docutils/test/test_parsers/test_rst/test_inline_markup.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 Tests for inline markup in docutils/parsers/rst/states.py.
  10 Interpreted text tests are in a separate module, test_interpreted.py.
  11 """
  12
  13 from __init__ import DocutilsTestSupport
  14
  15 def suite():
  16     s = DocutilsTestSupport.ParserTestSuite()
  17     s.generateTests(totest)
  18     return s
  19
  20 totest = {}
  21
  22 totest['emphasis'] = [
  23 ["""\
  24 *emphasis*
  25 """,
  26 """\
  27 <document source="test data">
  28     <paragraph>
  29         <emphasis>
  30             emphasis
  31 """],
  32 [u"""\
  33 l'*emphasis* with the *emphasis*' apostrophe.
  34 l\u2019*emphasis* with the *emphasis*\u2019 apostrophe.
  35 """,
  36 u"""\
  37 <document source="test data">
  38     <paragraph>
  39         l\'
  40         <emphasis>
  41             emphasis
  42          with the \n\
  43         <emphasis>
  44             emphasis
  45         \' apostrophe.
  46         l\u2019
  47         <emphasis>
  48             emphasis
  49          with the \n\
  50         <emphasis>
  51             emphasis
  52         \u2019 apostrophe.
  53 """],
  54 ["""\
  55 *emphasized sentence
  56 across lines*
  57 """,
  58 """\
  59 <document source="test data">
  60     <paragraph>
  61         <emphasis>
  62             emphasized sentence
  63             across lines
  64 """],
  65 ["""\
  66 *emphasis without closing asterisk
  67 """,
  68 """\
  69 <document source="test data">
  70     <paragraph>
  71         <problematic ids="id2" refid="id1">
  72             *
  73         emphasis without closing asterisk
  74     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
  75         <paragraph>
  76             Inline emphasis start-string without end-string.
  77 """],
  78 [r"""some punctuation is allowed around inline markup, e.g.
  79 /*emphasis*/, -*emphasis*-, and :*emphasis*: (delimiters),
  80 (*emphasis*), [*emphasis*], <*emphasis*>, {*emphasis*} (open/close pairs)
  81 *emphasis*., *emphasis*,, *emphasis*!, and *emphasis*\ (closing delimiters),
  82
  83 but not
  84 )*emphasis*(, ]*emphasis*[, >*emphasis*>, }*emphasis*{ (close/open pairs),
  85 (*), [*], '*' or '"*"' ("quoted" start-string),
  86 x*2* or 2*x* (alphanumeric char before),
  87 \*args or * (escaped, whitespace behind start-string),
  88 or *the\* *stars\* *inside* (escaped, whitespace before end-string).
  89
  90 However, '*args' will trigger a warning and may be problematic.
  91
  92 what about *this**?
  93 """,
  94 """\
  95 <document source="test data">
  96     <paragraph>
  97         some punctuation is allowed around inline markup, e.g.
  98         /
  99         <emphasis>
 100             emphasis
 101         /, -
 102         <emphasis>
 103             emphasis
 104         -, and :
 105         <emphasis>
 106             emphasis
 107         : (delimiters),
 108         (
 109         <emphasis>
 110             emphasis
 111         ), [
 112         <emphasis>
 113             emphasis
 114         ], <
 115         <emphasis>
 116             emphasis
 117         >, {
 118         <emphasis>
 119             emphasis
 120         } (open/close pairs)
 121         <emphasis>
 122             emphasis
 123         ., \n\
 124         <emphasis>
 125             emphasis
 126         ,, \n\
 127         <emphasis>
 128             emphasis
 129         !, and \n\
 130         <emphasis>
 131             emphasis
 132         (closing delimiters),
 133     <paragraph>
 134         but not
 135         )*emphasis*(, ]*emphasis*[, >*emphasis*>, }*emphasis*{ (close/open pairs),
 136         (*), [*], '*' or '"*"' ("quoted" start-string),
 137         x*2* or 2*x* (alphanumeric char before),
 138         *args or * (escaped, whitespace behind start-string),
 139         or \n\
 140         <emphasis>
 141             the* *stars* *inside
 142          (escaped, whitespace before end-string).
 143     <paragraph>
 144         However, '
 145         <problematic ids="id2" refid="id1">
 146             *
 147         args' will trigger a warning and may be problematic.
 148     <system_message backrefs="id2" ids="id1" level="2" line="13" source="test data" type="WARNING">
 149         <paragraph>
 150             Inline emphasis start-string without end-string.
 151     <paragraph>
 152         what about \n\
 153         <emphasis>
 154             this*
 155         ?
 156 """],
 157 [u"""\
 158 Quotes around inline markup:
 159
 160 '*emphasis*' "*emphasis*" Straight,
 161 ‘*emphasis*’ “*emphasis*” English, ...,
 162 « *emphasis* » ‹ *emphasis* › « *emphasis* » ‹ *emphasis* ›
 163 « *emphasis* » ‹ *emphasis* › French,
 164 „*emphasis*“ ‚*emphasis*‘ »*emphasis*« ›*emphasis*‹ German, Czech, ...,
 165 „*emphasis*” «*emphasis*» Romanian,
 166 “*emphasis*„ ‘*emphasis*‚ Greek,
 167 「*emphasis*」 『*emphasis*』traditional Chinese,
 168 ”*emphasis*” ’*emphasis*’ »*emphasis*» ›*emphasis*› Swedish, Finnish,
 169 „*emphasis*” ‚*emphasis*’ Polish,
 170 „*emphasis*” »*emphasis*« ’*emphasis*’ Hungarian,
 171 """,
 172 u"""\
 173 <document source="test data">
 174     <paragraph>
 175         Quotes around inline markup:
 176     <paragraph>
 177         \'
 178         <emphasis>
 179             emphasis
 180         \' "
 181         <emphasis>
 182             emphasis
 183         " Straight,
 184         \u2018
 185         <emphasis>
 186             emphasis
 187         \u2019 \u201c
 188         <emphasis>
 189             emphasis
 190         \u201d English, ...,
 191         \xab\u202f
 192         <emphasis>
 193             emphasis
 194         \u202f\xbb \u2039\u202f
 195         <emphasis>
 196             emphasis
 197         \u202f\u203a \xab\xa0
 198         <emphasis>
 199             emphasis
 200         \xa0\xbb \u2039\xa0
 201         <emphasis>
 202             emphasis
 203         \xa0\u203a
 204         \xab\u2005
 205         <emphasis>
 206             emphasis
 207         \u2005\xbb \u2039\u2005
 208         <emphasis>
 209             emphasis
 210         \u2005\u203a French,
 211         \u201e
 212         <emphasis>
 213             emphasis
 214         \u201c \u201a
 215         <emphasis>
 216             emphasis
 217         \u2018 \xbb
 218         <emphasis>
 219             emphasis
 220         \xab \u203a
 221         <emphasis>
 222             emphasis
 223         \u2039 German, Czech, ...,
 224         \u201e
 225         <emphasis>
 226             emphasis
 227         \u201d \xab
 228         <emphasis>
 229             emphasis
 230         \xbb Romanian,
 231         \u201c
 232         <emphasis>
 233             emphasis
 234         \u201e \u2018
 235         <emphasis>
 236             emphasis
 237         \u201a Greek,
 238         \u300c
 239         <emphasis>
 240             emphasis
 241         \u300d \u300e
 242         <emphasis>
 243             emphasis
 244         \u300ftraditional Chinese,
 245         \u201d
 246         <emphasis>
 247             emphasis
 248         \u201d \u2019
 249         <emphasis>
