test/test_parsers/test_rst/test_inline_markup.py

   1 #! /usr/bin/env python
   2 # -*- coding: utf8 -*-
   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 ["""\
 985 `embedded email address <jdoe@example.com>`__
 986
 987 `embedded email address broken across lines <jdoe
 988 @example.com>`__
 989 """,
 990 """\
 991 <document source="test data">
 992     <paragraph>
 993         <reference name="embedded email address" refuri="mailto:jdoe@example.com">
 994             embedded email address
 995     <paragraph>
 996         <reference name="embedded email address broken across lines" refuri="mailto:jdoe@example.com">
 997             embedded email address broken across lines
 998 """],
 999 [r"""
1000 `embedded URI with too much whitespace < http://example.com/
1001 long/path /and  /whitespace >`__
1002
1003 `embedded URI with too much whitespace at end <http://example.com/
1004 long/path /and  /whitespace >`__
1005
1006 `embedded URI with no preceding whitespace<http://example.com>`__
1007
1008 `escaped URI \<http://example.com>`__
1009
1010 See `HTML Anchors: \<a>`_.
1011 """,
1012 """\
1013 <document source="test data">
1014     <paragraph>
1015         <reference anonymous="1" name="embedded URI with too much whitespace < http://example.com/ long/path /and /whitespace >">
1016             embedded URI with too much whitespace < http://example.com/
1017             long/path /and  /whitespace >
1018     <paragraph>
1019         <reference anonymous="1" name="embedded URI with too much whitespace at end <http://example.com/ long/path /and /whitespace >">
1020             embedded URI with too much whitespace at end <http://example.com/
1021             long/path /and  /whitespace >
1022     <paragraph>
1023         <reference anonymous="1" name="embedded URI with no preceding whitespace<http://example.com>">
1024             embedded URI with no preceding whitespace<http://example.com>
1025     <paragraph>
1026         <reference anonymous="1" name="escaped URI <http://example.com>">
1027             escaped URI <http://example.com>
1028     <paragraph>
1029         See \n\
1030         <reference name="HTML Anchors: <a>" refname="html anchors: <a>">
1031             HTML Anchors: <a>
1032         .
1033 """],
1034 ["""\
1035 Relative URIs' reference text can be omitted:
1036
1037 `<reference>`_
1038
1039 `<anonymous>`__
1040 """,
1041 """\
1042 <document source="test data">
1043     <paragraph>
1044         Relative URIs' reference text can be omitted:
1045     <paragraph>
1046         <reference name="reference" refuri="reference">
1047             reference
1048         <target ids="reference" names="reference" refuri="reference">
1049     <paragraph>
1050         <reference name="anonymous" refuri="anonymous">
1051             anonymous
1052 """],
1053 ["""\
1054 Escape trailing low-line char in URIs:
1055
1056 `<reference\_>`_
1057
1058 `<anonymous\_>`__
1059 """,
1060 """\
1061 <document source="test data">
1062     <paragraph>
1063         Escape trailing low-line char in URIs:
1064     <paragraph>
1065         <reference name="reference_" refuri="reference_">
1066             reference_
1067         <target ids="reference" names="reference_" refuri="reference_">
1068     <paragraph>
1069         <reference name="anonymous_" refuri="anonymous_">
1070             anonymous_
1071 """],
1072 ]
1073
1074 totest['embedded_aliases'] = [
1075 ["""\
1076 `phrase reference <alias_>`_
1077 """,
1078 """\
1079 <document source="test data">
1080     <paragraph>
1081         <reference name="phrase reference" refname="alias">
1082             phrase reference
1083         <target names="phrase\ reference" refname="alias">
1084 """],
1085 ["""\
1086 `anonymous reference <alias_>`__
1087 """,
1088 """\
1089 <document source="test data">
1090     <paragraph>
1091         <reference name="anonymous reference" refname="alias">
1092             anonymous reference
1093 """],
1094 ["""\
1095 `embedded alias on next line
1096 <alias_>`__
1097 """,
1098 """\
1099 <document source="test data">
1100     <paragraph>
1101         <reference name="embedded alias on next line" refname="alias">
1102             embedded alias on next line
1103 """],
1104 ["""\
1105 `embedded alias across lines <alias
1106 phrase_>`__
1107 """,
1108 """\
1109 <document source="test data">
1110     <paragraph>
