docutils/test/test_nodes.py

   1 #! /usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3
   4 # $Id$
   5 # Author: David Goodger <goodger@python.org>
   6 # Copyright: This module has been placed in the public domain.
   7
   8 """
   9 Test module for nodes.py.
  10 """
  11
  12 import sys
  13 import unittest
  14 import types
  15 import DocutilsTestSupport              # must be imported before docutils
  16 from DocutilsTestSupport import nodes, utils
  17
  18 debug = False
  19
  20
  21 class TextTests(unittest.TestCase):
  22
  23     def setUp(self):
  24         self.text = nodes.Text('Line 1.\nLine 2.')
  25         self.unicode_text = nodes.Text(u'Möhren')
  26         self.longtext = nodes.Text('Mary had a little lamb whose '
  27                                    'fleece was white as snow and '
  28                                    'everwhere that Mary went the '
  29                                    'lamb was sure to go.')
  30
  31     def test_repr(self):
  32         self.assertEqual(repr(self.text), r"<#text: 'Line 1.\nLine 2.'>")
  33         self.assertEqual(self.text.shortrepr(),
  34                           r"<#text: 'Line 1.\nLine 2.'>")
  35         self.assertEqual(nodes.reprunicode('foo'), u'foo')
  36         if sys.version_info < (3,):
  37             self.assertEqual(repr(self.unicode_text), r"<#text: 'M\xf6hren'>")
  38         else:
  39             self.assertEqual(repr(self.unicode_text), u"<#text: 'Möhren'>")
  40
  41     def test_str(self):
  42         self.assertEqual(str(self.text), 'Line 1.\nLine 2.')
  43
  44     def test_unicode(self):
  45         self.assertEqual(unicode(self.unicode_text), u'Möhren')
  46         self.assertEqual(str(self.unicode_text), 'M\xf6hren')
  47
  48     def test_astext(self):
  49         self.assertTrue(isinstance(self.text.astext(), unicode))
  50         self.assertEqual(self.text.astext(), u'Line 1.\nLine 2.')
  51         self.assertEqual(self.unicode_text.astext(), u'Möhren')
  52
  53     def test_pformat(self):
  54         self.assertTrue(isinstance(self.text.pformat(), unicode))
  55         self.assertEqual(self.text.pformat(), u'Line 1.\nLine 2.\n')
  56
  57     def test_asciirestriction(self):
  58         if sys.version_info < (3,):
  59             self.assertRaises(UnicodeDecodeError, nodes.Text,
  60                               b'hol%s' % chr(224))
  61         else:
  62             # no bytes at all allowed
  63             self.assertRaises(TypeError, nodes.Text, b'hol')
  64
  65     def test_longrepr(self):
  66         self.assertEqual(repr(self.longtext), r"<#text: 'Mary had a "
  67                           r"little lamb whose fleece was white as snow "
  68                           r"and everwh ...'>")
  69         self.assertEqual(self.longtext.shortrepr(),
  70                           r"<#text: 'Mary had a lit ...'>")
  71
  72 class ElementTests(unittest.TestCase):
  73
  74     def test_empty(self):
  75         element = nodes.Element()
  76         self.assertEqual(repr(element), '<Element: >')
  77         self.assertEqual(str(element), '<Element/>')
  78         dom = element.asdom()
  79         self.assertEqual(dom.toxml(), '<Element/>')
  80         dom.unlink()
  81         element['attr'] = '1'
  82         self.assertEqual(repr(element), '<Element: >')
  83         self.assertEqual(str(element), '<Element attr="1"/>')
  84         dom = element.asdom()
  85         self.assertEqual(dom.toxml(), '<Element attr="1"/>')
  86         dom.unlink()
  87         self.assertEqual(element.pformat(), '<Element attr="1">\n')
  88         del element['attr']
  89         element['mark'] = u'\u2022'
  90         self.assertEqual(repr(element), '<Element: >')
  91         if sys.version_info < (3,):
  92             self.assertEqual(str(element), '<Element mark="\\u2022"/>')
  93         else:
  94             self.assertEqual(str(element), u'<Element mark="\u2022"/>')
  95         dom = element.asdom()
  96         self.assertEqual(dom.toxml(), u'<Element mark="\u2022"/>')
  97         dom.unlink()
  98         element['names'] = ['nobody', u'имя', u'näs']
  99         if sys.version_info < (3,):
 100             self.assertEqual(repr(element),
 101                 '<Element "nobody; \\u0438\\u043c\\u044f; n\\xe4s": >')
 102         else:
 103             self.assertEqual(repr(element), u'<Element "nobody; имя; näs": >')
 104         self.assertTrue(isinstance(repr(element), str))
 105
 106     def test_withtext(self):
 107         element = nodes.Element('text\nmore', nodes.Text('text\nmore'))
 108         uelement = nodes.Element(u'grün', nodes.Text(u'grün'))
 109         self.assertEqual(repr(element), r"<Element: <#text: 'text\nmore'>>")
 110         if sys.version_info < (3,):
