annotate genshi/path.py @ 333:b70563ade748

Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
author cmlenz
date Tue, 07 Nov 2006 17:50:28 +0000
parents 1dbeaef463e7
children caf7b68ab5dc 55cf81951686
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
330
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
124 if event[0] is START:
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
125 depth = 1
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
126 while depth > 0:
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
127 subevent = stream.next()
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
128 if subevent[0] is START:
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
129 depth += 1
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
130 elif subevent[0] is END:
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
131 depth -= 1
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
132 yield subevent
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
133 test(subevent, namespaces, variables,
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
134 updateonly=True)
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
135 elif result:
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
136 yield result
26
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
137 return Stream(_generate())
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
138
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
139 def test(self, ignore_context=False):
26
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
140 """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
141 a specific stream event.
039fc5b87405 * Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents: 25
diff changeset
142
305
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
143 The function returned expects the positional arguments `event`,
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
144 `namespaces` and `variables`. The first is a stream event, while the
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
145 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
146 of variable names to values, respectively. In addition, the function
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
147 accepts an `updateonly` keyword argument that default to `False`. If
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
148 it is set to `True`, the function only updates its internal state,
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
149 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
150
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
151 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
152 example, a `START` or `TEXT` event.) Otherwise, it returns `None`.
33
0e1fc0211416 Add doctests for path module.
cmlenz
parents: 27
diff changeset
153
230
24757b771651 Renamed Markup to Genshi in repository.
cmlenz
parents: 228
diff changeset
154 >>> from genshi.input import XML
33
0e1fc0211416 Add doctests for path module.
cmlenz
parents: 27
diff changeset
155 >>> xml = XML('<root><elem><child id="1"/></elem><child id="2"/></root>')
0e1fc0211416 Add doctests for path module.
cmlenz
parents: 27
diff changeset
156 >>> test = Path('child').test()
305
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
157 >>> for event in xml:
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
158 ... if test(event, {}, {}):
6e6950ac0e56 Various performance-oriented tweaks.
cmlenz
parents: 282
diff changeset
159 ... print event
326
08ada6b4b767 Fixed `__repr__` of the `QName`, `Attrs`, and `Expression` classes so that the output can be used as code to instantiate the object again.
cmlenz
parents: 306
diff changeset
160 ('START', (QName(u'child'), Attrs([(QName(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
161 """
333
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
162 paths = [(p, len(p), [0], [], [0] * len(p)) for p in [
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
163 (ignore_context and [_DOTSLASHSLASH] or []) + p for p in self.paths
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
164 ]]
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
165
306
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
166 def _test(event, namespaces, variables, updateonly=False):
330
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
167 kind, data, pos = event[:3]
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
168 retval = None
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
169 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
170 # 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
171 if kind is END:
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
172 if cursors:
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
173 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
174 continue
223
861105f3afe3 Fix typo introduced in [272].
cmlenz
parents: 217
diff changeset
175 elif kind is START:
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
176 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
177
306
3425c26d2c09 Minor optimization for XPath evaluation.
