annotate genshi/path.py @ 306:3425c26d2c09

Minor optimization for XPath evaluation.
author cmlenz
date Mon, 16 Oct 2006 16:41:06 +0000
parents 6e6950ac0e56
children 08ada6b4b767 ca91b55eb6df
rev   line source
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
1 # -*- coding: utf-8 -*-
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
2 #
66
822089ae65ce Switch copyright to Edgewall and URLs to markup.edgewall.org.
cmlenz
parents: 61
diff changeset
3 # Copyright (C) 2006 Edgewall Software
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
4 # All rights reserved.
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
5 #
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
6 # This software is licensed as described in the file COPYING, which
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
7 # you should have received as part of this distribution. The terms
230
24757b771651 Renamed Markup to Genshi in repository.
cmlenz
parents: 228
diff changeset
8 # are also available at http://genshi.edgewall.org/wiki/License.
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
9 #
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
10 # This software consists of voluntary contributions made by many
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
11 # individuals. For the exact contribution history, see the revision
230
24757b771651 Renamed Markup to Genshi in repository.
cmlenz
parents: 228
diff changeset
12 # history and logs, available at http://genshi.edgewall.org/log/.
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
13
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
14 """Basic support for evaluating XPath expressions against streams.
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
15
230
24757b771651 Renamed Markup to Genshi in repository.
cmlenz
parents: 228
diff changeset
16 >>> from genshi.input import XML
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
17 >>> doc = XML('''<doc>
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
18 ... <items count="2">
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
19 ... <item status="new">
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
20 ... <summary>Foo</summary>
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
21 ... </item>
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
22 ... <item status="closed">
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
23 ... <summary>Bar</summary>
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
24 ... </item>
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
25 ... </items>
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
26 ... </doc>''')
8a4d9064f363 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()')
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
28 Bar
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
29
8a4d9064f363 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
8a4d9064f363 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.
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
32 """
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
33
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
34 from math import ceil, floor
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
35 import re
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
36
230
24757b771651 Renamed Markup to Genshi in repository.
cmlenz
parents: 228
diff changeset
37 from genshi.core import Stream, Attrs, Namespace, QName
24757b771651 Renamed Markup to Genshi in repository.
cmlenz
parents: 228
diff changeset
38 from genshi.core import START, END, TEXT, COMMENT, PI
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
39
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
40 __all__ = ['Path', 'PathSyntaxError']
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
41
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
42
114
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
43 class Axis(object):
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
44 """Defines constants for the various supported XPath axes."""
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
45
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
46 ATTRIBUTE = 'attribute'
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
47 CHILD = 'child'
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
48 DESCENDANT = 'descendant'
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
49 DESCENDANT_OR_SELF = 'descendant-or-self'
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
50 SELF = 'self'
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
51
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
52 def forname(cls, name):
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
53 """Return the axis constant for the given name, or `None` if no such
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
54 axis was defined.
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
55 """
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
56 return getattr(cls, name.upper().replace('-', '_'), None)
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
57 forname = classmethod(forname)
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
58
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
59
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
60 ATTRIBUTE = Axis.ATTRIBUTE
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
61 CHILD = Axis.CHILD
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
62 DESCENDANT = Axis.DESCENDANT
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
63 DESCENDANT_OR_SELF = Axis.DESCENDANT_OR_SELF
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
64 SELF = Axis.SELF
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
65
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
66
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
67 class Path(object):
26
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
68 """Implements basic XPath support on streams.
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
69
26
039fc5b87405 * 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
039fc5b87405 * 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
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
72 substream matching that path.
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
73 """
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
74
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
75 def __init__(self, text, filename=None, lineno=-1):
26
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
76 """Create the path object from a string.
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
77
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
78 @param text: the path expression
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
79 """
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
80 self.source = text
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
81 self.paths = PathParser(text, filename, lineno).parse()
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
82
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
83 def __repr__(self):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
84 paths = []
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
85 for path in self.paths:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
86 steps = []
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
87 for axis, nodetest, predicates in path:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
88 steps.append('%s::%s' % (axis, nodetest))
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
89 for predicate in predicates:
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
90 steps[-1] += '[%s]' % predicate
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
91 paths.append('/'.join(steps))
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
92 return '<%s "%s">' % (self.__class__.__name__, '|'.join(paths))
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
93
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
94 def select(self, stream, namespaces=None, variables=None):
26
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
95 """Returns a substream of the given stream that matches the path.
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
96
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
97 If there are no matches, this method returns an empty stream.
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
98
230
24757b771651 Renamed Markup to Genshi in repository.
cmlenz
parents: 228
diff changeset
99 >>> from genshi.input import XML
33
0e1fc0211416 Add doctests for path module.
cmlenz
parents: 27
diff changeset
100 >>> xml = XML('<root><elem><child>Text</child></elem></root>')
61
33c2702cf6da Use a different namespace than Kid uses.
cmlenz
parents: 38
diff changeset
101
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
102 >>> print Path('.//child').select(xml)
33
0e1fc0211416 Add doctests for path module.
cmlenz
parents: 27
diff changeset
103 <child>Text</child>
0e1fc0211416 Add doctests for path module.
cmlenz
parents: 27
diff changeset
104
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
105 >>> print Path('.//child/text()').select(xml)
33
0e1fc0211416 Add doctests for path module.
cmlenz
parents: 27
diff changeset
106 Text
0e1fc0211416 Add doctests for path module.
