docutils/test/test_parsers/test_rst/test_inline_markup.py

   1 #! /usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3
   4 # $Id$
   5 # Author: David Goodger <goodger@python.org>
   6 # Copyright: This module has been placed in the public domain.
   7
   8 """
   9 Tests for inline markup in docutils/parsers/rst/states.py.
  10 Interpreted text tests are in a separate module, test_interpreted.py.
  11 """
  12
  13 from __init__ import DocutilsTestSupport
  14
  15 def suite():
  16     s = DocutilsTestSupport.ParserTestSuite()
  17     s.generateTests(totest)
  18     return s
  19
  20 totest = {}
  21
  22 totest['emphasis'] = [
  23 ["""\
  24 *emphasis*
  25 """,
  26 """\
  27 <document source="test data">
  28     <paragraph>
  29         <emphasis>
  30             emphasis
  31 """],
  32 [u"""\
  33 l'*emphasis* with the *emphasis*' apostrophe.
  34 l\u2019*emphasis* with the *emphasis*\u2019 apostrophe.
  35 """,
  36 u"""\
  37 <document source="test data">
  38     <paragraph>
  39         l\'
  40         <emphasis>
  41             emphasis
  42          with the \n\
  43         <emphasis>
  44             emphasis
  45         \' apostrophe.
  46         l\u2019
  47         <emphasis>
  48             emphasis
  49          with the \n\
  50         <emphasis>
  51             emphasis
  52         \u2019 apostrophe.
  53 """],
  54 ["""\
  55 *emphasized sentence
  56 across lines*
  57 """,
  58 """\
  59 <document source="test data">
  60     <paragraph>
  61         <emphasis>
  62             emphasized sentence
  63             across lines
  64 """],
  65 ["""\
  66 *emphasis without closing asterisk
  67 """,
  68 """\
  69 <document source="test data">
  70     <paragraph>
  71         <problematic ids="id2" refid="id1">
  72             *
  73         emphasis without closing asterisk
  74     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
  75         <paragraph>
  76             Inline emphasis start-string without end-string.
  77 """],
  78 [r"""some punctuation is allowed around inline markup, e.g.
  79 /*emphasis*/, -*emphasis*-, and :*emphasis*: (delimiters),
  80 (*emphasis*), [*emphasis*], <*emphasis*>, {*emphasis*} (open/close pairs)
  81 *emphasis*., *emphasis*,, *emphasis*!, and *emphasis*\ (closing delimiters),
  82
  83 but not
  84 )*emphasis*(, ]*emphasis*[, >*emphasis*>, }*emphasis*{ (close/open pairs),
  85 (*), [*], '*' or '"*"' ("quoted" start-string),
  86 x*2* or 2*x* (alphanumeric char before),
  87 \*args or * (escaped, whitespace behind start-string),
  88 or *the\* *stars\* *inside* (escaped, whitespace before end-string).
  89
  90 However, '*args' will trigger a warning and may be problematic.
  91
  92 what about *this**?
  93 """,
  94 """\
  95 <document source="test data">
  96     <paragraph>
  97         some punctuation is allowed around inline markup, e.g.
  98         /
  99         <emphasis>
 100             emphasis
 101         /, -
 102         <emphasis>
 103             emphasis
 104         -, and :
 105         <emphasis>
 106             emphasis
 107         : (delimiters),
 108         (
 109         <emphasis>
 110             emphasis
 111         ), [
 112         <emphasis>
 113             emphasis
 114         ], <
 115         <emphasis>
 116             emphasis
 117         >, {
 118         <emphasis>
 119             emphasis
 120         } (open/close pairs)
 121         <emphasis>
 122             emphasis
 123         ., \n\
 124         <emphasis>
 125             emphasis
 126         ,, \n\
 127         <emphasis>
 128             emphasis
 129         !, and \n\
 130         <emphasis>
 131             emphasis
 132         (closing delimiters),
 133     <paragraph>
 134         but not
 135         )*emphasis*(, ]*emphasis*[, >*emphasis*>, }*emphasis*{ (close/open pairs),
 136         (*), [*], '*' or '"*"' ("quoted" start-string),
 137         x*2* or 2*x* (alphanumeric char before),
 138         *args or * (escaped, whitespace behind start-string),
 139         or \n\
 140         <emphasis>
 141             the* *stars* *inside
 142          (escaped, whitespace before end-string).
 143     <paragraph>
 144         However, '
 145         <problematic ids="id2" refid="id1">
 146             *
 147         args' will trigger a warning and may be problematic.
 148     <system_message backrefs="id2" ids="id1" level="2" line="13" source="test data" type="WARNING">
 149         <paragraph>
 150             Inline emphasis start-string without end-string.
 151     <paragraph>
 152         what about \n\
 153         <emphasis>
 154             this*
 155         ?