 111             self.assertEqual(repr(uelement), "<Element: <#text: 'gr\\xfcn'>>")
 112         else:
 113             self.assertEqual(repr(uelement), u"<Element: <#text: 'grün'>>")
 114         self.assertTrue(isinstance(repr(uelement),str))
 115         self.assertEqual(str(element), '<Element>text\nmore</Element>')
 116         self.assertEqual(str(uelement), '<Element>gr\xfcn</Element>')
 117         dom = element.asdom()
 118         self.assertEqual(dom.toxml(), '<Element>text\nmore</Element>')
 119         dom.unlink()
 120         element['attr'] = '1'
 121         self.assertEqual(repr(element), r"<Element: <#text: 'text\nmore'>>")
 122         self.assertEqual(str(element),
 123                           '<Element attr="1">text\nmore</Element>')
 124         dom = element.asdom()
 125         self.assertEqual(dom.toxml(),
 126                           '<Element attr="1">text\nmore</Element>')
 127         dom.unlink()
 128         self.assertEqual(element.pformat(),
 129                           '<Element attr="1">\n    text\n    more\n')
 130
 131     def test_clear(self):
 132         element = nodes.Element()
 133         element += nodes.Element()
 134         self.assertTrue(len(element))
 135         element.clear()
 136         self.assertTrue(not len(element))
 137
 138     def test_normal_attributes(self):
 139         element = nodes.Element()
 140         self.assertTrue('foo' not in element)
 141         self.assertRaises(KeyError, element.__getitem__, 'foo')
 142         element['foo'] = 'sometext'
 143         self.assertEqual(element['foo'], 'sometext')
 144         del element['foo']
 145         self.assertRaises(KeyError, element.__getitem__, 'foo')
 146
 147     def test_default_attributes(self):
 148         element = nodes.Element()
 149         self.assertEqual(element['ids'], [])
 150         self.assertEqual(element.non_default_attributes(), {})
 151         self.assertTrue(not element.is_not_default('ids'))
 152         self.assertTrue(element['ids'] is not nodes.Element()['ids'])
 153         element['ids'].append('someid')
 154         self.assertEqual(element['ids'], ['someid'])
 155         self.assertEqual(element.non_default_attributes(),
 156                           {'ids': ['someid']})
 157         self.assertTrue(element.is_not_default('ids'))
 158
 159     def test_update_basic_atts(self):
 160         element1 = nodes.Element(ids=['foo', 'bar'], test=['test1'])
 161         element2 = nodes.Element(ids=['baz', 'qux'], test=['test2'])
 162         element1.update_basic_atts(element2)
 163         # 'ids' are appended because 'ids' is a basic attribute.
 164         self.assertEqual(element1['ids'], ['foo', 'bar', 'baz', 'qux'])
 165         # 'test' is not overwritten because it is not a basic attribute.
 166         self.assertEqual(element1['test'], ['test1'])
 167
 168     def test_update_all_atts(self):
 169         # Note: Also tests is_not_list_attribute and is_not_known_attribute
 170         # and various helpers
 171         ## Test for full attribute replacement
 172         element1 = nodes.Element(ids=['foo', 'bar'], parent_only='parent',
 173                                  all_nodes='mom')
 174         element2 = nodes.Element(ids=['baz', 'qux'], child_only='child',
 175                                  all_nodes='dad', source='source')
 176
 177         # Test for when same fields are replaced as well as source...
 178         element1.update_all_atts_consistantly(element2, True, True)
 179         # 'ids' are appended because 'ids' is a basic attribute.
 180         self.assertEquals(element1['ids'], ['foo', 'bar', 'baz', 'qux'])
 181         # 'parent_only' should remain unaffected.
 182         self.assertEquals(element1['parent_only'], 'parent')
 183         # 'all_nodes' is overwritten due to the second parameter == True.
 184         self.assertEquals(element1['all_nodes'], 'dad')
 185         # 'child_only' should have been added.
 186         self.assertEquals(element1['child_only'], 'child')
 187         # 'source' is also overwritten due to the third parameter == True.
 188         self.assertEquals(element1['source'], 'source')
 189
 190         # Test for when same fields are replaced but not source...
 191         element1 = nodes.Element(ids=['foo', 'bar'], parent_only='parent',
 192                                  all_nodes='mom')
 193         element1.update_all_atts_consistantly(element2)
 194         # 'ids' are appended because 'ids' is a basic attribute.
 195         self.assertEquals(element1['ids'], ['foo', 'bar', 'baz', 'qux'])
 196         # 'parent_only' should remain unaffected.
 197         self.assertEquals(element1['parent_only'], 'parent')
 198         # 'all_nodes' is overwritten due to the second parameter default of True.
