Mercurial > genshi > mirror
annotate genshi/path.py @ 719:4bc6741b2811 trunk
Fix copyright years.
author | cmlenz |
---|---|
date | Thu, 10 Apr 2008 19:47:27 +0000 |
parents | 1f73e534e715 |
children | 7fa3af78556f 25b8ffa7fe65 |
rev | line source |
---|---|
1 | 1 # -*- coding: utf-8 -*- |
2 # | |
719 | 3 # Copyright (C) 2006-2008 Edgewall Software |
1 | 4 # All rights reserved. |
5 # | |
6 # This software is licensed as described in the file COPYING, which | |
7 # you should have received as part of this distribution. The terms | |
230 | 8 # are also available at http://genshi.edgewall.org/wiki/License. |
1 | 9 # |
10 # This software consists of voluntary contributions made by many | |
11 # individuals. For the exact contribution history, see the revision | |
230 | 12 # history and logs, available at http://genshi.edgewall.org/log/. |
1 | 13 |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
14 """Basic support for evaluating XPath expressions against streams. |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
15 |
230 | 16 >>> from genshi.input import XML |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
17 >>> doc = XML('''<doc> |
516 | 18 ... <items count="4"> |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
19 ... <item status="new"> |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
20 ... <summary>Foo</summary> |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
21 ... </item> |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
22 ... <item status="closed"> |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
23 ... <summary>Bar</summary> |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
24 ... </item> |
516 | 25 ... <item status="closed" resolution="invalid"> |
26 ... <summary>Baz</summary> | |
27 ... </item> | |
28 ... <item status="closed" resolution="fixed"> | |
29 ... <summary>Waz</summary> | |
30 ... </item> | |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
31 ... </items> |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
32 ... </doc>''') |
516 | 33 >>> print doc.select('items/item[@status="closed" and ' |
34 ... '(@resolution="invalid" or not(@resolution))]/summary/text()') | |
35 BarBaz | |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
36 |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
37 Because the XPath engine operates on markup streams (as opposed to tree |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
38 structures), it only implements a subset of the full XPath 1.0 language. |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
39 """ |
1 | 40 |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
41 from math import ceil, floor |
593 | 42 import operator |
1 | 43 import re |
44 | |
230 | 45 from genshi.core import Stream, Attrs, Namespace, QName |
638
1f73e534e715
Fix for descendant-or-self XPath patterns when namespaces are involved.
cmlenz
parents:
605
diff
changeset
|
46 from genshi.core import START, END, TEXT, START_NS, END_NS, COMMENT, PI, \ |
1f73e534e715
Fix for descendant-or-self XPath patterns when namespaces are involved.
cmlenz
parents:
605
diff
changeset
|
47 START_CDATA, END_CDATA |
1 | 48 |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
49 __all__ = ['Path', 'PathSyntaxError'] |
425
073640758a42
Try to use proper reStructuredText for docstrings throughout.
cmlenz
parents:
386
diff
changeset
|
50 __docformat__ = 'restructuredtext en' |
1 | 51 |
52 | |
114 | 53 class Axis(object): |
54 """Defines constants for the various supported XPath axes.""" | |
55 | |
56 ATTRIBUTE = 'attribute' | |
57 CHILD = 'child' | |
58 DESCENDANT = 'descendant' | |
59 DESCENDANT_OR_SELF = 'descendant-or-self' | |
60 SELF = 'self' | |
61 | |
62 def forname(cls, name): | |
63 """Return the axis constant for the given name, or `None` if no such | |
64 axis was defined. | |
65 """ | |
66 return getattr(cls, name.upper().replace('-', '_'), None) | |
67 forname = classmethod(forname) | |
68 | |
69 | |
70 ATTRIBUTE = Axis.ATTRIBUTE | |
71 CHILD = Axis.CHILD | |
72 DESCENDANT = Axis.DESCENDANT | |
73 DESCENDANT_OR_SELF = Axis.DESCENDANT_OR_SELF | |
74 SELF = Axis.SELF | |
75 | |
76 | |
1 | 77 class Path(object): |
26
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
78 """Implements basic XPath support on streams. |
1 | 79 |
26
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
80 Instances of this class represent a "compiled" XPath expression, and provide |
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
81 methods for testing the path against a stream, as well as extracting a |
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
82 substream matching that path. |
1 | 83 """ |
84 | |
139
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
85 def __init__(self, text, filename=None, lineno=-1): |
26
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
86 """Create the path object from a string. |
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
87 |
425
073640758a42
Try to use proper reStructuredText for docstrings throughout.
cmlenz
parents:
386
diff
changeset
|
88 :param text: the path expression |
498 | 89 :param filename: the name of the file in which the path expression was |
90 found (used in error messages) | |
91 :param lineno: the line on which the expression was found | |
26
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
92 """ |
1 | 93 self.source = text |
139
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
94 self.paths = PathParser(text, filename, lineno).parse() |
1 | 95 |
96 def __repr__(self): | |
137 | 97 paths = [] |
98 for path in self.paths: | |
99 steps = [] | |
100 for axis, nodetest, predicates in path: | |
101 steps.append('%s::%s' % (axis, nodetest)) | |
102 for predicate in predicates: | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
103 steps[-1] += '[%s]' % predicate |
137 | 104 paths.append('/'.join(steps)) |
105 return '<%s "%s">' % (self.__class__.__name__, '|'.join(paths)) | |
1 | 106 |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
107 def select(self, stream, namespaces=None, variables=None): |
26
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
108 """Returns a substream of the given stream that matches the path. |
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
109 |
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
110 If there are no matches, this method returns an empty stream. |
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
111 |
230 | 112 >>> from genshi.input import XML |
33 | 113 >>> xml = XML('<root><elem><child>Text</child></elem></root>') |
61 | 114 |
216
636fe6766b4d
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
|
115 >>> print Path('.//child').select(xml) |
33 | 116 <child>Text</child> |
117 | |
216
636fe6766b4d
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
|
118 >>> print Path('.//child/text()').select(xml) |
33 | 119 Text |
120 | |
425
073640758a42
Try to use proper reStructuredText for docstrings throughout.
cmlenz
parents:
386
diff
changeset
|
121 :param stream: the stream to select from |
073640758a42
Try to use proper reStructuredText for docstrings throughout.
cmlenz
parents:
386
diff
changeset
|
122 :param namespaces: (optional) a mapping of namespace prefixes to URIs |
073640758a42
Try to use proper reStructuredText for docstrings throughout.
cmlenz
parents:
386
diff
changeset
|
123 :param variables: (optional) a mapping of variable names to values |
073640758a42
Try to use proper reStructuredText for docstrings throughout.