 156 """],
 157 [u"""\
 158 Quotes around inline markup:
 159
 160 '*emphasis*' "*emphasis*" Straight,
 161 ‘*emphasis*’ “*emphasis*” English, ...,
 162 « *emphasis* » ‹ *emphasis* › « *emphasis* » ‹ *emphasis* ›
 163 « *emphasis* » ‹ *emphasis* › French,
 164 „*emphasis*“ ‚*emphasis*‘ »*emphasis*« ›*emphasis*‹ German, Czech, ...,
 165 „*emphasis*” «*emphasis*» Romanian,
 166 “*emphasis*„ ‘*emphasis*‚ Greek,
 167 「*emphasis*」 『*emphasis*』traditional Chinese,
 168 ”*emphasis*” ’*emphasis*’ »*emphasis*» ›*emphasis*› Swedish, Finnish,
 169 „*emphasis*” ‚*emphasis*’ Polish,
 170 „*emphasis*” »*emphasis*« ’*emphasis*’ Hungarian,
 171 """,
 172 u"""\
 173 <document source="test data">
 174     <paragraph>
 175         Quotes around inline markup:
 176     <paragraph>
 177         \'
 178         <emphasis>
 179             emphasis
 180         \' "
 181         <emphasis>
 182             emphasis
 183         " Straight,
 184         \u2018
 185         <emphasis>
 186             emphasis
 187         \u2019 \u201c
 188         <emphasis>
 189             emphasis
 190         \u201d English, ...,
 191         \xab\u202f
 192         <emphasis>
 193             emphasis
 194         \u202f\xbb \u2039\u202f
 195         <emphasis>
 196             emphasis
 197         \u202f\u203a \xab\xa0
 198         <emphasis>
 199             emphasis
 200         \xa0\xbb \u2039\xa0
 201         <emphasis>
 202             emphasis
 203         \xa0\u203a
 204         \xab\u2005
 205         <emphasis>
 206             emphasis
 207         \u2005\xbb \u2039\u2005
 208         <emphasis>
 209             emphasis
 210         \u2005\u203a French,
 211         \u201e
 212         <emphasis>
 213             emphasis
 214         \u201c \u201a
 215         <emphasis>
 216             emphasis
 217         \u2018 \xbb
 218         <emphasis>
 219             emphasis
 220         \xab \u203a
 221         <emphasis>
 222             emphasis
 223         \u2039 German, Czech, ...,
 224         \u201e
 225         <emphasis>
 226             emphasis
 227         \u201d \xab
 228         <emphasis>
 229             emphasis
 230         \xbb Romanian,
 231         \u201c
 232         <emphasis>
 233             emphasis
 234         \u201e \u2018
 235         <emphasis>
 236             emphasis
 237         \u201a Greek,
 238         \u300c
 239         <emphasis>
 240             emphasis
 241         \u300d \u300e
 242         <emphasis>
 243             emphasis
 244         \u300ftraditional Chinese,
 245         \u201d
 246         <emphasis>
 247             emphasis
 248         \u201d \u2019
 249         <emphasis>
 250             emphasis
 251         \u2019 \xbb
 252         <emphasis>
 253             emphasis
 254         \xbb \u203a
 255         <emphasis>
 256             emphasis
 257         \u203a Swedish, Finnish,
 258         \u201e
 259         <emphasis>
 260             emphasis
 261         \u201d \u201a
 262         <emphasis>
 263             emphasis
 264         \u2019 Polish,
 265         \u201e
 266         <emphasis>
 267             emphasis
 268         \u201d \xbb
 269         <emphasis>
 270             emphasis
 271         \xab \u2019
 272         <emphasis>
 273             emphasis
 274         \u2019 Hungarian,
 275 """],
 276 [r"""
 277 Emphasized asterisk: *\**
 278
 279 Emphasized double asterisk: *\***
 280 """,
 281 """\
 282 <document source="test data">
 283     <paragraph>
 284         Emphasized asterisk: \n\
 285         <emphasis>
 286             *
 287     <paragraph>
 288         Emphasized double asterisk: \n\
 289         <emphasis>
 290             **
 291 """],
 292 ]
 293
 294 totest['strong'] = [
 295 ["""\
 296 **strong**
 297 """,
 298 """\
 299 <document source="test data">
 300     <paragraph>
 301         <strong>
 302             strong
 303 """],
 304 [u"""\
 305 l'**strong** and l\u2019**strong** with apostrophe
 306 """,
 307 u"""\
 308 <document source="test data">
 309     <paragraph>
 310         l'
 311         <strong>
 312             strong
 313          and l\u2019
 314         <strong>
 315             strong
 316          with apostrophe
 317 """],
 318 [u"""\
 319 quoted '**strong**', quoted "**strong**",
 320 quoted \u2018**strong**\u2019, quoted \u201c**strong**\u201d,
 321 quoted \xab**strong**\xbb
 322 """,
 323 u"""\
 324 <document source="test data">
 325     <paragraph>
 326         quoted '
 327         <strong>
 328             strong
 329         ', quoted "
 330         <strong>
 331             strong
 332         ",
 333         quoted \u2018
 334         <strong>
 335             strong
 336         \u2019, quoted \u201c
 337         <strong>
 338             strong
 339         \u201d,
 340         quoted \xab
 341         <strong>
 342             strong
 343         \xbb
 344 """],
 345 [r"""
 346 (**strong**) but not (**) or '(** ' or x**2 or \**kwargs or **
 347
 348 (however, '**kwargs' will trigger a warning and may be problematic)
 349 """,
 350 """\
 351 <document source="test data">
 352     <paragraph>
 353         (
 354         <strong>
 355             strong
 356         ) but not (**) or '(** ' or x**2 or **kwargs or **
 357     <paragraph>
 358         (however, '
 359         <problematic ids="id2" refid="id1">
 360             **
 361         kwargs' will trigger a warning and may be problematic)
 362     <system_message backrefs="id2" ids="id1" level="2" line="4" source="test data" type="WARNING">
 363         <paragraph>
 364             Inline strong start-string without end-string.
 365 """],
 366 ["""\
 367 Strong asterisk: *****
 368
 369 Strong double asterisk: ******
 370 """,
 371 """\
 372 <document source="test data">
 373     <paragraph>
 374         Strong asterisk: \n\
 375         <strong>
 376             *
 377     <paragraph>
 378         Strong double asterisk: \n\
 379         <strong>
 380             **
 381 """],
 382 ["""\
 383 **strong without closing asterisks
 384 """,
 385 """\
 386 <document source="test data">
 387     <paragraph>
 388         <problematic ids="id2" refid="id1">
 389             **
 390         strong without closing asterisks
 391     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 392         <paragraph>