 250             emphasis
 251         \u2019 \xbb
 252         <emphasis>
 253             emphasis
 254         \xbb \u203a
 255         <emphasis>
 256             emphasis
 257         \u203a Swedish, Finnish,
 258         \u201e
 259         <emphasis>
 260             emphasis
 261         \u201d \u201a
 262         <emphasis>
 263             emphasis
 264         \u2019 Polish,
 265         \u201e
 266         <emphasis>
 267             emphasis
 268         \u201d \xbb
 269         <emphasis>
 270             emphasis
 271         \xab \u2019
 272         <emphasis>
 273             emphasis
 274         \u2019 Hungarian,
 275 """],
 276 [r"""
 277 Emphasized asterisk: *\**
 278
 279 Emphasized double asterisk: *\***
 280 """,
 281 """\
 282 <document source="test data">
 283     <paragraph>
 284         Emphasized asterisk: \n\
 285         <emphasis>
 286             *
 287     <paragraph>
 288         Emphasized double asterisk: \n\
 289         <emphasis>
 290             **
 291 """],
 292 ]
 293
 294 totest['strong'] = [
 295 ["""\
 296 **strong**
 297 """,
 298 """\
 299 <document source="test data">
 300     <paragraph>
 301         <strong>
 302             strong
 303 """],
 304 [u"""\
 305 l'**strong** and l\u2019**strong** with apostrophe
 306 """,
 307 u"""\
 308 <document source="test data">
 309     <paragraph>
 310         l'
 311         <strong>
 312             strong
 313          and l\u2019
 314         <strong>
 315             strong
 316          with apostrophe
 317 """],
 318 [u"""\
 319 quoted '**strong**', quoted "**strong**",
 320 quoted \u2018**strong**\u2019, quoted \u201c**strong**\u201d,
 321 quoted \xab**strong**\xbb
 322 """,
 323 u"""\
 324 <document source="test data">
 325     <paragraph>
 326         quoted '
 327         <strong>
 328             strong
 329         ', quoted "
 330         <strong>
 331             strong
 332         ",
 333         quoted \u2018
 334         <strong>
 335             strong
 336         \u2019, quoted \u201c
 337         <strong>
 338             strong
 339         \u201d,
 340         quoted \xab
 341         <strong>
 342             strong
 343         \xbb
 344 """],
 345 [r"""
 346 (**strong**) but not (**) or '(** ' or x**2 or \**kwargs or **
 347
 348 (however, '**kwargs' will trigger a warning and may be problematic)
 349 """,
 350 """\
 351 <document source="test data">
 352     <paragraph>
 353         (
 354         <strong>
 355             strong
 356         ) but not (**) or '(** ' or x**2 or **kwargs or **
 357     <paragraph>
 358         (however, '
 359         <problematic ids="id2" refid="id1">
 360             **
 361         kwargs' will trigger a warning and may be problematic)
 362     <system_message backrefs="id2" ids="id1" level="2" line="4" source="test data" type="WARNING">
 363         <paragraph>
 364             Inline strong start-string without end-string.
 365 """],
 366 ["""\
 367 Strong asterisk: *****
 368
 369 Strong double asterisk: ******
 370 """,
 371 """\
 372 <document source="test data">
 373     <paragraph>
 374         Strong asterisk: \n\
 375         <strong>
 376             *
 377     <paragraph>
 378         Strong double asterisk: \n\
 379         <strong>
 380             **
 381 """],
 382 ["""\
 383 **strong without closing asterisks
 384 """,
 385 """\
 386 <document source="test data">
 387     <paragraph>
 388         <problematic ids="id2" refid="id1">
 389             **
 390         strong without closing asterisks
 391     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 392         <paragraph>
 393             Inline strong start-string without end-string.
 394 """],
 395 ]
 396
 397 totest['literal'] = [
 398 ["""\
 399 ``literal``
 400 """,
 401 """\
 402 <document source="test data">
 403     <paragraph>
 404         <literal>
 405             literal
 406 """],
 407 [r"""
 408 ``\literal``
 409 """,
 410 """\
 411 <document source="test data">
 412     <paragraph>
 413         <literal>
 414             \\literal
 415 """],
 416 [r"""
 417 ``lite\ral``
 418 """,
 419 """\
 420 <document source="test data">
 421     <paragraph>
 422         <literal>
 423             lite\\ral
 424 """],
 425 [r"""
 426 ``literal\``
 427 """,
 428 """\
 429 <document source="test data">
 430     <paragraph>
 431         <literal>
 432             literal\\
 433 """],
 434 [u"""\
 435 l'``literal`` and l\u2019``literal`` with apostrophe
 436 """,
 437 u"""\
 438 <document source="test data">
 439     <paragraph>
 440         l'
 441         <literal>
 442             literal
 443          and l\u2019
 444         <literal>
 445             literal
 446          with apostrophe
 447 """],
 448 [u"""\
 449 quoted '``literal``', quoted "``literal``",
 450 quoted \u2018``literal``\u2019, quoted \u201c``literal``\u201d,
 451 quoted \xab``literal``\xbb
 452 """,
 453 u"""\
 454 <document source="test data">
 455     <paragraph>
 456         quoted '
 457         <literal>
 458             literal
 459         ', quoted "
 460         <literal>
 461             literal
 462         ",
 463         quoted \u2018
 464         <literal>
 465             literal
 466         \u2019, quoted \u201c
 467         <literal>
 468             literal
 469         \u201d,
 470         quoted \xab
 471         <literal>
 472             literal
 473         \xbb
 474 """],
 475 [u"""\
 476 ``'literal'`` with quotes, ``"literal"`` with quotes,
 477 ``\u2018literal\u2019`` with quotes, ``\u201cliteral\u201d`` with quotes,
 478 ``\xabliteral\xbb`` with quotes
 479 """,
 480 u"""\
 481 <document source="test data">
 482     <paragraph>
 483         <literal>
 484             'literal'
 485          with quotes, \n\
 486         <literal>
 487             "literal"
 488          with quotes,
 489         <literal>
 490             \u2018literal\u2019
 491          with quotes, \n\
 492         <literal>
 493             \u201cliteral\u201d
 494          with quotes,
 495         <literal>
 496             \xabliteral\xbb
 497          with quotes
 498 """],
 499 [r"""
 500 ``literal ``TeX quotes'' & \backslash`` but not "``" or ``
 501
 502 (however, ``standalone TeX quotes'' will trigger a warning
 503 and may be problematic)
 504 """,
 505 """\
 506 <document source="test data">
 507     <paragraph>
 508         <literal>
 509             literal ``TeX quotes'' & \\backslash
 510          but not "``" or ``
 511     <paragraph>
 512         (however, \n\
 513         <problematic ids="id2" refid="id1">
 514             ``
 515         standalone TeX quotes'' will trigger a warning
 516         and may be problematic)
 517     <system_message backrefs="id2" ids="id1" level="2" line="4" source="test data" type="WARNING">
 518         <paragraph>
 519             Inline literal start-string without end-string.
 520 """],
 521 ["""\
 522 Find the ```interpreted text``` in this paragraph!
 523 """,
 524 """\
 525 <document source="test data">
 526     <paragraph>
 527         Find the \n\
 528         <literal>
 529             `interpreted text`
 530          in this paragraph!