cmlenz
parents:
386
diff
changeset
|
124 :return: the substream matching the path, or an empty stream |
498 | 125 :rtype: `Stream` |
26
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
126 """ |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
127 if namespaces is None: |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
128 namespaces = {} |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
129 if variables is None: |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
130 variables = {} |
1 | 131 stream = iter(stream) |
26
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
132 def _generate(): |
1 | 133 test = self.test() |
305 | 134 for event in stream: |
135 result = test(event, namespaces, variables) | |
1 | 136 if result is True: |
305 | 137 yield event |
330
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
138 if event[0] is START: |
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
139 depth = 1 |
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
140 while depth > 0: |
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
141 subevent = stream.next() |
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
142 if subevent[0] is START: |
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
143 depth += 1 |
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
144 elif subevent[0] is END: |
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
145 depth -= 1 |
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
146 yield subevent |
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
147 test(subevent, namespaces, variables, |
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
148 updateonly=True) |
1 | 149 elif result: |
150 yield result | |
605
d0345c64da65
Text templates now default to rendering as plain text; it is no longer necessary to explicitly specify the "text" method to the `render()` or `serialize()` method of the generated markup stream. See tickets #62 and #118.
cmlenz
parents:
593
diff
changeset
|
151 return Stream(_generate(), |
d0345c64da65
Text templates now default to rendering as plain text; it is no longer necessary to explicitly specify the "text" method to the `render()` or `serialize()` method of the generated markup stream. See tickets #62 and #118.
cmlenz
parents:
593
diff
changeset
|
152 serializer=getattr(stream, 'serializer', None)) |
1 | 153 |
38
ee669cb9cccc
Fix for #2 (incorrect context node in path expressions). Still some paths that produce incorrect results, but the common case seems to work now.
cmlenz
parents:
37
diff
changeset
|
154 def test(self, ignore_context=False): |
26
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
155 """Returns a function that can be used to track whether the path matches |
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
156 a specific stream event. |
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
157 |
425
073640758a42
Try to use proper reStructuredText for docstrings throughout.
cmlenz
parents:
386
diff
changeset
|
158 The function returned expects the positional arguments ``event``, |
073640758a42
Try to use proper reStructuredText for docstrings throughout.
cmlenz
parents:
386
diff
changeset
|
159 ``namespaces`` and ``variables``. The first is a stream event, while the |
305 | 160 latter two are a mapping of namespace prefixes to URIs, and a mapping |
306 | 161 of variable names to values, respectively. In addition, the function |
425
073640758a42
Try to use proper reStructuredText for docstrings throughout.
cmlenz
parents:
386
diff
changeset
|
162 accepts an ``updateonly`` keyword argument that default to ``False``. If |
073640758a42
Try to use proper reStructuredText for docstrings throughout.
cmlenz
parents:
386
diff
changeset
|
163 it is set to ``True``, the function only updates its internal state, |
306 | 164 but does not perform any tests or return a result. |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
165 |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
166 If the path matches the event, the function returns the match (for |
425
073640758a42
Try to use proper reStructuredText for docstrings throughout.
cmlenz
parents:
386
diff
changeset
|
167 example, a `START` or `TEXT` event.) Otherwise, it returns ``None``. |
33 | 168 |
230 | 169 >>> from genshi.input import XML |
33 | 170 >>> xml = XML('<root><elem><child id="1"/></elem><child id="2"/></root>') |
171 >>> test = Path('child').test() | |
305 | 172 >>> for event in xml: |
173 ... if test(event, {}, {}): | |
386 | 174 ... print event[0], repr(event[1]) |
175 START (QName(u'child'), Attrs([(QName(u'id'), u'2')])) | |
498 | 176 |
177 :param ignore_context: if `True`, the path is interpreted like a pattern | |
178 in XSLT, meaning for example that it will match | |
179 at any depth | |
180 :return: a function that can be used to test individual events in a | |
181 stream against the path | |
182 :rtype: ``function`` | |
26
3c1a022be04c
* Split out the XPath tests into a separate `unittest`-based file.
cmlenz
parents:
25
diff
changeset
|
183 """ |
333
e5ba8c5a30fe
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
|
184 paths = [(p, len(p), [0], [], [0] * len(p)) for p in [ |
e5ba8c5a30fe
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
|
185 (ignore_context and [_DOTSLASHSLASH] or []) + p for p in self.paths |
e5ba8c5a30fe
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
|
186 ]] |
1 | 187 |
306 | 188 def _test(event, namespaces, variables, updateonly=False): |
330
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
189 kind, data, pos = event[:3] |
259
fe8dbe9066c1
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
|
190 retval = None |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
191 for steps, size, cursors, cutoff, counter in paths: |
216
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
192 # Manage the stack that tells us "where we are" in the stream |
211
e5151983df0d
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
|
193 if kind is END: |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
194 if cursors: |
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
195 cursors.pop() |
211
e5151983df0d
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
|
196 continue |
223 | 197 elif kind is START: |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
198 cursors.append(cursors and cursors[-1] or 0) |
638
1f73e534e715
Fix for descendant-or-self XPath patterns when namespaces are involved.
cmlenz
parents:
605
diff
changeset
|
199 elif kind is START_NS or kind is END_NS \ |
1f73e534e715
Fix for descendant-or-self XPath patterns when namespaces are involved.
cmlenz
parents:
605
diff
changeset
|
200 or kind is START_CDATA or kind is END_CDATA: |
1f73e534e715
Fix for descendant-or-self XPath patterns when namespaces are involved.
cmlenz
parents:
605
diff
changeset
|
201 continue |
259
fe8dbe9066c1
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
|
202 |
306 | 203 if updateonly or retval or not cursors: |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
204 continue |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
205 cursor = cursors[-1] |
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
206 depth = len(cursors) |
1 | 207 |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
208 if cutoff and depth + int(kind is not START) > cutoff[0]: |
216
636fe6766b4d
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 continue |
636fe6766b4d
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 |
636fe6766b4d
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
|
211 ctxtnode = not ignore_context and kind is START \ |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
212 and depth == 2 |
259
fe8dbe9066c1
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
|
213 matched = None |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
214 while 1: |
216
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
215 # Fetch the next location step |
137 | 216 axis, nodetest, predicates = steps[cursor] |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
217 |
216
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
218 # If this is the start event for the context node, and the |
636fe6766b4d
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 # axis of the location step doesn't include the current |
636fe6766b4d
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 # element, skip the test |
636fe6766b4d
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 if ctxtnode and (axis is CHILD or axis is DESCENDANT): |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
222 break |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
223 |
216
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
224 # Is this the last step of the location path? |
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
225 last_step = cursor + 1 == size |
636fe6766b4d
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 |
636fe6766b4d
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 # Perform the actual node test |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
228 matched = nodetest(kind, data, pos, namespaces, variables) |
216
636fe6766b4d
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 |
636fe6766b4d
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
|
230 # The node test matched |
636fe6766b4d
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 if matched: |
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
232 |
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
233 # Check all the predicates for this step |
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
234 if predicates: |
636fe6766b4d
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
|
235 for predicate in predicates: |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
236 pretval = predicate(kind, data, pos, namespaces, |
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
237 variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
238 if type(pretval) is float: # FIXME <- need to |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
239 # check this for |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
240 # other types that |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
241 # can be coerced to |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
242 # float |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
243 counter[cursor] += 1 |
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
244 if counter[cursor] != int(pretval): |
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
245 pretval = False |
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
246 if not pretval: |
216
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
247 matched = None |
636fe6766b4d
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
|
248 break |
636fe6766b4d
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 |
636fe6766b4d
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 # Both the node test and the predicates matched |
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
251 if matched: |
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
252 if last_step: |
217
f150cff4da18
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
|
253 if not ctxtnode or kind is not START \ |
f150cff4da18
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
|
254 or axis is ATTRIBUTE or axis is SELF: |
216
636fe6766b4d
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
|
255 retval = matched |
636fe6766b4d
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
|
256 elif not ctxtnode or axis is SELF \ |
636fe6766b4d
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
|
257 or axis is DESCENDANT_OR_SELF: |
636fe6766b4d
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
|
258 cursor += 1 |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
259 cursors[-1] = cursor |
216
636fe6766b4d
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
|
260 cutoff[:] = [] |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
261 |
333
e5ba8c5a30fe
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 kind is START: |
e5ba8c5a30fe
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 if last_step and not (axis is DESCENDANT or |
e5ba8c5a30fe
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 axis is DESCENDANT_OR_SELF): |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
265 cutoff[:] = [depth] |
216
636fe6766b4d
Many fixes to XPath evaluation. Among other things, this should get rid of the bug that attributes were getting ?pulled up? by `py:match` directives using `py:attrs="select('@*')"` (see #50).