 393             Inline strong start-string without end-string.
 394 """],
 395 ]
 396
 397 totest['literal'] = [
 398 ["""\
 399 ``literal``
 400 """,
 401 """\
 402 <document source="test data">
 403     <paragraph>
 404         <literal>
 405             literal
 406 """],
 407 [r"""
 408 ``\literal``
 409 """,
 410 """\
 411 <document source="test data">
 412     <paragraph>
 413         <literal>
 414             \\literal
 415 """],
 416 [r"""
 417 ``lite\ral``
 418 """,
 419 """\
 420 <document source="test data">
 421     <paragraph>
 422         <literal>
 423             lite\\ral
 424 """],
 425 [r"""
 426 ``literal\``
 427 """,
 428 """\
 429 <document source="test data">
 430     <paragraph>
 431         <literal>
 432             literal\\
 433 """],
 434 [u"""\
 435 l'``literal`` and l\u2019``literal`` with apostrophe
 436 """,
 437 u"""\
 438 <document source="test data">
 439     <paragraph>
 440         l'
 441         <literal>
 442             literal
 443          and l\u2019
 444         <literal>
 445             literal
 446          with apostrophe
 447 """],
 448 [u"""\
 449 quoted '``literal``', quoted "``literal``",
 450 quoted \u2018``literal``\u2019, quoted \u201c``literal``\u201d,
 451 quoted \xab``literal``\xbb
 452 """,
 453 u"""\
 454 <document source="test data">
 455     <paragraph>
 456         quoted '
 457         <literal>
 458             literal
 459         ', quoted "
 460         <literal>
 461             literal
 462         ",
 463         quoted \u2018
 464         <literal>
 465             literal
 466         \u2019, quoted \u201c
 467         <literal>
 468             literal
 469         \u201d,
 470         quoted \xab
 471         <literal>
 472             literal
 473         \xbb
 474 """],
 475 [u"""\
 476 ``'literal'`` with quotes, ``"literal"`` with quotes,
 477 ``\u2018literal\u2019`` with quotes, ``\u201cliteral\u201d`` with quotes,
 478 ``\xabliteral\xbb`` with quotes
 479 """,
 480 u"""\
 481 <document source="test data">
 482     <paragraph>
 483         <literal>
 484             'literal'
 485          with quotes, \n\
 486         <literal>
 487             "literal"
 488          with quotes,
 489         <literal>
 490             \u2018literal\u2019
 491          with quotes, \n\
 492         <literal>
 493             \u201cliteral\u201d
 494          with quotes,
 495         <literal>
 496             \xabliteral\xbb
 497          with quotes
 498 """],
 499 [r"""
 500 ``literal ``TeX quotes'' & \backslash`` but not "``" or ``
 501
 502 (however, ``standalone TeX quotes'' will trigger a warning
 503 and may be problematic)
 504 """,
 505 """\
 506 <document source="test data">
 507     <paragraph>
 508         <literal>
 509             literal ``TeX quotes'' & \\backslash
 510          but not "``" or ``
 511     <paragraph>
 512         (however, \n\
 513         <problematic ids="id2" refid="id1">
 514             ``
 515         standalone TeX quotes'' will trigger a warning
 516         and may be problematic)
 517     <system_message backrefs="id2" ids="id1" level="2" line="4" source="test data" type="WARNING">
 518         <paragraph>
 519             Inline literal start-string without end-string.
 520 """],
 521 ["""\
 522 Find the ```interpreted text``` in this paragraph!
 523 """,
 524 """\
 525 <document source="test data">
 526     <paragraph>
 527         Find the \n\
 528         <literal>
 529             `interpreted text`
 530          in this paragraph!
 531 """],
 532 ["""\
 533 ``literal without closing backquotes
 534 """,
 535 """\
 536 <document source="test data">
 537     <paragraph>
 538         <problematic ids="id2" refid="id1">
 539             ``
 540         literal without closing backquotes
 541     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 542         <paragraph>
 543             Inline literal start-string without end-string.
 544 """],
 545 [r"""
 546 Python ``list``\s use square bracket syntax.
 547 """,
 548 """\
 549 <document source="test data">
 550     <paragraph>
 551         Python \n\
 552         <literal>
 553             list
 554         s use square bracket syntax.
 555 """],
 556 [r"""
 557 Blank after opening `` not allowed.
 558 """,
 559 """\
 560 <document source="test data">
 561     <paragraph>
 562         Blank after opening `` not allowed.
 563 """],
 564 [r"""
 565 no blank ``after closing``continues`` literal.
 566 """,
 567 """\
 568 <document source="test data">
 569     <paragraph>
 570         no blank \n\
 571         <literal>
 572             after closing``continues
 573          literal.
 574 """],
 575 [r"""
 576 dot ``after closing``. is possible.
 577 """,
 578 """\
 579 <document source="test data">
 580     <paragraph>
 581         dot \n\
 582         <literal>
 583             after closing
 584         . is possible.
 585 """],
 586 ]
 587
 588 totest['references'] = [
 589 ["""\
 590 ref_
 591 """,
 592 """\
 593 <document source="test data">
 594     <paragraph>
 595         <reference name="ref" refname="ref">
 596             ref
 597 """],
 598 [u"""\
 599 l'ref_ and l\u2019ref_ with apostrophe
 600 """,
 601 u"""\
 602 <document source="test data">
 603     <paragraph>
 604         l'
 605         <reference name="ref" refname="ref">
 606             ref
 607          and l\u2019
 608         <reference name="ref" refname="ref">
 609             ref
 610          with apostrophe
 611 """],
 612 [u"""\
 613 quoted 'ref_', quoted "ref_",
 614 quoted \u2018ref_\u2019, quoted \u201cref_\u201d,
 615 quoted \xabref_\xbb,
 616 but not 'ref ref'_, "ref ref"_, \u2018ref ref\u2019_,
 617 \u201cref ref\u201d_, or \xabref ref\xbb_
 618 """,
 619 u"""\
 620 <document source="test data">
 621     <paragraph>
 622         quoted '
 623         <reference name="ref" refname="ref">
 624             ref
 625         ', quoted "
 626         <reference name="ref" refname="ref">
 627             ref
 628         ",
 629         quoted \u2018
 630         <reference name="ref" refname="ref">
 631             ref
 632         \u2019, quoted \u201c
 633         <reference name="ref" refname="ref">
 634             ref
 635         \u201d,
 636         quoted \xab
 637         <reference name="ref" refname="ref">
 638             ref
 639         \xbb,
 640         but not 'ref ref'_, "ref ref"_, \u2018ref ref\u2019_,
 641         \u201cref ref\u201d_, or \xabref ref\xbb_
 642 """],
 643 ["""\
 644 ref__
 645 """,
 646 """\
 647 <document source="test data">
 648     <paragraph>
 649         <reference anonymous="1" name="ref">
 650             ref
 651 """],
 652 [u"""\
 653 l'ref__ and l\u2019ref__ with apostrophe
 654 """,
 655 u"""\
 656 <document source="test data">
 657     <paragraph>
 658         l'
 659         <reference anonymous="1" name="ref">
 660             ref
 661          and l\u2019
 662         <reference anonymous="1" name="ref">
 663             ref
 664          with apostrophe
 665 """],
 666 [u"""\
 667 quoted 'ref__', quoted "ref__",
 668 quoted \u2018ref__\u2019, quoted \u201cref__\u201d,
 669 quoted \xabref__\xbb,
 670 but not 'ref ref'__, "ref ref"__, \u2018ref ref\u2019__,
 671 \u201cref ref\u201d__, or \xabref ref\xbb__
 672 """,
 673 u"""\
 674 <document source="test data">
 675     <paragraph>
 676         quoted '
 677         <reference anonymous="1" name="ref">
 678             ref
 679         ', quoted "
 680         <reference anonymous="1" name="ref">
 681             ref
 682         ",
 683         quoted \u2018
 684         <reference anonymous="1" name="ref">
 685             ref
 686         \u2019, quoted \u201c
 687         <reference anonymous="1" name="ref">