cmlenz
parents: 27
diff changeset
107
26
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
108 @param stream: the stream to select from
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
109 @param namespaces: (optional) a mapping of namespace prefixes to URIs
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
110 @param variables: (optional) a mapping of variable names to values
26
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
111 @return: the substream matching the path, or an empty stream
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
112 """
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
113 if namespaces is None:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
114 namespaces = {}
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
115 if variables is None:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
116 variables = {}
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
117 stream = iter(stream)
26
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
118 def _generate():
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
119 test = self.test()
305
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
120 for event in stream:
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
121 result = test(event, namespaces, variables)
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
122 if result is True:
305
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
123 yield event
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
124 depth = 1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
125 while depth > 0:
305
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
126 subevent = stream.next()
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
127 if subevent[0] is START:
73
b0fd16111f2e Some more performance tweaks.
cmlenz
parents: 70
diff changeset
128 depth += 1
305
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
129 elif subevent[0] is END:
73
b0fd16111f2e Some more performance tweaks.
cmlenz
parents: 70
diff changeset
130 depth -= 1
305
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
131 yield subevent
306
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
132 test(subevent, namespaces, variables, updateonly=True)
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
133 elif result:
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
134 yield result
26
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
135 return Stream(_generate())
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
136
38
fec9f4897415 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
137 def test(self, ignore_context=False):
26
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
138 """Returns a function that can be used to track whether the path matches
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
139 a specific stream event.
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
140
305
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
141 The function returned expects the positional arguments `event`,
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
142 `namespaces` and `variables`. The first is a stream event, while the
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
143 latter two are a mapping of namespace prefixes to URIs, and a mapping
306
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
144 of variable names to values, respectively. In addition, the function
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
145 accepts an `updateonly` keyword argument that default to `False`. If
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
146 it is set to `True`, the function only updates its internal state,
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
147 but does not perform any tests or return a result.
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
148
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
149 If the path matches the event, the function returns the match (for
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
150 example, a `START` or `TEXT` event.) Otherwise, it returns `None`.
33
0e1fc0211416 Add doctests for path module.
cmlenz
parents: 27
diff changeset
151
230
24757b771651 Renamed Markup to Genshi in repository.
cmlenz
parents: 228
diff changeset
152 >>> from genshi.input import XML
33
0e1fc0211416 Add doctests for path module.
cmlenz
parents: 27
diff changeset
153 >>> xml = XML('<root><elem><child id="1"/></elem><child id="2"/></root>')
0e1fc0211416 Add doctests for path module.
cmlenz
parents: 27
diff changeset
154 >>> test = Path('child').test()
305
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
155 >>> for event in xml:
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
156 ... if test(event, {}, {}):
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
157 ... print event
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
158 ('START', (u'child', [(u'id', u'2')]), (None, 1, 34))
26
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
159 """
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
160 paths = [(p, len(p), [0], [], [0] * len(p)) for p in self.paths]
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
161
306
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
162 def _test(event, namespaces, variables, updateonly=False):
305
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
163 kind, data, pos = event
259
6f11ad260890 Fix bug in evaluating XPath expressions using the union operator `|`, which caused any path but the first to get out of sync with the event stream, and the whole thing returning too few results.
cmlenz
parents: 250
diff changeset
164 retval = None
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
165 for steps, size, cursors, cutoff, counter in paths:
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
166 # Manage the stack that tells us "where we are" in the stream
211
0a14c2a06be3 Fix another regression introduced in [258]: some kinds of cascaded match templates were broken, for example in the TurboGears example app.
cmlenz
parents: 179
diff changeset
167 if kind is END:
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
168 if cursors:
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
169 cursors.pop()
211
0a14c2a06be3 Fix another regression introduced in [258]: some kinds of cascaded match templates were broken, for example in the TurboGears example app.
cmlenz
parents: 179
diff changeset
170 continue
223
861105f3afe3 Fix typo introduced in [272].
cmlenz
parents: 217
diff changeset
171 elif kind is START:
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
172 cursors.append(cursors and cursors[-1] or 0)
259
6f11ad260890 Fix bug in evaluating XPath expressions using the union operator `|`, which caused any path but the first to get out of sync with the event stream, and the whole thing returning too few results.
cmlenz
parents: 250
diff changeset
173
306
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
174 if updateonly or retval or not cursors:
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
175 continue
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
176 cursor = cursors[-1]
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
177 depth = len(cursors)
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
178
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
179 if cutoff and depth + int(kind is not START) > cutoff[0]:
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
180 continue
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
181
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
182 ctxtnode = not ignore_context and kind is START \
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
183 and depth == 2
259
6f11ad260890 Fix bug in evaluating XPath expressions using the union operator `|`, which caused any path but the first to get out of sync with the event stream, and the whole thing returning too few results.
cmlenz
parents: 250
diff changeset
184 matched = None
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
185 while 1:
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
186 # Fetch the next location step
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
187 axis, nodetest, predicates = steps[cursor]
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
188
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
189 # If this is the start event for the context node, and the
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
190 # axis of the location step doesn't include the current
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
191 # element, skip the test
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
192 if ctxtnode and (axis is CHILD or axis is DESCENDANT):
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
193 break
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
194
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
195 # Is this the last step of the location path?