1111         <reference name="embedded alias across lines" refname="alias phrase">
1112             embedded alias across lines
1113 """],
1114 ["""\
1115 `embedded alias with whitespace <alias
1116 long  phrase_>`__
1117 """,
1118 """\
1119 <document source="test data">
1120     <paragraph>
1121         <reference name="embedded alias with whitespace" refname="alias long phrase">
1122             embedded alias with whitespace
1123 """],
1124 [r"""
1125 `embedded alias with too much whitespace < alias_ >`__
1126
1127 `embedded alias with no preceding whitespace<alias_>`__
1128 """,
1129 """\
1130 <document source="test data">
1131     <paragraph>
1132         <reference anonymous="1" name="embedded alias with too much whitespace < alias_ >">
1133             embedded alias with too much whitespace < alias_ >
1134     <paragraph>
1135         <reference anonymous="1" name="embedded alias with no preceding whitespace<alias_>">
1136             embedded alias with no preceding whitespace<alias_>
1137 """],
1138 ]
1139
1140 totest['inline_targets'] = [
1141 ["""\
1142 _`target`
1143
1144 Here is _`another target` in some text. And _`yet
1145 another target`, spanning lines.
1146
1147 _`Here is  a    TaRgeT` with case and spacial difficulties.
1148 """,
1149 """\
1150 <document source="test data">
1151     <paragraph>
1152         <target ids="target" names="target">
1153             target
1154     <paragraph>
1155         Here is \n\
1156         <target ids="another-target" names="another\ target">
1157             another target
1158          in some text. And \n\
1159         <target ids="yet-another-target" names="yet\ another\ target">
1160             yet
1161             another target
1162         , spanning lines.
1163     <paragraph>
1164         <target ids="here-is-a-target" names="here\ is\ a\ target">
1165             Here is  a    TaRgeT
1166          with case and spacial difficulties.
1167 """],
1168 [u"""\
1169 l'_`target1` and l\u2019_`target2` with apostrophe
1170 """,
1171 u"""\
1172 <document source="test data">
1173     <paragraph>
1174         l'
1175         <target ids="target1" names="target1">
1176             target1
1177          and l\u2019
1178         <target ids="target2" names="target2">
1179             target2
1180          with apostrophe
1181 """],
1182 [u"""\
1183 quoted '_`target1`', quoted "_`target2`",
1184 quoted \u2018_`target3`\u2019, quoted \u201c_`target4`\u201d,
1185 quoted \xab_`target5`\xbb
1186 """,
1187 u"""\
1188 <document source="test data">
1189     <paragraph>
1190         quoted '
1191         <target ids="target1" names="target1">
1192             target1
1193         ', quoted "
1194         <target ids="target2" names="target2">
1195             target2
1196         ",
1197         quoted \u2018
1198         <target ids="target3" names="target3">
1199             target3
1200         \u2019, quoted \u201c
1201         <target ids="target4" names="target4">
1202             target4
1203         \u201d,
1204         quoted \xab
1205         <target ids="target5" names="target5">
1206             target5
1207         \xbb
1208 """],
1209 [u"""\
1210 _`'target1'` with quotes, _`"target2"` with quotes,
1211 _`\u2018target3\u2019` with quotes, _`\u201ctarget4\u201d` with quotes,
1212 _`\xabtarget5\xbb` with quotes
1213 """,
1214 u"""\
1215 <document source="test data">
1216     <paragraph>
1217         <target ids="target1" names="'target1'">
1218             'target1'
1219          with quotes, \n\
1220         <target ids="target2" names=""target2"">
1221             "target2"
1222          with quotes,
1223         <target ids="target3" names="\u2018target3\u2019">
1224             \u2018target3\u2019
1225          with quotes, \n\
1226         <target ids="target4" names="\u201ctarget4\u201d">
1227             \u201ctarget4\u201d
1228          with quotes,
1229         <target ids="target5" names="\xabtarget5\xbb">
1230             \xabtarget5\xbb
1231          with quotes
1232 """],
1233 ["""\
1234 But this isn't a _target; targets require backquotes.
1235
1236 And _`this`_ is just plain confusing.
1237 """,
1238 """\
1239 <document source="test data">
1240     <paragraph>
1241         But this isn't a _target; targets require backquotes.
1242     <paragraph>
1243         And \n\
1244         <problematic ids="id2" refid="id1">
1245             _`
1246         this`_ is just plain confusing.