cmlenz
parents:
215
diff
changeset
|
266 |
333
e5ba8c5a30fe
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
|
267 elif steps[cursor][0] is ATTRIBUTE: |
216
636fe6766b4d
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
|
268 # If the axis of the next location step is the |
363
37e4b4bb0b53
Parse template includes at parse time to avoid some runtime overhead.
cmlenz
parents:
333
diff
changeset
|
269 # attribute axis, we need to move on to processing |
37e4b4bb0b53
Parse template includes at parse time to avoid some runtime overhead.
cmlenz
parents:
333
diff
changeset
|
270 # that step without waiting for the next markup |
37e4b4bb0b53
Parse template includes at parse time to avoid some runtime overhead.
cmlenz
parents:
333
diff
changeset
|
271 # event |
216
636fe6766b4d
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
|
272 continue |
636fe6766b4d
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
|
273 |
636fe6766b4d
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
|
274 # We're done with this step if it's the last step or the |
636fe6766b4d
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
|
275 # axis isn't "self" |
384 | 276 if not matched or last_step or not ( |
277 axis is SELF or axis is DESCENDANT_OR_SELF): | |
216
636fe6766b4d
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
|
278 break |
638
1f73e534e715
Fix for descendant-or-self XPath patterns when namespaces are involved.
cmlenz
parents:
605
diff
changeset
|
279 if ctxtnode and axis is DESCENDANT_OR_SELF: |
1f73e534e715
Fix for descendant-or-self XPath patterns when namespaces are involved.
cmlenz
parents:
605
diff
changeset
|
280 ctxtnode = False |
216
636fe6766b4d
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
|
281 |
333
e5ba8c5a30fe
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
|
282 if (retval or not matched) and kind is START and \ |
e5ba8c5a30fe
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
|
283 not (axis is DESCENDANT or axis is DESCENDANT_OR_SELF): |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
284 # If this step is not a closure, it cannot be matched until |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
285 # the current element is closed... so we need to move the |
114 | 286 # cursor back to the previous closure and retest that |
287 # against the current element | |
333
e5ba8c5a30fe
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
|
288 backsteps = [(i, k, d, p) for i, (k, d, p) |
e5ba8c5a30fe
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
|
289 in enumerate(steps[:cursor]) |
215 | 290 if k is DESCENDANT or k is DESCENDANT_OR_SELF] |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
291 backsteps.reverse() |
333
e5ba8c5a30fe
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
|
292 for cursor, axis, nodetest, predicates in backsteps: |
e5ba8c5a30fe
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
|
293 if nodetest(kind, data, pos, namespaces, variables): |
e5ba8c5a30fe
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
|
294 cutoff[:] = [] |
e5ba8c5a30fe
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
|
295 break |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
296 cursors[-1] = cursor |
1 | 297 |
259
fe8dbe9066c1
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
|
298 return retval |
1 | 299 |
300 return _test | |
301 | |
302 | |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
303 class PathSyntaxError(Exception): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
304 """Exception raised when an XPath expression is syntactically incorrect.""" |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
305 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
306 def __init__(self, message, filename=None, lineno=-1, offset=-1): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
307 if filename: |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
308 message = '%s (%s, line %d)' % (message, filename, lineno) |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
309 Exception.__init__(self, message) |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
310 self.filename = filename |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
311 self.lineno = lineno |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
312 self.offset = offset |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
313 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
314 |
137 | 315 class PathParser(object): |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
316 """Tokenizes and parses an XPath expression.""" |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
317 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
318 _QUOTES = (("'", "'"), ('"', '"')) |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
319 _TOKENS = ('::', ':', '..', '.', '//', '/', '[', ']', '()', '(', ')', '@', |
179
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
320 '=', '!=', '!', '|', ',', '>=', '>', '<=', '<', '$') |
163
9c023c395e44
Support for XPath number literals including decimal places.
cmlenz
parents:
162
diff
changeset
|
321 _tokenize = re.compile('("[^"]*")|(\'[^\']*\')|((?:\d+)?\.\d+)|(%s)|([^%s\s]+)|\s+' % ( |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
322 '|'.join([re.escape(t) for t in _TOKENS]), |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
323 ''.join([re.escape(t[0]) for t in _TOKENS]))).findall |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
324 |
139
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
325 def __init__(self, text, filename=None, lineno=-1): |
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
326 self.filename = filename |
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
327 self.lineno = lineno |
163
9c023c395e44
Support for XPath number literals including decimal places.
cmlenz
parents:
162
diff
changeset
|
328 self.tokens = filter(None, [dqstr or sqstr or number or token or name |
9c023c395e44
Support for XPath number literals including decimal places.