 688             ref
 689         \u201d,
 690         quoted \xab
 691         <reference anonymous="1" name="ref">
 692             ref
 693         \xbb,
 694         but not 'ref ref'__, "ref ref"__, \u2018ref ref\u2019__,
 695         \u201cref ref\u201d__, or \xabref ref\xbb__
 696 """],
 697 ["""\
 698 ref_, r_, r_e-f_, -ref_, and anonymousref__,
 699 but not _ref_ or __attr__ or object.__attr__
 700 """,
 701 """\
 702 <document source="test data">
 703     <paragraph>
 704         <reference name="ref" refname="ref">
 705             ref
 706         , \n\
 707         <reference name="r" refname="r">
 708             r
 709         , \n\
 710         <reference name="r_e-f" refname="r_e-f">
 711             r_e-f
 712         , -
 713         <reference name="ref" refname="ref">
 714             ref
 715         , and \n\
 716         <reference anonymous="1" name="anonymousref">
 717             anonymousref
 718         ,
 719         but not _ref_ or __attr__ or object.__attr__
 720 """],
 721 ]
 722
 723 totest['phrase_references'] = [
 724 ["""\
 725 `phrase reference`_
 726 """,
 727 """\
 728 <document source="test data">
 729     <paragraph>
 730         <reference name="phrase reference" refname="phrase reference">
 731             phrase reference
 732 """],
 733 [u"""\
 734 l'`phrase reference`_ and l\u2019`phrase reference`_ with apostrophe
 735 """,
 736 u"""\
 737 <document source="test data">
 738     <paragraph>
 739         l'
 740         <reference name="phrase reference" refname="phrase reference">
 741             phrase reference
 742          and l\u2019
 743         <reference name="phrase reference" refname="phrase reference">
 744             phrase reference
 745          with apostrophe
 746 """],
 747 [u"""\
 748 quoted '`phrase reference`_', quoted "`phrase reference`_",
 749 quoted \u2018`phrase reference`_\u2019,
 750 quoted \u201c`phrase reference`_\u201d,
 751 quoted \xab`phrase reference`_\xbb
 752 """,
 753 u"""\
 754 <document source="test data">
 755     <paragraph>
 756         quoted '
 757         <reference name="phrase reference" refname="phrase reference">
 758             phrase reference
 759         ', quoted "
 760         <reference name="phrase reference" refname="phrase reference">
 761             phrase reference
 762         ",
 763         quoted \u2018
 764         <reference name="phrase reference" refname="phrase reference">
 765             phrase reference
 766         \u2019,
 767         quoted \u201c
 768         <reference name="phrase reference" refname="phrase reference">
 769             phrase reference
 770         \u201d,
 771         quoted \xab
 772         <reference name="phrase reference" refname="phrase reference">
 773             phrase reference
 774         \xbb
 775 """],
 776 [u"""\
 777 `'phrase reference'`_ with quotes, `"phrase reference"`_ with quotes,
 778 `\u2018phrase reference\u2019`_ with quotes,
 779 `\u201cphrase reference\u201d`_ with quotes,
 780 `\xabphrase reference\xbb`_ with quotes
 781 """,
 782 u"""\
 783 <document source="test data">
 784     <paragraph>
 785         <reference name="'phrase reference'" refname="'phrase reference'">
 786             'phrase reference'
 787          with quotes, \n\
 788         <reference name=""phrase reference"" refname=""phrase reference"">
 789             "phrase reference"
 790          with quotes,
 791         <reference name="\u2018phrase reference\u2019" refname="\u2018phrase reference\u2019">
 792             \u2018phrase reference\u2019
 793          with quotes,
 794         <reference name="\u201cphrase reference\u201d" refname="\u201cphrase reference\u201d">
 795             \u201cphrase reference\u201d
 796          with quotes,
 797         <reference name="\xabphrase reference\xbb" refname="\xabphrase reference\xbb">
 798             \xabphrase reference\xbb
 799          with quotes
 800 """],
 801 ["""\
 802 `anonymous reference`__
 803 """,
 804 """\
 805 <document source="test data">
 806     <paragraph>
 807         <reference anonymous="1" name="anonymous reference">
 808             anonymous reference
 809 """],
 810 [u"""\
 811 l'`anonymous reference`__ and l\u2019`anonymous reference`__ with apostrophe
 812 """,
 813 u"""\
 814 <document source="test data">
 815     <paragraph>
 816         l'
 817         <reference anonymous="1" name="anonymous reference">
 818             anonymous reference
 819          and l\u2019
 820         <reference anonymous="1" name="anonymous reference">
 821             anonymous reference
 822          with apostrophe
 823 """],
 824 [u"""\
 825 quoted '`anonymous reference`__', quoted "`anonymous reference`__",
 826 quoted \u2018`anonymous reference`__\u2019,
 827 quoted \u201c`anonymous reference`__\u201d,
 828 quoted \xab`anonymous reference`__\xbb
 829 """,
 830 u"""\
 831 <document source="test data">
 832     <paragraph>
 833         quoted '
 834         <reference anonymous="1" name="anonymous reference">
 835             anonymous reference
 836         ', quoted "
 837         <reference anonymous="1" name="anonymous reference">
 838             anonymous reference
 839         ",
 840         quoted \u2018
 841         <reference anonymous="1" name="anonymous reference">
 842             anonymous reference
 843         \u2019,
 844         quoted \u201c
 845         <reference anonymous="1" name="anonymous reference">
 846             anonymous reference
 847         \u201d,
 848         quoted \xab
 849         <reference anonymous="1" name="anonymous reference">
 850             anonymous reference
 851         \xbb
 852 """],
 853 [u"""\
 854 `'anonymous reference'`__ with quotes, `"anonymous reference"`__ with quotes,
 855 `\u2018anonymous reference\u2019`__ with quotes,
 856 `\u201canonymous reference\u201d`__ with quotes,
 857 `\xabanonymous reference\xbb`__ with quotes
 858 """,
 859 u"""\
 860 <document source="test data">
 861     <paragraph>
 862         <reference anonymous="1" name="'anonymous reference'">
 863             'anonymous reference'
 864          with quotes, \n\
 865         <reference anonymous="1" name=""anonymous reference"">
 866             "anonymous reference"
 867          with quotes,
 868         <reference anonymous="1" name="\u2018anonymous reference\u2019">
 869             \u2018anonymous reference\u2019
 870          with quotes,
 871         <reference anonymous="1" name="\u201canonymous reference\u201d">
 872             \u201canonymous reference\u201d
 873          with quotes,
 874         <reference anonymous="1" name="\xabanonymous reference\xbb">
 875             \xabanonymous reference\xbb
 876          with quotes
 877 """],
 878 ["""\
 879 `phrase reference
 880 across lines`_
 881 """,
 882 """\
 883 <document source="test data">
 884     <paragraph>
 885         <reference name="phrase reference across lines" refname="phrase reference across lines">
 886             phrase reference
 887             across lines
 888 """],
 889 ["""\
 890 `phrase\`_ reference`_
 891 """,
 892 """\
 893 <document source="test data">
 894     <paragraph>
 895         <reference name="phrase`_ reference" refname="phrase`_ reference">
 896             phrase`_ reference
 897 """],
 898 ["""\
 899 Invalid phrase reference:
 900
 901 :role:`phrase reference`_
 902 """,
 903 """\
 904 <document source="test data">
 905     <paragraph>
 906         Invalid phrase reference:
 907     <paragraph>
 908         <problematic ids="id2" refid="id1">
 909             :role:`phrase reference`_
 910     <system_message backrefs="id2" ids="id1" level="2" line="3" source="test data" type="WARNING">