1247     <system_message backrefs="id2" ids="id1" level="2" line="3" source="test data" type="WARNING">
1248         <paragraph>
1249             Inline target start-string without end-string.
1250 """],
1251 ["""\
1252 _`inline target without closing backquote
1253 """,
1254 """\
1255 <document source="test data">
1256     <paragraph>
1257         <problematic ids="id2" refid="id1">
1258             _`
1259         inline target without closing backquote
1260     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
1261         <paragraph>
1262             Inline target start-string without end-string.
1263 """],
1264 ]
1265
1266 totest['footnote_reference'] = [
1267 ["""\
1268 [1]_
1269 """,
1270 """\
1271 <document source="test data">
1272     <paragraph>
1273         <footnote_reference ids="id1" refname="1">
1274             1
1275 """],
1276 ["""\
1277 [#]_
1278 """,
1279 """\
1280 <document source="test data">
1281     <paragraph>
1282         <footnote_reference auto="1" ids="id1">
1283 """],
1284 ["""\
1285 [#label]_
1286 """,
1287 """\
1288 <document source="test data">
1289     <paragraph>
1290         <footnote_reference auto="1" ids="id1" refname="label">
1291 """],
1292 ["""\
1293 [*]_
1294 """,
1295 """\
1296 <document source="test data">
1297     <paragraph>
1298         <footnote_reference auto="*" ids="id1">
1299 """],
1300 ["""\
1301 Adjacent footnote refs are not possible: [*]_[#label]_ [#]_[2]_ [1]_[*]_
1302 """,
1303 """\
1304 <document source="test data">
1305     <paragraph>
1306         Adjacent footnote refs are not possible: [*]_[#label]_ [#]_[2]_ [1]_[*]_
1307 """],
1308 ]
1309
1310 totest['citation_reference'] = [
1311 ["""\
1312 [citation]_
1313 """,
1314 """\
1315 <document source="test data">
1316     <paragraph>
1317         <citation_reference ids="id1" refname="citation">
1318             citation
1319 """],
1320 ["""\
1321 [citation]_ and [cit-ation]_ and [cit.ation]_ and [CIT1]_ but not [CIT 1]_
1322 """,
1323 """\
1324 <document source="test data">
1325     <paragraph>
1326         <citation_reference ids="id1" refname="citation">
1327             citation
1328          and \n\
1329         <citation_reference ids="id2" refname="cit-ation">
1330             cit-ation
1331          and \n\
1332         <citation_reference ids="id3" refname="cit.ation">
1333             cit.ation
1334          and \n\
1335         <citation_reference ids="id4" refname="cit1">
1336             CIT1
1337          but not [CIT 1]_
1338 """],
1339 ["""\
1340 Adjacent citation refs are not possible: [citation]_[CIT1]_
1341 """,
1342 """\
1343 <document source="test data">
1344     <paragraph>
1345         Adjacent citation refs are not possible: [citation]_[CIT1]_
1346 """],
1347 ]