cmlenz
parents:
162
diff
changeset
|
329 for dqstr, sqstr, number, token, name in |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
330 self._tokenize(text)]) |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
331 self.pos = 0 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
332 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
333 # Tokenizer |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
334 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
335 at_end = property(lambda self: self.pos == len(self.tokens) - 1) |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
336 cur_token = property(lambda self: self.tokens[self.pos]) |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
337 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
338 def next_token(self): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
339 self.pos += 1 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
340 return self.tokens[self.pos] |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
341 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
342 def peek_token(self): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
343 if not self.at_end: |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
344 return self.tokens[self.pos + 1] |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
345 return None |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
346 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
347 # Recursive descent parser |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
348 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
349 def parse(self): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
350 """Parses the XPath expression and returns a list of location path |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
351 tests. |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
352 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
353 For union expressions (such as `*|text()`), this function returns one |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
354 test for each operand in the union. For patch expressions that don't |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
355 use the union operator, the function always returns a list of size 1. |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
356 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
357 Each path test in turn is a sequence of tests that correspond to the |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
358 location steps, each tuples of the form `(axis, testfunc, predicates)` |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
359 """ |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
360 paths = [self._location_path()] |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
361 while self.cur_token == '|': |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
362 self.next_token() |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
363 paths.append(self._location_path()) |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
364 if not self.at_end: |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
365 raise PathSyntaxError('Unexpected token %r after end of expression' |
139
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
366 % self.cur_token, self.filename, self.lineno) |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
367 return paths |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
368 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
369 def _location_path(self): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
370 steps = [] |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
371 while True: |
215 | 372 if self.cur_token.startswith('/'): |
373 if self.cur_token == '//': | |
374 steps.append((DESCENDANT_OR_SELF, NodeTest(), [])) | |
375 elif not steps: | |
376 raise PathSyntaxError('Absolute location paths not ' | |
377 'supported', self.filename, | |
378 self.lineno) | |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
379 self.next_token() |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
380 |
137 | 381 axis, nodetest, predicates = self._location_step() |
382 if not axis: | |
145
47bbd9d2a5af
* Fix error in expression evaluation when the expression evaluates to an iterable that does not produce event tuples.
cmlenz
parents:
139
diff
changeset
|
383 axis = CHILD |
137 | 384 steps.append((axis, nodetest, predicates)) |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
385 |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
386 if self.at_end or not self.cur_token.startswith('/'): |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
387 break |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
388 |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
389 return steps |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
390 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
391 def _location_step(self): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
392 if self.cur_token == '@': |
114 | 393 axis = ATTRIBUTE |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
394 self.next_token() |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
395 elif self.cur_token == '.': |
114 | 396 axis = SELF |
137 | 397 elif self.cur_token == '..': |
139
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
398 raise PathSyntaxError('Unsupported axis "parent"', self.filename, |
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
399 self.lineno) |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
400 elif self.peek_token() == '::': |
114 | 401 axis = Axis.forname(self.cur_token) |
402 if axis is None: | |
139
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
403 raise PathSyntaxError('Unsupport axis "%s"' % axis, |
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
404 self.filename, self.lineno) |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
405 self.next_token() |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
406 self.next_token() |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
407 else: |
137 | 408 axis = None |
409 nodetest = self._node_test(axis or CHILD) | |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
410 predicates = [] |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
411 while self.cur_token == '[': |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
412 predicates.append(self._predicate()) |
137 | 413 return axis, nodetest, predicates |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
414 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
415 def _node_test(self, axis=None): |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
416 test = prefix = None |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
417 next_token = self.peek_token() |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
418 if next_token in ('(', '()'): # Node type test |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
419 test = self._node_type() |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
420 |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
421 elif next_token == ':': # Namespace prefix |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
422 prefix = self.cur_token |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
423 self.next_token() |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
424 localname = self.next_token() |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
425 if localname == '*': |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
426 test = QualifiedPrincipalTypeTest(axis, prefix) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
427 else: |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
428 test = QualifiedNameTest(axis, prefix, localname) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
429 |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
430 else: # Name test |
137 | 431 if self.cur_token == '*': |
432 test = PrincipalTypeTest(axis) | |
433 elif self.cur_token == '.': | |
434 test = NodeTest() | |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
435 else: |
137 | 436 test = LocalNameTest(axis, self.cur_token) |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
437 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
438 if not self.at_end: |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
439 self.next_token() |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
440 return test |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
441 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
442 def _node_type(self): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
443 name = self.cur_token |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
444 self.next_token() |
137 | 445 |
446 args = [] | |
447 if self.cur_token != '()': | |
448 # The processing-instruction() function optionally accepts the | |
449 # name of the PI as argument, which must be a literal string | |
450 self.next_token() # ( | |
451 if self.cur_token != ')': | |
452 string = self.cur_token | |
453 if (string[0], string[-1]) in self._QUOTES: | |
454 string = string[1:-1] | |
455 args.append(string) | |
456 | |
457 cls = _nodetest_map.get(name) | |
458 if not cls: | |
139
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
459 raise PathSyntaxError('%s() not allowed here' % name, self.filename, |
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
460 self.lineno) |
137 | 461 return cls(*args) |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
462 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
463 def _predicate(self): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
464 assert self.cur_token == '[' |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
465 self.next_token() |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
466 expr = self._or_expr() |
121
062e51ad7b19
Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`.
cmlenz
parents:
114
diff
changeset
|
467 if self.cur_token != ']': |
062e51ad7b19
Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`.
cmlenz
parents:
114
diff
changeset
|
468 raise PathSyntaxError('Expected "]" to close predicate, ' |
139
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
469 'but found "%s"' % self.cur_token, |
8332287b5508
Implement position reporting for XPath syntax errors. Closes #20.
cmlenz
parents:
137
diff
changeset
|
470 self.filename, self.lineno) |
111
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
471 if not self.at_end: |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
472 self.next_token() |
2368c3becc52
Some fixes and more unit tests for the XPath engine.
cmlenz
parents:
106
diff
changeset
|
473 return expr |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
474 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
475 def _or_expr(self): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
476 expr = self._and_expr() |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
477 while self.cur_token == 'or': |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
478 self.next_token() |
137 | 479 expr = OrOperator(expr, self._and_expr()) |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
480 return expr |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
481 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
482 def _and_expr(self): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
483 expr = self._equality_expr() |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
484 while self.cur_token == 'and': |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
485 self.next_token() |
137 | 486 expr = AndOperator(expr, self._equality_expr()) |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
487 return expr |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
488 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
489 def _equality_expr(self): |
162
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
490 expr = self._relational_expr() |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
491 while self.cur_token in ('=', '!='): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
492 op = _operator_map[self.cur_token] |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
493 self.next_token() |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
494 expr = op(expr, self._relational_expr()) |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
495 return expr |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
496 |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
497 def _relational_expr(self): |
516 | 498 expr = self._sub_expr() |
162
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
499 while self.cur_token in ('>', '>=', '<', '>='): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
500 op = _operator_map[self.cur_token] |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
501 self.next_token() |
516 | 502 expr = op(expr, self._sub_expr()) |
503 return expr | |
504 | |
505 def _sub_expr(self): | |
506 token = self.cur_token | |
507 if token != '(': | |
508 return self._primary_expr() | |
509 self.next_token() | |
510 expr = self._or_expr() | |
511 if self.cur_token != ')': | |
512 raise PathSyntaxError('Expected ")" to close sub-expression, ' | |
513 'but found "%s"' % self.cur_token, | |
514 self.filename, self.lineno) | |
515 self.next_token() | |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
516 return expr |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
517 |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
518 def _primary_expr(self): |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
519 token = self.cur_token |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
520 if len(token) > 1 and (token[0], token[-1]) in self._QUOTES: |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
521 self.next_token() |
137 | 522 return StringLiteral(token[1:-1]) |
163
9c023c395e44
Support for XPath number literals including decimal places.