 199         self.assertEquals(element1['all_nodes'], 'dad')
 200         # 'child_only' should have been added.
 201         self.assertEquals(element1['child_only'], 'child')
 202         # 'source' remains unset due to the third parameter default of False.
 203         self.assertEquals(element1.get('source'), None)
 204
 205         # Test for when fields are NOT replaced but source is...
 206         element1 = nodes.Element(ids=['foo', 'bar'], parent_only='parent',
 207                                  all_nodes='mom')
 208         element1.update_all_atts_consistantly(element2, False, True)
 209         # 'ids' are appended because 'ids' is a basic attribute.
 210         self.assertEquals(element1['ids'], ['foo', 'bar', 'baz', 'qux'])
 211         # 'parent_only' should remain unaffected.
 212         self.assertEquals(element1['parent_only'], 'parent')
 213         # 'all_nodes' is preserved due to the second parameter == False.
 214         self.assertEquals(element1['all_nodes'], 'mom')
 215         # 'child_only' should have been added.
 216         self.assertEquals(element1['child_only'], 'child')
 217         # 'source' is added due to the third parameter == True.
 218         self.assertEquals(element1['source'], 'source')
 219         element1 = nodes.Element(source='destination')
 220         element1.update_all_atts_consistantly(element2, False, True)
 221         # 'source' remains unchanged due to the second parameter == False.
 222         self.assertEquals(element1['source'], 'destination')
 223
 224         # Test for when same fields are replaced but not source...
 225         element1 = nodes.Element(ids=['foo', 'bar'], parent_only='parent',
 226                                  all_nodes='mom')
 227         element1.update_all_atts_consistantly(element2, False)
 228         # 'ids' are appended because 'ids' is a basic attribute.
 229         self.assertEquals(element1['ids'], ['foo', 'bar', 'baz', 'qux'])
 230         # 'parent_only' should remain unaffected.
 231         self.assertEquals(element1['parent_only'], 'parent')
 232         # 'all_nodes' is preserved due to the second parameter == False.
 233         self.assertEquals(element1['all_nodes'], 'mom')
 234         # 'child_only' should have been added.
 235         self.assertEquals(element1['child_only'], 'child')
 236         # 'source' remains unset due to the third parameter default of False.
 237         self.assertEquals(element1.get('source'), None)
 238
 239         ## Test for List attribute merging
 240         # Attribute Concatination
 241         element1 = nodes.Element(ss='a', sl='1', ls=['I'], ll=['A'])
 242         element2 = nodes.Element(ss='b', sl=['2'], ls='II', ll=['B'])
 243         element1.update_all_atts_concatenating(element2)
 244         # 'ss' is replaced because non-list
 245         self.assertEquals(element1['ss'], 'b')
 246         # 'sl' is replaced because they are both not lists
 247         self.assertEquals(element1['sl'], ['2'])
 248         # 'ls' is replaced because they are both not lists
 249         self.assertEquals(element1['ls'], 'II')
 250         # 'll' is extended because they are both lists
 251         self.assertEquals(element1['ll'], ['A', 'B'])
 252
 253         # Attribute Coercion
 254         element1 = nodes.Element(ss='a', sl='1', ls=['I'], ll=['A'])
 255         element2 = nodes.Element(ss='b', sl=['2'], ls='II', ll=['B'])
 256         element1.update_all_atts_coercion(element2)
 257         # 'ss' is replaced because non-list
 258         self.assertEquals(element1['ss'], 'b')
 259         # 'sl' is converted to a list and appended because element2 has a list
 260         self.assertEquals(element1['sl'], ['1', '2'])
 261         # 'ls' has element2's value appended to the list
 262         self.assertEquals(element1['ls'], ['I', 'II'])
 263         # 'll' is extended because they are both lists
 264         self.assertEquals(element1['ll'], ['A', 'B'])
 265
 266         # Attribute Conversion
 267         element1 = nodes.Element(ss='a', sl='1', ls=['I'], ll=['A'])
 268         element2 = nodes.Element(ss='b', sl=['2'], ls='II', ll=['B'])
 269         element1.update_all_atts_convert(element2)
 270         # 'ss' is converted to a list with the values from each element
 271         self.assertEquals(element1['ss'], ['a', 'b'])
 272         # 'sl' is converted to a list and appended
 273         self.assertEquals(element1['sl'], ['1', '2'])
 274         # 'ls' has element2's value appended to the list
 275         self.assertEquals(element1['ls'], ['I', 'II'])
 276         # 'll' is extended
 277         self.assertEquals(element1['ll'], ['A', 'B'])
 278
 279     def test_replace_self(self):
 280         parent = nodes.Element(ids=['parent'])
 281         child1 = nodes.Element(ids=['child1'])
 282         grandchild = nodes.Element(ids=['grandchild'])
 283         child1 += grandchild
 284         child2 = nodes.Element(ids=['child2'])
 285         twins = [nodes.Element(ids=['twin%s' % i]) for i in (1, 2)]
 286         child2 += twins
 287         child3 = nodes.Element(ids=['child3'])
 288         child4 = nodes.Element(ids=['child4'])
 289         parent += [child1, child2, child3, child4]
 290         self.assertEqual(parent.pformat(), """\
 291 <Element ids="parent">
 292     <Element ids="child1">
 293         <Element ids="grandchild">
 294     <Element ids="child2">
 295         <Element ids="twin1">
 296         <Element ids="twin2">
 297     <Element ids="child3">
 298     <Element ids="child4">
 299 """)
 300         # Replace child1 with the grandchild.