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
196 last_step = cursor + 1 == size
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
197
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
198 # Perform the actual node test
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
199 matched = nodetest(kind, data, pos, namespaces, variables)
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
200
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
201 # The node test matched
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
202 if matched:
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
203
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
204 # Check all the predicates for this step
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
205 if predicates:
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
206 for predicate in predicates:
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
207 pretval = predicate(kind, data, pos, namespaces,
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
208 variables)
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
209 if type(pretval) is float:
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
210 counter[cursor] += 1
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
211 if counter[cursor] != int(pretval):
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
212 pretval = False
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
213 if not pretval:
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
214 matched = None
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
215 break
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
217 # Both the node test and the predicates matched
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
218 if matched:
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
219 if last_step:
217
d8b195b22a44 Fix `py:match` directive which would screw up in some scenarios due to incorrect handling of the substream. Closes #49.
cmlenz
parents: 216
diff changeset
220 if not ctxtnode or kind is not START \
d8b195b22a44 Fix `py:match` directive which would screw up in some scenarios due to incorrect handling of the substream. Closes #49.
cmlenz
parents: 216
diff changeset
221 or axis is ATTRIBUTE or axis is SELF:
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
222 retval = matched
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
223 elif not ctxtnode or axis is SELF \
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
224 or axis is DESCENDANT_OR_SELF:
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
225 cursor += 1
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
226 cursors[-1] = cursor
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
227 cutoff[:] = []
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
228
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
229 elif not ignore_context and kind is START:
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
230 cutoff[:] = [depth]
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
231
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
232 if last_step and not ignore_context and kind is START:
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
233 if (axis is not DESCENDANT and
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
234 axis is not DESCENDANT_OR_SELF):
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
235 cutoff[:] = [depth]
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
236
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
237 if kind is START and not last_step:
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
238 next_axis = steps[cursor][0]
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
239 if next_axis is ATTRIBUTE:
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
240 # If the axis of the next location step is the
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
241 # attribute axis, we need to move on to processing
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
242 # that step without waiting for the next markup
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
243 # event
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
244 continue
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
245
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
246 # We're done with this step if it's the last step or the
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
247 # axis isn't "self"
217
d8b195b22a44 Fix `py:match` directive which would screw up in some scenarios due to incorrect handling of the substream. Closes #49.
cmlenz
parents: 216
diff changeset
248 if last_step or (axis is not SELF and
d8b195b22a44 Fix `py:match` directive which would screw up in some scenarios due to incorrect handling of the substream. Closes #49.
cmlenz
parents: 216
diff changeset
249 axis is not DESCENDANT_OR_SELF):
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
250 break
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
251
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
252 if kind is START and axis is not DESCENDANT \
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
253 and axis is not DESCENDANT_OR_SELF:
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
254 # If this step is not a closure, it cannot be matched until
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
255 # the current element is closed... so we need to move the
114
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
256 # cursor back to the previous closure and retest that
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
257 # against the current element
215
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
258 backsteps = [(k, d, p) for k, d, p in steps[:cursor]
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
259 if k is DESCENDANT or k is DESCENDANT_OR_SELF]
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
260 backsteps.reverse()
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
261 for axis, nodetest, predicates in backsteps:
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
262 matched = nodetest(kind, data, pos, namespaces,
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
263 variables)
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
264 if not matched:
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
265 cursor -= 1
216
0a01371cecbc Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents: 215
diff changeset
266 cutoff[:] = []
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
267 break
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
268 cursors[-1] = cursor
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
269
259
6f11ad260890 Fix bug in evaluating XPath expressions using the union operator `|`, which caused any path but the first to get out of sync with the event stream, and the whole thing returning too few results.
cmlenz
parents: 250
diff changeset
270 return retval
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
271
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
272 return _test
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
273
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
274
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
275 class PathSyntaxError(Exception):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
276 """Exception raised when an XPath expression is syntactically incorrect."""
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
277
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
278 def __init__(self, message, filename=None, lineno=-1, offset=-1):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
279 if filename:
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
280 message = '%s (%s, line %d)' % (message, filename, lineno)
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
281 Exception.__init__(self, message)
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
282 self.filename = filename
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
283 self.lineno = lineno
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
284 self.offset = offset
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
285
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
286
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
287 class PathParser(object):
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
288 """Tokenizes and parses an XPath expression."""
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
289
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
290 _QUOTES = (("'", "'"), ('"', '"'))
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
291 _TOKENS = ('::', ':', '..', '.', '//', '/', '[', ']', '()', '(', ')', '@',
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
292 '=', '!=', '!', '|', ',', '>=', '>', '<=', '<', '$')
163
9df6f057efd3 Support for XPath number literals including decimal places.
cmlenz
parents: 162
diff changeset
293 _tokenize = re.compile('("[^"]*")|(\'[^\']*\')|((?:\d+)?\.\d+)|(%s)|([^%s\s]+)|\s+' % (
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
294 '|'.join([re.escape(t) for t in _TOKENS]),
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
295 ''.join([re.escape(t[0]) for t in _TOKENS]))).findall
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
296
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
297 def __init__(self, text, filename=None, lineno=-1):
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
298 self.filename = filename
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
299 self.lineno = lineno
163
9df6f057efd3 Support for XPath number literals including decimal places.
cmlenz
parents: 162
diff changeset
300 self.tokens = filter(None, [dqstr or sqstr or number or token or name
9df6f057efd3 Support for XPath number literals including decimal places.
cmlenz
parents: 162
diff changeset
301 for dqstr, sqstr, number, token, name in
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
302 self._tokenize(text)])
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
303 self.pos = 0
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
304
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
305 # Tokenizer
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
306
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
307 at_end = property(lambda self: self.pos == len(self.tokens) - 1)
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
308 cur_token = property(lambda self: self.tokens[self.pos])
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
309
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
310 def next_token(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
311 self.pos += 1
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
312 return self.tokens[self.pos]
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
313
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
314 def peek_token(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
315 if not self.at_end:
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
316 return self.tokens[self.pos + 1]
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
317 return None
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
318
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
319 # Recursive descent parser
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
320
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
321 def parse(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
322 """Parses the XPath expression and returns a list of location path
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
323 tests.