cmlenz
parents: 305
diff changeset
178 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
179 continue
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
180 cursor = cursors[-1]
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
181 depth = len(cursors)
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
182
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
183 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
184 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
185
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 ctxtnode = not ignore_context and kind is START \
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
187 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
188 matched = None
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
189 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
190 # Fetch the next location step
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
191 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
192
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
193 # 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
194 # 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
195 # 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
196 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
197 break
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
198
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
199 # 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
200 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
201
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 # Perform the actual node test
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
203 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
204
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 # 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
206 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
207
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
208 # 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
209 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
210 for predicate in predicates:
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
211 pretval = predicate(kind, data, pos, namespaces,
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
212 variables)
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
213 if type(pretval) is float:
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
214 counter[cursor] += 1
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
215 if counter[cursor] != int(pretval):
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
216 pretval = False
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
217 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
218 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
219 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
220
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
221 # 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
222 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
223 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
224 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
225 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
226 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
227 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
228 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
229 cursor += 1
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
230 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
231 cutoff[:] = []
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
232
333
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
233 if kind is START:
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
234 if last_step and not (axis is DESCENDANT or
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
235 axis is DESCENDANT_OR_SELF):
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
236 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
237
333
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
238 elif steps[cursor][0] is ATTRIBUTE:
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
239 # If the axis of the next location step is the
333
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
240 # attribute axis, we need to move on to
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
241 # processing that step without waiting for the
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
242 # next markup event
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
243 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
244
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 # 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
246 # axis isn't "self"
333
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
247 if last_step or not (axis is SELF or
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
248 axis is 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
249 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
250
333
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
251 if (retval or not matched) and kind is START and \
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
252 not (axis is DESCENDANT or axis is DESCENDANT_OR_SELF):
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
253 # 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
254 # 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
255 # cursor back to the previous closure and retest that
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
256 # against the current element
333
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
257 backsteps = [(i, k, d, p) for i, (k, d, p)
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
258 in enumerate(steps[:cursor])
215
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()
333
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
261 for cursor, axis, nodetest, predicates in backsteps:
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
262 if nodetest(kind, data, pos, namespaces, variables):
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
263 cutoff[:] = []
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
264 break
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
265 cursors[-1] = cursor
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
266
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
267 return retval
1
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
268
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
269 return _test
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
270
821114ec4f69 Initial import.
cmlenz
parents:
diff changeset
271
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
272 class PathSyntaxError(Exception):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
273 """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
274
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
275 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
276 if filename:
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
277 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
278 Exception.__init__(self, message)
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
279 self.filename = filename
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
280 self.lineno = lineno
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
281 self.offset = offset
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
282
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
283
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
284 class PathParser(object):
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
285 """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
286
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
287 _QUOTES = (("'", "'"), ('"', '"'))
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
288 _TOKENS = ('::', ':', '..', '.', '//', '/', '[', ']', '()', '(', ')', '@',
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
289 '=', '!=', '!', '|', ',', '>=', '>', '<=', '<', '$')
163
9df6f057efd3 Support for XPath number literals including decimal places.
cmlenz
parents: 162
diff changeset
290 _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
291 '|'.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
292 ''.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
293
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
294 def __init__(self, text, filename=None, lineno=-1):
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
295 self.filename = filename
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
296 self.lineno = lineno
163
9df6f057efd3 Support for XPath number literals including decimal places.
cmlenz
parents: 162
diff changeset
297 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
298 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
299 self._tokenize(text)])
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
300 self.pos = 0
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
301
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
302 # Tokenizer
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
303
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
304 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
305 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
306
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
307 def next_token(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
308 self.pos += 1
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
309 return self.tokens[self.pos]
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
310
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
311 def peek_token(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
312 if not self.at_end:
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
313 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
314 return None
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
315
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
316 # Recursive descent parser
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
317
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
318 def parse(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
319 """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
320 tests.
