parse_node_common(class_, block)
        self._parse_constructors(class_.constructors)
        self._parse_methods(class_, class_.methods)
        self._parse_vfuncs(class_, class_.virtual_methods)
        self._parse_methods(class_, class_.static_methods)
        self._parse_properties(class_, class_.properties)
        self._parse_signals(class_, class_.signals)
        self._parse_fields(class_, class_.fields)
        if block:
            class_.doc = block.comment

    def _parse_interface(self, interface):
        block = self._blocks.get(interface.type_name)
        self._parse_node_common(interface, block)
        self._parse_methods(interface, interface.methods)
        self._parse_vfuncs(interface, interface.virtual_methods)
        self._parse_properties(interface, interface.properties)
        self._parse_signals(interface, interface.signals)
        self._parse_fields(interface, interface.fields)
        if block:
            interface.doc = block.comment

    def _parse_record(self, record):
        block = self._blocks.get(record.symbol)
        self._parse_node_common(record, block)
        self._parse_constructors(record.constructors)
        self._parse_methods(record, record.methods)
        self._parse_fields(record, record.fields, block)
        if block:
            record.doc = block.comment

    def _parse_boxed(self, boxed):
        block = self._blocks.get(boxed.name)
        self._parse_node_common(boxed, block)
        self._parse_constructors(boxed.constructors)
        self._parse_methods(boxed, boxed.methods)
        if block:
            boxed.doc = block.comment

    def _parse_union(self, union):
        block = self._blocks.get(union.name)
        self._parse_node_common(union, block)
        self._parse_fields(union, union.fields, block)
        self._parse_constructors(union.constructors)
        self._parse_methods(union, union.methods)
        if block:
            union.doc = block.comment

    def _parse_enum(self, enum):
        block = self._blocks.get(enum.symbol)
        self._parse_node_common(enum, block)
        if block:
            enum.doc = block.comment
            type_opt = block.options.get(OPT_TYPE)
            if type_opt and type_opt.one() == OPT_VAL_BITFIELD:
                # This is hack, but hey, it works :-)
                enum.__class__ = Bitfield

    def _parse_bitfield(self, bitfield):
        block = self._blocks.get(bitfield.symbol)
        self._parse_node_common(bitfield, block)
        if block:
            bitfield.doc = block.comment

    def _parse_constructors(self, constructors):
        for ctor in constructors:
            self._parse_function(ctor)

    def _parse_fields(self, parent, fields, block=None):
        for field in fields:
            self._parse_field(parent, field, block)

    def _parse_properties(self, parent, properties):
        for prop in properties:
            self._parse_property(parent, prop)

    def _parse_methods(self, parent, methods):
        for method in methods:
            self._parse_method(parent, method)

    def _parse_vfuncs(self, parent, vfuncs):
        for vfunc in vfuncs:
            self._parse_vfunc(parent, vfunc)

    def _parse_signals(self, parent, signals):
        for signal in signals:
            self._parse_signal(parent, signal)

    def _parse_property(self, parent, prop):
        block = self._blocks.get('%s:%s' % (parent.type_name, prop.name))
        self._parse_node_common(prop, block)
        if block:
            prop.doc = block.comment

    def _parse_callback(self, callback):
        block = self._blocks.get(callback.ctype)
        self._parse_node_common(callback, block)
        self._parse_params(callback, callback.parameters, block)
        self._parse_return(callback, callback.retval, block)
        if block:
            callback.doc = block.comment

    def _parse_callable(self, callable, block):
        self._parse_node_common(callable, block)
        for i, param in enumerate(callable.parameters):
            if (param.type.ctype != 'GDestroyNotify' and
                param.type.name != 'GLib.DestroyNotify'):
                continue
            if i < 2:
                break
            callback_param = callable.parameters[i-2]
            if callback_param.closure_index != -1:
                callback_param.scope = OPT_SCOPE_NOTIFIED
                callback_param.transfer = PARAM_TRANSFER_NONE

        self._parse_params(callable, callable.parameters, block)
        self._parse_return(callable, callable.retval, block)
        if block:
            callable.doc = block.comment

    def _parse_function(self, func):
        block = self._blocks.get(func.symbol)
        self._parse_callable(func, block)
        self._parse_rename_to_func(func, block)

    def _parse_signal(self, parent, signal):
        block = self._blocks.get('%s::%s' % (parent.type_name, signal.name))
        self._parse_node_common(signal, block)
        # We're only attempting to name the signal parameters if
        # the number of parameter tags (@foo) is the same or greater
        # than the number of signal parameters
        if block and len(block.tags) > len(signal.parameters):
            names = block.tags.items()
        else:
            names = []
        for i, param in enumerate(signal.parameters):
            if names:
                name, tag = names[i+1]
                param.name = name
                options = getattr(tag, 'options', {})
                param_type = options.get(OPT_TYPE)
                if param_type:
                    param.type = self._resolve(param_type.one(), param.type)
            else:
                tag = None
            self._parse_param(signal, param, tag)
        self._parse_return(signal, signal.retval, block)
        if block:
            signal.doc = block.comment