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
324
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
325 For union expressions (such as `*|text()`), this function returns one
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
326 test for each operand in the union. For patch expressions that don't
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
327 use the union operator, the function always returns a list of size 1.
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
328
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
329 Each path test in turn is a sequence of tests that correspond to the
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
330 location steps, each tuples of the form `(axis, testfunc, predicates)`
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
331 """
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
332 paths = [self._location_path()]
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
333 while self.cur_token == '|':
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
334 self.next_token()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
335 paths.append(self._location_path())
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
336 if not self.at_end:
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
337 raise PathSyntaxError('Unexpected token %r after end of expression'
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
338 % self.cur_token, self.filename, self.lineno)
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
339 return paths
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
340
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
341 def _location_path(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
342 steps = []
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
343 while True:
215
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
344 if self.cur_token.startswith('/'):
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
345 if self.cur_token == '//':
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
346 steps.append((DESCENDANT_OR_SELF, NodeTest(), []))
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
347 elif not steps:
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
348 raise PathSyntaxError('Absolute location paths not '
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
349 'supported', self.filename,
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
350 self.lineno)
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
351 self.next_token()
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
352
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
353 axis, nodetest, predicates = self._location_step()
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
354 if not axis:
145
56d534eb53f9 * Fix error in expression evaluation when the expression evaluates to an iterable that does not produce event tuples.
cmlenz
parents: 139
diff changeset
355 axis = CHILD
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
356 steps.append((axis, nodetest, predicates))
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
357
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
358 if self.at_end or not self.cur_token.startswith('/'):
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
359 break
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
360
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
361 return steps
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
362
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
363 def _location_step(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
364 if self.cur_token == '@':
114
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
365 axis = ATTRIBUTE
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
366 self.next_token()
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
367 elif self.cur_token == '.':
114
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
368 axis = SELF
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
369 elif self.cur_token == '..':
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
370 raise PathSyntaxError('Unsupported axis "parent"', self.filename,
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
371 self.lineno)
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
372 elif self.peek_token() == '::':
114
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
373 axis = Axis.forname(self.cur_token)
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
374 if axis is None:
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
375 raise PathSyntaxError('Unsupport axis "%s"' % axis,
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
376 self.filename, self.lineno)
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
377 self.next_token()
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
378 self.next_token()
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
379 else:
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
380 axis = None
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
381 nodetest = self._node_test(axis or CHILD)
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
382 predicates = []
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
383 while self.cur_token == '[':
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
384 predicates.append(self._predicate())
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
385 return axis, nodetest, predicates
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
386
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
387 def _node_test(self, axis=None):
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
388 test = prefix = None
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
389 next_token = self.peek_token()
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
390 if next_token in ('(', '()'): # Node type test
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
391 test = self._node_type()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
392
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
393 elif next_token == ':': # Namespace prefix
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
394 prefix = self.cur_token
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
395 self.next_token()
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
396 localname = self.next_token()
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
397 if localname == '*':
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
398 test = QualifiedPrincipalTypeTest(axis, prefix)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
399 else:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
400 test = QualifiedNameTest(axis, prefix, localname)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
401
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
402 else: # Name test
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
403 if self.cur_token == '*':
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
404 test = PrincipalTypeTest(axis)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
405 elif self.cur_token == '.':
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
406 test = NodeTest()
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
407 else:
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
408 test = LocalNameTest(axis, self.cur_token)
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
409
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
410 if not self.at_end:
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
411 self.next_token()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
412 return test
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
413
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
414 def _node_type(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
415 name = self.cur_token
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
416 self.next_token()
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
417
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
418 args = []
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
419 if self.cur_token != '()':
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
420 # The processing-instruction() function optionally accepts the
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
421 # name of the PI as argument, which must be a literal string
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
422 self.next_token() # (
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
423 if self.cur_token != ')':
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
424 string = self.cur_token
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
425 if (string[0], string[-1]) in self._QUOTES:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
426 string = string[1:-1]
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
427 args.append(string)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
428
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
429 cls = _nodetest_map.get(name)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
430 if not cls:
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
431 raise PathSyntaxError('%s() not allowed here' % name, self.filename,
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
432 self.lineno)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
433 return cls(*args)
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
434
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
435 def _predicate(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
436 assert self.cur_token == '['
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
437 self.next_token()
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
438 expr = self._or_expr()
121
22a7080ed242 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`.
cmlenz
parents: 114
diff changeset
439 if self.cur_token != ']':
22a7080ed242 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`.