 531 """],
 532 ["""\
 533 ``literal without closing backquotes
 534 """,
 535 """\
 536 <document source="test data">
 537     <paragraph>
 538         <problematic ids="id2" refid="id1">
 539             ``
 540         literal without closing backquotes
 541     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 542         <paragraph>
 543             Inline literal start-string without end-string.
 544 """],
 545 [r"""
 546 Python ``list``\s use square bracket syntax.
 547 """,
 548 """\
 549 <document source="test data">
 550     <paragraph>
 551         Python \n\
 552         <literal>
 553             list
 554         s use square bracket syntax.
 555 """],
 556 ]
 557
 558 totest['references'] = [
 559 ["""\
 560 ref_
 561 """,
 562 """\
 563 <document source="test data">
 564     <paragraph>
 565         <reference name="ref" refname="ref">
 566             ref
 567 """],
 568 [u"""\
 569 l'ref_ and l\u2019ref_ with apostrophe
 570 """,
 571 u"""\
 572 <document source="test data">
 573     <paragraph>
 574         l'
 575         <reference name="ref" refname="ref">
 576             ref
 577          and l\u2019
 578         <reference name="ref" refname="ref">
 579             ref
 580          with apostrophe
 581 """],
 582 [u"""\
 583 quoted 'ref_', quoted "ref_",
 584 quoted \u2018ref_\u2019, quoted \u201cref_\u201d,
 585 quoted \xabref_\xbb,
 586 but not 'ref ref'_, "ref ref"_, \u2018ref ref\u2019_,
 587 \u201cref ref\u201d_, or \xabref ref\xbb_
 588 """,
 589 u"""\
 590 <document source="test data">
 591     <paragraph>
 592         quoted '
 593         <reference name="ref" refname="ref">
 594             ref
 595         ', quoted "
 596         <reference name="ref" refname="ref">
 597             ref
 598         ",
 599         quoted \u2018
 600         <reference name="ref" refname="ref">
 601             ref
 602         \u2019, quoted \u201c
 603         <reference name="ref" refname="ref">
 604             ref
 605         \u201d,
 606         quoted \xab
 607         <reference name="ref" refname="ref">
 608             ref
 609         \xbb,
 610         but not 'ref ref'_, "ref ref"_, \u2018ref ref\u2019_,
 611         \u201cref ref\u201d_, or \xabref ref\xbb_
 612 """],
 613 ["""\
 614 ref__
 615 """,
 616 """\
 617 <document source="test data">
 618     <paragraph>
 619         <reference anonymous="1" name="ref">
 620             ref
 621 """],
 622 [u"""\
 623 l'ref__ and l\u2019ref__ with apostrophe
 624 """,
 625 u"""\
 626 <document source="test data">
 627     <paragraph>
 628         l'
 629         <reference anonymous="1" name="ref">
 630             ref
 631          and l\u2019
 632         <reference anonymous="1" name="ref">
 633             ref
 634          with apostrophe
 635 """],
 636 [u"""\
 637 quoted 'ref__', quoted "ref__",
 638 quoted \u2018ref__\u2019, quoted \u201cref__\u201d,
 639 quoted \xabref__\xbb,
 640 but not 'ref ref'__, "ref ref"__, \u2018ref ref\u2019__,
 641 \u201cref ref\u201d__, or \xabref ref\xbb__
 642 """,
 643 u"""\
 644 <document source="test data">
 645     <paragraph>
 646         quoted '
 647         <reference anonymous="1" name="ref">
 648             ref
 649         ', quoted "
 650         <reference anonymous="1" name="ref">
 651             ref
 652         ",
 653         quoted \u2018
 654         <reference anonymous="1" name="ref">
 655             ref
 656         \u2019, quoted \u201c
 657         <reference anonymous="1" name="ref">
 658             ref
 659         \u201d,
 660         quoted \xab
 661         <reference anonymous="1" name="ref">
 662             ref
 663         \xbb,
 664         but not 'ref ref'__, "ref ref"__, \u2018ref ref\u2019__,
 665         \u201cref ref\u201d__, or \xabref ref\xbb__
 666 """],
 667 ["""\
 668 ref_, r_, r_e-f_, -ref_, and anonymousref__,
 669 but not _ref_ or __attr__ or object.__attr__
 670 """,
 671 """\
 672 <document source="test data">
 673     <paragraph>
 674         <reference name="ref" refname="ref">
 675             ref
 676         , \n\
 677         <reference name="r" refname="r">
 678             r
 679         , \n\
 680         <reference name="r_e-f" refname="r_e-f">
 681             r_e-f
 682         , -
 683         <reference name="ref" refname="ref">
 684             ref
 685         , and \n\
 686         <reference anonymous="1" name="anonymousref">
 687             anonymousref
 688         ,
 689         but not _ref_ or __attr__ or object.__attr__
 690 """],
 691 ]
 692
 693 totest['phrase_references'] = [
 694 ["""\
 695 `phrase reference`_
 696 """,
 697 """\
 698 <document source="test data">
 699     <paragraph>
 700         <reference name="phrase reference" refname="phrase reference">
 701             phrase reference
 702 """],
 703 [u"""\
 704 l'`phrase reference`_ and l\u2019`phrase reference`_ with apostrophe
 705 """,
 706 u"""\
 707 <document source="test data">
 708     <paragraph>
 709         l'
 710         <reference name="phrase reference" refname="phrase reference">
 711             phrase reference
 712          and l\u2019
 713         <reference name="phrase reference" refname="phrase reference">
 714             phrase reference
 715          with apostrophe
 716 """],
 717 [u"""\
 718 quoted '`phrase reference`_', quoted "`phrase reference`_",
 719 quoted \u2018`phrase reference`_\u2019,
 720 quoted \u201c`phrase reference`_\u201d,
 721 quoted \xab`phrase reference`_\xbb
 722 """,
 723 u"""\
 724 <document source="test data">
 725     <paragraph>
 726         quoted '
 727         <reference name="phrase reference" refname="phrase reference">
 728             phrase reference
 729         ', quoted "
 730         <reference name="phrase reference" refname="phrase reference">
 731             phrase reference
 732         ",
 733         quoted \u2018
 734         <reference name="phrase reference" refname="phrase reference">
 735             phrase reference
 736         \u2019,
 737         quoted \u201c
 738         <reference name="phrase reference" refname="phrase reference">
 739             phrase reference
 740         \u201d,
 741         quoted \xab
 742         <reference name="phrase reference" refname="phrase reference">
 743             phrase reference
 744         \xbb
 745 """],
 746 [u"""\
 747 `'phrase reference'`_ with quotes, `"phrase reference"`_ with quotes,