 301         child1.replace_self(child1[0])
 302         self.assertEqual(parent[0], grandchild)
 303         # Assert that 'ids' have been updated.
 304         self.assertEqual(grandchild['ids'], ['grandchild', 'child1'])
 305         # Replace child2 with its children.
 306         child2.replace_self(child2[:])
 307         self.assertEqual(parent[1:3], twins)
 308         # Assert that 'ids' have been propagated to first child.
 309         self.assertEqual(twins[0]['ids'], ['twin1', 'child2'])
 310         self.assertEqual(twins[1]['ids'], ['twin2'])
 311         # Replace child3 with new child.
 312         newchild = nodes.Element(ids=['newchild'])
 313         child3.replace_self(newchild)
 314         self.assertEqual(parent[3], newchild)
 315         self.assertEqual(newchild['ids'], ['newchild', 'child3'])
 316         # Crazy but possible case: Substitute child4 for itself.
 317         child4.replace_self(child4)
 318         # Make sure the 'child4' ID hasn't been duplicated.
 319         self.assertEqual(child4['ids'], ['child4'])
 320         self.assertEqual(len(parent), 5)
 321
 322     def test_unicode(self):
 323         node = nodes.Element(u'Möhren', nodes.Text(u'Möhren', u'Möhren'))
 324         self.assertEqual(unicode(node), u'<Element>Möhren</Element>')
 325
 326
 327 class MiscTests(unittest.TestCase):
 328
 329     def test_reprunicode(self):
 330         # return `unicode` instance
 331         self.assertTrue(isinstance(nodes.reprunicode('foo'), unicode))
 332         self.assertEqual(nodes.reprunicode('foo'), u'foo')
 333         self.assertEqual(nodes.reprunicode(u'Möhre'), u'Möhre')
 334         if sys.version_info < (3,): # strip leading "u" from representation
 335             self.assertEqual(repr(nodes.reprunicode(u'Möhre')),
 336                              repr(u'Möhre')[1:])
 337         else: # no change to `unicode` under Python 3k
 338             self.assertEqual(repr(nodes.reprunicode(u'Möhre')), repr(u'Möhre'))
 339
 340     def test_ensure_str(self):
 341         self.assertTrue(isinstance(nodes.ensure_str(u'über'), str))
 342         self.assertEqual(nodes.ensure_str('over'), 'over')
 343         if sys.version_info < (3,): # strip leading "u" from representation
 344             self.assertEqual(nodes.ensure_str(u'über'), r'\xfcber')
 345         else:
 346             self.assertEqual(nodes.ensure_str(u'über'), r'über')
 347
 348     def test_node_class_names(self):
 349         node_class_names = []
 350         for x in dir(nodes):
 351             c = getattr(nodes, x)
 352             if isinstance(c, (type, types.ClassType)) and \
 353                    issubclass(c, nodes.Node) and len(c.__bases__) > 1:
 354                 node_class_names.append(x)
 355         node_class_names.sort()
 356         nodes.node_class_names.sort()
 357         self.assertEqual(node_class_names, nodes.node_class_names)
 358
 359     ids = [(u'a', 'a'), ('A', 'a'), ('', ''), ('a b \n c', 'a-b-c'),
 360            ('a.b.c', 'a-b-c'), (' - a - b - c - ', 'a-b-c'), (' - ', ''),
 361            (u'\u2020\u2066', ''), (u'a \xa7 b \u2020 c', 'a-b-c'),
 362            ('1', ''), ('1abc', 'abc'),
 363           ]
 364     ids_unicode_all = [
 365             (u'\u00f8 o with stroke', 'o-o-with-stroke'),
 366             (u'\u0111 d with stroke', 'd-d-with-stroke'),
 367             (u'\u0127 h with stroke', 'h-h-with-stroke'),
 368             (u'\u0131 dotless i', 'i-dotless-i'),
 369             (u'\u0142 l with stroke', 'l-l-with-stroke'),
 370             (u'\u0167 t with stroke', 't-t-with-stroke'),
 371            # From Latin Extended-B
 372             (u'\u0180 b with stroke', 'b-b-with-stroke'),
 373             (u'\u0183 b with topbar', 'b-b-with-topbar'),
 374             (u'\u0188 c with hook', 'c-c-with-hook'),