 911         <paragraph>
 912             Mismatch: both interpreted text role prefix and reference suffix.
 913 """],
 914 ["""\
 915 Invalid phrase reference:
 916
 917 `phrase reference`:role:_
 918 """,
 919 """\
 920 <document source="test data">
 921     <paragraph>
 922         Invalid phrase reference:
 923     <paragraph>
 924         <problematic ids="id2" refid="id1">
 925             `phrase reference`:role:_
 926     <system_message backrefs="id2" ids="id1" level="2" line="3" source="test data" type="WARNING">
 927         <paragraph>
 928             Mismatch: both interpreted text role suffix and reference suffix.
 929 """],
 930 ["""\
 931 `phrase reference_ without closing backquote
 932 """,
 933 """\
 934 <document source="test data">
 935     <paragraph>
 936         <problematic ids="id2" refid="id1">
 937             `
 938         phrase \n\
 939         <reference name="reference" refname="reference">
 940             reference
 941          without closing backquote
 942     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 943         <paragraph>
 944             Inline interpreted text or phrase reference start-string without end-string.
 945 """],
 946 ["""\
 947 `anonymous phrase reference__ without closing backquote
 948 """,
 949 """\
 950 <document source="test data">
 951     <paragraph>
 952         <problematic ids="id2" refid="id1">
 953             `
 954         anonymous phrase \n\
 955         <reference anonymous="1" name="reference">
 956             reference
 957          without closing backquote
 958     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 959         <paragraph>
 960             Inline interpreted text or phrase reference start-string without end-string.
 961 """],
 962 ]
 963
 964 totest['embedded_URIs'] = [
 965 ["""\
 966 `phrase reference <http://example.com>`_
 967 """,
 968 """\
 969 <document source="test data">
 970     <paragraph>
 971         <reference name="phrase reference" refuri="http://example.com">
 972             phrase reference
 973         <target ids="phrase-reference" names="phrase\ reference" refuri="http://example.com">
 974 """],
 975 ["""\
 976 `anonymous reference <http://example.com>`__
 977 """,
 978 """\
 979 <document source="test data">
 980     <paragraph>
 981         <reference name="anonymous reference" refuri="http://example.com">
 982             anonymous reference
 983 """],
 984 ["""\
 985 `embedded URI on next line
 986 <http://example.com>`__
 987 """,
 988 """\
 989 <document source="test data">
 990     <paragraph>
 991         <reference name="embedded URI on next line" refuri="http://example.com">
 992             embedded URI on next line
 993 """],
 994 ["""\
 995 `embedded URI across lines <http://example.com/
 996 long/path>`__
 997 """,
 998 """\
 999 <document source="test data">
1000     <paragraph>
1001         <reference name="embedded URI across lines" refuri="http://example.com/long/path">
1002             embedded URI across lines
1003 """],
1004 ["""\
1005 `embedded URI with whitespace <http://example.com/
1006 long/path /and  /whitespace>`__
1007 """,
1008 """\
1009 <document source="test data">
1010     <paragraph>
1011         <reference name="embedded URI with whitespace" refuri="http://example.com/long/path/and/whitespace">
1012             embedded URI with whitespace
1013 """],
1014 [r"""
1015 `embedded URI with escaped whitespace <http://example.com/a\
1016 long/path\ and/some\ escaped\ whitespace>`__
1017
1018 `<omitted\ reference\ text\ with\ escaped\ whitespace>`__
1019 """,
1020 """\
1021 <document source="test data">
1022     <paragraph>
1023         <reference name="embedded URI with escaped whitespace" refuri="http://example.com/a long/path and/some escaped whitespace">
1024             embedded URI with escaped whitespace
1025     <paragraph>
1026         <reference name="omitted reference text with escaped whitespace" refuri="omitted reference text with escaped whitespace">
1027             omitted reference text with escaped whitespace
1028 """],
1029 ["""\
1030 `embedded email address <jdoe@example.com>`__
1031
1032 `embedded email address broken across lines <jdoe
1033 @example.com>`__
1034 """,
1035 """\
1036 <document source="test data">
1037     <paragraph>
1038         <reference name="embedded email address" refuri="mailto:jdoe@example.com">
1039             embedded email address
1040     <paragraph>
1041         <reference name="embedded email address broken across lines" refuri="mailto:jdoe@example.com">
1042             embedded email address broken across lines
1043 """],
1044 [r"""
1045 `embedded URI with too much whitespace < http://example.com/
1046 long/path /and  /whitespace >`__
1047
1048 `embedded URI with too much whitespace at end <http://example.com/
1049 long/path /and  /whitespace >`__
1050
1051 `embedded URI with no preceding whitespace<http://example.com>`__
1052
1053 `escaped URI \<http://example.com>`__
1054
1055 See `HTML Anchors: \<a>`_.
1056 """,
1057 """\
1058 <document source="test data">
1059     <paragraph>
1060         <reference anonymous="1" name="embedded URI with too much whitespace < http://example.com/ long/path /and /whitespace >">
1061             embedded URI with too much whitespace < http://example.com/
1062             long/path /and  /whitespace >
1063     <paragraph>
1064         <reference anonymous="1" name="embedded URI with too much whitespace at end <http://example.com/ long/path /and /whitespace >">
1065             embedded URI with too much whitespace at end <http://example.com/
1066             long/path /and  /whitespace >
1067     <paragraph>
1068         <reference anonymous="1" name="embedded URI with no preceding whitespace<http://example.com>">
1069             embedded URI with no preceding whitespace<http://example.com>
1070     <paragraph>
1071         <reference anonymous="1" name="escaped URI <http://example.com>">
1072             escaped URI <http://example.com>
1073     <paragraph>
1074         See \n\
1075         <reference name="HTML Anchors: <a>" refname="html anchors: <a>">
1076             HTML Anchors: <a>
1077         .