cmlenz
parents:
162
diff
changeset
|
523 elif token[0].isdigit() or token[0] == '.': |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
524 self.next_token() |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
525 return NumberLiteral(as_float(token)) |
179
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
526 elif token == '$': |
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
527 token = self.next_token() |
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
528 self.next_token() |
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
529 return VariableReference(token) |
121
062e51ad7b19
Added support for the XPath functions `name()`, `namespace-uri()`, `local-name()`, and `not()`.
cmlenz
parents:
114
diff
changeset
|
530 elif not self.at_end and self.peek_token().startswith('('): |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
531 return self._function_call() |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
532 else: |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
533 axis = None |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
534 if token == '@': |
114 | 535 axis = ATTRIBUTE |
106
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
536 self.next_token() |
f9473bdc93b2
Complete rewrite of the XPath parsing, which was a mess before. Closes #19.
cmlenz
parents:
77
diff
changeset
|
537 return self._node_test(axis) |
137 | 538 |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
539 def _function_call(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
540 name = self.cur_token |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
541 if self.next_token() == '()': |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
542 args = [] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
543 else: |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
544 assert self.cur_token == '(' |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
545 self.next_token() |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
546 args = [self._or_expr()] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
547 while self.cur_token == ',': |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
548 self.next_token() |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
549 args.append(self._or_expr()) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
550 if not self.cur_token == ')': |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
551 raise PathSyntaxError('Expected ")" to close function argument ' |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
552 'list, but found "%s"' % self.cur_token, |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
553 self.filename, self.lineno) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
554 self.next_token() |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
555 cls = _function_map.get(name) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
556 if not cls: |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
557 raise PathSyntaxError('Unsupported function "%s"' % name, |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
558 self.filename, self.lineno) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
559 return cls(*args) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
560 |
137 | 561 |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
562 # Type coercion |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
563 |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
564 def as_scalar(value): |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
565 """Convert value to a scalar. If a single element Attrs() object is passed |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
566 the value of the single attribute will be returned.""" |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
567 if isinstance(value, Attrs): |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
568 assert len(value) == 1 |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
569 return value[0][1] |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
570 else: |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
571 return value |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
572 |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
573 def as_float(value): |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
574 # FIXME - if value is a bool it will be coerced to 0.0 and consequently |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
575 # compared as a float. This is probably not ideal. |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
576 return float(as_scalar(value)) |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
577 |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
578 def as_long(value): |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
579 return long(as_scalar(value)) |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
580 |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
581 def as_string(value): |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
582 value = as_scalar(value) |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
583 if value is False: |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
584 return u'' |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
585 return unicode(value) |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
586 |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
587 def as_bool(value): |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
588 return bool(as_scalar(value)) |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
589 |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
590 |
137 | 591 # Node tests |
592 | |
593 class PrincipalTypeTest(object): | |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
594 """Node test that matches any event with the given principal type.""" |
137 | 595 __slots__ = ['principal_type'] |
596 def __init__(self, principal_type): | |
597 self.principal_type = principal_type | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
598 def __call__(self, kind, data, pos, namespaces, variables): |
137 | 599 if kind is START: |
600 if self.principal_type is ATTRIBUTE: | |
601 return data[1] or None | |
602 else: | |
603 return True | |
604 def __repr__(self): | |
605 return '*' | |
606 | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
607 class QualifiedPrincipalTypeTest(object): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
608 """Node test that matches any event with the given principal type in a |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
609 specific namespace.""" |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
610 __slots__ = ['principal_type', 'prefix'] |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
611 def __init__(self, principal_type, prefix): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
612 self.principal_type = principal_type |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
613 self.prefix = prefix |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
614 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
615 namespace = Namespace(namespaces.get(self.prefix)) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
616 if kind is START: |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
617 if self.principal_type is ATTRIBUTE and data[1]: |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
618 return Attrs([(name, value) for name, value in data[1] |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
619 if name in namespace]) or None |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
620 else: |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
621 return data[0] in namespace |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
622 def __repr__(self): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
623 return '%s:*' % self.prefix |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
624 |
137 | 625 class LocalNameTest(object): |
364
7dabedbb53fb
Fix for #77: match templates were matching their own output.
cmlenz
parents:
363
diff
changeset
|
626 """Node test that matches any event with the given principal type and |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
627 local name. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
628 """ |
137 | 629 __slots__ = ['principal_type', 'name'] |
630 def __init__(self, principal_type, name): | |
631 self.principal_type = principal_type | |
632 self.name = name | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
633 def __call__(self, kind, data, pos, namespaces, variables): |
137 | 634 if kind is START: |
635 if self.principal_type is ATTRIBUTE and self.name in data[1]: | |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
636 return Attrs([(self.name, data[1].get(self.name))]) |
137 | 637 else: |
638 return data[0].localname == self.name | |
639 def __repr__(self): | |
640 return self.name | |
641 | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
642 class QualifiedNameTest(object): |
364
7dabedbb53fb
Fix for #77: match templates were matching their own output.
cmlenz
parents:
363
diff
changeset
|
643 """Node test that matches any event with the given principal type and |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
644 qualified name. |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
645 """ |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
646 __slots__ = ['principal_type', 'prefix', 'name'] |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
647 def __init__(self, principal_type, prefix, name): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
648 self.principal_type = principal_type |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
649 self.prefix = prefix |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
650 self.name = name |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
651 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
652 qname = QName('%s}%s' % (namespaces.get(self.prefix), self.name)) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
653 if kind is START: |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
654 if self.principal_type is ATTRIBUTE and qname in data[1]: |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
655 return Attrs([(self.name, data[1].get(self.name))]) |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
656 else: |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
657 return data[0] == qname |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
658 def __repr__(self): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
659 return '%s:%s' % (self.prefix, self.name) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
660 |
137 | 661 class CommentNodeTest(object): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
662 """Node test that matches any comment events.""" |
137 | 663 __slots__ = [] |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
664 def __call__(self, kind, data, pos, namespaces, variables): |
330
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
665 return kind is COMMENT |
137 | 666 def __repr__(self): |
667 return 'comment()' | |
668 | |
669 class NodeTest(object): | |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
670 """Node test that matches any node.""" |
137 | 671 __slots__ = [] |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
672 def __call__(self, kind, data, pos, namespaces, variables): |
137 | 673 if kind is START: |
674 return True | |
675 return kind, data, pos | |
676 def __repr__(self): | |
677 return 'node()' | |
678 | |
679 class ProcessingInstructionNodeTest(object): | |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
680 """Node test that matches any processing instruction event.""" |
137 | 681 __slots__ = ['target'] |
682 def __init__(self, target=None): | |
683 self.target = target | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
684 def __call__(self, kind, data, pos, namespaces, variables): |
330
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
685 return kind is PI and (not self.target or data[0] == self.target) |
137 | 686 def __repr__(self): |
687 arg = '' | |
688 if self.target: | |
689 arg = '"' + self.target + '"' | |
690 return 'processing-instruction(%s)' % arg | |
691 | |
692 class TextNodeTest(object): | |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
693 """Node test that matches any text event.""" |
137 | 694 __slots__ = [] |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
695 def __call__(self, kind, data, pos, namespaces, variables): |
330
3e749eaa3100
XPath tests should never return event tuples, just values or booleans.