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
321
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
322 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
323 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
324 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
325
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
326 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
327 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
328 """
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
329 paths = [self._location_path()]
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
330 while self.cur_token == '|':
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
331 self.next_token()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
332 paths.append(self._location_path())
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
333 if not self.at_end:
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
334 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
335 % 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
336 return paths
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
337
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
338 def _location_path(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
339 steps = []
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
340 while True:
215
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
341 if self.cur_token.startswith('/'):
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
342 if self.cur_token == '//':
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
343 steps.append((DESCENDANT_OR_SELF, NodeTest(), []))
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
344 elif not steps:
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
345 raise PathSyntaxError('Absolute location paths not '
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
346 'supported', self.filename,
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
347 self.lineno)
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
348 self.next_token()
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
349
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
350 axis, nodetest, predicates = self._location_step()
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
351 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
352 axis = CHILD
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
353 steps.append((axis, nodetest, predicates))
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
354
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
355 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
356 break
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
357
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
358 return steps
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
359
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
360 def _location_step(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
361 if self.cur_token == '@':
114
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
362 axis = ATTRIBUTE
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
363 self.next_token()
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
364 elif self.cur_token == '.':
114
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
365 axis = SELF
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
366 elif self.cur_token == '..':
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
367 raise PathSyntaxError('Unsupported axis "parent"', self.filename,
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
368 self.lineno)
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
369 elif self.peek_token() == '::':
114
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
370 axis = Axis.forname(self.cur_token)
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
371 if axis is None:
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
372 raise PathSyntaxError('Unsupport axis "%s"' % axis,
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
373 self.filename, self.lineno)
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
374 self.next_token()
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
375 self.next_token()
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
376 else:
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
377 axis = None
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
378 nodetest = self._node_test(axis or CHILD)
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
379 predicates = []
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
380 while self.cur_token == '[':
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
381 predicates.append(self._predicate())
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
382 return axis, nodetest, predicates
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
383
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
384 def _node_test(self, axis=None):
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
385 test = prefix = None
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
386 next_token = self.peek_token()
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
387 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
388 test = self._node_type()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
389
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
390 elif next_token == ':': # Namespace prefix
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
391 prefix = self.cur_token
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
392 self.next_token()
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
393 localname = self.next_token()
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
394 if localname == '*':
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
395 test = QualifiedPrincipalTypeTest(axis, prefix)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
396 else:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
397 test = QualifiedNameTest(axis, prefix, localname)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
398
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
399 else: # Name test
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
400 if self.cur_token == '*':
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
401 test = PrincipalTypeTest(axis)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
402 elif self.cur_token == '.':
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
403 test = NodeTest()
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
404 else:
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
405 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
406
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
407 if not self.at_end:
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
408 self.next_token()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
409 return test
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
410
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
411 def _node_type(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
412 name = self.cur_token
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
413 self.next_token()
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
414
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
415 args = []
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
416 if self.cur_token != '()':
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
417 # The processing-instruction() function optionally accepts the
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
418 # name of the PI as argument, which must be a literal string
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
419 self.next_token() # (
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
420 if self.cur_token != ')':
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
421 string = self.cur_token
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
422 if (string[0], string[-1]) in self._QUOTES:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
423 string = string[1:-1]
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
424 args.append(string)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
425
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
426 cls = _nodetest_map.get(name)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
427 if not cls:
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
428 raise PathSyntaxError('%s() not allowed here' % name, self.filename,
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
429 self.lineno)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
430 return cls(*args)
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
431
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
432 def _predicate(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
433 assert self.cur_token == '['
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
434 self.next_token()
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
435 expr = self._or_expr()
121
22a7080ed242 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`.
cmlenz
parents: 114
diff changeset
436 if self.cur_token != ']':
22a7080ed242 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`.
cmlenz
parents: 114
diff changeset
437 raise PathSyntaxError('Expected "]" to close predicate, '
139
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
438 'but found "%s"' % self.cur_token,
54131cbb91a5 Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents: 137
diff changeset
439 self.filename, self.lineno)
111
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
440 if not self.at_end:
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
441 self.next_token()
8a4d9064f363 Some fixes and more unit tests for the XPath engine.
cmlenz
parents: 106
diff changeset
442 return expr
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
443
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
444 def _or_expr(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
445 expr = self._and_expr()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
446 while self.cur_token == 'or':
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
447 self.next_token()
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
448 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
449 return expr
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
450
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
451 def _and_expr(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
452 expr = self._equality_expr()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
453 while self.cur_token == 'and':
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
454 self.next_token()
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
455 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
456 return expr
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
457
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
458 def _equality_expr(self):
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
459 expr = self._relational_expr()
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
460 while self.cur_token in ('=', '!='):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
461 op = _operator_map[self.cur_token]
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
462 self.next_token()
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
463 expr = op(expr, self._relational_expr())
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
464 return expr
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
465
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
466 def _relational_expr(self):
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
467 expr = self._primary_expr()
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
468 while self.cur_token in ('>', '>=', '<', '>='):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
469 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
470 self.next_token()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
471 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
472 return expr
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
473
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
474 def _primary_expr(self):
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
475 token = self.cur_token
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
476 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
477 self.next_token()
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
478 return StringLiteral(token[1:-1])
163
9df6f057efd3 Support for XPath number literals including decimal places.