1078 """],
1079 ["""\
1080 Relative URIs' reference text can be omitted:
1081
1082 `<reference>`_
1083
1084 `<anonymous>`__
1085 """,
1086 """\
1087 <document source="test data">
1088     <paragraph>
1089         Relative URIs' reference text can be omitted:
1090     <paragraph>
1091         <reference name="reference" refuri="reference">
1092             reference
1093         <target ids="reference" names="reference" refuri="reference">
1094     <paragraph>
1095         <reference name="anonymous" refuri="anonymous">
1096             anonymous
1097 """],
1098 [r"""
1099 Escape trailing low-line char in URIs:
1100
1101 `<reference\_>`_
1102
1103 `<anonymous\_>`__
1104 """,
1105 """\
1106 <document source="test data">
1107     <paragraph>
1108         Escape trailing low-line char in URIs:
1109     <paragraph>
1110         <reference name="reference_" refuri="reference_">
1111             reference_
1112         <target ids="reference" names="reference_" refuri="reference_">
1113     <paragraph>
1114         <reference name="anonymous_" refuri="anonymous_">
1115             anonymous_
1116 """],
1117 ["""\
1118 Escape other char in URIs:
1119
1120 `<reference\:1>`_
1121
1122 `<anonymous\\call>`__
1123 """,
1124 """\
1125 <document source="test data">
1126     <paragraph>
1127         Escape other char in URIs:
1128     <paragraph>
1129         <reference name="reference:1" refuri="reference:1">
1130             reference:1
1131         <target ids="reference-1" names="reference:1" refuri="reference:1">
1132     <paragraph>
1133         <reference name="anonymouscall" refuri="anonymouscall">
1134             anonymouscall
1135 """],
1136 ]
1137
1138 totest['embedded_aliases'] = [
1139 ["""\
1140 `phrase reference <alias_>`_
1141 """,
1142 """\
1143 <document source="test data">
1144     <paragraph>
1145         <reference name="phrase reference" refname="alias">
1146             phrase reference
1147         <target names="phrase\ reference" refname="alias">
1148 """],
1149 ["""\
1150 `anonymous reference <alias_>`__
1151 """,
1152 """\
1153 <document source="test data">
1154     <paragraph>
1155         <reference name="anonymous reference" refname="alias">
1156             anonymous reference
1157 """],
1158 ["""\
1159 `embedded alias on next line
1160 <alias_>`__
1161 """,
1162 """\
1163 <document source="test data">
1164     <paragraph>
1165         <reference name="embedded alias on next line" refname="alias">
1166             embedded alias on next line
1167 """],
1168 ["""\
1169 `embedded alias across lines <alias
1170 phrase_>`__
1171 """,
1172 """\
1173 <document source="test data">
1174     <paragraph>
1175         <reference name="embedded alias across lines" refname="alias phrase">
1176             embedded alias across lines
1177 """],
1178 ["""\
1179 `embedded alias with whitespace <alias
1180 long  phrase_>`__
1181 """,
1182 """\
1183 <document source="test data">
1184     <paragraph>
1185         <reference name="embedded alias with whitespace" refname="alias long phrase">
1186             embedded alias with whitespace
1187 """],
1188 ["""\
1189 `<embedded alias with whitespace_>`__
1190 """,
1191 """\
1192 <document source="test data">
1193     <paragraph>
1194         <reference name="embedded alias with whitespace" refname="embedded alias with whitespace">
1195             embedded alias with whitespace
1196 """],
1197 [r"""
1198 `no embedded alias (whitespace inside bracket) < alias_ >`__
1199
1200 `no embedded alias (no preceding whitespace)<alias_>`__
1201 """,
1202 """\
1203 <document source="test data">
1204     <paragraph>
1205         <reference anonymous="1" name="no embedded alias (whitespace inside bracket) < alias_ >">
1206             no embedded alias (whitespace inside bracket) < alias_ >
1207     <paragraph>
1208         <reference anonymous="1" name="no embedded alias (no preceding whitespace)<alias_>">
1209             no embedded alias (no preceding whitespace)<alias_>
1210 """],
1211 [r"""
1212 `anonymous reference <alias\ with\\ escaped \:characters_>`__
1213 """,
1214 """\
1215 <document source="test data">
1216     <paragraph>
1217         <reference name="anonymous reference" refname="aliaswith\ escaped :characters">
1218             anonymous reference
1219 """],
1220 [r"""
1221 `anonymous reference <alias\ with\\ escaped \:characters_>`__
1222 """,
1223 """\
1224 <document source="test data">
1225     <paragraph>
1226         <reference name="anonymous reference" refname="aliaswith\ escaped :characters">
1227             anonymous reference
1228 """],
1229 ]
1230
1231 totest['inline_targets'] = [
1232 ["""\
1233 _`target`
1234
1235 Here is _`another target` in some text. And _`yet
1236 another target`, spanning lines.
1237
1238 _`Here is  a    TaRgeT` with case and spacial difficulties.
1239 """,
1240 """\
1241 <document source="test data">
1242     <paragraph>
1243         <target ids="target" names="target">
1244             target
1245     <paragraph>
1246         Here is \n\
1247         <target ids="another-target" names="another\ target">
1248             another target
1249          in some text. And \n\
1250         <target ids="yet-another-target" names="yet\ another\ target">
1251             yet
1252             another target
1253         , spanning lines.
1254     <paragraph>
1255         <target ids="here-is-a-target" names="here\ is\ a\ target">
1256             Here is  a    TaRgeT
1257          with case and spacial difficulties.
