genshi/mirror: markup/path.py annotate

annotate markup/path.py @ 121:062e51ad7b19 trunk

Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`.

author	cmlenz
date	Wed, 02 Aug 2006 12:55:05 +0000
parents	4c4e81d12649
children	6c5c6f67d3e8

rev	line source
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	1 # -- coding: utf-8 --
5479aae32f5a Initial import. cmlenz parents: diff changeset	2 #
66 59eb24184e9c Switch copyright to Edgewall and URLs to markup.edgewall.org. cmlenz parents: 61 diff changeset	3 # Copyright (C) 2006 Edgewall Software
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	4 # All rights reserved.
5479aae32f5a Initial import. cmlenz parents: diff changeset	5 #
5479aae32f5a Initial import. cmlenz parents: diff changeset	6 # This software is licensed as described in the file COPYING, which
5479aae32f5a Initial import. cmlenz parents: diff changeset	7 # you should have received as part of this distribution. The terms
66 59eb24184e9c Switch copyright to Edgewall and URLs to markup.edgewall.org. cmlenz parents: 61 diff changeset	8 # are also available at http://markup.edgewall.org/wiki/License.
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	9 #
5479aae32f5a Initial import. cmlenz parents: diff changeset	10 # This software consists of voluntary contributions made by many
5479aae32f5a Initial import. cmlenz parents: diff changeset	11 # individuals. For the exact contribution history, see the revision
66 59eb24184e9c Switch copyright to Edgewall and URLs to markup.edgewall.org. cmlenz parents: 61 diff changeset	12 # history and logs, available at http://markup.edgewall.org/log/.
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	13
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	14 """Basic support for evaluating XPath expressions against streams.
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	15
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	16 >>> from markup.input import XML
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	17 >>> doc = XML('''<doc>
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	18 ... <items count="2">
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	19 ... <item status="new">
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	20 ... <summary>Foo</summary>
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	21 ... </item>
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	22 ... <item status="closed">
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	23 ... <summary>Bar</summary>
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	24 ... </item>
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	25 ... </items>
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	26 ... </doc>''')
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	27 >>> print doc.select('items/item[@status="closed"]/summary/text()')
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	28 Bar
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	29
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	30 Because the XPath engine operates on markup streams (as opposed to tree
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	31 structures), it only implements a subset of the full XPath 1.0 language.
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	32 """
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	33
5479aae32f5a Initial import. cmlenz parents: diff changeset	34 import re
5479aae32f5a Initial import. cmlenz parents: diff changeset	35
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	36 from markup.core import QName, Stream, START, END, TEXT, COMMENT, PI
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	37
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	38 __all__ = ['Path', 'PathSyntaxError']
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	39
5479aae32f5a Initial import. cmlenz parents: diff changeset	40
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	41 class Axis(object):
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	42 """Defines constants for the various supported XPath axes."""
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	43
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	44 ATTRIBUTE = 'attribute'
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	45 CHILD = 'child'
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	46 DESCENDANT = 'descendant'
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	47 DESCENDANT_OR_SELF = 'descendant-or-self'
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	48 NAMESPACE = 'namespace'
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	49 SELF = 'self'
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	50
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	51 def forname(cls, name):
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	52 """Return the axis constant for the given name, or `None` if no such
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	53 axis was defined.
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	54 """
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	55 return getattr(cls, name.upper().replace('-', '_'), None)
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	56 forname = classmethod(forname)
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	57
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	58
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	59 ATTRIBUTE = Axis.ATTRIBUTE
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	60 CHILD = Axis.CHILD
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	61 DESCENDANT = Axis.DESCENDANT
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	62 DESCENDANT_OR_SELF = Axis.DESCENDANT_OR_SELF
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	63 NAMESPACE = Axis.NAMESPACE
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	64 SELF = Axis.SELF
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	65
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	66
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	67 class Path(object):
26 3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	68 """Implements basic XPath support on streams.
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	69
26 3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	70 Instances of this class represent a "compiled" XPath expression, and provide
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	71 methods for testing the path against a stream, as well as extracting a
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	72 substream matching that path.