cmlenz
parents: 162
diff changeset
479 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
480 self.next_token()
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
481 return NumberLiteral(float(token))
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
482 elif token == '$':
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
483 token = self.next_token()
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
484 self.next_token()
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
485 return VariableReference(token)
121
22a7080ed242 Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`.
cmlenz
parents: 114
diff changeset
486 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
487 return self._function_call()
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
488 else:
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
489 axis = None
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
490 if token == '@':
114
8f53c3ad385c Use constants for axes in XPath engine.
cmlenz
parents: 111
diff changeset
491 axis = ATTRIBUTE
106
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
492 self.next_token()
61fa4cadb766 Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents: 77
diff changeset
493 return self._node_test(axis)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
494
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
495 def _function_call(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
496 name = self.cur_token
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
497 if self.next_token() == '()':
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
498 args = []
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
499 else:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
500 assert self.cur_token == '('
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
501 self.next_token()
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
502 args = [self._or_expr()]
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
503 while 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.append(self._or_expr())
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
506 if not self.cur_token == ')':
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
507 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
508 '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
509 self.filename, self.lineno)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
510 self.next_token()
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
511 cls = _function_map.get(name)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
512 if not cls:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
513 raise PathSyntaxError('Unsupported function "%s"' % name,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
514 self.filename, self.lineno)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
515 return cls(*args)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
516
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
517
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
518 # Node tests
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
519
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
520 class PrincipalTypeTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
521 """Node test that matches any event with the given principal type."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
522 __slots__ = ['principal_type']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
523 def __init__(self, principal_type):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
524 self.principal_type = principal_type
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
525 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
526 if kind is START:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
527 if self.principal_type is ATTRIBUTE:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
528 return data[1] or None
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
529 else:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
530 return True
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
531 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
532 return '*'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
533
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
534 class QualifiedPrincipalTypeTest(object):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
535 """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
536 specific namespace."""
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
537 __slots__ = ['principal_type', 'prefix']
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
538 def __init__(self, principal_type, prefix):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
539 self.principal_type = principal_type
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
540 self.prefix = prefix
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
541 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
542 namespace = Namespace(namespaces.get(self.prefix))
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
543 if kind is START:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
544 if self.principal_type is ATTRIBUTE and data[1]:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
545 return Attrs([(name, value) for name, value in data[1]
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
546 if name in namespace]) or None
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
547 else:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
548 return data[0] in namespace
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
549 def __repr__(self):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
550 return '%s:*' % self.prefix
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
551
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
552 class LocalNameTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
553 """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
554 local name.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
555 """
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
556 __slots__ = ['principal_type', 'name']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
557 def __init__(self, principal_type, name):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
558 self.principal_type = principal_type
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
559 self.name = name
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
560 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
561 if kind is START:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
562 if self.principal_type is ATTRIBUTE and self.name in data[1]:
234
4501f2fef82f * Minor simplification of XPath engine.
cmlenz
parents: 230
diff changeset
563 return data[1].get(self.name)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
564 else:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
565 return data[0].localname == self.name
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
566 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
567 return self.name
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
568
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
569 class QualifiedNameTest(object):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
570 """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
571 qualified name.
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
572 """
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
573 __slots__ = ['principal_type', 'prefix', 'name']
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
574 def __init__(self, principal_type, prefix, name):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
575 self.principal_type = principal_type
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
576 self.prefix = prefix
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
577 self.name = name
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
578 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
579 qname = QName('%s}%s' % (namespaces.get(self.prefix), self.name))
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
580 if kind is START:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
581 if self.principal_type is ATTRIBUTE and qname in data[1]:
234
4501f2fef82f * Minor simplification of XPath engine.