    def _parse_method(self, parent, meth):
        block = self._blocks.get(meth.symbol)
        self._parse_function(meth)
        virtual = self._get_tag(block, TAG_VFUNC)
        if virtual:
            invoker_name = virtual.value
            matched = False
            for vfunc in parent.virtual_methods:
                if vfunc.name == invoker_name:
                    matched = True
                    vfunc.invoker = meth.name
                    break
            if not matched:
                print "warning: unmatched virtual invoker %r for method %r" % \
                    (invoker_name, meth.symbol)

    def _parse_vfunc(self, parent, vfunc):
        key = '%s::%s' % (parent.type_name, vfunc.name)
        self._parse_callable(vfunc, self._blocks.get(key))

    def _parse_field(self, parent, field, block=None):
        if isinstance(field, Callback):
            self._parse_callback(field)
        else:
            if not block:
                return
            tag = block.get(field.name)
            if not tag:
                return
            t = tag.options.get('type')
            if not t:
                return
            field.type.name = self._transformer.resolve_type_name(t.one())

    def _check_arg_annotations(self, parent, params, block):
        if block is None:
            return
        for tag in block.tags.keys():
            if tag == TAG_RETURNS:
                continue
            for param in params:
                if param.name == tag:
                    break
            else:
                return
                #print 'WARNING: annotation for "%s" refers to unknown ' \
                #        'argument "%s"' % (parent.name, tag)

    def _parse_params(self, parent, params, block):
        self._check_arg_annotations(parent, params, block)
        for param in params:
            tag = self._get_tag(block, param.name)
            self._parse_param(parent, param, tag)

    def _parse_return(self, parent, return_, block):
        tag = self._get_tag(block, TAG_RETURNS)
        self._parse_param_ret_common(parent, return_, tag)

    def _parse_param(self, parent, param, tag):
        options = getattr(tag, 'options', {})
        if isinstance(parent, Function):
            scope = options.get(OPT_SCOPE)
            if scope:
                param.scope = scope.one()
                param.transfer = PARAM_TRANSFER_NONE
            elif (param.type.ctype == 'GAsyncReadyCallback' or
                  param.type.name == 'Gio.AsyncReadyCallback'):
                param.scope = OPT_SCOPE_ASYNC
                param.transfer = PARAM_TRANSFER_NONE

            destroy = options.get(OPT_DESTROY)
            if destroy:
                param.destroy_index = parent.get_parameter_index(destroy.one())
                self._fixup_param_destroy(parent, param)
            closure = options.get(OPT_CLOSURE)
            if closure:
                param.closure_index = parent.get_parameter_index(closure.one())
                self._fixup_param_closure(parent, param)
        if isinstance(parent, Callback):
            if OPT_CLOSURE in options:
                param.closure_index = parent.get_parameter_index(param.name)
                self._fixup_param_closure(parent, param)

        self._parse_param_ret_common(parent, param, tag)

    def _fixup_param_destroy(self, parent, param):
        for p in parent.parameters:
            if p is not param and p.destroy_index == param.destroy_index:
                p.destroy_index = None

    def _fixup_param_closure(self, parent, param):
        for p in parent.parameters:
            if p is not param and p.closure_index == param.closure_index:
                p.closure_index = None

    def _parse_param_ret_common(self, parent, node, tag):
        options = getattr(tag, 'options', {})
        node.direction = self._extract_direction(node, options)
        container_type = self._extract_container_type(
            parent, node, options)
        if container_type is not None:
            node.type = container_type
        if node.direction is None:
            node.direction = self._guess_direction(node)
        node.transfer = self._extract_transfer(parent, node, options)
        param_type = options.get(OPT_TYPE)
        if param_type:
            node.type = self._resolve(param_type.one(), node.type)

        if (OPT_ALLOW_NONE in options or
            node.type.ctype == 'GCancellable*'):
            node.allow_none = True

        assert node.transfer is not None
        if tag is not None and tag.comment is not None:
            node.doc = tag.comment

    def _extract_direction(self, node, options):
        if (OPT_INOUT in options or
            OPT_INOUT_ALT in options):
            direction = PARAM_DIRECTION_INOUT
        elif OPT_OUT in options:
            direction = PARAM_DIRECTION_OUT
        elif OPT_IN in options:
            direction = PARAM_DIRECTION_IN
        else:
            direction = node.direction
        return direction

    def _guess_array(self, node):
        ctype = node.type.ctype
        if ctype is None:
            return False
        if not ctype.endswith('*'):
            return False
        if node.type.canonical in default_array_types:
            return True
        return False

    def _extract_container_type(self, parent, node, options):
        has_element_type = OPT_ELEMENT_TYPE in options
        has_array = OPT_ARRAY in options

        if not has_array:
            has_array = \
                node.type.name in ['GLib.Array', 'GLib.PtrArray',
                                   'GLib.ByteArray'] or \
                node.type.ctype in ['GArray*', 'GPtrArray*', 'GByteArray*']