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	73 """
5479aae32f5a Initial import. cmlenz parents: diff changeset	74
5479aae32f5a Initial import. cmlenz parents: diff changeset	75 def __init__(self, text):
26 3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	76 """Create the path object from a string.
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	77
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	78 @param text: the path expression
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	79 """
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	80 self.source = text
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	81 self.paths = _PathParser(text).parse()
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	82
5479aae32f5a Initial import. cmlenz parents: diff changeset	83 def __repr__(self):
5479aae32f5a Initial import. cmlenz parents: diff changeset	84 return '<%s "%s">' % (self.__class__.__name__, self.source)
5479aae32f5a Initial import. cmlenz parents: diff changeset	85
5479aae32f5a Initial import. cmlenz parents: diff changeset	86 def select(self, stream):
26 3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	87 """Returns a substream of the given stream that matches the path.
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	88
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	89 If there are no matches, this method returns an empty stream.
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	90
33 1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	91 >>> from markup.input import XML
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	92 >>> xml = XML('<root><elem><child>Text</child></elem></root>')
61 448792ab1303 Use a different namespace than Kid uses. cmlenz parents: 38 diff changeset	93
33 1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	94 >>> print Path('child').select(xml)
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	95 <child>Text</child>
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	96
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	97 >>> print Path('child/text()').select(xml)
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	98 Text
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	99
26 3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	100 @param stream: the stream to select from
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	101 @return: the substream matching the path, or an empty stream
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	102 """
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	103 stream = iter(stream)
26 3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	104 def _generate():
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	105 test = self.test()
5479aae32f5a Initial import. cmlenz parents: diff changeset	106 for kind, data, pos in stream:
5479aae32f5a Initial import. cmlenz parents: diff changeset	107 result = test(kind, data, pos)
5479aae32f5a Initial import. cmlenz parents: diff changeset	108 if result is True:
5479aae32f5a Initial import. cmlenz parents: diff changeset	109 yield kind, data, pos
5479aae32f5a Initial import. cmlenz parents: diff changeset	110 depth = 1
5479aae32f5a Initial import. cmlenz parents: diff changeset	111 while depth > 0:
73 1da51d718391 Some more performance tweaks. cmlenz parents: 70 diff changeset	112 subkind, subdata, subpos = stream.next()
1da51d718391 Some more performance tweaks. cmlenz parents: 70 diff changeset	113 if subkind is START:
1da51d718391 Some more performance tweaks. cmlenz parents: 70 diff changeset	114 depth += 1
1da51d718391 Some more performance tweaks. cmlenz parents: 70 diff changeset	115 elif subkind is END:
1da51d718391 Some more performance tweaks. cmlenz parents: 70 diff changeset	116 depth -= 1
1da51d718391 Some more performance tweaks. cmlenz parents: 70 diff changeset	117 yield subkind, subdata, subpos
1da51d718391 Some more performance tweaks. cmlenz parents: 70 diff changeset	118 test(subkind, subdata, subpos)
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	119 elif result:
5479aae32f5a Initial import. cmlenz parents: diff changeset	120 yield result
26 3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	121 return Stream(_generate())
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	122
38 ee669cb9cccc Fix for #2 (incorrect context node in path expressions). Still some paths that produce incorrect results, but the common case seems to work now. cmlenz parents: 37 diff changeset	123 def test(self, ignore_context=False):
26 3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	124 """Returns a function that can be used to track whether the path matches
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	125 a specific stream event.
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	126
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	127 The function returned expects the positional arguments `kind`, `data`,
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	128 and `pos`, i.e. basically an unpacked stream event. If the path matches
3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	129 the event, the function returns the match (for example, a `START` or
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	130 `TEXT` event.) Otherwise, it returns `None`.