cmlenz
parents: 114
diff changeset
440 raise PathSyntaxError('Expected "]" to close predicate, '
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
441 'but found "%s"' % self.cur_token,
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
442 self.filename, self.lineno)
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
443 if not self.at_end:
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
444 self.next_token()
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
445 return expr
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
446
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
447 def _or_expr(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
448 expr = self._and_expr()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
449 while self.cur_token == 'or':
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
450 self.next_token()
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
451 expr = OrOperator(expr, self._and_expr())
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
452 return expr
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
453
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
454 def _and_expr(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
455 expr = self._equality_expr()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
456 while self.cur_token == 'and':
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
457 self.next_token()
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
458 expr = AndOperator(expr, self._equality_expr())
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
459 return expr
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
460
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
461 def _equality_expr(self):
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
462 expr = self._relational_expr()
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
463 while self.cur_token in ('=', '!='):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
464 op = _operator_map[self.cur_token]
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
465 self.next_token()
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
466 expr = op(expr, self._relational_expr())
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
467 return expr
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
468
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
469 def _relational_expr(self):
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
470 expr = self._primary_expr()
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
471 while self.cur_token in ('>', '>=', '<', '>='):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
472 op = _operator_map[self.cur_token]
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
473 self.next_token()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
474 expr = op(expr, self._primary_expr())
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
475 return expr
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
476
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
477 def _primary_expr(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
478 token = self.cur_token
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
479 if len(token) > 1 and (token[0], token[-1]) in self._QUOTES:
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
480 self.next_token()
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
481 return StringLiteral(token[1:-1])
163
9df6f057efd3 Support for XPath number literals including decimal places.
cmlenz
parents: 162
diff changeset
482 elif token[0].isdigit() or token[0] == '.':
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
483 self.next_token()
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
484 return NumberLiteral(float(token))
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
485 elif token == '$':
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
486 token = self.next_token()
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
487 self.next_token()
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
488 return VariableReference(token)
121
22a7080ed242 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`.
cmlenz
parents: 114
diff changeset
489 elif not self.at_end and self.peek_token().startswith('('):
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
490 return self._function_call()
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
491 else:
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
492 axis = None
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
493 if token == '@':
114
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
494 axis = ATTRIBUTE
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
495 self.next_token()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
496 return self._node_test(axis)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
497
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
498 def _function_call(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
499 name = self.cur_token
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
500 if self.next_token() == '()':
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
501 args = []
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
502 else:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
503 assert self.cur_token == '('
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
504 self.next_token()
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
505 args = [self._or_expr()]
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
506 while self.cur_token == ',':
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
507 self.next_token()
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
508 args.append(self._or_expr())
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
509 if not self.cur_token == ')':
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
510 raise PathSyntaxError('Expected ")" to close function argument '
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
511 'list, but found "%s"' % self.cur_token,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
512 self.filename, self.lineno)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
513 self.next_token()
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
514 cls = _function_map.get(name)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
515 if not cls:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
516 raise PathSyntaxError('Unsupported function "%s"' % name,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
517 self.filename, self.lineno)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
518 return cls(*args)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
519
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
520
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
521 # Node tests
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
522
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
523 class PrincipalTypeTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
524 """Node test that matches any event with the given principal type."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
525 __slots__ = ['principal_type']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
526 def __init__(self, principal_type):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
527 self.principal_type = principal_type
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
528 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
529 if kind is START:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
530 if self.principal_type is ATTRIBUTE:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
531 return data[1] or None
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
532 else:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
533 return True
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
534 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
535 return '*'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
536
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
537 class QualifiedPrincipalTypeTest(object):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
538 """Node test that matches any event with the given principal type in a
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
539 specific namespace."""
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
540 __slots__ = ['principal_type', 'prefix']
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
541 def __init__(self, principal_type, prefix):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
542 self.principal_type = principal_type
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
543 self.prefix = prefix
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
544 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
545 namespace = Namespace(namespaces.get(self.prefix))
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
546 if kind is START:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
547 if self.principal_type is ATTRIBUTE and data[1]:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
548 return Attrs([(name, value) for name, value in data[1]
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
549 if name in namespace]) or None
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
550 else:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
551 return data[0] in namespace
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
552 def __repr__(self):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
553 return '%s:*' % self.prefix
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
554
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
555 class LocalNameTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
556 """Node test that matches any event with the given prinipal type and
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
557 local name.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
558 """
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
559 __slots__ = ['principal_type', 'name']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
560 def __init__(self, principal_type, name):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
561 self.principal_type = principal_type
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
562 self.name = name
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
563 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
564 if kind is START:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
565 if self.principal_type is ATTRIBUTE and self.name in data[1]:
234
4501f2fef82f * Minor simplification of XPath engine.
cmlenz
parents: 230
diff changeset
566 return data[1].get(self.name)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
567 else:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
568 return data[0].localname == self.name
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
569 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
570 return self.name
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
571
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
572 class QualifiedNameTest(object):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
573 """Node test that matches any event with the given prinipal type and
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
574 qualified name.