cmlenz
parents: 230
diff changeset
582 return data[1].get(qname)
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
583 else:
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
584 return data[0] == qname
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
585 def __repr__(self):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
586 return '%s:%s' % (self.prefix, self.name)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
587
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
588 class CommentNodeTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
589 """Node test that matches any comment events."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
590 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
591 def __call__(self, kind, data, pos, namespaces, variables):
330
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
592 return kind is COMMENT
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
593 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
594 return 'comment()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
595
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
596 class NodeTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
597 """Node test that matches any node."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
598 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
599 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
600 if kind is START:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
601 return True
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
602 return kind, data, pos
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
603 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
604 return 'node()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
605
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
606 class ProcessingInstructionNodeTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
607 """Node test that matches any processing instruction event."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
608 __slots__ = ['target']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
609 def __init__(self, target=None):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
610 self.target = target
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
611 def __call__(self, kind, data, pos, namespaces, variables):
330
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
612 return kind is PI and (not self.target or data[0] == self.target)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
613 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
614 arg = ''
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
615 if self.target:
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
616 arg = '"' + self.target + '"'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
617 return 'processing-instruction(%s)' % arg
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
618
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
619 class TextNodeTest(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
620 """Node test that matches any text event."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
621 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
622 def __call__(self, kind, data, pos, namespaces, variables):
330
1dbeaef463e7 XPath tests should never return event tuples, just values or booleans.
cmlenz
parents: 326
diff changeset
623 return kind is TEXT
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
624 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
625 return 'text()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
626
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
627 _nodetest_map = {'comment': CommentNodeTest, 'node': NodeTest,
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
628 'processing-instruction': ProcessingInstructionNodeTest,
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
629 'text': TextNodeTest}
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
630
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
631 # Functions
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
632
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
633 class Function(object):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
634 """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
635
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
636 class BooleanFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
637 """The `boolean` function, which converts its argument to a boolean
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
638 value.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
639 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
640 __slots__ = ['expr']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
641 def __init__(self, expr):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
642 self.expr = expr
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
643 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
644 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
645 return bool(val)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
646 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
647 return 'boolean(%r)' % self.expr
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
648
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
649 class CeilingFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
650 """The `ceiling` function, which returns the nearest lower integer number
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
651 for the given number.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
652 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
653 __slots__ = ['number']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
654 def __init__(self, number):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
655 self.number = number
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
656 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
657 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
658 return ceil(float(number))
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
659 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
660 return 'ceiling(%r)' % self.number
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
661
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
662 class ConcatFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
663 """The `concat` function, which concatenates (joins) the variable number of
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
664 strings it gets as arguments.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
665 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
666 __slots__ = ['exprs']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
667 def __init__(self, *exprs):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
668 self.exprs = exprs
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
669 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
670 strings = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
671 for item in [expr(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
672 for expr in self.exprs]:
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
673 strings.append(item)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
674 return u''.join(strings)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
675 def __repr__(self):
169
ce0bfdff334c Fix syntax error in `path` module.
cmlenz
parents: 164
diff changeset
676 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
677
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
678 class ContainsFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
679 """The `contains` function, which returns whether a string contains a given
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
680 substring.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
681 """
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
682 __slots__ = ['string1', 'string2']
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
683 def __init__(self, string1, string2):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
684 self.string1 = string1
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
685 self.string2 = string2
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
686 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
687 string1 = self.string1(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
688 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
689 return string2 in string1
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
690 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
691 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
692
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
693 class FalseFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
694 """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
695 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
696 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
697 return False
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
698 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
699 return 'false()'
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
700
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
701 class FloorFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
702 """The `ceiling` function, which returns the nearest higher integer number
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
703 for the given number.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
704 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
705 __slots__ = ['number']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
706 def __init__(self, number):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
707 self.number = number
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
708 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
709 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
710 return floor(float(number))
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
711 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
712 return 'floor(%r)' % self.number
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
713
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
714 class LocalNameFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
715 """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
716 element.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
717 """
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
718 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
719 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
720 if kind is START:
234
4501f2fef82f * Minor simplification of XPath engine.
cmlenz
parents: 230
diff changeset
721 return data[0].localname
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
722 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
723 return 'local-name()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
724
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
725 class NameFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
726 """The `name` function, which returns the qualified name of the current
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
727 element.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
728 """
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
729 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
730 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
731 if kind is START:
234
4501f2fef82f * Minor simplification of XPath engine.