33 1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	131
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	132 >>> from markup.input import XML
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	133 >>> xml = XML('<root><elem><child id="1"/></elem><child id="2"/></root>')
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	134 >>> test = Path('child').test()
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	135 >>> for kind, data, pos in xml:
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	136 ... if test(kind, data, pos):
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	137 ... print kind, data
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	138 START (u'child', [(u'id', u'1')])
1fdb7054fb42 Add doctests for path module. cmlenz parents: 27 diff changeset	139 START (u'child', [(u'id', u'2')])
26 3c1a022be04c * Split out the XPath tests into a separate `unittest`-based file. cmlenz parents: 25 diff changeset	140 """
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	141 paths = [(idx, steps, len(steps), [0])
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	142 for idx, steps in enumerate(self.paths)]
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	143
5479aae32f5a Initial import. cmlenz parents: diff changeset	144 def _test(kind, data, pos):
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	145 for idx, steps, size, stack in paths:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	146 if not stack:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	147 continue
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	148 cursor = stack[-1]
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	149
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	150 if kind is END:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	151 stack.pop()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	152 continue
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	153
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	154 elif kind is START:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	155 stack.append(cursor)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	156
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	157 matched = None
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	158 while 1:
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	159 axis, node_test, predicates = steps[cursor]
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	160
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	161 matched = node_test(kind, data, pos)
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	162 if matched and predicates:
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	163 for predicate in predicates:
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	164 if not predicate(kind, data, pos):
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	165 matched = None
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	166 break
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	167
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	168 if matched:
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	169 if cursor + 1 == size: # the last location step
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	170 if ignore_context or \
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	171 kind is not START or \
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	172 axis in (ATTRIBUTE, SELF) or \
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	173 len(stack) > 2:
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	174 return matched
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	175 else:
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	176 cursor += 1
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	177 stack[-1] = cursor
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	178
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	179 if axis is not SELF:
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	180 break
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	181
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	182 if not matched and kind is START \
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	183 and axis not in (DESCENDANT, DESCENDANT_OR_SELF):
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	184 # If this step is not a closure, it cannot be matched until
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	185 # the current element is closed... so we need to move the
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	186 # cursor back to the previous closure and retest that
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	187 # against the current element
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	188 backsteps = [step for step in steps[:cursor]
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	189 if step[0] in (DESCENDANT, DESCENDANT_OR_SELF)]
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	190 backsteps.reverse()
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	191 for axis, node_test, predicates in backsteps:
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	192 matched = node_test(kind, data, pos)
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	193 if not matched:
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	194 cursor -= 1
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	195 break
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	196 stack[-1] = cursor
1 5479aae32f5a Initial import. cmlenz parents: diff changeset	197
5479aae32f5a Initial import. cmlenz parents: diff changeset	198 return None
5479aae32f5a Initial import. cmlenz parents: diff changeset	199
5479aae32f5a Initial import. cmlenz parents: diff changeset	200 return _test
5479aae32f5a Initial import. cmlenz parents: diff changeset	201
5479aae32f5a Initial import. cmlenz parents: diff changeset	202
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	203 def _node_test_current_element():
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	204 def _node_test_current_element(kind, *_):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	205 return kind is START
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	206 return _node_test_current_element
77 f5ec6d4a61e4 * Simplify implementation of the individual XPath tests (use closures instead of callable classes) cmlenz parents: 73 diff changeset	207
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	208 def _node_test_any_child_element():
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	209 def _node_test_any_child_element(kind, *_):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	210 return kind is START
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	211 return _node_test_any_child_element
77 f5ec6d4a61e4 * Simplify implementation of the individual XPath tests (use closures instead of callable classes) cmlenz parents: 73 diff changeset	212
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	213 def _node_test_child_element_by_name(name):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	214 def _node_test_child_element_by_name(kind, data, _):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	215 return kind is START and data[0].localname == name
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	216 return _node_test_child_element_by_name
77 f5ec6d4a61e4 * Simplify implementation of the individual XPath tests (use closures instead of callable classes) cmlenz parents: 73 diff changeset	217
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	218 def _node_test_any_attribute():
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	219 def _node_test_any_attribute(kind, data, _):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	220 if kind is START and data[1]:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	221 return data[1]
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	222 return _node_test_any_attribute
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	223
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	224 def _node_test_attribute_by_name(name):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	225 def _node_test_attribute_by_name(kind, data, pos):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	226 if kind is START and name in data[1]:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	227 return TEXT, data[1].get(name), pos
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	228 return _node_test_attribute_by_name
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	229
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	230 def _function_comment():
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	231 def _function_comment(kind, data, pos):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	232 return kind is COMMENT and (kind, data, pos)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	233 return _function_comment
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	234