 748 `\u2018phrase reference\u2019`_ with quotes,
 749 `\u201cphrase reference\u201d`_ with quotes,
 750 `\xabphrase reference\xbb`_ with quotes
 751 """,
 752 u"""\
 753 <document source="test data">
 754     <paragraph>
 755         <reference name="'phrase reference'" refname="'phrase reference'">
 756             'phrase reference'
 757          with quotes, \n\
 758         <reference name=""phrase reference"" refname=""phrase reference"">
 759             "phrase reference"
 760          with quotes,
 761         <reference name="\u2018phrase reference\u2019" refname="\u2018phrase reference\u2019">
 762             \u2018phrase reference\u2019
 763          with quotes,
 764         <reference name="\u201cphrase reference\u201d" refname="\u201cphrase reference\u201d">
 765             \u201cphrase reference\u201d
 766          with quotes,
 767         <reference name="\xabphrase reference\xbb" refname="\xabphrase reference\xbb">
 768             \xabphrase reference\xbb
 769          with quotes
 770 """],
 771 ["""\
 772 `anonymous reference`__
 773 """,
 774 """\
 775 <document source="test data">
 776     <paragraph>
 777         <reference anonymous="1" name="anonymous reference">
 778             anonymous reference
 779 """],
 780 [u"""\
 781 l'`anonymous reference`__ and l\u2019`anonymous reference`__ with apostrophe
 782 """,
 783 u"""\
 784 <document source="test data">
 785     <paragraph>
 786         l'
 787         <reference anonymous="1" name="anonymous reference">
 788             anonymous reference
 789          and l\u2019
 790         <reference anonymous="1" name="anonymous reference">
 791             anonymous reference
 792          with apostrophe
 793 """],
 794 [u"""\
 795 quoted '`anonymous reference`__', quoted "`anonymous reference`__",
 796 quoted \u2018`anonymous reference`__\u2019,
 797 quoted \u201c`anonymous reference`__\u201d,
 798 quoted \xab`anonymous reference`__\xbb
 799 """,
 800 u"""\
 801 <document source="test data">
 802     <paragraph>
 803         quoted '
 804         <reference anonymous="1" name="anonymous reference">
 805             anonymous reference
 806         ', quoted "
 807         <reference anonymous="1" name="anonymous reference">
 808             anonymous reference
 809         ",
 810         quoted \u2018
 811         <reference anonymous="1" name="anonymous reference">
 812             anonymous reference
 813         \u2019,
 814         quoted \u201c
 815         <reference anonymous="1" name="anonymous reference">
 816             anonymous reference
 817         \u201d,
 818         quoted \xab
 819         <reference anonymous="1" name="anonymous reference">
 820             anonymous reference
 821         \xbb
 822 """],
 823 [u"""\
 824 `'anonymous reference'`__ with quotes, `"anonymous reference"`__ with quotes,
 825 `\u2018anonymous reference\u2019`__ with quotes,
 826 `\u201canonymous reference\u201d`__ with quotes,
 827 `\xabanonymous reference\xbb`__ with quotes
 828 """,
 829 u"""\
 830 <document source="test data">
 831     <paragraph>
 832         <reference anonymous="1" name="'anonymous reference'">
 833             'anonymous reference'
 834          with quotes, \n\
 835         <reference anonymous="1" name=""anonymous reference"">
 836             "anonymous reference"
 837          with quotes,
 838         <reference anonymous="1" name="\u2018anonymous reference\u2019">
 839             \u2018anonymous reference\u2019
 840          with quotes,
 841         <reference anonymous="1" name="\u201canonymous reference\u201d">
 842             \u201canonymous reference\u201d
 843          with quotes,
 844         <reference anonymous="1" name="\xabanonymous reference\xbb">
 845             \xabanonymous reference\xbb
 846          with quotes
 847 """],
 848 ["""\
 849 `phrase reference
 850 across lines`_
 851 """,
 852 """\
 853 <document source="test data">
 854     <paragraph>
 855         <reference name="phrase reference across lines" refname="phrase reference across lines">
 856             phrase reference
 857             across lines
 858 """],
 859 ["""\
 860 `phrase\`_ reference`_
 861 """,
 862 """\
 863 <document source="test data">
 864     <paragraph>
 865         <reference name="phrase`_ reference" refname="phrase`_ reference">
 866             phrase`_ reference
 867 """],
 868 ["""\
 869 Invalid phrase reference:
 870
 871 :role:`phrase reference`_
 872 """,
 873 """\
 874 <document source="test data">
 875     <paragraph>
 876         Invalid phrase reference:
 877     <paragraph>
 878         <problematic ids="id2" refid="id1">
 879             :role:`phrase reference`_
 880     <system_message backrefs="id2" ids="id1" level="2" line="3" source="test data" type="WARNING">
 881         <paragraph>
 882             Mismatch: both interpreted text role prefix and reference suffix.
 883 """],
 884 ["""\
 885 Invalid phrase reference:
 886
 887 `phrase reference`:role:_
 888 """,
 889 """\
 890 <document source="test data">
 891     <paragraph>
 892         Invalid phrase reference:
 893     <paragraph>
 894         <problematic ids="id2" refid="id1">
 895             `phrase reference`:role:_
 896     <system_message backrefs="id2" ids="id1" level="2" line="3" source="test data" type="WARNING">
 897         <paragraph>
 898             Mismatch: both interpreted text role suffix and reference suffix.
 899 """],
 900 ["""\
 901 `phrase reference_ without closing backquote
 902 """,
 903 """\
 904 <document source="test data">
 905     <paragraph>
 906         <problematic ids="id2" refid="id1">
 907             `
 908         phrase \n\
 909         <reference name="reference" refname="reference">
 910             reference
 911          without closing backquote
 912     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 913         <paragraph>
 914             Inline interpreted text or phrase reference start-string without end-string.
 915 """],
 916 ["""\
 917 `anonymous phrase reference__ without closing backquote
 918 """,
 919 """\
 920 <document source="test data">
 921     <paragraph>
 922         <problematic ids="id2" refid="id1">
 923             `
 924         anonymous phrase \n\
 925         <reference anonymous="1" name="reference">
 926             reference
 927          without closing backquote
 928     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 929         <paragraph>