cmlenz
parents: 230
diff changeset
732 return data[0]
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
733 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
734 return 'name()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
735
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
736 class NamespaceUriFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
737 """The `namespace-uri` function, which returns the namespace URI of the
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
738 current element.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
739 """
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
740 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
741 def __call__(self, kind, data, pos, namespaces, variables):
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
742 if kind is START:
234
4501f2fef82f * Minor simplification of XPath engine.
cmlenz
parents: 230
diff changeset
743 return data[0].namespace
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
744 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
745 return 'namespace-uri()'
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
746
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
747 class NotFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
748 """The `not` function, which returns the negated boolean value of its
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
749 argument.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
750 """
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
751 __slots__ = ['expr']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
752 def __init__(self, expr):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
753 self.expr = expr
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
754 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
755 return not self.expr(kind, data, pos, namespaces, variables)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
756 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
757 return 'not(%s)' % self.expr
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
758
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
759 class NormalizeSpaceFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
760 """The `normalize-space` function, which removes leading and trailing
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
761 whitespace in the given string, and replaces multiple adjacent whitespace
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
762 characters inside the string with a single space.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
763 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
764 __slots__ = ['expr']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
765 _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
766 def __init__(self, expr):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
767 self.expr = expr
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
768 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
769 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
770 return self._normalize(' ', string.strip())
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
771 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
772 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
773
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
774 class NumberFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
775 """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
776 __slots__ = ['expr']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
777 def __init__(self, expr):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
778 self.expr = expr
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
779 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
780 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
781 return float(val)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
782 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
783 return 'number(%r)' % self.expr
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
784
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
785 class RoundFunction(Function):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
786 """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
787 given number.
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
788 """
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
789 __slots__ = ['number']
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
790 def __init__(self, number):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
791 self.number = number
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
792 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
793 number = self.number(kind, data, pos, namespaces, variables)
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
794 return round(float(number))
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
795 def __repr__(self):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
796 return 'round(%r)' % self.number
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
797
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
798 class StartsWithFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
799 """The `starts-with` function that returns whether one string starts with
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
800 a given substring.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
801 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
802 __slots__ = ['string1', 'string2']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
803 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
804 self.string1 = string1
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
805 self.string2 = string2
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
806 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
807 string1 = self.string1(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
808 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
809 return string1.startswith(string2)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
810 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
811 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
812
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
813 class StringLengthFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
814 """The `string-length` function that returns the length of the given
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
815 string.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
816 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
817 __slots__ = ['expr']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
818 def __init__(self, expr):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
819 self.expr = expr
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
820 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
821 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
822 return len(string)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
823 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
824 return 'string-length(%r)' % self.expr
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
825
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
826 class SubstringFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
827 """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
828 at the given offset, and optionally limited to the given length.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
829 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
830 __slots__ = ['string', 'start', 'length']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
831 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
832 self.string = string
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
833 self.start = start
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
834 self.length = length
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
835 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
836 string = self.string(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
837 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
838 length = 0
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
839 if self.length is not None:
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
840 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
841 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
842 def __repr__(self):
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:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
844 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
845 self.length)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
846 else:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
847 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
848
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
849 class SubstringAfterFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
850 """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
851 is found after the given substring.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
852 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
853 __slots__ = ['string1', 'string2']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
854 def __init__(self, string1, string2):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
855 self.string1 = string1
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
856 self.string2 = string2
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
857 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
858 string1 = self.string1(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
859 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
860 index = string1.find(string2)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
861 if index >= 0:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
862 return string1[index + len(string2):]
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
863 return u''
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
864 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
865 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
866
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
867 class SubstringBeforeFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
868 """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
869 is found before the given substring.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
870 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
871 __slots__ = ['string1', 'string2']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
872 def __init__(self, string1, string2):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
873 self.string1 = string1
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
874 self.string2 = string2
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
875 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
876 string1 = self.string1(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
877 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
878 index = string1.find(string2)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
879 if index >= 0:
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
880 return string1[:index]
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
881 return u''
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
882 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
883 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
884
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
885 class TranslateFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
886 """The `translate` function that translates a set of characters in a
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
887 string to target set of characters.