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
575 """
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
576 __slots__ = ['principal_type', 'prefix', 'name']
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
577 def __init__(self, principal_type, prefix, name):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
578 self.principal_type = principal_type
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
579 self.prefix = prefix
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
580 self.name = name
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
581 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
582 qname = QName('%s}%s' % (namespaces.get(self.prefix), self.name))
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
583 if kind is START:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
584 if self.principal_type is ATTRIBUTE and qname in data[1]:
234
4501f2fef82f * Minor simplification of XPath engine.
cmlenz
parents: 230
diff changeset
585 return data[1].get(qname)
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
586 else:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
587 return data[0] == qname
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
588 def __repr__(self):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
589 return '%s:%s' % (self.prefix, self.name)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
590
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
591 class CommentNodeTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
592 """Node test that matches any comment events."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
593 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
594 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
595 return kind is COMMENT and (kind, data, pos)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
596 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
597 return 'comment()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
598
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
599 class NodeTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
600 """Node test that matches any node."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
601 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
602 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
603 if kind is START:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
604 return True
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
605 return kind, data, pos
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
606 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
607 return 'node()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
608
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
609 class ProcessingInstructionNodeTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
610 """Node test that matches any processing instruction event."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
611 __slots__ = ['target']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
612 def __init__(self, target=None):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
613 self.target = target
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
614 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
615 if kind is PI and (not self.target or data[0] == self.target):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
616 return (kind, data, pos)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
617 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
618 arg = ''
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
619 if self.target:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
620 arg = '"' + self.target + '"'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
621 return 'processing-instruction(%s)' % arg
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
622
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
623 class TextNodeTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
624 """Node test that matches any text event."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
625 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
626 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
627 return kind is TEXT and (kind, data, pos)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
628 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
629 return 'text()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
630
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
631 _nodetest_map = {'comment': CommentNodeTest, 'node': NodeTest,
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
632 'processing-instruction': ProcessingInstructionNodeTest,
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
633 'text': TextNodeTest}
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
634
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
635 # Functions
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
636
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
637 class Function(object):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
638 """Base class for function nodes in XPath expressions."""
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
639
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
640 class BooleanFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
641 """The `boolean` function, which converts its argument to a boolean
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
642 value.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
643 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
644 __slots__ = ['expr']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
645 def __init__(self, expr):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
646 self.expr = expr
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
647 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
648 val = self.expr(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
649 return bool(val)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
650 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
651 return 'boolean(%r)' % self.expr
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
652
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
653 class CeilingFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
654 """The `ceiling` function, which returns the nearest lower integer number
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
655 for the given number.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
656 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
657 __slots__ = ['number']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
658 def __init__(self, number):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
659 self.number = number
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
660 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
661 number = self.number(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
662 return ceil(float(number))
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
663 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
664 return 'ceiling(%r)' % self.number
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
665
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
666 class ConcatFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
667 """The `concat` function, which concatenates (joins) the variable number of
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
668 strings it gets as arguments.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
669 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
670 __slots__ = ['exprs']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
671 def __init__(self, *exprs):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
672 self.exprs = exprs
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
673 def __call__(self, kind, data, pos, namespaces, variables):
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
674 strings = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
675 for item in [expr(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
676 for expr in self.exprs]:
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
677 strings.append(item)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
678 return u''.join(strings)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
679 def __repr__(self):
169
ce0bfdff334c Fix syntax error in `path` module.
cmlenz
parents: 164
diff changeset
680 return 'concat(%s)' % ', '.join([repr(expr) for expr in self.exprs])
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
681
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
682 class ContainsFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
683 """The `contains` function, which returns whether a string contains a given
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
684 substring.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
685 """
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
686 __slots__ = ['string1', 'string2']
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
687 def __init__(self, string1, string2):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
688 self.string1 = string1
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
689 self.string2 = string2
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
690 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
691 string1 = self.string1(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
692 string2 = self.string2(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
693 return string2 in string1
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
694 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
695 return 'contains(%r, %r)' % (self.string1, self.string2)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
696
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
697 class FalseFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
698 """The `false` function, which always returns the boolean `false` value."""
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
699 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
700 def __call__(self, kind, data, pos, namespaces, variables):
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
701 return False
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
702 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
703 return 'false()'
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
704
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
705 class FloorFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
706 """The `ceiling` function, which returns the nearest higher integer number
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
707 for the given number.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
708 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
709 __slots__ = ['number']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
710 def __init__(self, number):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
711 self.number = number
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
712 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
713 number = self.number(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
714 return floor(float(number))
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
715 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
716 return 'floor(%r)' % self.number
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
717
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
718 class LocalNameFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
719 """The `local-name` function, which returns the local name of the current
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
720 element.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
721 """
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
722 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
723 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
724 if kind is START:
234
4501f2fef82f * Minor simplification of XPath engine.
cmlenz
parents: 230
diff changeset
725 return data[0].localname
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
726 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
727 return 'local-name()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
728
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
729 class NameFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
730 """The `name` function, which returns the qualified name of the current
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
731 element.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
732 """
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
733 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
734 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
735 if kind is START:
234
4501f2fef82f * Minor simplification of XPath engine.
cmlenz
parents: 230
diff changeset
736 return data[0]
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
737 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
738 return 'name()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
739
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
740 class NamespaceUriFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
741 """The `namespace-uri` function, which returns the namespace URI of the
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
742 current element.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
743 """
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
744 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
745 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
746 if kind is START:
234
4501f2fef82f * Minor simplification of XPath engine.
cmlenz
parents: 230
diff changeset
747 return data[0].namespace
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
748 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
749 return 'namespace-uri()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
750
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
751 class NotFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
752 """The `not` function, which returns the negated boolean value of its
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
753 argument.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
754 """
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
755 __slots__ = ['expr']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
756 def __init__(self, expr):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
757 self.expr = expr
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
758 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
759 return not self.expr(kind, data, pos, namespaces, variables)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
760 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
761 return 'not(%s)' % self.expr
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
762
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
763 class NormalizeSpaceFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
764 """The `normalize-space` function, which removes leading and trailing
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
765 whitespace in the given string, and replaces multiple adjacent whitespace
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
766 characters inside the string with a single space.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
767 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
768 __slots__ = ['expr']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
769 _normalize = re.compile(r'\s{2,}').sub
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
770 def __init__(self, expr):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
771 self.expr = expr
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
772 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
773 string = self.expr(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
774 return self._normalize(' ', string.strip())
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
775 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
776 return 'normalize-space(%s)' % repr(self.expr)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
777
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
778 class NumberFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
779 """The `number` function that converts its argument to a number."""