 930             Inline interpreted text or phrase reference start-string without end-string.
 931 """],
 932 ]
 933
 934 totest['embedded_URIs'] = [
 935 ["""\
 936 `phrase reference <http://example.com>`_
 937 """,
 938 """\
 939 <document source="test data">
 940     <paragraph>
 941         <reference name="phrase reference" refuri="http://example.com">
 942             phrase reference
 943         <target ids="phrase-reference" names="phrase\ reference" refuri="http://example.com">
 944 """],
 945 ["""\
 946 `anonymous reference <http://example.com>`__
 947 """,
 948 """\
 949 <document source="test data">
 950     <paragraph>
 951         <reference name="anonymous reference" refuri="http://example.com">
 952             anonymous reference
 953 """],
 954 ["""\
 955 `embedded URI on next line
 956 <http://example.com>`__
 957 """,
 958 """\
 959 <document source="test data">
 960     <paragraph>
 961         <reference name="embedded URI on next line" refuri="http://example.com">
 962             embedded URI on next line
 963 """],
 964 ["""\
 965 `embedded URI across lines <http://example.com/
 966 long/path>`__
 967 """,
 968 """\
 969 <document source="test data">
 970     <paragraph>
 971         <reference name="embedded URI across lines" refuri="http://example.com/long/path">
 972             embedded URI across lines
 973 """],
 974 ["""\
 975 `embedded URI with whitespace <http://example.com/
 976 long/path /and  /whitespace>`__
 977 """,
 978 """\
 979 <document source="test data">
 980     <paragraph>
 981         <reference name="embedded URI with whitespace" refuri="http://example.com/long/path/and/whitespace">
 982             embedded URI with whitespace
 983 """],
 984 [r"""
 985 `embedded URI with escaped whitespace <http://example.com/a\
 986 long/path\ and/some\ escaped\ whitespace>`__
 987
 988 `<omitted\ reference\ text\ with\ escaped\ whitespace>`__
 989 """,
 990 """\
 991 <document source="test data">
 992     <paragraph>
 993         <reference name="embedded URI with escaped whitespace" refuri="http://example.com/a long/path and/some escaped whitespace">
 994             embedded URI with escaped whitespace
 995     <paragraph>
 996         <reference name="omitted reference text with escaped whitespace" refuri="omitted reference text with escaped whitespace">
 997             omitted reference text with escaped whitespace
 998 """],
 999 ["""\
1000 `embedded email address <jdoe@example.com>`__
1001
1002 `embedded email address broken across lines <jdoe
1003 @example.com>`__
1004 """,
1005 """\
1006 <document source="test data">
1007     <paragraph>
1008         <reference name="embedded email address" refuri="mailto:jdoe@example.com">
1009             embedded email address
1010     <paragraph>
1011         <reference name="embedded email address broken across lines" refuri="mailto:jdoe@example.com">
1012             embedded email address broken across lines
1013 """],
1014 [r"""
1015 `embedded URI with too much whitespace < http://example.com/
1016 long/path /and  /whitespace >`__
1017
1018 `embedded URI with too much whitespace at end <http://example.com/
1019 long/path /and  /whitespace >`__
1020
1021 `embedded URI with no preceding whitespace<http://example.com>`__
1022
1023 `escaped URI \<http://example.com>`__
1024
1025 See `HTML Anchors: \<a>`_.
1026 """,
1027 """\
1028 <document source="test data">
1029     <paragraph>
1030         <reference anonymous="1" name="embedded URI with too much whitespace < http://example.com/ long/path /and /whitespace >">
1031             embedded URI with too much whitespace < http://example.com/
1032             long/path /and  /whitespace >
1033     <paragraph>
1034         <reference anonymous="1" name="embedded URI with too much whitespace at end <http://example.com/ long/path /and /whitespace >">
1035             embedded URI with too much whitespace at end <http://example.com/
1036             long/path /and  /whitespace >
1037     <paragraph>
1038         <reference anonymous="1" name="embedded URI with no preceding whitespace<http://example.com>">
1039             embedded URI with no preceding whitespace<http://example.com>
1040     <paragraph>
1041         <reference anonymous="1" name="escaped URI <http://example.com>">
1042             escaped URI <http://example.com>
1043     <paragraph>
1044         See \n\
1045         <reference name="HTML Anchors: <a>" refname="html anchors: <a>">
1046             HTML Anchors: <a>
1047         .
1048 """],
1049 ["""\
1050 Relative URIs' reference text can be omitted:
1051
1052 `<reference>`_
1053
1054 `<anonymous>`__
1055 """,
1056 """\
1057 <document source="test data">
1058     <paragraph>
1059         Relative URIs' reference text can be omitted:
1060     <paragraph>
1061         <reference name="reference" refuri="reference">
1062             reference
1063         <target ids="reference" names="reference" refuri="reference">
1064     <paragraph>
1065         <reference name="anonymous" refuri="anonymous">
1066             anonymous
1067 """],
1068 [r"""
1069 Escape trailing low-line char in URIs:
1070
1071 `<reference\_>`_
1072
1073 `<anonymous\_>`__
1074 """,
1075 """\
1076 <document source="test data">
1077     <paragraph>
1078         Escape trailing low-line char in URIs:
1079     <paragraph>
1080         <reference name="reference_" refuri="reference_">
1081             reference_
1082         <target ids="reference" names="reference_" refuri="reference_">
1083     <paragraph>
1084         <reference name="anonymous_" refuri="anonymous_">
1085             anonymous_
1086 """],
1087 ["""\
1088 Escape other char in URIs:
1089
1090 `<reference\:1>`_
1091
1092 `<anonymous\\call>`__
1093 """,
1094 """\
1095 <document source="test data">
1096     <paragraph>
1097         Escape other char in URIs:
1098     <paragraph>
1099         <reference name="reference:1" refuri="reference:1">
1100             reference:1
1101         <target ids="reference-1" names="reference:1" refuri="reference:1">
1102     <paragraph>
1103         <reference name="anonymouscall" refuri="anonymouscall">
1104             anonymouscall
1105 """],
1106 ]
1107
1108 totest['embedded_aliases'] = [
1109 ["""\
1110 `phrase reference <alias_>`_
1111 """,
1112 """\
1113 <document source="test data">
1114     <paragraph>
1115         <reference name="phrase reference" refname="alias">
1116             phrase reference
1117         <target names="phrase\ reference" refname="alias">
1118 """],
1119 ["""\
1120 `anonymous reference <alias_>`__
1121 """,
1122 """\
1123 <document source="test data">
1124     <paragraph>
1125         <reference name="anonymous reference" refname="alias">
1126             anonymous reference
1127 """],
1128 ["""\
1129 `embedded alias on next line
1130 <alias_>`__
1131 """,
1132 """\
1133 <document source="test data">
1134     <paragraph>
1135         <reference name="embedded alias on next line" refname="alias">
1136             embedded alias on next line
1137 """],
1138 ["""\
1139 `embedded alias across lines <alias
1140 phrase_>`__
1141 """,
1142 """\
1143 <document source="test data">
1144     <paragraph>
1145         <reference name="embedded alias across lines" refname="alias phrase">
1146             embedded alias across lines
1147 """],
1148 ["""\
1149 `embedded alias with whitespace <alias
1150 long  phrase_>`__
1151 """,
1152 """\
1153 <document source="test data">
1154     <paragraph>
1155         <reference name="embedded alias with whitespace" refname="alias long phrase">
1156             embedded alias with whitespace
1157 """],
1158 ["""\
1159 `<embedded alias with whitespace_>`__
1160 """,
1161 """\
1162 <document source="test data">
1163     <paragraph>
1164         <reference name="embedded alias with whitespace" refname="embedded alias with whitespace">
1165             embedded alias with whitespace
1166 """],
1167 [r"""
1168 `no embedded alias (whitespace inside bracket) < alias_ >`__
1169
1170 `no embedded alias (no preceding whitespace)<alias_>`__
1171 """,
1172 """\
1173 <document source="test data">
1174     <paragraph>
1175         <reference anonymous="1" name="no embedded alias (whitespace inside bracket) < alias_ >">
1176             no embedded alias (whitespace inside bracket) < alias_ >
1177     <paragraph>
1178         <reference anonymous="1" name="no embedded alias (no preceding whitespace)<alias_>">
1179             no embedded alias (no preceding whitespace)<alias_>
1180 """],
1181 [r"""
1182 `anonymous reference <alias\ with\\ escaped \:characters_>`__
1183 """,
1184 """\
1185 <document source="test data">
1186     <paragraph>
1187         <reference name="anonymous reference" refname="aliaswith\ escaped :characters">
1188             anonymous reference
1189 """],
1190 [r"""
1191 `anonymous reference <alias\ with\\ escaped \:characters_>`__
1192 """,
1193 """\
1194 <document source="test data">
1195     <paragraph>
1196         <reference name="anonymous reference" refname="aliaswith\ escaped :characters">
1197             anonymous reference
1198 """],
1199 ]
1200
1201 totest['inline_targets'] = [
1202 ["""\
1203 _`target`
1204
1205 Here is _`another target` in some text. And _`yet
1206 another target`, spanning lines.