cmlenz
parents:
326
diff
changeset
|
696 return kind is TEXT |
137 | 697 def __repr__(self): |
698 return 'text()' | |
699 | |
700 _nodetest_map = {'comment': CommentNodeTest, 'node': NodeTest, | |
701 'processing-instruction': ProcessingInstructionNodeTest, | |
702 'text': TextNodeTest} | |
703 | |
704 # Functions | |
705 | |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
706 class Function(object): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
707 """Base class for function nodes in XPath expressions.""" |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
708 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
709 class BooleanFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
710 """The `boolean` function, which converts its argument to a boolean |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
711 value. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
712 """ |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
713 __slots__ = ['expr'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
714 def __init__(self, expr): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
715 self.expr = expr |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
716 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
717 val = self.expr(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
718 return as_bool(val) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
719 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
720 return 'boolean(%r)' % self.expr |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
721 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
722 class CeilingFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
723 """The `ceiling` function, which returns the nearest lower integer number |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
724 for the given number. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
725 """ |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
726 __slots__ = ['number'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
727 def __init__(self, number): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
728 self.number = number |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
729 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
730 number = self.number(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
731 return ceil(as_float(number)) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
732 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
733 return 'ceiling(%r)' % self.number |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
734 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
735 class ConcatFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
736 """The `concat` function, which concatenates (joins) the variable number of |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
737 strings it gets as arguments. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
738 """ |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
739 __slots__ = ['exprs'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
740 def __init__(self, *exprs): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
741 self.exprs = exprs |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
742 def __call__(self, kind, data, pos, namespaces, variables): |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
743 strings = [] |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
744 for item in [expr(kind, data, pos, namespaces, variables) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
745 for expr in self.exprs]: |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
746 strings.append(as_string(item)) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
747 return u''.join(strings) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
748 def __repr__(self): |
169 | 749 return 'concat(%s)' % ', '.join([repr(expr) for expr in self.exprs]) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
750 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
751 class ContainsFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
752 """The `contains` function, which returns whether a string contains a given |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
753 substring. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
754 """ |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
755 __slots__ = ['string1', 'string2'] |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
756 def __init__(self, string1, string2): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
757 self.string1 = string1 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
758 self.string2 = string2 |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
759 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
760 string1 = self.string1(kind, data, pos, namespaces, variables) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
761 string2 = self.string2(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
762 return as_string(string2) in as_string(string1) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
763 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
764 return 'contains(%r, %r)' % (self.string1, self.string2) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
765 |
534
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
766 class MatchesFunction(Function): |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
767 """The `matches` function, which returns whether a string matches a regular |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
768 expression. |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
769 """ |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
770 __slots__ = ['string1', 'string2'] |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
771 flag_mapping = {'s': re.S, 'm': re.M, 'i': re.I, 'x': re.X} |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
772 |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
773 def __init__(self, string1, string2, flags=''): |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
774 self.string1 = string1 |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
775 self.string2 = string2 |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
776 self.flags = self._map_flags(flags) |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
777 def __call__(self, kind, data, pos, namespaces, variables): |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
778 string1 = as_string(self.string1(kind, data, pos, namespaces, variables)) |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
779 string2 = as_string(self.string2(kind, data, pos, namespaces, variables)) |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
780 return re.search(string2, string1, self.flags) |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
781 def _map_flags(self, flags): |
593 | 782 return reduce(operator.or_, |
534
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
783 [self.flag_map[flag] for flag in flags], re.U) |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
784 def __repr__(self): |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
785 return 'contains(%r, %r)' % (self.string1, self.string2) |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
786 |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
787 class FalseFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
788 """The `false` function, which always returns the boolean `false` value.""" |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
789 __slots__ = [] |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
790 def __call__(self, kind, data, pos, namespaces, variables): |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
791 return False |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
792 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
793 return 'false()' |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
794 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
795 class FloorFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
796 """The `ceiling` function, which returns the nearest higher integer number |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
797 for the given number. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
798 """ |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
799 __slots__ = ['number'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
800 def __init__(self, number): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
801 self.number = number |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
802 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
803 number = self.number(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
804 return floor(as_float(number)) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
805 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
806 return 'floor(%r)' % self.number |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
807 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
808 class LocalNameFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
809 """The `local-name` function, which returns the local name of the current |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
810 element. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
811 """ |
137 | 812 __slots__ = [] |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
813 def __call__(self, kind, data, pos, namespaces, variables): |
137 | 814 if kind is START: |
234 | 815 return data[0].localname |
137 | 816 def __repr__(self): |
817 return 'local-name()' | |
818 | |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
819 class NameFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
820 """The `name` function, which returns the qualified name of the current |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
821 element. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
822 """ |
137 | 823 __slots__ = [] |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
824 def __call__(self, kind, data, pos, namespaces, variables): |
137 | 825 if kind is START: |
234 | 826 return data[0] |
137 | 827 def __repr__(self): |
828 return 'name()' | |
829 | |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
830 class NamespaceUriFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
831 """The `namespace-uri` function, which returns the namespace URI of the |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
832 current element. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
833 """ |
137 | 834 __slots__ = [] |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
835 def __call__(self, kind, data, pos, namespaces, variables): |
137 | 836 if kind is START: |
234 | 837 return data[0].namespace |
137 | 838 def __repr__(self): |
839 return 'namespace-uri()' | |
840 | |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
841 class NotFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
842 """The `not` function, which returns the negated boolean value of its |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
843 argument. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
844 """ |
137 | 845 __slots__ = ['expr'] |
846 def __init__(self, expr): | |
847 self.expr = expr | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
848 def __call__(self, kind, data, pos, namespaces, variables): |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
849 return not as_bool(self.expr(kind, data, pos, namespaces, variables)) |
137 | 850 def __repr__(self): |
851 return 'not(%s)' % self.expr | |
852 | |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
853 class NormalizeSpaceFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
854 """The `normalize-space` function, which removes leading and trailing |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
855 whitespace in the given string, and replaces multiple adjacent whitespace |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
856 characters inside the string with a single space. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
857 """ |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
858 __slots__ = ['expr'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
859 _normalize = re.compile(r'\s{2,}').sub |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
860 def __init__(self, expr): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
861 self.expr = expr |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
862 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
863 string = self.expr(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
864 return self._normalize(' ', as_string(string).strip()) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
865 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
866 return 'normalize-space(%s)' % repr(self.expr) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
867 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
868 class NumberFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
869 """The `number` function that converts its argument to a number.""" |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
870 __slots__ = ['expr'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
871 def __init__(self, expr): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
872 self.expr = expr |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
873 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
874 val = self.expr(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
875 return as_float(val) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
876 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
877 return 'number(%r)' % self.expr |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
878 |
162
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
879 class RoundFunction(Function): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
880 """The `round` function, which returns the nearest integer number for the |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
881 given number. |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
882 """ |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
883 __slots__ = ['number'] |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
884 def __init__(self, number): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
885 self.number = number |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
886 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
887 number = self.number(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
888 return round(as_float(number)) |
162
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
889 def __repr__(self): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
890 return 'round(%r)' % self.number |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
891 |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
892 class StartsWithFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
893 """The `starts-with` function that returns whether one string starts with |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
894 a given substring. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
895 """ |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
896 __slots__ = ['string1', 'string2'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
897 def __init__(self, string1, string2): |
282
24b3cbbc1b1b
Fix `starts-with()` XPath function so that it actually compares the two strings. Closes #61.