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
780 __slots__ = ['expr']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
781 def __init__(self, expr):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
782 self.expr = expr
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
783 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
784 val = self.expr(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
785 return float(val)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
786 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
787 return 'number(%r)' % self.expr
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
788
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
789 class RoundFunction(Function):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
790 """The `round` function, which returns the nearest integer number for the
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
791 given number.
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
792 """
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
793 __slots__ = ['number']
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
794 def __init__(self, number):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
795 self.number = number
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
796 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
797 number = self.number(kind, data, pos, namespaces, variables)
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
798 return round(float(number))
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
799 def __repr__(self):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
800 return 'round(%r)' % self.number
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
801
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
802 class StartsWithFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
803 """The `starts-with` function that returns whether one string starts with
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
804 a given substring.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
805 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
806 __slots__ = ['string1', 'string2']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
807 def __init__(self, string1, string2):
282
db5bdeca9192 Fix `starts-with()` XPath function so that it actually compares the two strings. Closes #61.
cmlenz
parents: 259
diff changeset
808 self.string1 = string1
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
809 self.string2 = string2
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
810 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
811 string1 = self.string1(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
812 string2 = self.string2(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
813 return string1.startswith(string2)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
814 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
815 return 'starts-with(%r, %r)' % (self.string1, self.string2)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
816
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
817 class StringLengthFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
818 """The `string-length` function that returns the length of the given
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
819 string.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
820 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
821 __slots__ = ['expr']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
822 def __init__(self, expr):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
823 self.expr = expr
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
824 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
825 string = self.expr(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
826 return len(string)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
827 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
828 return 'string-length(%r)' % self.expr
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
829
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
830 class SubstringFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
831 """The `substring` function that returns the part of a string that starts
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
832 at the given offset, and optionally limited to the given length.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
833 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
834 __slots__ = ['string', 'start', 'length']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
835 def __init__(self, string, start, length=None):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
836 self.string = string
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
837 self.start = start
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
838 self.length = length
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
839 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
840 string = self.string(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
841 start = self.start(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
842 length = 0
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
843 if self.length is not None:
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
844 length = self.length(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
845 return string[int(start):len(string) - int(length)]
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
846 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
847 if self.length is not None:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
848 return 'substring(%r, %r, %r)' % (self.string, self.start,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
849 self.length)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
850 else:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
851 return 'substring(%r, %r)' % (self.string, self.start)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
852
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
853 class SubstringAfterFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
854 """The `substring-after` function that returns the part of a string that
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
855 is found after the given substring.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
856 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
857 __slots__ = ['string1', 'string2']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
858 def __init__(self, string1, string2):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
859 self.string1 = string1
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
860 self.string2 = string2
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
861 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
862 string1 = self.string1(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
863 string2 = self.string2(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
864 index = string1.find(string2)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
865 if index >= 0:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
866 return string1[index + len(string2):]
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
867 return u''
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
868 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
869 return 'substring-after(%r, %r)' % (self.string1, self.string2)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
870
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
871 class SubstringBeforeFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
872 """The `substring-before` function that returns the part of a string that
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
873 is found before the given substring.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
874 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
875 __slots__ = ['string1', 'string2']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
876 def __init__(self, string1, string2):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
877 self.string1 = string1
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
878 self.string2 = string2
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
879 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
880 string1 = self.string1(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
881 string2 = self.string2(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
882 index = string1.find(string2)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
883 if index >= 0:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
884 return string1[:index]
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
885 return u''
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
886 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
887 return 'substring-after(%r, %r)' % (self.string1, self.string2)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
888
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
889 class TranslateFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
890 """The `translate` function that translates a set of characters in a
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
891 string to target set of characters.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
892 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
893 __slots__ = ['string', 'fromchars', 'tochars']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
894 def __init__(self, string, fromchars, tochars):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
895 self.string = string
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
896 self.fromchars = fromchars
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
897 self.tochars = tochars
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
898 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
899 string = self.string(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
900 fromchars = self.fromchars(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
901 tochars = self.tochars(kind, data, pos, namespaces, variables)
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
902 table = dict(zip([ord(c) for c in fromchars],
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
903 [ord(c) for c in tochars]))
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
904 return string.translate(table)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
905 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
906 return 'translate(%r, %r, %r)' % (self.string, self.fromchars,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
907 self.tochars)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
908
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
909 class TrueFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
910 """The `true` function, which always returns the boolean `true` value."""