 375             (u'\u018c d with topbar', 'd-d-with-topbar'),
 376             (u'\u0192 f with hook', 'f-f-with-hook'),
 377             (u'\u0199 k with hook', 'k-k-with-hook'),
 378             (u'\u019a l with bar', 'l-l-with-bar'),
 379             (u'\u019e n with long right leg', 'n-n-with-long-right-leg'),
 380             (u'\u01a5 p with hook', 'p-p-with-hook'),
 381             (u'\u01ab t with palatal hook', 't-t-with-palatal-hook'),
 382             (u'\u01ad t with hook', 't-t-with-hook'),
 383             (u'\u01b4 y with hook', 'y-y-with-hook'),
 384             (u'\u01b6 z with stroke', 'z-z-with-stroke'),
 385             (u'\u01e5 g with stroke', 'g-g-with-stroke'),
 386             (u'\u0225 z with hook', 'z-z-with-hook'),
 387             (u'\u0234 l with curl', 'l-l-with-curl'),
 388             (u'\u0235 n with curl', 'n-n-with-curl'),
 389             (u'\u0236 t with curl', 't-t-with-curl'),
 390             (u'\u0237 dotless j', 'j-dotless-j'),
 391             (u'\u023c c with stroke', 'c-c-with-stroke'),
 392             (u'\u023f s with swash tail', 's-s-with-swash-tail'),
 393             (u'\u0240 z with swash tail', 'z-z-with-swash-tail'),
 394             (u'\u0247 e with stroke', 'e-e-with-stroke'),
 395             (u'\u0249 j with stroke', 'j-j-with-stroke'),
 396             (u'\u024b q with hook tail', 'q-q-with-hook-tail'),
 397             (u'\u024d r with stroke', 'r-r-with-stroke'),
 398             (u'\u024f y with stroke', 'y-y-with-stroke'),
 399            # From Latin-1 Supplements
 400             (u'\u00e0: a with grave', 'a-a-with-grave'),
 401             (u'\u00e1 a with acute', 'a-a-with-acute'),
 402             (u'\u00e2 a with circumflex', 'a-a-with-circumflex'),
 403             (u'\u00e3 a with tilde', 'a-a-with-tilde'),
 404             (u'\u00e4 a with diaeresis', 'a-a-with-diaeresis'),
 405             (u'\u00e5 a with ring above', 'a-a-with-ring-above'),
 406             (u'\u00e7 c with cedilla', 'c-c-with-cedilla'),
 407             (u'\u00e8 e with grave', 'e-e-with-grave'),
 408             (u'\u00e9 e with acute', 'e-e-with-acute'),
 409             (u'\u00ea e with circumflex', 'e-e-with-circumflex'),
 410             (u'\u00eb e with diaeresis', 'e-e-with-diaeresis'),
 411             (u'\u00ec i with grave', 'i-i-with-grave'),
 412             (u'\u00ed i with acute', 'i-i-with-acute'),
 413             (u'\u00ee i with circumflex', 'i-i-with-circumflex'),
 414             (u'\u00ef i with diaeresis', 'i-i-with-diaeresis'),
 415             (u'\u00f1 n with tilde', 'n-n-with-tilde'),
 416             (u'\u00f2 o with grave', 'o-o-with-grave'),
 417             (u'\u00f3 o with acute', 'o-o-with-acute'),
 418             (u'\u00f4 o with circumflex', 'o-o-with-circumflex'),
 419             (u'\u00f5 o with tilde', 'o-o-with-tilde'),
 420             (u'\u00f6 o with diaeresis', 'o-o-with-diaeresis'),
 421             (u'\u00f9 u with grave', 'u-u-with-grave'),
 422             (u'\u00fa u with acute', 'u-u-with-acute'),
 423             (u'\u00fb u with circumflex', 'u-u-with-circumflex'),
 424             (u'\u00fc u with diaeresis', 'u-u-with-diaeresis'),
 425             (u'\u00fd y with acute', 'y-y-with-acute'),
 426             (u'\u00ff y with diaeresis', 'y-y-with-diaeresis'),
 427            # From Latin Extended-A
 428             (u'\u0101 a with macron', 'a-a-with-macron'),
 429             (u'\u0103 a with breve', 'a-a-with-breve'),
 430             (u'\u0105 a with ogonek', 'a-a-with-ogonek'),
 431             (u'\u0107 c with acute', 'c-c-with-acute'),
 432             (u'\u0109 c with circumflex', 'c-c-with-circumflex'),
 433             (u'\u010b c with dot above', 'c-c-with-dot-above'),
 434             (u'\u010d c with caron', 'c-c-with-caron'),
 435             (u'\u010f d with caron', 'd-d-with-caron'),