        # FIXME: This is a hack :-(
        if (not isinstance(node, Field) and
            (not has_element_type and
             (node.direction is None
              or isinstance(node, Return)
              or node.direction == PARAM_DIRECTION_IN))):
            if self._guess_array(node):
                has_array = True

        if has_array:
            container_type = self._parse_array(parent, node, options)
        elif has_element_type:
            container_type = self._parse_element_type(parent, node, options)
        else:
            container_type = None

        return container_type

    def _parse_array(self, parent, node, options):
        array_opt = options.get(OPT_ARRAY)
        if array_opt:
            array_values = array_opt.all()
        else:
            array_values = {}

        element_type = options.get(OPT_ELEMENT_TYPE)
        if element_type is not None:
            element_type_node = self._resolve(element_type.one())
        else:
            element_type_node = Type(node.type.name) # erase ctype

        container_type = Array(node.type.ctype,
                               element_type_node)
        container_type.is_const = node.type.is_const
        if OPT_ARRAY_ZERO_TERMINATED in array_values:
            container_type.zeroterminated = array_values.get(
                OPT_ARRAY_ZERO_TERMINATED) == '1'
        length = array_values.get(OPT_ARRAY_LENGTH)
        if length is not None:
            param_index = parent.get_parameter_index(length)
            container_type.length_param_index = param_index
            # For in parameters we're incorrectly deferring
            # char/unsigned char to utf8 when a length annotation
            # is specified.
            if (isinstance(node, Parameter) and
                node.type.name == 'utf8' and
                self._guess_direction(node) == PARAM_DIRECTION_IN and
                element_type is None):
                # FIXME: unsigned char/guchar should be uint8
                container_type.element_type = Type('int8')
        container_type.size = array_values.get(OPT_ARRAY_FIXED_SIZE)
        return container_type

    def _resolve(self, type_str, orig_node=None):
        def grab_one(type_str, resolver, top_combiner, combiner):
            """Return a complete type, and the trailing string part after it.
            Use resolver() on each identifier, and combiner() on the parts of
            each complete type. (top_combiner is used on the top-most type.)"""
            bits = re.split(r'([,<>])', type_str, 1)
            first, sep, rest = [bits[0], '', ''] if (len(bits)==1) else bits
            args = [resolver(first)]
            if sep == '<':
                while sep != '>':
                    next, rest = grab_one(rest, resolver, combiner, combiner)
                    args.append(next)
                    sep, rest = rest[0], rest[1:]
            else:
                rest = sep + rest
            return top_combiner(*args), rest
        def resolver(ident):
            return self._transformer.resolve_param_type(Type(ident))
        def combiner(base, *rest):
            if not rest:
                return base
            if (base.name in ['GLib.List', 'GLib.SList'] or
                base.ctype in ['GList*', 'GSList*']) and len(rest)==1:
                return List(base.name, base.ctype, *rest)
            if (base.name in ['GLib.HashTable'] or
                base.ctype in ['GHashTable*']) and len(rest)==2:
                return Map(base.name, base.ctype, *rest)
            print "WARNING: throwing away type parameters:", type_str
            return base
        def top_combiner(base, *rest):
            """For the top most type, recycle orig_node if possible."""
            if orig_node is not None:
                orig_node.name = base.name
                base = orig_node # preserve other properties of orig_node
            return combiner(base, *rest)

        result, rest = grab_one(type_str, resolver, top_combiner, combiner)
        if rest:
            print "WARNING: throwing away trailing part of type:", type_str
        return result

    def _parse_element_type(self, parent, node, options):
        element_type_opt = options.get(OPT_ELEMENT_TYPE)
        element_type = element_type_opt.flat()
        if (node.type.name in ['GLib.List', 'GLib.SList'] or
            node.type.ctype in ['GList*', 'GSList*']):
            assert len(element_type) == 1
            container_type = List(
                node.type.name,
                node.type.ctype,
                self._resolve(element_type[0]))
        elif (node.type.name in ['GLib.HashTable'] or
              node.type.ctype in ['GHashTable*']):
            assert len(element_type) == 2
            container_type = Map(
                node.type.name,
                node.type.ctype,
                self._resolve(element_type[0]),
                self._resolve(element_type[1]))
        else:
            print 'FIXME: unhandled element-type container:', node
        return container_type