1207
1208 _`Here is  a    TaRgeT` with case and spacial difficulties.
1209 """,
1210 """\
1211 <document source="test data">
1212     <paragraph>
1213         <target ids="target" names="target">
1214             target
1215     <paragraph>
1216         Here is \n\
1217         <target ids="another-target" names="another\ target">
1218             another target
1219          in some text. And \n\
1220         <target ids="yet-another-target" names="yet\ another\ target">
1221             yet
1222             another target
1223         , spanning lines.
1224     <paragraph>
1225         <target ids="here-is-a-target" names="here\ is\ a\ target">
1226             Here is  a    TaRgeT
1227          with case and spacial difficulties.
1228 """],
1229 [u"""\
1230 l'_`target1` and l\u2019_`target2` with apostrophe
1231 """,
1232 u"""\
1233 <document source="test data">
1234     <paragraph>
1235         l'
1236         <target ids="target1" names="target1">
1237             target1
1238          and l\u2019
1239         <target ids="target2" names="target2">
1240             target2
1241          with apostrophe
1242 """],
1243 [u"""\
1244 quoted '_`target1`', quoted "_`target2`",
1245 quoted \u2018_`target3`\u2019, quoted \u201c_`target4`\u201d,
1246 quoted \xab_`target5`\xbb
1247 """,
1248 u"""\
1249 <document source="test data">
1250     <paragraph>
1251         quoted '
1252         <target ids="target1" names="target1">
1253             target1
1254         ', quoted "
1255         <target ids="target2" names="target2">
1256             target2
1257         ",
1258         quoted \u2018
1259         <target ids="target3" names="target3">
1260             target3
1261         \u2019, quoted \u201c
1262         <target ids="target4" names="target4">
1263             target4
1264         \u201d,
1265         quoted \xab
1266         <target ids="target5" names="target5">
1267             target5
1268         \xbb
1269 """],
1270 [u"""\
1271 _`'target1'` with quotes, _`"target2"` with quotes,
1272 _`\u2018target3\u2019` with quotes, _`\u201ctarget4\u201d` with quotes,
1273 _`\xabtarget5\xbb` with quotes
1274 """,
1275 u"""\
1276 <document source="test data">
1277     <paragraph>
1278         <target ids="target1" names="'target1'">
1279             'target1'
1280          with quotes, \n\
1281         <target ids="target2" names=""target2"">
1282             "target2"
1283          with quotes,
1284         <target ids="target3" names="\u2018target3\u2019">
1285             \u2018target3\u2019
1286          with quotes, \n\
1287         <target ids="target4" names="\u201ctarget4\u201d">
1288             \u201ctarget4\u201d
1289          with quotes,
1290         <target ids="target5" names="\xabtarget5\xbb">
1291             \xabtarget5\xbb
1292          with quotes
1293 """],
1294 ["""\
1295 But this isn't a _target; targets require backquotes.
1296
1297 And _`this`_ is just plain confusing.
1298 """,
1299 """\
1300 <document source="test data">
1301     <paragraph>
1302         But this isn't a _target; targets require backquotes.
1303     <paragraph>
1304         And \n\
1305         <problematic ids="id2" refid="id1">
1306             _`
1307         this`_ is just plain confusing.
1308     <system_message backrefs="id2" ids="id1" level="2" line="3" source="test data" type="WARNING">
1309         <paragraph>
1310             Inline target start-string without end-string.
1311 """],
1312 ["""\
1313 _`inline target without closing backquote
1314 """,
1315 """\
1316 <document source="test data">
1317     <paragraph>
1318         <problematic ids="id2" refid="id1">
1319             _`
1320         inline target without closing backquote
1321     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
1322         <paragraph>
1323             Inline target start-string without end-string.
1324 """],
1325 ]
1326
1327 totest['footnote_reference'] = [
1328 ["""\
1329 [1]_
1330 """,
1331 """\
1332 <document source="test data">
1333     <paragraph>
1334         <footnote_reference ids="id1" refname="1">
1335             1
1336 """],
1337 ["""\
1338 [#]_
1339 """,
1340 """\
1341 <document source="test data">
1342     <paragraph>
1343         <footnote_reference auto="1" ids="id1">
1344 """],
1345 ["""\
1346 [#label]_
1347 """,
1348 """\
1349 <document source="test data">
1350     <paragraph>
1351         <footnote_reference auto="1" ids="id1" refname="label">
1352 """],
1353 ["""\
1354 [*]_
1355 """,
1356 """\
1357 <document source="test data">
1358     <paragraph>
1359         <footnote_reference auto="*" ids="id1">
1360 """],
1361 ["""\
1362 Adjacent footnote refs are not possible: [*]_[#label]_ [#]_[2]_ [1]_[*]_
1363 """,
1364 """\
1365 <document source="test data">
1366     <paragraph>
1367         Adjacent footnote refs are not possible: [*]_[#label]_ [#]_[2]_ [1]_[*]_
1368 """],
1369 ]
1370
1371 totest['citation_reference'] = [
1372 ["""\
1373 [citation]_
1374 """,
1375 """\
1376 <document source="test data">
1377     <paragraph>
1378         <citation_reference ids="id1" refname="citation">
1379             citation
1380 """],
1381 ["""\
1382 [citation]_ and [cit-ation]_ and [cit.ation]_ and [CIT1]_ but not [CIT 1]_
1383 """,
1384 """\
1385 <document source="test data">
1386     <paragraph>
1387         <citation_reference ids="id1" refname="citation">
1388             citation
1389          and \n\
1390         <citation_reference ids="id2" refname="cit-ation">
1391             cit-ation
1392          and \n\
1393         <citation_reference ids="id3" refname="cit.ation">
1394             cit.ation
1395          and \n\
1396         <citation_reference ids="id4" refname="cit1">
1397             CIT1
1398          but not [CIT 1]_
1399 """],
1400 ["""\
1401 Adjacent citation refs are not possible: [citation]_[CIT1]_
1402 """,
1403 """\
1404 <document source="test data">
1405     <paragraph>
1406         Adjacent citation refs are not possible: [citation]_[CIT1]_
1407 """],
1408 ]