 436             (u'\u0113 e with macron', 'e-e-with-macron'),
 437             (u'\u0115 e with breve', 'e-e-with-breve'),
 438             (u'\u0117 e with dot above', 'e-e-with-dot-above'),
 439             (u'\u0119 e with ogonek', 'e-e-with-ogonek'),
 440             (u'\u011b e with caron', 'e-e-with-caron'),
 441             (u'\u011d g with circumflex', 'g-g-with-circumflex'),
 442             (u'\u011f g with breve', 'g-g-with-breve'),
 443             (u'\u0121 g with dot above', 'g-g-with-dot-above'),
 444             (u'\u0123 g with cedilla', 'g-g-with-cedilla'),
 445             (u'\u0125 h with circumflex', 'h-h-with-circumflex'),
 446             (u'\u0129 i with tilde', 'i-i-with-tilde'),
 447             (u'\u012b i with macron', 'i-i-with-macron'),
 448             (u'\u012d i with breve', 'i-i-with-breve'),
 449             (u'\u012f i with ogonek', 'i-i-with-ogonek'),
 450             (u'\u0133 ligature ij', 'ij-ligature-ij'),
 451             (u'\u0135 j with circumflex', 'j-j-with-circumflex'),
 452             (u'\u0137 k with cedilla', 'k-k-with-cedilla'),
 453             (u'\u013a l with acute', 'l-l-with-acute'),
 454             (u'\u013c l with cedilla', 'l-l-with-cedilla'),
 455             (u'\u013e l with caron', 'l-l-with-caron'),
 456             (u'\u0140 l with middle dot', 'l-l-with-middle-dot'),
 457             (u'\u0144 n with acute', 'n-n-with-acute'),
 458             (u'\u0146 n with cedilla', 'n-n-with-cedilla'),
 459             (u'\u0148 n with caron', 'n-n-with-caron'),
 460             (u'\u014d o with macron', 'o-o-with-macron'),
 461             (u'\u014f o with breve', 'o-o-with-breve'),
 462             (u'\u0151 o with double acute', 'o-o-with-double-acute'),
 463             (u'\u0155 r with acute', 'r-r-with-acute'),
 464             (u'\u0157 r with cedilla', 'r-r-with-cedilla'),
 465             (u'\u0159 r with caron', 'r-r-with-caron'),
 466             (u'\u015b s with acute', 's-s-with-acute'),
 467             (u'\u015d s with circumflex', 's-s-with-circumflex'),
 468             (u'\u015f s with cedilla', 's-s-with-cedilla'),
 469             (u'\u0161 s with caron', 's-s-with-caron'),
 470             (u'\u0163 t with cedilla', 't-t-with-cedilla'),
 471             (u'\u0165 t with caron', 't-t-with-caron'),
 472             (u'\u0169 u with tilde', 'u-u-with-tilde'),
 473             (u'\u016b u with macron', 'u-u-with-macron'),
 474             (u'\u016d u with breve', 'u-u-with-breve'),
 475             (u'\u016f u with ring above', 'u-u-with-ring-above'),
 476             (u'\u0171 u with double acute', 'u-u-with-double-acute'),
 477             (u'\u0173 u with ogonek', 'u-u-with-ogonek'),
 478             (u'\u0175 w with circumflex', 'w-w-with-circumflex'),
 479             (u'\u0177 y with circumflex', 'y-y-with-circumflex'),
 480             (u'\u017a z with acute', 'z-z-with-acute'),
 481             (u'\u017c z with dot above', 'z-z-with-dot-above'),
 482             (u'\u017e z with caron', 'z-z-with-caron'),
 483            # From Latin Extended-B
 484             (u'\u01a1 o with horn', 'o-o-with-horn'),
 485             (u'\u01b0 u with horn', 'u-u-with-horn'),
 486             (u'\u01c6 dz with caron', 'dz-dz-with-caron'),
 487             (u'\u01c9 lj', 'lj-lj'),
 488             (u'\u01cc nj', 'nj-nj'),
 489             (u'\u01ce a with caron', 'a-a-with-caron'),
 490             (u'\u01d0 i with caron', 'i-i-with-caron'),
 491             (u'\u01d2 o with caron', 'o-o-with-caron'),
 492             (u'\u01d4 u with caron', 'u-u-with-caron'),
 493             (u'\u01e7 g with caron', 'g-g-with-caron'),
 494             (u'\u01e9 k with caron', 'k-k-with-caron'),
 495             (u'\u01eb o with ogonek', 'o-o-with-ogonek'),