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
888 """
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
889 __slots__ = ['string', 'fromchars', 'tochars']
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
890 def __init__(self, string, fromchars, tochars):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
891 self.string = string
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
892 self.fromchars = fromchars
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
893 self.tochars = tochars
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
894 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
895 string = self.string(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
896 fromchars = self.fromchars(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
897 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
898 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
899 [ord(c) for c in tochars]))
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
900 return string.translate(table)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
901 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
902 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
903 self.tochars)
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
904
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
905 class TrueFunction(Function):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
906 """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
907 __slots__ = []
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
908 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
909 return True
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
910 def __repr__(self):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
911 return 'true()'
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
912
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
913 _function_map = {'boolean': BooleanFunction, 'ceiling': CeilingFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
914 'concat': ConcatFunction, 'contains': ContainsFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
915 'false': FalseFunction, 'floor': FloorFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
916 'local-name': LocalNameFunction, 'name': NameFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
917 'namespace-uri': NamespaceUriFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
918 'normalize-space': NormalizeSpaceFunction, 'not': NotFunction,
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
919 'number': NumberFunction, 'round': RoundFunction,
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
920 'starts-with': StartsWithFunction,
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
921 'string-length': StringLengthFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
922 'substring': SubstringFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
923 'substring-after': SubstringAfterFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
924 'substring-before': SubstringBeforeFunction,
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
925 'translate': TranslateFunction, 'true': TrueFunction}
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
926
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
927 # Literals & Variables
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
928
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
929 class Literal(object):
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
930 """Abstract base class for literal nodes."""
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
931
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
932 class StringLiteral(Literal):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
933 """A string literal node."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
934 __slots__ = ['text']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
935 def __init__(self, text):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
936 self.text = text
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
937 def __call__(self, kind, data, pos, namespaces, variables):
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
938 return self.text
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
939 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
940 return '"%s"' % self.text
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
941
155
50d4b08017df * String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents: 145
diff changeset
942 class NumberLiteral(Literal):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
943 """A number literal node."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
944 __slots__ = ['number']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
945 def __init__(self, number):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
946 self.number = number
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
947 def __call__(self, kind, data, pos, namespaces, variables):
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
948 return self.number
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
949 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
950 return str(self.number)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
951
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
952 class VariableReference(Literal):
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
953 """A variable reference node."""
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
954 __slots__ = ['name']
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
955 def __init__(self, name):
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
956 self.name = name
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
957 def __call__(self, kind, data, pos, namespaces, variables):
228
f79b20a50919 Add support for position predicates in XPath expressions.
cmlenz
parents: 224
diff changeset
958 return variables.get(self.name)
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
959 def __repr__(self):
215
e92135672812 A couple of minor XPath fixes.
cmlenz
parents: 211
diff changeset
960 return str(self.name)
179
a2e0a7986d19 Implemented support for XPath variables in predicates (#31).
cmlenz
parents: 169
diff changeset
961
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
962 # Operators
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
963
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
964 class AndOperator(object):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
965 """The boolean operator `and`."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
966 __slots__ = ['lval', 'rval']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
967 def __init__(self, lval, rval):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
968 self.lval = lval
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
969 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
970 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
971 lval = self.lval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
972 if not lval:
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
973 return False
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
974 rval = self.rval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
975 return bool(rval)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
976 def __repr__(self):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
977 return '%s and %s' % (self.lval, self.rval)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
978
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
979 class EqualsOperator(object):
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
980 """The equality operator `=`."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
981 __slots__ = ['lval', 'rval']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
982 def __init__(self, lval, rval):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
983 self.lval = lval
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
984 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
985 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
986 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
987 rval = self.rval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
988 return lval == rval
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
989 def __repr__(self):
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
990 return '%s=%s' % (self.lval, self.rval)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
991
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
992 class NotEqualsOperator(object):
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
993 """The equality operator `!=`."""