cmlenz
parents:
259
diff
changeset
|
898 self.string1 = string1 |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
899 self.string2 = string2 |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
900 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
901 string1 = self.string1(kind, data, pos, namespaces, variables) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
902 string2 = self.string2(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
903 return as_string(string1).startswith(as_string(string2)) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
904 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
905 return 'starts-with(%r, %r)' % (self.string1, self.string2) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
906 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
907 class StringLengthFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
908 """The `string-length` function that returns the length of the given |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
909 string. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
910 """ |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
911 __slots__ = ['expr'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
912 def __init__(self, expr): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
913 self.expr = expr |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
914 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
915 string = self.expr(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
916 return len(as_string(string)) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
917 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
918 return 'string-length(%r)' % self.expr |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
919 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
920 class SubstringFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
921 """The `substring` function that returns the part of a string that starts |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
922 at the given offset, and optionally limited to the given length. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
923 """ |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
924 __slots__ = ['string', 'start', 'length'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
925 def __init__(self, string, start, length=None): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
926 self.string = string |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
927 self.start = start |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
928 self.length = length |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
929 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
930 string = self.string(kind, data, pos, namespaces, variables) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
931 start = self.start(kind, data, pos, namespaces, variables) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
932 length = 0 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
933 if self.length is not None: |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
934 length = self.length(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
935 return string[as_long(start):len(as_string(string)) - as_long(length)] |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
936 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
937 if self.length is not None: |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
938 return 'substring(%r, %r, %r)' % (self.string, self.start, |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
939 self.length) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
940 else: |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
941 return 'substring(%r, %r)' % (self.string, self.start) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
942 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
943 class SubstringAfterFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
944 """The `substring-after` function that returns the part of a string that |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
945 is found after the given substring. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
946 """ |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
947 __slots__ = ['string1', 'string2'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
948 def __init__(self, string1, string2): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
949 self.string1 = string1 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
950 self.string2 = string2 |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
951 def __call__(self, kind, data, pos, namespaces, variables): |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
952 string1 = as_string(self.string1(kind, data, pos, namespaces, variables)) |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
953 string2 = as_string(self.string2(kind, data, pos, namespaces, variables)) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
954 index = string1.find(string2) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
955 if index >= 0: |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
956 return string1[index + len(string2):] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
957 return u'' |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
958 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
959 return 'substring-after(%r, %r)' % (self.string1, self.string2) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
960 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
961 class SubstringBeforeFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
962 """The `substring-before` function that returns the part of a string that |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
963 is found before the given substring. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
964 """ |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
965 __slots__ = ['string1', 'string2'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
966 def __init__(self, string1, string2): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
967 self.string1 = string1 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
968 self.string2 = string2 |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
969 def __call__(self, kind, data, pos, namespaces, variables): |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
970 string1 = as_string(self.string1(kind, data, pos, namespaces, variables)) |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
971 string2 = as_string(self.string2(kind, data, pos, namespaces, variables)) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
972 index = string1.find(string2) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
973 if index >= 0: |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
974 return string1[:index] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
975 return u'' |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
976 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
977 return 'substring-after(%r, %r)' % (self.string1, self.string2) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
978 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
979 class TranslateFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
980 """The `translate` function that translates a set of characters in a |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
981 string to target set of characters. |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
982 """ |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
983 __slots__ = ['string', 'fromchars', 'tochars'] |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
984 def __init__(self, string, fromchars, tochars): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
985 self.string = string |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
986 self.fromchars = fromchars |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
987 self.tochars = tochars |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
988 def __call__(self, kind, data, pos, namespaces, variables): |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
989 string = as_string(self.string(kind, data, pos, namespaces, variables)) |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
990 fromchars = as_string(self.fromchars(kind, data, pos, namespaces, variables)) |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
991 tochars = as_string(self.tochars(kind, data, pos, namespaces, variables)) |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
992 table = dict(zip([ord(c) for c in fromchars], |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
993 [ord(c) for c in tochars])) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
994 return string.translate(table) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
995 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
996 return 'translate(%r, %r, %r)' % (self.string, self.fromchars, |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
997 self.tochars) |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
998 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
999 class TrueFunction(Function): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1000 """The `true` function, which always returns the boolean `true` value.""" |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1001 __slots__ = [] |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1002 def __call__(self, kind, data, pos, namespaces, variables): |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1003 return True |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1004 def __repr__(self): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1005 return 'true()' |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1006 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1007 _function_map = {'boolean': BooleanFunction, 'ceiling': CeilingFunction, |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1008 'concat': ConcatFunction, 'contains': ContainsFunction, |
534
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
1009 'matches': MatchesFunction, 'false': FalseFunction, 'floor': |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
1010 FloorFunction, 'local-name': LocalNameFunction, 'name': |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
1011 NameFunction, 'namespace-uri': NamespaceUriFunction, |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1012 'normalize-space': NormalizeSpaceFunction, 'not': NotFunction, |
162
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1013 'number': NumberFunction, 'round': RoundFunction, |
534
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
1014 'starts-with': StartsWithFunction, 'string-length': |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
1015 StringLengthFunction, 'substring': SubstringFunction, |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
1016 'substring-after': SubstringAfterFunction, 'substring-before': |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
1017 SubstringBeforeFunction, 'translate': TranslateFunction, |
57b5d5138f1a
Add XPath `matches()` function which, of course, supports the Python regular
athomas
parents:
518
diff
changeset
|
1018 'true': TrueFunction} |
137 | 1019 |
179
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
1020 # Literals & Variables |
137 | 1021 |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1022 class Literal(object): |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1023 """Abstract base class for literal nodes.""" |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1024 |
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1025 class StringLiteral(Literal): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1026 """A string literal node.""" |
137 | 1027 __slots__ = ['text'] |
1028 def __init__(self, text): | |
1029 self.text = text | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1030 def __call__(self, kind, data, pos, namespaces, variables): |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
1031 return self.text |
137 | 1032 def __repr__(self): |
1033 return '"%s"' % self.text | |
1034 | |
155
9a5aedda1099
* String literals in XPath expressions that contains spaces are now tokenizes correctly.