 496             (u'\u01ed o with ogonek and macron', 'o-o-with-ogonek-and-macron'),
 497             (u'\u01f0 j with caron', 'j-j-with-caron'),
 498             (u'\u01f3 dz', 'dz-dz'),
 499             (u'\u01f5 g with acute', 'g-g-with-acute'),
 500             (u'\u01f9 n with grave', 'n-n-with-grave'),
 501             (u'\u0201 a with double grave', 'a-a-with-double-grave'),
 502             (u'\u0203 a with inverted breve', 'a-a-with-inverted-breve'),
 503             (u'\u0205 e with double grave', 'e-e-with-double-grave'),
 504             (u'\u0207 e with inverted breve', 'e-e-with-inverted-breve'),
 505             (u'\u0209 i with double grave', 'i-i-with-double-grave'),
 506             (u'\u020b i with inverted breve', 'i-i-with-inverted-breve'),
 507             (u'\u020d o with double grave', 'o-o-with-double-grave'),
 508             (u'\u020f o with inverted breve', 'o-o-with-inverted-breve'),
 509             (u'\u0211 r with double grave', 'r-r-with-double-grave'),
 510             (u'\u0213 r with inverted breve', 'r-r-with-inverted-breve'),
 511             (u'\u0215 u with double grave', 'u-u-with-double-grave'),
 512             (u'\u0217 u with inverted breve', 'u-u-with-inverted-breve'),
 513             (u'\u0219 s with comma below', 's-s-with-comma-below'),
 514             (u'\u021b t with comma below', 't-t-with-comma-below'),
 515             (u'\u021f h with caron', 'h-h-with-caron'),
 516             (u'\u0227 a with dot above', 'a-a-with-dot-above'),
 517             (u'\u0229 e with cedilla', 'e-e-with-cedilla'),
 518             (u'\u022f o with dot above', 'o-o-with-dot-above'),
 519             (u'\u0233 y with macron', 'y-y-with-macron'),
 520            # digraphs From Latin-1 Supplements
 521             (u'\u00df: ligature sz', 'sz-ligature-sz'),
 522             (u'\u00e6 ae', 'ae-ae'),
 523             (u'\u0153 ligature oe', 'oe-ligature-oe'),
 524             (u'\u0238 db digraph', 'db-db-digraph'),
 525             (u'\u0239 qp digraph', 'qp-qp-digraph'),
 526             ]
 527
 528     def test_make_id(self):
 529         failures = []
 530         tests = self.ids + self.ids_unicode_all
 531         for input, expect in tests:
 532             output = nodes.make_id(input)
 533             if expect != output:
 534                 failures.append("'%s' != '%s'" % (expect, output))
 535         if failures:
 536             self.fail("%d failures in %d\n%s" % (len(failures), len(self.ids), "\n".join(failures)))
 537
 538     def test_traverse(self):
 539         e = nodes.Element()
 540         e += nodes.Element()
 541         e[0] += nodes.Element()
 542         e[0] += nodes.TextElement()
 543         e[0][1] += nodes.Text('some text')
 544         e += nodes.Element()
 545         e += nodes.Element()
 546         self.assertEqual(list(e.traverse()),
 547                           [e, e[0], e[0][0], e[0][1], e[0][1][0], e[1], e[2]])
 548         self.assertEqual(list(e.traverse(include_self=False)),
 549                           [e[0], e[0][0], e[0][1], e[0][1][0], e[1], e[2]])
 550         self.assertEqual(list(e.traverse(descend=False)),
 551                           [e])
 552         self.assertEqual(list(e[0].traverse(descend=False, ascend=True)),
 553                           [e[0], e[1], e[2]])
 554         self.assertEqual(list(e[0][0].traverse(descend=False, ascend=True)),
 555                           [e[0][0], e[0][1], e[1], e[2]])
 556         self.assertEqual(list(e[0][0].traverse(descend=False, siblings=True)),
 557                           [e[0][0], e[0][1]])
 558         self.testlist = e[0:2]
 559         self.assertEqual(list(e.traverse(condition=self.not_in_testlist)),
 560                           [e, e[0][0], e[0][1], e[0][1][0], e[2]])
 561         # Return siblings despite siblings=False because ascend is true.