1258 """],
1259 [u"""\
1260 l'_`target1` and l\u2019_`target2` with apostrophe
1261 """,
1262 u"""\
1263 <document source="test data">
1264     <paragraph>
1265         l'
1266         <target ids="target1" names="target1">
1267             target1
1268          and l\u2019
1269         <target ids="target2" names="target2">
1270             target2
1271          with apostrophe
1272 """],
1273 [u"""\
1274 quoted '_`target1`', quoted "_`target2`",
1275 quoted \u2018_`target3`\u2019, quoted \u201c_`target4`\u201d,
1276 quoted \xab_`target5`\xbb
1277 """,
1278 u"""\
1279 <document source="test data">
1280     <paragraph>
1281         quoted '
1282         <target ids="target1" names="target1">
1283             target1
1284         ', quoted "
1285         <target ids="target2" names="target2">
1286             target2
1287         ",
1288         quoted \u2018
1289         <target ids="target3" names="target3">
1290             target3
1291         \u2019, quoted \u201c
1292         <target ids="target4" names="target4">
1293             target4
1294         \u201d,
1295         quoted \xab
1296         <target ids="target5" names="target5">
1297             target5
1298         \xbb
1299 """],
1300 [u"""\
1301 _`'target1'` with quotes, _`"target2"` with quotes,
1302 _`\u2018target3\u2019` with quotes, _`\u201ctarget4\u201d` with quotes,
1303 _`\xabtarget5\xbb` with quotes
1304 """,
1305 u"""\
1306 <document source="test data">
1307     <paragraph>
1308         <target ids="target1" names="'target1'">
1309             'target1'
1310          with quotes, \n\
1311         <target ids="target2" names=""target2"">
1312             "target2"
1313          with quotes,
1314         <target ids="target3" names="\u2018target3\u2019">
1315             \u2018target3\u2019
1316          with quotes, \n\
1317         <target ids="target4" names="\u201ctarget4\u201d">
1318             \u201ctarget4\u201d
1319          with quotes,
1320         <target ids="target5" names="\xabtarget5\xbb">
1321             \xabtarget5\xbb
1322          with quotes
1323 """],
1324 ["""\
1325 But this isn't a _target; targets require backquotes.
1326
1327 And _`this`_ is just plain confusing.
1328 """,
1329 """\
1330 <document source="test data">
1331     <paragraph>
1332         But this isn't a _target; targets require backquotes.
1333     <paragraph>
1334         And \n\
1335         <problematic ids="id2" refid="id1">
1336             _`
1337         this`_ is just plain confusing.
1338     <system_message backrefs="id2" ids="id1" level="2" line="3" source="test data" type="WARNING">
1339         <paragraph>
1340             Inline target start-string without end-string.
1341 """],
1342 ["""\
1343 _`inline target without closing backquote
1344 """,
1345 """\
1346 <document source="test data">
1347     <paragraph>
1348         <problematic ids="id2" refid="id1">
1349             _`
1350         inline target without closing backquote
1351     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
1352         <paragraph>
1353             Inline target start-string without end-string.
1354 """],
1355 ]
1356
1357 totest['footnote_reference'] = [
1358 ["""\
1359 [1]_
1360 """,
1361 """\
1362 <document source="test data">
1363     <paragraph>
1364         <footnote_reference ids="id1" refname="1">
1365             1
1366 """],
1367 ["""\
1368 [#]_
1369 """,
1370 """\
1371 <document source="test data">
1372     <paragraph>
1373         <footnote_reference auto="1" ids="id1">
1374 """],
1375 ["""\
1376 [#label]_
1377 """,
1378 """\
1379 <document source="test data">
1380     <paragraph>
1381         <footnote_reference auto="1" ids="id1" refname="label">
1382 """],
1383 ["""\
1384 [*]_
1385 """,
1386 """\
1387 <document source="test data">
1388     <paragraph>
1389         <footnote_reference auto="*" ids="id1">
1390 """],
1391 ["""\
1392 Adjacent footnote refs are not possible: [*]_[#label]_ [#]_[2]_ [1]_[*]_
1393 """,
1394 """\
1395 <document source="test data">
1396     <paragraph>
1397         Adjacent footnote refs are not possible: [*]_[#label]_ [#]_[2]_ [1]_[*]_
1398 """],
1399 ]
1400
1401 totest['citation_reference'] = [
1402 ["""\
1403 [citation]_
1404 """,
1405 """\
1406 <document source="test data">
1407     <paragraph>
1408         <citation_reference ids="id1" refname="citation">
1409             citation
1410 """],
1411 ["""\
1412 [citation]_ and [cit-ation]_ and [cit.ation]_ and [CIT1]_ but not [CIT 1]_
1413 """,
1414 """\
1415 <document source="test data">
1416     <paragraph>
1417         <citation_reference ids="id1" refname="citation">
1418             citation
1419          and \n\
1420         <citation_reference ids="id2" refname="cit-ation">
1421             cit-ation
1422          and \n\
1423         <citation_reference ids="id3" refname="cit.ation">
1424             cit.ation
1425          and \n\
1426         <citation_reference ids="id4" refname="cit1">
1427             CIT1
1428          but not [CIT 1]_
1429 """],
1430 ["""\
1431 Adjacent citation refs are not possible: [citation]_[CIT1]_
1432 """,
1433 """\
1434 <document source="test data">
1435     <paragraph>
1436         Adjacent citation refs are not possible: [citation]_[CIT1]_
1437 """],
1438 ]
1439
1440 totest['substitution_references'] = [
1441 ["""\
1442 |subref|
1443 """,
1444 """\
1445 <document source="test data">
1446     <paragraph>
1447         <substitution_reference refname="subref">
1448             subref
1449 """],
1450 ["""\
1451 |subref|_ and |subref|__
1452 """,
1453 """\
1454 <document source="test data">
1455     <paragraph>
1456         <reference refname="subref">
1457             <substitution_reference refname="subref">
1458                 subref
1459          and \n\
1460         <reference anonymous="1">
1461             <substitution_reference refname="subref">
1462                 subref
1463 """],
1464 ["""\
1465 |substitution reference|
1466 """,
1467 """\
1468 <document source="test data">
1469     <paragraph>
1470         <substitution_reference refname="substitution reference">
1471             substitution reference
1472 """],
1473 ["""\
1474 |substitution
1475 reference|
1476 """,
1477 """\
1478 <document source="test data">
1479     <paragraph>
1480         <substitution_reference refname="substitution reference">
1481             substitution
1482             reference
1483 """],
1484 ["""\
1485 |substitution reference without closing verbar
1486 """,
1487 """\
1488 <document source="test data">
1489     <paragraph>
1490         <problematic ids="id2" refid="id1">
1491             |
1492         substitution reference without closing verbar
1493     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
1494         <paragraph>
1495             Inline substitution_reference start-string without end-string.
1496 """],
1497 ["""\
1498 first | then || and finally |||
1499 """,
1500 """\
1501 <document source="test data">
1502     <paragraph>
1503         first | then || and finally |||
1504 """],
1505 ]
1506
1507 totest['standalone_hyperlink'] = [
1508 ["""\
1509 http://www.standalone.hyperlink.com
1510
1511 http:/one-slash-only.absolute.path
1512
1513 [http://example.com]
1514
1515 (http://example.com)
1516
1517 <http://example.com>
1518
1519 http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html
1520
1521 http://[3ffe:2a00:100:7031::1] (the final "]" is ambiguous in text)
1522
1523 http://[3ffe:2a00:100:7031::1]/
1524
1525 mailto:someone@somewhere.com
1526
1527 news:comp.lang.python
1528
1529 An email address in a sentence: someone@somewhere.com.
1530
1531 ftp://ends.with.a.period.