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	235 def _function_local_name():
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	236 def _function_local_name(kind, data, pos):
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	237 if kind is START:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	238 return TEXT, data[0].localname, pos
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	239 return kind, data, pos
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	240 return _function_local_name
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	241
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	242 def _function_name():
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	243 def _function_name(kind, data, pos):
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	244 if kind is START:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	245 return TEXT, data[0], pos
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	246 return kind, data, pos
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	247 return _function_name
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	248
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	249 def _function_namespace_uri():
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	250 def _function_namespace_uri(kind, data, pos):
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	251 if kind is START:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	252 return TEXT, data[0].namespace, pos
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	253 return kind, data, pos
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	254 return _function_namespace_uri
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	255
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	256 def _function_node():
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	257 def _function_node(kind, data, pos):
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	258 if kind is START:
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	259 return True
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	260 return kind, data, pos
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	261 return _function_node
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	262
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	263 def _function_not(expr):
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	264 def _function_not(kind, data, pos):
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	265 return not expr(kind, data, pos)
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	266 return _function_not
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	267
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	268 def _function_processing_instruction(name=None):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	269 def _function_processing_instruction(kind, data, pos):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	270 if kind is PI and (not name or data[0] == name):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	271 return (kind, data, pos)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	272 return _function_processing_instruction
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	273
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	274 def _function_text():
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	275 def _function_text(kind, data, pos):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	276 return kind is TEXT and (kind, data, pos)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	277 return _function_text
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	278
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	279 def _literal_string(text):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	280 def _literal_string(*_):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	281 return TEXT, text, (None, -1, -1)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	282 return _literal_string
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	283
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	284 def _operator_eq(lval, rval):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	285 def _operator_eq(kind, data, pos):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	286 lv = lval(kind, data, pos)
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	287 if type(lv) is tuple:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	288 lv = lv[1]
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	289 rv = rval(kind, data, pos)
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	290 if type(rv) is tuple:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	291 rv = rv[1]
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	292 return lv == rv
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	293 return _operator_eq
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	294
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	295 def _operator_neq(lval, rval):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	296 def _operator_neq(kind, data, pos):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	297 lv = lval(kind, data, pos)
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	298 if type(lv) is tuple:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	299 lv = lv[1]
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	300 rv = rval(kind, data, pos)
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	301 if type(rv) is tuple:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	302 rv = rv[1]
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	303 return lv != rv
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	304 return _operator_neq
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	305
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	306 def _operator_and(lval, rval):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	307 def _operator_and(kind, data, pos):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	308 lv = lval(kind, data, pos)
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	309 if type(lv) is tuple:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	310 lv = lv[1]
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	311 if not lv:
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	312 return False
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	313 rv = rval(kind, data, pos)
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	314 if type(rv) is tuple:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	315 rv = rv[1]
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	316 return bool(rv)
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	317 return _operator_and
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	318
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	319 def _operator_or(lval, rval):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	320 def _operator_or(kind, data, pos):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	321 lv = lval(kind, data, pos)
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	322 if type(lv) is tuple:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	323 lv = lv[1]
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	324 if lv:
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	325 return True
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	326 rv = rval(kind, data, pos)
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	327 if type(rv) is tuple:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	328 rv = rv[1]
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	329 return bool(rv)
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	330 return _operator_or
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	331
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	332
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	333 class PathSyntaxError(Exception):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	334 """Exception raised when an XPath expression is syntactically incorrect."""
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	335
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	336 def __init__(self, message, filename=None, lineno=-1, offset=-1):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	337 if filename:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	338 message = '%s (%s, line %d)' % (message, filename, lineno)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	339 Exception.__init__(self, message)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	340 self.filename = filename
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	341 self.lineno = lineno
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	342 self.offset = offset
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	343
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	344
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	345 class _PathParser(object):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	346 """Tokenizes and parses an XPath expression."""