1348
1349 totest['substitution_references'] = [
1350 ["""\
1351 |subref|
1352 """,
1353 """\
1354 <document source="test data">
1355     <paragraph>
1356         <substitution_reference refname="subref">
1357             subref
1358 """],
1359 ["""\
1360 |subref|_ and |subref|__
1361 """,
1362 """\
1363 <document source="test data">
1364     <paragraph>
1365         <reference refname="subref">
1366             <substitution_reference refname="subref">
1367                 subref
1368          and \n\
1369         <reference anonymous="1">
1370             <substitution_reference refname="subref">
1371                 subref
1372 """],
1373 ["""\
1374 |substitution reference|
1375 """,
1376 """\
1377 <document source="test data">
1378     <paragraph>
1379         <substitution_reference refname="substitution reference">
1380             substitution reference
1381 """],
1382 ["""\
1383 |substitution
1384 reference|
1385 """,
1386 """\
1387 <document source="test data">
1388     <paragraph>
1389         <substitution_reference refname="substitution reference">
1390             substitution
1391             reference
1392 """],
1393 ["""\
1394 |substitution reference without closing verbar
1395 """,
1396 """\
1397 <document source="test data">
1398     <paragraph>
1399         <problematic ids="id2" refid="id1">
1400             |
1401         substitution reference without closing verbar
1402     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
1403         <paragraph>
1404             Inline substitution_reference start-string without end-string.
1405 """],
1406 ["""\
1407 first | then || and finally |||
1408 """,
1409 """\
1410 <document source="test data">
1411     <paragraph>
1412         first | then || and finally |||
1413 """],
1414 ]
1415
1416 totest['standalone_hyperlink'] = [
1417 ["""\
1418 http://www.standalone.hyperlink.com
1419
1420 http:/one-slash-only.absolute.path
1421
1422 [http://example.com]
1423
1424 (http://example.com)
1425
1426 <http://example.com>
1427
1428 http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html
1429
1430 http://[3ffe:2a00:100:7031::1] (the final "]" is ambiguous in text)
1431
1432 http://[3ffe:2a00:100:7031::1]/
1433
1434 mailto:someone@somewhere.com
1435
1436 news:comp.lang.python
1437
1438 An email address in a sentence: someone@somewhere.com.
1439
1440 ftp://ends.with.a.period.
1441
1442 (a.question.mark@end?)
1443 """,
1444 """\
1445 <document source="test data">
1446     <paragraph>
1447         <reference refuri="http://www.standalone.hyperlink.com">
1448             http://www.standalone.hyperlink.com
1449     <paragraph>
1450         <reference refuri="http:/one-slash-only.absolute.path">
1451             http:/one-slash-only.absolute.path
1452     <paragraph>
1453         [
1454         <reference refuri="http://example.com">
1455             http://example.com
1456         ]
1457     <paragraph>
1458         (
1459         <reference refuri="http://example.com">
1460             http://example.com
1461         )
1462     <paragraph>
1463         <
1464         <reference refuri="http://example.com">
1465             http://example.com
1466         >
1467     <paragraph>
1468         <reference refuri="http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html">
1469             http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html
1470     <paragraph>
1471         <reference refuri="http://[3ffe:2a00:100:7031::1">
1472             http://[3ffe:2a00:100:7031::1
1473         ] (the final "]" is ambiguous in text)
1474     <paragraph>
1475         <reference refuri="http://[3ffe:2a00:100:7031::1]/">
1476             http://[3ffe:2a00:100:7031::1]/
1477     <paragraph>
1478         <reference refuri="mailto:someone@somewhere.com">
1479             mailto:someone@somewhere.com
1480     <paragraph>
1481         <reference refuri="news:comp.lang.python">
1482             news:comp.lang.python
1483     <paragraph>
1484         An email address in a sentence: \n\
1485         <reference refuri="mailto:someone@somewhere.com">
1486             someone@somewhere.com
1487         .
1488     <paragraph>
1489         <reference refuri="ftp://ends.with.a.period">
1490             ftp://ends.with.a.period
1491         .
1492     <paragraph>
1493         (
1494         <reference refuri="mailto:a.question.mark@end">
1495             a.question.mark@end
1496         ?)
1497 """],
1498 [r"""
1499 Valid URLs with escaped markup characters:
1500
1501 http://example.com/\*content\*/whatever
1502
1503 http://example.com/\*content*/whatever
1504 """,
1505 """\
1506 <document source="test data">
1507     <paragraph>
1508         Valid URLs with escaped markup characters:
1509     <paragraph>
1510         <reference refuri="http://example.com/*content*/whatever">
1511             http://example.com/*content*/whatever
1512     <paragraph>
1513         <reference refuri="http://example.com/*content*/whatever">
1514             http://example.com/*content*/whatever
1515 """],
1516 ["""\
1517 Valid URLs may end with punctuation inside "<>":
1518
1519 <http://example.org/ends-with-dot.>
1520 """,
1521 """\
1522 <document source="test data">
1523     <paragraph>
1524         Valid URLs may end with punctuation inside "<>":
1525     <paragraph>
1526         <
1527         <reference refuri="http://example.org/ends-with-dot.">
1528             http://example.org/ends-with-dot.