    def _extract_transfer(self, parent, node, options):
        transfer_opt = options.get(OPT_TRANSFER)
        if transfer_opt is None:
            transfer = self._guess_transfer(node, options)
        else:
            transfer = transfer_opt.one()
            if transfer is None:
                transfer = PARAM_TRANSFER_FULL
            if transfer not in [PARAM_TRANSFER_NONE,
                                PARAM_TRANSFER_CONTAINER,
                                PARAM_TRANSFER_FULL]:
                raise InvalidAnnotationError(
                    "transfer for %s of %r is invalid (%r), must be one of "
                    "none, container, full." % (node, parent.name, transfer))
        return transfer

    def _parse_node_common(self, node, block):
        self._parse_version(node, block)
        self._parse_deprecated(node, block)
        self._parse_attributes(node, block)
        self._parse_skip(node, block)

    def _parse_version(self, node, block):
        since_tag = self._get_tag(block, TAG_SINCE)
        if since_tag is None:
            return
        node.version = since_tag.value

    def _parse_deprecated(self, node, block):
        deprecated_tag = self._get_tag(block, TAG_DEPRECATED)
        if deprecated_tag is None:
            return
        value = deprecated_tag.value
        if ': ' in value:
            version, desc = value.split(': ')
        else:
            desc = value
            version = None
        node.deprecated = desc
        if version is not None:
            node.deprecated_version = version

    def _parse_attributes(self, node, block):
        annos_tag = self._get_tag(block, TAG_ATTRIBUTES)
        if annos_tag is None:
            return
        for key, value in annos_tag.options.iteritems():
            if value:
                node.attributes.append((key, value.one()))

    def _parse_skip(self, node, block):
        if block is not None:
            if OPT_SKIP in block.options:
                node.skip = True

    def _parse_rename_to_func(self, node, block):
        rename_to_tag = self._get_tag(block, TAG_RENAME_TO)
        if rename_to_tag is None:
            return
        new_name = rename_to_tag.value

        shadowed = []

        def shadowed_filter(n):
            if isinstance(n, Function) and n.symbol == new_name:
                shadowed.append(n)
                return False
            return True

        self._namespace.remove_matching(shadowed_filter)
        if len(shadowed) == 1:
            # method override; use the same (stripped) name as the overloaded
            # method referenced.
            # Note that 'g_timeout_add_full' may specify a new_name of
            # 'g_timeout_add' but the *real* name desired is the stripped name
            # of 'g_timeout_add' which is 'timeout_add' (for example).
            node.name = shadowed[0].name
        elif len(shadowed) == 0:
            # literal rename, to force a particular prefix strip or whatever
            # Example: the "nm-utils" package uses a "NM" prefix in most places
            # but some functions have an "nm_utils_" prefix; the 'Rename To:'
            # annotation in this case is used to strip the 'utils_' part off.
            node.name = new_name
        else:
            assert False # more than two shadowed methods?  Shouldn't happen.

    def _guess_direction(self, node):
        if node.direction:
            return node.direction
        is_pointer = False
        if node.type.ctype:
            is_pointer = '*' in node.type.ctype

        if is_pointer and node.type.name in BASIC_GIR_TYPES:
            return PARAM_DIRECTION_OUT

        return PARAM_DIRECTION_IN

    def _guess_transfer(self, node, options):
        if node.transfer is not None:
            return node.transfer

        # Anything with 'const' gets none
        if node.type.is_const:
            return PARAM_TRANSFER_NONE
        elif node.type.name in [TYPE_NONE, TYPE_ANY]:
            return PARAM_TRANSFER_NONE
        elif isinstance(node.type, Varargs):
            return PARAM_TRANSFER_NONE
        elif isinstance(node, Parameter):
            if node.direction in [PARAM_DIRECTION_INOUT,
                                  PARAM_DIRECTION_OUT]:
                return PARAM_TRANSFER_FULL
            # This one is a hack for compatibility; the transfer
            # for string parameters really has no defined meaning.
            elif node.type.canonical == 'utf8':
                return PARAM_TRANSFER_FULL
            else:
                return PARAM_TRANSFER_NONE
        elif isinstance(node, Return):
            if (node.type.canonical in BASIC_GIR_TYPES or
                (node.type.canonical in [TYPE_NONE, TYPE_ANY] and
                 node.type.is_const)):
                return PARAM_TRANSFER_NONE
            else:
                return PARAM_TRANSFER_FULL
        elif isinstance(node, Field):
            return PARAM_TRANSFER_NONE
        else:
            raise AssertionError(node)