cmlenz
parents:
145
diff
changeset
|
1035 class NumberLiteral(Literal): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1036 """A number literal node.""" |
137 | 1037 __slots__ = ['number'] |
1038 def __init__(self, number): | |
1039 self.number = number | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1040 def __call__(self, kind, data, pos, namespaces, variables): |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
1041 return self.number |
137 | 1042 def __repr__(self): |
1043 return str(self.number) | |
1044 | |
179
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
1045 class VariableReference(Literal): |
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
1046 """A variable reference node.""" |
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
1047 __slots__ = ['name'] |
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
1048 def __init__(self, name): |
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
1049 self.name = name |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1050 def __call__(self, kind, data, pos, namespaces, variables): |
228
a5b38b459cbb
Add support for position predicates in XPath expressions.
cmlenz
parents:
224
diff
changeset
|
1051 return variables.get(self.name) |
179
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
1052 def __repr__(self): |
215 | 1053 return str(self.name) |
179
13909179e5e1
Implemented support for XPath variables in predicates (#31).
cmlenz
parents:
169
diff
changeset
|
1054 |
137 | 1055 # Operators |
1056 | |
1057 class AndOperator(object): | |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1058 """The boolean operator `and`.""" |
137 | 1059 __slots__ = ['lval', 'rval'] |
1060 def __init__(self, lval, rval): | |
1061 self.lval = lval | |
1062 self.rval = rval | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1063 def __call__(self, kind, data, pos, namespaces, variables): |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1064 lval = as_bool(self.lval(kind, data, pos, namespaces, variables)) |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1065 if not lval: |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1066 return False |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1067 rval = self.rval(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1068 return as_bool(rval) |
137 | 1069 def __repr__(self): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1070 return '%s and %s' % (self.lval, self.rval) |
137 | 1071 |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1072 class EqualsOperator(object): |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1073 """The equality operator `=`.""" |
137 | 1074 __slots__ = ['lval', 'rval'] |
1075 def __init__(self, lval, rval): | |
1076 self.lval = lval | |
1077 self.rval = rval | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1078 def __call__(self, kind, data, pos, namespaces, variables): |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1079 lval = as_scalar(self.lval(kind, data, pos, namespaces, variables)) |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1080 rval = as_scalar(self.rval(kind, data, pos, namespaces, variables)) |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1081 return lval == rval |
137 | 1082 def __repr__(self): |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1083 return '%s=%s' % (self.lval, self.rval) |
137 | 1084 |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1085 class NotEqualsOperator(object): |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1086 """The equality operator `!=`.""" |
137 | 1087 __slots__ = ['lval', 'rval'] |
1088 def __init__(self, lval, rval): | |
1089 self.lval = lval | |
1090 self.rval = rval | |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1091 def __call__(self, kind, data, pos, namespaces, variables): |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1092 lval = as_scalar(self.lval(kind, data, pos, namespaces, variables)) |
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1093 rval = as_scalar(self.rval(kind, data, pos, namespaces, variables)) |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1094 return lval != rval |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1095 def __repr__(self): |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1096 return '%s!=%s' % (self.lval, self.rval) |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1097 |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1098 class OrOperator(object): |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1099 """The boolean operator `or`.""" |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1100 __slots__ = ['lval', 'rval'] |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1101 def __init__(self, lval, rval): |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1102 self.lval = lval |
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1103 self.rval = rval |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1104 def __call__(self, kind, data, pos, namespaces, variables): |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1105 lval = as_bool(self.lval(kind, data, pos, namespaces, variables)) |
161
7b1f07496bf7
Various docstring additions and other cosmetic changes.
cmlenz
parents:
156
diff
changeset
|
1106 if lval: |
137 | 1107 return True |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1108 rval = self.rval(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1109 return as_bool(rval) |
137 | 1110 def __repr__(self): |
1111 return '%s or %s' % (self.lval, self.rval) | |
1112 | |
162
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1113 class GreaterThanOperator(object): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1114 """The relational operator `>` (greater than).""" |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1115 __slots__ = ['lval', 'rval'] |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1116 def __init__(self, lval, rval): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1117 self.lval = lval |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1118 self.rval = rval |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1119 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1120 lval = self.lval(kind, data, pos, namespaces, variables) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1121 rval = self.rval(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1122 return as_float(lval) > as_float(rval) |
162
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1123 def __repr__(self): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1124 return '%s>%s' % (self.lval, self.rval) |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1125 |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1126 class GreaterThanOrEqualOperator(object): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1127 """The relational operator `>=` (greater than or equal).""" |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1128 __slots__ = ['lval', 'rval'] |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1129 def __init__(self, lval, rval): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1130 self.lval = lval |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1131 self.rval = rval |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1132 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1133 lval = self.lval(kind, data, pos, namespaces, variables) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1134 rval = self.rval(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1135 return as_float(lval) >= as_float(rval) |
162
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1136 def __repr__(self): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1137 return '%s>=%s' % (self.lval, self.rval) |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1138 |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1139 class LessThanOperator(object): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1140 """The relational operator `<` (less than).""" |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1141 __slots__ = ['lval', 'rval'] |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1142 def __init__(self, lval, rval): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1143 self.lval = lval |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1144 self.rval = rval |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1145 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1146 lval = self.lval(kind, data, pos, namespaces, variables) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1147 rval = self.rval(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1148 return as_float(lval) < as_float(rval) |
162
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1149 def __repr__(self): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1150 return '%s<%s' % (self.lval, self.rval) |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1151 |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1152 class LessThanOrEqualOperator(object): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1153 """The relational operator `<=` (less than or equal).""" |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1154 __slots__ = ['lval', 'rval'] |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1155 def __init__(self, lval, rval): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1156 self.lval = lval |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1157 self.rval = rval |
224
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1158 def __call__(self, kind, data, pos, namespaces, variables): |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1159 lval = self.lval(kind, data, pos, namespaces, variables) |
90d62225f411
Implement support for namespace prefixes in XPath expressions.
cmlenz
parents:
223
diff
changeset
|
1160 rval = self.rval(kind, data, pos, namespaces, variables) |
518
c1f2a859fd75
Attributes selected with an XPath are now returned as an `Attrs()` object in
athomas
parents:
516
diff
changeset
|
1161 return as_float(lval) <= as_float(rval) |
162
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1162 def __repr__(self): |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1163 return '%s<=%s' % (self.lval, self.rval) |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1164 |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1165 _operator_map = {'=': EqualsOperator, '!=': NotEqualsOperator, |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1166 '>': GreaterThanOperator, '>=': GreaterThanOrEqualOperator, |
456039594db9
Implement the XPath relational operators and the `round()` function.
cmlenz
parents:
161
diff
changeset
|
1167 '<': LessThanOperator, '>=': LessThanOrEqualOperator} |
333
e5ba8c5a30fe
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
|
1168 |
e5ba8c5a30fe
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
|
1169 |
e5ba8c5a30fe
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
|
1170 _DOTSLASHSLASH = (DESCENDANT_OR_SELF, PrincipalTypeTest(None), ()) |