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
911 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
912 def __call__(self, kind, data, pos, namespaces, variables):
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
913 return True
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
914 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
915 return 'true()'
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
916
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
917 _function_map = {'boolean': BooleanFunction, 'ceiling': CeilingFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
918 'concat': ConcatFunction, 'contains': ContainsFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
919 'false': FalseFunction, 'floor': FloorFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
920 'local-name': LocalNameFunction, 'name': NameFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
921 'namespace-uri': NamespaceUriFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
922 'normalize-space': NormalizeSpaceFunction, 'not': NotFunction,
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
923 'number': NumberFunction, 'round': RoundFunction,
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
924 'starts-with': StartsWithFunction,
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
925 'string-length': StringLengthFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
926 'substring': SubstringFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
927 'substring-after': SubstringAfterFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
928 'substring-before': SubstringBeforeFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
929 'translate': TranslateFunction, 'true': TrueFunction}
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
930
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
931 # Literals & Variables
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
932
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
933 class Literal(object):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
934 """Abstract base class for literal nodes."""
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
935
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
936 class StringLiteral(Literal):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
937 """A string literal node."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
938 __slots__ = ['text']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
939 def __init__(self, text):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
940 self.text = text
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
941 def __call__(self, kind, data, pos, namespaces, variables):
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
942 return self.text
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
943 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
944 return '"%s"' % self.text
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
945
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
946 class NumberLiteral(Literal):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
947 """A number literal node."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
948 __slots__ = ['number']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
949 def __init__(self, number):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
950 self.number = number
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
951 def __call__(self, kind, data, pos, namespaces, variables):
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
952 return self.number
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
953 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
954 return str(self.number)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
955
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
956 class VariableReference(Literal):
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
957 """A variable reference node."""
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
958 __slots__ = ['name']
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
959 def __init__(self, name):
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
960 self.name = name
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
961 def __call__(self, kind, data, pos, namespaces, variables):
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
962 return variables.get(self.name)
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
963 def __repr__(self):
215
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
964 return str(self.name)
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
965
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
966 # Operators
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
967
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
968 class AndOperator(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
969 """The boolean operator `and`."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
970 __slots__ = ['lval', 'rval']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
971 def __init__(self, lval, rval):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
972 self.lval = lval
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
973 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
974 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
975 lval = self.lval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
976 if not lval:
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
977 return False
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
978 rval = self.rval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
979 return bool(rval)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
980 def __repr__(self):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
981 return '%s and %s' % (self.lval, self.rval)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
982
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
983 class EqualsOperator(object):
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
984 """The equality operator `=`."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
985 __slots__ = ['lval', 'rval']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
986 def __init__(self, lval, rval):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
987 self.lval = lval
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
988 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
989 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
990 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
991 rval = self.rval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
992 return lval == rval
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
993 def __repr__(self):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
994 return '%s=%s' % (self.lval, self.rval)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
995
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
996 class NotEqualsOperator(object):
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
997 """The equality operator `!=`."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
998 __slots__ = ['lval', 'rval']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
999 def __init__(self, lval, rval):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
1000 self.lval = lval
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
1001 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1002 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1003 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1004 rval = self.rval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1005 return lval != rval
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1006 def __repr__(self):
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1007 return '%s!=%s' % (self.lval, self.rval)
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1008
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1009 class OrOperator(object):
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1010 """The boolean operator `or`."""
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1011 __slots__ = ['lval', 'rval']
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1012 def __init__(self, lval, rval):
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1013 self.lval = lval
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1014 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1015 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1016 lval = self.lval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1017 if lval:
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
1018 return True
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1019 rval = self.rval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1020 return bool(rval)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
1021 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
1022 return '%s or %s' % (self.lval, self.rval)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
1023
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1024 class GreaterThanOperator(object):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1025 """The relational operator `>` (greater than)."""
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1026 __slots__ = ['lval', 'rval']
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1027 def __init__(self, lval, rval):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1028 self.lval = lval
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1029 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1030 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1031 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1032 rval = self.rval(kind, data, pos, namespaces, variables)
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1033 return float(lval) > float(rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1034 def __repr__(self):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1035 return '%s>%s' % (self.lval, self.rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1036
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1037 class GreaterThanOrEqualOperator(object):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1038 """The relational operator `>=` (greater than or equal)."""
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1039 __slots__ = ['lval', 'rval']
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1040 def __init__(self, lval, rval):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1041 self.lval = lval
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1042 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1043 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1044 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1045 rval = self.rval(kind, data, pos, namespaces, variables)
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1046 return float(lval) >= float(rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1047 def __repr__(self):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1048 return '%s>=%s' % (self.lval, self.rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1049
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1050 class LessThanOperator(object):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1051 """The relational operator `<` (less than)."""
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1052 __slots__ = ['lval', 'rval']
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1053 def __init__(self, lval, rval):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1054 self.lval = lval
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1055 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1056 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1057 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1058 rval = self.rval(kind, data, pos, namespaces, variables)
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1059 return float(lval) < float(rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1060 def __repr__(self):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1061 return '%s<%s' % (self.lval, self.rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1062
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1063 class LessThanOrEqualOperator(object):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1064 """The relational operator `<=` (less than or equal)."""
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1065 __slots__ = ['lval', 'rval']
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1066 def __init__(self, lval, rval):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1067 self.lval = lval
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1068 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1069 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1070 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1071 rval = self.rval(kind, data, pos, namespaces, variables)
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1072 return float(lval) <= float(rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1073 def __repr__(self):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1074 return '%s<=%s' % (self.lval, self.rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1075
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1076 _operator_map = {'=': EqualsOperator, '!=': NotEqualsOperator,
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1077 '>': GreaterThanOperator, '>=': GreaterThanOrEqualOperator,
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1078 '<': LessThanOperator, '>=': LessThanOrEqualOperator}
Copyright (C) 2012-2017 Edgewall Software