1409
1410 totest['substitution_references'] = [
1411 ["""\
1412 |subref|
1413 """,
1414 """\
1415 <document source="test data">
1416     <paragraph>
1417         <substitution_reference refname="subref">
1418             subref
1419 """],
1420 ["""\
1421 |subref|_ and |subref|__
1422 """,
1423 """\
1424 <document source="test data">
1425     <paragraph>
1426         <reference refname="subref">
1427             <substitution_reference refname="subref">
1428                 subref
1429          and \n\
1430         <reference anonymous="1">
1431             <substitution_reference refname="subref">
1432                 subref
1433 """],
1434 ["""\
1435 |substitution reference|
1436 """,
1437 """\
1438 <document source="test data">
1439     <paragraph>
1440         <substitution_reference refname="substitution reference">
1441             substitution reference
1442 """],
1443 ["""\
1444 |substitution
1445 reference|
1446 """,
1447 """\
1448 <document source="test data">
1449     <paragraph>
1450         <substitution_reference refname="substitution reference">
1451             substitution
1452             reference
1453 """],
1454 ["""\
1455 |substitution reference without closing verbar
1456 """,
1457 """\
1458 <document source="test data">
1459     <paragraph>
1460         <problematic ids="id2" refid="id1">
1461             |
1462         substitution reference without closing verbar
1463     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
1464         <paragraph>
1465             Inline substitution_reference start-string without end-string.
1466 """],
1467 ["""\
1468 first | then || and finally |||
1469 """,
1470 """\
1471 <document source="test data">
1472     <paragraph>
1473         first | then || and finally |||
1474 """],
1475 ]
1476
1477 totest['standalone_hyperlink'] = [
1478 ["""\
1479 http://www.standalone.hyperlink.com
1480
1481 http:/one-slash-only.absolute.path
1482
1483 [http://example.com]
1484
1485 (http://example.com)
1486
1487 <http://example.com>
1488
1489 http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html
1490
1491 http://[3ffe:2a00:100:7031::1] (the final "]" is ambiguous in text)
1492
1493 http://[3ffe:2a00:100:7031::1]/
1494
1495 mailto:someone@somewhere.com
1496
1497 news:comp.lang.python
1498
1499 An email address in a sentence: someone@somewhere.com.
1500
1501 ftp://ends.with.a.period.
1502
1503 (a.question.mark@end?)
1504 """,
1505 """\
1506 <document source="test data">
1507     <paragraph>
1508         <reference refuri="http://www.standalone.hyperlink.com">
1509             http://www.standalone.hyperlink.com
1510     <paragraph>
1511         <reference refuri="http:/one-slash-only.absolute.path">
1512             http:/one-slash-only.absolute.path
1513     <paragraph>
1514         [
1515         <reference refuri="http://example.com">
1516             http://example.com
1517         ]
1518     <paragraph>
1519         (
1520         <reference refuri="http://example.com">
1521             http://example.com
1522         )
1523     <paragraph>
1524         <
1525         <reference refuri="http://example.com">
1526             http://example.com
1527         >
1528     <paragraph>
1529         <reference refuri="http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html">
1530             http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html
1531     <paragraph>
1532         <reference refuri="http://[3ffe:2a00:100:7031::1">
1533             http://[3ffe:2a00:100:7031::1
1534         ] (the final "]" is ambiguous in text)
1535     <paragraph>
1536         <reference refuri="http://[3ffe:2a00:100:7031::1]/">
1537             http://[3ffe:2a00:100:7031::1]/
1538     <paragraph>
1539         <reference refuri="mailto:someone@somewhere.com">
1540             mailto:someone@somewhere.com
1541     <paragraph>
1542         <reference refuri="news:comp.lang.python">
1543             news:comp.lang.python
1544     <paragraph>
1545         An email address in a sentence: \n\
1546         <reference refuri="mailto:someone@somewhere.com">
1547             someone@somewhere.com
1548         .
1549     <paragraph>
1550         <reference refuri="ftp://ends.with.a.period">
1551             ftp://ends.with.a.period
1552         .
1553     <paragraph>
1554         (
1555         <reference refuri="mailto:a.question.mark@end">
1556             a.question.mark@end
1557         ?)
1558 """],
1559 [r"""
1560 Valid URLs with escaped markup characters:
1561
1562 http://example.com/\*content\*/whatever
1563
1564 http://example.com/\*content*/whatever
1565 """,
1566 """\
1567 <document source="test data">
1568     <paragraph>
1569         Valid URLs with escaped markup characters:
1570     <paragraph>
1571         <reference refuri="http://example.com/*content*/whatever">
1572             http://example.com/*content*/whatever
1573     <paragraph>
1574         <reference refuri="http://example.com/*content*/whatever">
1575             http://example.com/*content*/whatever
1576 """],
1577 ["""\
1578 Valid URLs may end with punctuation inside "<>":
1579
1580 <http://example.org/ends-with-dot.>
1581 """,
1582 """\
1583 <document source="test data">
1584     <paragraph>
1585         Valid URLs may end with punctuation inside "<>":
1586     <paragraph>
1587         <
1588         <reference refuri="http://example.org/ends-with-dot.">
1589             http://example.org/ends-with-dot.
1590         >
1591 """],
1592 ["""\
1593 Valid URLs with interesting endings:
1594
1595 http://example.org/ends-with-pluses++
1596 """,
1597 """\
1598 <document source="test data">
1599     <paragraph>
1600         Valid URLs with interesting endings:
1601     <paragraph>
1602         <reference refuri="http://example.org/ends-with-pluses++">
1603             http://example.org/ends-with-pluses++
1604 """],
1605 ["""\
1606 None of these are standalone hyperlinks (their "schemes"
1607 are not recognized): signal:noise, a:b.
1608 """,
1609 """\
1610 <document source="test data">
1611     <paragraph>
1612         None of these are standalone hyperlinks (their "schemes"
1613         are not recognized): signal:noise, a:b.
1614 """],
1615 ["""\
1616 Escaped email addresses are not recognized: test\@example.org
1617 """,
1618 """\
1619 <document source="test data">
1620     <paragraph>
1621         Escaped email addresses are not recognized: test@example.org
1622 """],
1623 ]
1624
1625 totest['markup recognition rules'] = [
1626 ["""\
1627 __This__ should be left alone.
1628 """,
1629 """\
1630 <document source="test data">
1631     <paragraph>
1632         __This__ should be left alone.
1633 """],
1634 [r"""
1635 Character-level m\ *a*\ **r**\ ``k``\ `u`:title:\p
1636 with backslash-escaped whitespace, including new\
1637 lines.
1638 """,
1639 """\
1640 <document source="test data">
1641     <paragraph>
1642         Character-level m
1643         <emphasis>
1644             a
1645         <strong>
1646             r
1647         <literal>
1648             k
1649         <title_reference>
1650             u
1651         p
1652         with backslash-escaped whitespace, including newlines.
1653 """],
1654 [u"""\
1655 text-*separated*\u2010*by*\u2011*various*\u2012*dashes*\u2013*and*\u2014*hyphens*.
1656 \u00bf*punctuation*? \u00a1*examples*!\u00a0*\u00a0no-break-space\u00a0*.