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
994 __slots__ = ['lval', 'rval']
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
995 def __init__(self, lval, rval):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
996 self.lval = lval
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
997 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
998 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
999 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1000 rval = self.rval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1001 return lval != rval
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1002 def __repr__(self):
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1003 return '%s!=%s' % (self.lval, self.rval)
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1004
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1005 class OrOperator(object):
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1006 """The boolean operator `or`."""
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1007 __slots__ = ['lval', 'rval']
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1008 def __init__(self, lval, rval):
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1009 self.lval = lval
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1010 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1011 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1012 lval = self.lval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1013 if lval:
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
1014 return True
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1015 rval = self.rval(kind, data, pos, namespaces, variables)
161
a25f9fc5787d Various docstring additions and other cosmetic changes.
cmlenz
parents: 156
diff changeset
1016 return bool(rval)
137
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
1017 def __repr__(self):
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
1018 return '%s or %s' % (self.lval, self.rval)
38ddb21b6fa4 Further cleanup of XPath engine.
cmlenz
parents: 122
diff changeset
1019
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1020 class GreaterThanOperator(object):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1021 """The relational operator `>` (greater than)."""
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1022 __slots__ = ['lval', 'rval']
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1023 def __init__(self, lval, rval):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1024 self.lval = lval
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1025 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1026 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1027 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1028 rval = self.rval(kind, data, pos, namespaces, variables)
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1029 return float(lval) > float(rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1030 def __repr__(self):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1031 return '%s>%s' % (self.lval, self.rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1032
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1033 class GreaterThanOrEqualOperator(object):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1034 """The relational operator `>=` (greater than or equal)."""
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1035 __slots__ = ['lval', 'rval']
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1036 def __init__(self, lval, rval):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1037 self.lval = lval
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1038 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1039 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1040 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1041 rval = self.rval(kind, data, pos, namespaces, variables)
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1042 return float(lval) >= float(rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1043 def __repr__(self):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1044 return '%s>=%s' % (self.lval, self.rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1045
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1046 class LessThanOperator(object):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1047 """The relational operator `<` (less than)."""
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1048 __slots__ = ['lval', 'rval']
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1049 def __init__(self, lval, rval):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1050 self.lval = lval
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1051 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1052 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1053 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1054 rval = self.rval(kind, data, pos, namespaces, variables)
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1055 return float(lval) < float(rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1056 def __repr__(self):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1057 return '%s<%s' % (self.lval, self.rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1058
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1059 class LessThanOrEqualOperator(object):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1060 """The relational operator `<=` (less than or equal)."""
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1061 __slots__ = ['lval', 'rval']
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1062 def __init__(self, lval, rval):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1063 self.lval = lval
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1064 self.rval = rval
224
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1065 def __call__(self, kind, data, pos, namespaces, variables):
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1066 lval = self.lval(kind, data, pos, namespaces, variables)
e4dad1145f84 Implement support for namespace prefixes in XPath expressions.
cmlenz
parents: 223
diff changeset
1067 rval = self.rval(kind, data, pos, namespaces, variables)
162
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1068 return float(lval) <= float(rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1069 def __repr__(self):
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1070 return '%s<=%s' % (self.lval, self.rval)
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1071
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1072 _operator_map = {'=': EqualsOperator, '!=': NotEqualsOperator,
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1073 '>': GreaterThanOperator, '>=': GreaterThanOrEqualOperator,
f767cf98e3e3 Implement the XPath relational operators and the `round()` function.
cmlenz
parents: 161
diff changeset
1074 '<': LessThanOperator, '>=': LessThanOrEqualOperator}
333
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
1075
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
1076
b70563ade748 Fix XPath traversal in match templates. Previously, `div/p` would be treated the same as `div//p`, i.e. it would match all descendants and not just the immediate children.
cmlenz
parents: 330
diff changeset
1077 _DOTSLASHSLASH = (DESCENDANT_OR_SELF, PrincipalTypeTest(None), ())
Copyright (C) 2012-2017 Edgewall Software