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	347
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	348 _QUOTES = (("'", "'"), ('"', '"'))
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	349 _TOKENS = ('::', ':', '..', '.', '//', '/', '[', ']', '()', '(', ')', '@',
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	350 '=', '!=', '!', '\|', ',')
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	351 _tokenize = re.compile('(%s)\|([^%s\s]+)\|\s+' % (
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	352 '\|'.join([re.escape(t) for t in _TOKENS]),
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	353 ''.join([re.escape(t[0]) for t in _TOKENS]))).findall
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	354
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	355 def __init__(self, text):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	356 self.tokens = filter(None, [a or b for a, b in self._tokenize(text)])
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	357 self.pos = 0
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	358
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	359 # Tokenizer
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	360
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	361 at_end = property(lambda self: self.pos == len(self.tokens) - 1)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	362 cur_token = property(lambda self: self.tokens[self.pos])
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	363
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	364 def next_token(self):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	365 self.pos += 1
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	366 return self.tokens[self.pos]
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	367
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	368 def peek_token(self):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	369 if not self.at_end:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	370 return self.tokens[self.pos + 1]
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	371 return None
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	372
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	373 # Recursive descent parser
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	374
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	375 def parse(self):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	376 """Parses the XPath expression and returns a list of location path
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	377 tests.
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	378
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	379 For union expressions (such as `*\|text()`), this function returns one
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	380 test for each operand in the union. For patch expressions that don't
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	381 use the union operator, the function always returns a list of size 1.
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	382
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	383 Each path test in turn is a sequence of tests that correspond to the
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	384 location steps, each tuples of the form `(axis, testfunc, predicates)`
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	385 """
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	386 paths = [self._location_path()]
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	387 while self.cur_token == '\|':
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	388 self.next_token()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	389 paths.append(self._location_path())
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	390 if not self.at_end:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	391 raise PathSyntaxError('Unexpected token %r after end of expression'
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	392 % self.cur_token)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	393 return paths
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	394
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	395 def _location_path(self):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	396 next_is_closure = True
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	397 steps = []
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	398 while True:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	399 if self.cur_token == '//':
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	400 next_is_closure = True
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	401 self.next_token()
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	402 elif self.cur_token == '/' and not steps:
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	403 raise PathSyntaxError('Absolute location paths not supported')
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	404
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	405 axis, node_test, predicates = self._location_step()
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	406 if axis is CHILD and next_is_closure:
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	407 axis = DESCENDANT_OR_SELF
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	408 steps.append((axis, node_test, predicates))
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	409 next_is_closure = False
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	410
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	411 if self.at_end or not self.cur_token.startswith('/'):
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	412 break
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	413 self.next_token()
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	414
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	415 return steps
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	416
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	417 def _location_step(self):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	418 if self.cur_token == '@':
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	419 axis = ATTRIBUTE
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	420 self.next_token()
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	421 elif self.cur_token == '.':
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	422 axis = SELF
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	423 elif self.peek_token() == '::':
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	424 axis = Axis.forname(self.cur_token)
4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	425 if axis is None:
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	426 raise PathSyntaxError('Unsupport axis "%s"' % axis)
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	427 self.next_token()
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	428 self.next_token()
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	429 else:
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	430 axis = CHILD
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	431 node_test = self._node_test(axis)
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	432 predicates = []
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	433 while self.cur_token == '[':
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	434 predicates.append(self._predicate())
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	435 return axis, node_test, predicates
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	436
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	437 def _node_test(self, axis=None):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	438 test = None
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	439 if self.peek_token() in ('(', '()'): # Node type test
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	440 test = self._node_type()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	441
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	442 else: # Name test
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	443 if axis is ATTRIBUTE:
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	444 if self.cur_token == '*':
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	445 test = _node_test_any_attribute()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	446 else:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	447 test = _node_test_attribute_by_name(self.cur_token)
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	448 elif axis is SELF:
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	449 test = _node_test_current_element()