1529         >
1530 """],
1531 ["""\
1532 Valid URLs with interesting endings:
1533
1534 http://example.org/ends-with-pluses++
1535 """,
1536 """\
1537 <document source="test data">
1538     <paragraph>
1539         Valid URLs with interesting endings:
1540     <paragraph>
1541         <reference refuri="http://example.org/ends-with-pluses++">
1542             http://example.org/ends-with-pluses++
1543 """],
1544 ["""\
1545 None of these are standalone hyperlinks (their "schemes"
1546 are not recognized): signal:noise, a:b.
1547 """,
1548 """\
1549 <document source="test data">
1550     <paragraph>
1551         None of these are standalone hyperlinks (their "schemes"
1552         are not recognized): signal:noise, a:b.
1553 """],
1554 ["""\
1555 Escaped email addresses are not recognized: test\@example.org
1556 """,
1557 """\
1558 <document source="test data">
1559     <paragraph>
1560         Escaped email addresses are not recognized: test@example.org
1561 """],
1562 ]
1563
1564 totest['markup recognition rules'] = [
1565 ["""\
1566 __This__ should be left alone.
1567 """,
1568 """\
1569 <document source="test data">
1570     <paragraph>
1571         __This__ should be left alone.
1572 """],
1573 [r"""
1574 Character-level m\ *a*\ **r**\ ``k``\ `u`:title:\p
1575 with backslash-escaped whitespace, including new\
1576 lines.
1577 """,
1578 """\
1579 <document source="test data">
1580     <paragraph>
1581         Character-level m
1582         <emphasis>
1583             a
1584         <strong>
1585             r
1586         <literal>
1587             k
1588         <title_reference>
1589             u
1590         p
1591         with backslash-escaped whitespace, including newlines.
1592 """],
1593 [u"""\
1594 text-*separated*\u2010*by*\u2011*various*\u2012*dashes*\u2013*and*\u2014*hyphens*.
1595 \u00bf*punctuation*? \u00a1*examples*!\u00a0*\u00a0no-break-space\u00a0*.
1596 """,
1597 u"""\
1598 <document source="test data">
1599     <paragraph>
1600         text-
1601         <emphasis>
1602             separated
1603         \u2010
1604         <emphasis>
1605             by
1606         \u2011
1607         <emphasis>
1608             various
1609         \u2012
1610         <emphasis>
1611             dashes
1612         \u2013
1613         <emphasis>
1614             and
1615         \u2014
1616         <emphasis>
1617             hyphens
1618         .
1619         \xbf
1620         <emphasis>
1621             punctuation
1622         ? \xa1
1623         <emphasis>
1624             examples
1625         !\xa0
1626         <emphasis>
1627             \u00a0no-break-space\u00a0
1628         .