 562         self.assertEqual(list(e[1].traverse(ascend=True, siblings=False)),
 563                           [e[1], e[2]])
 564         self.assertEqual(list(e[0].traverse()),
 565                           [e[0], e[0][0], e[0][1], e[0][1][0]])
 566         self.testlist = [e[0][0], e[0][1]]
 567         self.assertEqual(list(e[0].traverse(condition=self.not_in_testlist)),
 568                                [e[0], e[0][1][0]])
 569         self.testlist.append(e[0][1][0])
 570         self.assertEqual(list(e[0].traverse(condition=self.not_in_testlist)),
 571                                [e[0]])
 572         self.assertEqual(list(e.traverse(nodes.TextElement)), [e[0][1]])
 573
 574     def test_next_node(self):
 575         e = nodes.Element()
 576         e += nodes.Element()
 577         e[0] += nodes.Element()
 578         e[0] += nodes.TextElement()
 579         e[0][1] += nodes.Text('some text')
 580         e += nodes.Element()
 581         e += nodes.Element()
 582         self.testlist = [e[0], e[0][1], e[1]]
 583         compare = [(e, e[0][0]),
 584                    (e[0], e[0][0]),
 585                    (e[0][0], e[0][1][0]),
 586                    (e[0][1], e[0][1][0]),
 587                    (e[0][1][0], e[2]),
 588                    (e[1], e[2]),
 589                    (e[2], None)]
 590         for node, next_node in compare:
 591             self.assertEqual(node.next_node(self.not_in_testlist, ascend=True),
 592                               next_node)
 593         self.assertEqual(e[0][0].next_node(ascend=True), e[0][1])
 594         self.assertEqual(e[2].next_node(), None)
 595
 596     def not_in_testlist(self, x):
 597         return x not in self.testlist
 598
 599     def test_copy(self):
 600         grandchild = nodes.Text('rawsource')
 601         child = nodes.emphasis('rawsource', grandchild, att='child')
 602         e = nodes.Element('rawsource', child, att='e')
 603         # Shallow copy:
 604         e_copy = e.copy()
 605         self.assertTrue(e is not e_copy)
 606         # Internal attributes (like `rawsource`) are also copied.
 607         self.assertEqual(e.rawsource, 'rawsource')
 608         self.assertEqual(e_copy.rawsource, e.rawsource)
 609         self.assertEqual(e_copy['att'], 'e')
 610         # Children are not copied.
 611         self.assertEqual(len(e_copy), 0)
 612         # Deep copy:
 613         e_deepcopy = e.deepcopy()
 614         self.assertEqual(e_deepcopy.rawsource, e.rawsource)
 615         self.assertEqual(e_deepcopy['att'], 'e')
 616         # Children are copied recursively.
 617         self.assertEqual(e_deepcopy[0][0], grandchild)
 618         self.assertTrue(e_deepcopy[0][0] is not grandchild)
 619         self.assertEqual(e_deepcopy[0]['att'], 'child')
 620
 621
 622 class TreeCopyVisitorTests(unittest.TestCase):
 623
 624     def setUp(self):
 625         document = utils.new_document('test data')
 626         document += nodes.paragraph('', 'Paragraph 1.')
 627         blist = nodes.bullet_list()
 628         for i in range(1, 6):
 629             item = nodes.list_item()
 630             for j in range(1, 4):
 631                 item += nodes.paragraph('', 'Item %s, paragraph %s.' % (i, j))
 632             blist += item
 633         document += blist
 634         self.document = document
 635
 636     def compare_trees(self, one, two):
 637         self.assertEqual(one.__class__, two.__class__)
 638         self.assertNotEqual(id(one), id(two))
 639         self.assertEqual(len(one.children), len(two.children))
 640         for i in range(len(one.children)):
 641             self.compare_trees(one.children[i], two.children[i])
 642
 643     def test_copy_whole(self):
 644         visitor = nodes.TreeCopyVisitor(self.document)
 645         self.document.walkabout(visitor)
 646         newtree = visitor.get_tree_copy()
 647         self.assertEqual(self.document.pformat(), newtree.pformat())
 648         self.compare_trees(self.document, newtree)
 649
 650
 651 class MiscFunctionTests(unittest.TestCase):
 652
 653     names = [('a', 'a'), ('A', 'a'), ('A a A', 'a a a'),
 654              ('A  a  A  a', 'a a a a'),
 655              ('  AaA\n\r\naAa\tAaA\t\t', 'aaa aaa aaa')]
 656
 657     def test_normalize_name(self):
 658         for input, output in self.names:
 659             normed = nodes.fully_normalize_name(input)
 660             self.assertEqual(normed, output)
 661
 662     def test_set_id_default(self):
 663         # Default prefixes.
 664         document = utils.new_document('test')
 665         # From name.
 666         element = nodes.Element(names=['test'])
 667         document.set_id(element)
 668         self.assertEqual(element['ids'], ['test'])
 669         # Auto-generated.
 670         element = nodes.Element()
 671         document.set_id(element)
 672         self.assertEqual(element['ids'], ['id1'])
 673
 674     def test_set_id_custom(self):
 675         # Custom prefixes.
 676         document = utils.new_document('test')
 677         # Change settings.
 678         document.settings.id_prefix = 'prefix'
 679         document.settings.auto_id_prefix = 'auto'
 680         # From name.
 681         element = nodes.Element(names=['test'])
 682         document.set_id(element)
 683         self.assertEqual(element['ids'], ['prefixtest'])
 684         # Auto-generated.
 685         element = nodes.Element()
 686         document.set_id(element)
 687         self.assertEqual(element['ids'], ['prefixauto1'])
 688
 689
 690 if __name__ == '__main__':
 691     unittest.main()