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