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	450 else:
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	451 if self.cur_token == '*':
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	452 test = _node_test_any_child_element()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	453 else:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	454 test = _node_test_child_element_by_name(self.cur_token)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	455
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	456 if not self.at_end:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	457 self.next_token()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	458 return test
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	459
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	460 def _node_type(self):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	461 name = self.cur_token
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	462 self.next_token()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	463 if name == 'comment':
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	464 return _function_comment()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	465 elif name == 'node':
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	466 return _function_node()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	467 elif name == 'processing-instruction':
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	468 args = []
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	469 if self.cur_token != '()':
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	470 # The processing-instruction() function optionally accepts the
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	471 # name of the PI as argument, which must be a literal string
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	472 self.next_token() # (
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	473 if self.cur_token != ')':
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	474 string = self.cur_token
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	475 if (string[0], string[-1]) in self._QUOTES:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	476 string = string[1:-1]
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	477 args.append(string)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	478 return _function_processing_instruction(*args)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	479 elif name == 'text':
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	480 return _function_text()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	481 else:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	482 raise PathSyntaxError('%s() not allowed here' % name)
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	483
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	484 def _predicate(self):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	485 assert self.cur_token == '['
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	486 self.next_token()
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	487 expr = self._or_expr()
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	488 if self.cur_token != ']':
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	489 raise PathSyntaxError('Expected "]" to close predicate, '
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	490 'but found "%s"' % self.cur_token)
111 2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	491 if not self.at_end:
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	492 self.next_token()
2368c3becc52 Some fixes and more unit tests for the XPath engine. cmlenz parents: 106 diff changeset	493 return expr
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	494
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	495 def _or_expr(self):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	496 expr = self._and_expr()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	497 while self.cur_token == 'or':
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	498 self.next_token()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	499 expr = _operator_or(expr, self._and_expr())
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	500 return expr
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	501
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	502 def _and_expr(self):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	503 expr = self._equality_expr()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	504 while self.cur_token == 'and':
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	505 self.next_token()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	506 expr = _operator_and(expr, self._equality_expr())
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	507 return expr
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	508
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	509 def _equality_expr(self):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	510 expr = self._primary_expr()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	511 while self.cur_token in ('=', '!='):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	512 op = {'=': _operator_eq, '!=': _operator_neq}[self.cur_token]
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	513 self.next_token()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	514 expr = op(expr, self._primary_expr())
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	515 return expr
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	516
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	517 def _primary_expr(self):
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	518 token = self.cur_token
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	519 if len(token) > 1 and (token[0], token[-1]) in self._QUOTES:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	520 self.next_token()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	521 return _literal_string(token[1:-1])
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	522 elif token[0].isdigit():
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	523 self.next_token()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	524 return _literal_number(float(token))
121 062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	525 elif not self.at_end and self.peek_token().startswith('('):
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	526 if self.next_token() == '()':
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	527 args = []
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	528 else:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	529 self.next_token()
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	530 args = [self._or_expr()]
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	531 while self.cur_token not in (',', ')'):
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	532 args.append(self._or_expr())
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	533 self.next_token()
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	534 if token == 'local-name':
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	535 return _function_local_name(*args)
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	536 elif token == 'name':
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	537 return _function_name(*args)
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	538 elif token == 'namespace-uri':
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	539 return _function_namespace_uri(*args)
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	540 elif token == 'not':
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	541 return _function_not(*args)
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	542 else:
062e51ad7b19 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`. cmlenz parents: 114 diff changeset	543 raise PathSyntaxError('Unsupported function "%s"' % token)
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	544 else:
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	545 axis = None
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	546 if token == '@':
114 4c4e81d12649 Use constants for axes in XPath engine. cmlenz parents: 111 diff changeset	547 axis = ATTRIBUTE
106 f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	548 self.next_token()
f9473bdc93b2 Complete rewrite of the XPath parsing, which was a mess before. Closes #19. cmlenz parents: 77 diff changeset	549 return self._node_test(axis)

Mercurial > genshi > mirror

annotate markup/path.py @ 121:062e51ad7b19 trunk