1657 """,
1658 u"""\
1659 <document source="test data">
1660     <paragraph>
1661         text-
1662         <emphasis>
1663             separated
1664         \u2010
1665         <emphasis>
1666             by
1667         \u2011
1668         <emphasis>
1669             various
1670         \u2012
1671         <emphasis>
1672             dashes
1673         \u2013
1674         <emphasis>
1675             and
1676         \u2014
1677         <emphasis>
1678             hyphens
1679         .
1680         \xbf
1681         <emphasis>
1682             punctuation
1683         ? \xa1
1684         <emphasis>
1685             examples
1686         !\xa0
1687         <emphasis>
1688             \u00a0no-break-space\u00a0
1689         .
1690 """],
1691 # Whitespace characters:
1692 #  \u180e*MONGOLIAN VOWEL SEPARATOR*\u180e,   fails in Python 2.4
1693 [u"""\
1694 text separated by
1695 *newline*
1696 or *space* or one of
1697 \xa0*NO-BREAK SPACE*\xa0,
1698 \u1680*OGHAM SPACE MARK*\u1680,
1699 \u2000*EN QUAD*\u2000,
1700 \u2001*EM QUAD*\u2001,
1701 \u2002*EN SPACE*\u2002,
1702 \u2003*EM SPACE*\u2003,
1703 \u2004*THREE-PER-EM SPACE*\u2004,
1704 \u2005*FOUR-PER-EM SPACE*\u2005,
1705 \u2006*SIX-PER-EM SPACE*\u2006,
1706 \u2007*FIGURE SPACE*\u2007,
1707 \u2008*PUNCTUATION SPACE*\u2008,
1708 \u2009*THIN SPACE*\u2009,
1709 \u200a*HAIR SPACE*\u200a,
1710 \u202f*NARROW NO-BREAK SPACE*\u202f,
1711 \u205f*MEDIUM MATHEMATICAL SPACE*\u205f,
1712 \u3000*IDEOGRAPHIC SPACE*\u3000,
1713 \u2028*LINE SEPARATOR*\u2028
1714 """,
1715 u"""\
1716 <document source="test data">
1717     <paragraph>
1718         text separated by
1719         <emphasis>
1720             newline
1721         \n\
1722         or \n\
1723         <emphasis>
1724             space
1725          or one of
1726         \xa0
1727         <emphasis>
1728             NO-BREAK SPACE
1729         \xa0,
1730         \u1680
1731         <emphasis>
1732             OGHAM SPACE MARK
1733         \u1680,
1734         \u2000
1735         <emphasis>
1736             EN QUAD
1737         \u2000,
1738         \u2001
1739         <emphasis>
1740             EM QUAD
1741         \u2001,
1742         \u2002
1743         <emphasis>
1744             EN SPACE
1745         \u2002,
1746         \u2003
1747         <emphasis>
1748             EM SPACE
1749         \u2003,
1750         \u2004
1751         <emphasis>
1752             THREE-PER-EM SPACE
1753         \u2004,
1754         \u2005
1755         <emphasis>
1756             FOUR-PER-EM SPACE
1757         \u2005,
1758         \u2006
1759         <emphasis>
1760             SIX-PER-EM SPACE
1761         \u2006,
1762         \u2007
1763         <emphasis>
1764             FIGURE SPACE
1765         \u2007,
1766         \u2008
1767         <emphasis>
1768             PUNCTUATION SPACE
1769         \u2008,
1770         \u2009
1771         <emphasis>
1772             THIN SPACE
1773         \u2009,
1774         \u200a
1775         <emphasis>
1776             HAIR SPACE
1777         \u200a,
1778         \u202f
1779         <emphasis>
1780             NARROW NO-BREAK SPACE
1781         \u202f,
1782         \u205f
1783         <emphasis>
1784             MEDIUM MATHEMATICAL SPACE
1785         \u205f,
1786         \u3000
1787         <emphasis>
1788             IDEOGRAPHIC SPACE
1789         \u3000,
1790     <paragraph>
1791         <emphasis>
1792             LINE SEPARATOR
1793 """],
1794 [u"""\
1795 inline markup separated by non-ASCII whitespace
1796 \xa0**NO-BREAK SPACE**\xa0, \xa0``NO-BREAK SPACE``\xa0, \xa0`NO-BREAK SPACE`\xa0,
1797 \u2000**EN QUAD**\u2000, \u2000``EN QUAD``\u2000, \u2000`EN QUAD`\u2000,
1798 \u202f**NARROW NBSP**\u202f, \u202f``NARROW NBSP``\u202f, \u202f`NARROW NBSP`\u202f,
1799 """,
1800 u"""\
1801 <document source="test data">
1802     <paragraph>
1803         inline markup separated by non-ASCII whitespace
1804         \xa0
1805         <strong>
1806             NO-BREAK SPACE
1807         \xa0, \xa0
1808         <literal>
1809             NO-BREAK SPACE
1810         \xa0, \xa0
1811         <title_reference>
1812             NO-BREAK SPACE
1813         \xa0,
1814         \u2000
1815         <strong>
1816             EN QUAD
1817         \u2000, \u2000
1818         <literal>
1819             EN QUAD
1820         \u2000, \u2000
1821         <title_reference>
1822             EN QUAD
1823         \u2000,
1824         \u202f
1825         <strong>
1826             NARROW NBSP
1827         \u202f, \u202f
1828         <literal>
1829             NARROW NBSP
1830         \u202f, \u202f
1831         <title_reference>
1832             NARROW NBSP
1833         \u202f,
1834 """],
1835 # « * » ‹ * › « * » ‹ * › « * » ‹ * › French,
1836 [u"""\
1837 "Quoted" markup start-string (matched openers & closers) -> no markup:
1838
1839 '*' "*" (*) <*> [*] {*}
1840 ⁅*⁆
1841
1842 Some international quoting styles:
1843 ‘*’ “*” English, ...,
1844 „*“ ‚*‘ »*« ›*‹ German, Czech, ...,
1845 „*” «*» Romanian,
1846 “*„ ‘*‚ Greek,
1847 「*」 『*』traditional Chinese,
1848 ”*” ’*’ »*» ›*› Swedish, Finnish,
1849 „*” ‚*’ Polish,
1850 „*” »*« ’*’ Hungarian,
1851
1852 But this is „*’ emphasized »*‹.
1853 """,
1854 u"""\
1855 <document source="test data">
1856     <paragraph>
1857         "Quoted" markup start-string (matched openers & closers) -> no markup:
1858     <paragraph>
1859         '*' "*" (*) <*> [*] {*}
1860         ⁅*⁆
1861     <paragraph>
1862         Some international quoting styles:
1863         ‘*’ “*” English, ...,
1864         „*“ ‚*‘ »*« ›*‹ German, Czech, ...,
1865         „*” «*» Romanian,
1866         “*„ ‘*‚ Greek,
1867         「*」 『*』traditional Chinese,
1868         ”*” ’*’ »*» ›*› Swedish, Finnish,
1869         „*” ‚*’ Polish,
1870         „*” »*« ’*’ Hungarian,
1871     <paragraph>
1872         But this is „
1873         <emphasis>
1874             ’ emphasized »
1875         ‹.
1876 """],
1877 ]
1878
1879
1880 if __name__ == '__main__':
1881     import unittest
1882     unittest.main(defaultTest='suite')