1629 """],
1630 # Whitespace characters:
1631 #  \u180e*MONGOLIAN VOWEL SEPARATOR*\u180e,   fails in Python 2.4
1632 [u"""\
1633 text separated by
1634 *newline*
1635 or *space* or one of
1636 \xa0*NO-BREAK SPACE*\xa0,
1637 \u1680*OGHAM SPACE MARK*\u1680,
1638 \u2000*EN QUAD*\u2000,
1639 \u2001*EM QUAD*\u2001,
1640 \u2002*EN SPACE*\u2002,
1641 \u2003*EM SPACE*\u2003,
1642 \u2004*THREE-PER-EM SPACE*\u2004,
1643 \u2005*FOUR-PER-EM SPACE*\u2005,
1644 \u2006*SIX-PER-EM SPACE*\u2006,
1645 \u2007*FIGURE SPACE*\u2007,
1646 \u2008*PUNCTUATION SPACE*\u2008,
1647 \u2009*THIN SPACE*\u2009,
1648 \u200a*HAIR SPACE*\u200a,
1649 \u202f*NARROW NO-BREAK SPACE*\u202f,
1650 \u205f*MEDIUM MATHEMATICAL SPACE*\u205f,
1651 \u3000*IDEOGRAPHIC SPACE*\u3000,
1652 \u2028*LINE SEPARATOR*\u2028
1653 """,
1654 u"""\
1655 <document source="test data">
1656     <paragraph>
1657         text separated by
1658         <emphasis>
1659             newline
1660         \n\
1661         or \n\
1662         <emphasis>
1663             space
1664          or one of
1665         \xa0
1666         <emphasis>
1667             NO-BREAK SPACE
1668         \xa0,
1669         \u1680
1670         <emphasis>
1671             OGHAM SPACE MARK
1672         \u1680,
1673         \u2000
1674         <emphasis>
1675             EN QUAD
1676         \u2000,
1677         \u2001
1678         <emphasis>
1679             EM QUAD
1680         \u2001,
1681         \u2002
1682         <emphasis>
1683             EN SPACE
1684         \u2002,
1685         \u2003
1686         <emphasis>
1687             EM SPACE
1688         \u2003,
1689         \u2004
1690         <emphasis>
1691             THREE-PER-EM SPACE
1692         \u2004,
1693         \u2005
1694         <emphasis>
1695             FOUR-PER-EM SPACE
1696         \u2005,
1697         \u2006
1698         <emphasis>
1699             SIX-PER-EM SPACE
1700         \u2006,
1701         \u2007
1702         <emphasis>
1703             FIGURE SPACE
1704         \u2007,
1705         \u2008
1706         <emphasis>
1707             PUNCTUATION SPACE
1708         \u2008,
1709         \u2009
1710         <emphasis>
1711             THIN SPACE
1712         \u2009,
1713         \u200a
1714         <emphasis>
1715             HAIR SPACE
1716         \u200a,
1717         \u202f
1718         <emphasis>
1719             NARROW NO-BREAK SPACE
1720         \u202f,
1721         \u205f
1722         <emphasis>
1723             MEDIUM MATHEMATICAL SPACE
1724         \u205f,
1725         \u3000
1726         <emphasis>
1727             IDEOGRAPHIC SPACE
1728         \u3000,
1729     <paragraph>
1730         <emphasis>
1731             LINE SEPARATOR
1732 """],
1733 # « * » ‹ * › « * » ‹ * › « * » ‹ * › French,
1734 [u"""\
1735 "Quoted" markup start-string (matched openers & closers) -> no markup:
1736
1737 '*' "*" (*) <*> [*] {*}
1738 ⁅*⁆
1739
1740 Some international quoting styles:
1741 ‘*’ “*” English, ...,
1742 „*“ ‚*‘ »*« ›*‹ German, Czech, ...,
1743 „*” «*» Romanian,
1744 “*„ ‘*‚ Greek,
1745 「*」 『*』traditional Chinese,
1746 ”*” ’*’ »*» ›*› Swedish, Finnish,
1747 „*” ‚*’ Polish,
1748 „*” »*« ’*’ Hungarian,
1749
1750 But this is „*’ emphasized »*‹.
1751 """,
1752 u"""\
1753 <document source="test data">
1754     <paragraph>
1755         "Quoted" markup start-string (matched openers & closers) -> no markup:
1756     <paragraph>
1757         '*' "*" (*) <*> [*] {*}
1758         ⁅*⁆
1759     <paragraph>
1760         Some international quoting styles:
1761         ‘*’ “*” English, ...,
1762         „*“ ‚*‘ »*« ›*‹ German, Czech, ...,
1763         „*” «*» Romanian,
1764         “*„ ‘*‚ Greek,
1765         「*」 『*』traditional Chinese,
1766         ”*” ’*’ »*» ›*› Swedish, Finnish,
1767         „*” ‚*’ Polish,
1768         „*” »*« ’*’ Hungarian,
1769     <paragraph>
1770         But this is „
1771         <emphasis>
1772             ’ emphasized »
1773         ‹.
1774 """],
1775 ]
1776
1777
1778 if __name__ == '__main__':
1779     import unittest
1780     unittest.main(defaultTest='suite')