1532
1533 (a.question.mark@end?)
1534 """,
1535 """\
1536 <document source="test data">
1537     <paragraph>
1538         <reference refuri="http://www.standalone.hyperlink.com">
1539             http://www.standalone.hyperlink.com
1540     <paragraph>
1541         <reference refuri="http:/one-slash-only.absolute.path">
1542             http:/one-slash-only.absolute.path
1543     <paragraph>
1544         [
1545         <reference refuri="http://example.com">
1546             http://example.com
1547         ]
1548     <paragraph>
1549         (
1550         <reference refuri="http://example.com">
1551             http://example.com
1552         )
1553     <paragraph>
1554         <
1555         <reference refuri="http://example.com">
1556             http://example.com
1557         >
1558     <paragraph>
1559         <reference refuri="http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html">
1560             http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html
1561     <paragraph>
1562         <reference refuri="http://[3ffe:2a00:100:7031::1">
1563             http://[3ffe:2a00:100:7031::1
1564         ] (the final "]" is ambiguous in text)
1565     <paragraph>
1566         <reference refuri="http://[3ffe:2a00:100:7031::1]/">
1567             http://[3ffe:2a00:100:7031::1]/
1568     <paragraph>
1569         <reference refuri="mailto:someone@somewhere.com">
1570             mailto:someone@somewhere.com
1571     <paragraph>
1572         <reference refuri="news:comp.lang.python">
1573             news:comp.lang.python
1574     <paragraph>
1575         An email address in a sentence: \n\
1576         <reference refuri="mailto:someone@somewhere.com">
1577             someone@somewhere.com
1578         .
1579     <paragraph>
1580         <reference refuri="ftp://ends.with.a.period">
1581             ftp://ends.with.a.period
1582         .
1583     <paragraph>
1584         (
1585         <reference refuri="mailto:a.question.mark@end">
1586             a.question.mark@end
1587         ?)
1588 """],
1589 [r"""
1590 Valid URLs with escaped markup characters:
1591
1592 http://example.com/\*content\*/whatever
1593
1594 http://example.com/\*content*/whatever
1595 """,
1596 """\
1597 <document source="test data">
1598     <paragraph>
1599         Valid URLs with escaped markup characters:
1600     <paragraph>
1601         <reference refuri="http://example.com/*content*/whatever">
1602             http://example.com/*content*/whatever
1603     <paragraph>
1604         <reference refuri="http://example.com/*content*/whatever">
1605             http://example.com/*content*/whatever
1606 """],
1607 ["""\
1608 Valid URLs may end with punctuation inside "<>":
1609
1610 <http://example.org/ends-with-dot.>
1611 """,
1612 """\
1613 <document source="test data">
1614     <paragraph>
1615         Valid URLs may end with punctuation inside "<>":
1616     <paragraph>
1617         <
1618         <reference refuri="http://example.org/ends-with-dot.">
1619             http://example.org/ends-with-dot.
1620         >
1621 """],
1622 ["""\
1623 Valid URLs with interesting endings:
1624
1625 http://example.org/ends-with-pluses++
1626 """,
1627 """\
1628 <document source="test data">
1629     <paragraph>
1630         Valid URLs with interesting endings:
1631     <paragraph>
1632         <reference refuri="http://example.org/ends-with-pluses++">
1633             http://example.org/ends-with-pluses++
1634 """],
1635 ["""\
1636 None of these are standalone hyperlinks (their "schemes"
1637 are not recognized): signal:noise, a:b.
1638 """,
1639 """\
1640 <document source="test data">
1641     <paragraph>
1642         None of these are standalone hyperlinks (their "schemes"
1643         are not recognized): signal:noise, a:b.
1644 """],
1645 ["""\
1646 Escaped email addresses are not recognized: test\@example.org
1647 """,
1648 """\
1649 <document source="test data">
1650     <paragraph>
1651         Escaped email addresses are not recognized: test@example.org
1652 """],
1653 ]
1654
1655 totest['markup recognition rules'] = [
1656 ["""\
1657 __This__ should be left alone.
1658 """,
1659 """\
1660 <document source="test data">
1661     <paragraph>
1662         __This__ should be left alone.
1663 """],
1664 [r"""
1665 Character-level m\ *a*\ **r**\ ``k``\ `u`:title:\p
1666 with backslash-escaped whitespace, including new\
1667 lines.
1668 """,
1669 """\
1670 <document source="test data">
1671     <paragraph>
1672         Character-level m
1673         <emphasis>
1674             a
1675         <strong>
1676             r
1677         <literal>
1678             k
1679         <title_reference>
1680             u
1681         p
1682         with backslash-escaped whitespace, including newlines.
1683 """],
1684 [u"""\
1685 text-*separated*\u2010*by*\u2011*various*\u2012*dashes*\u2013*and*\u2014*hyphens*.
1686 \u00bf*punctuation*? \u00a1*examples*!\u00a0*no-break-space*\u00a0.
1687 """,
1688 u"""\
1689 <document source="test data">
1690     <paragraph>
1691         text-
1692         <emphasis>
1693             separated
1694         \u2010
1695         <emphasis>
1696             by
1697         \u2011
1698         <emphasis>
1699             various
1700         \u2012
1701         <emphasis>
1702             dashes
1703         \u2013
1704         <emphasis>
1705             and
1706         \u2014
1707         <emphasis>
1708             hyphens
1709         .
1710         \xbf
1711         <emphasis>
1712             punctuation
1713         ? \xa1
1714         <emphasis>
1715             examples
1716         !\xa0
1717         <emphasis>
1718             no-break-space
1719         \u00a0.
1720 """],
1721 # Whitespace characters:
1722 #  \u180e*MONGOLIAN VOWEL SEPARATOR*\u180e,   fails in Python 2.6
1723 [u"""\
1724 text separated by
1725 *newline*
1726 or *space* or one of
1727 \xa0*NO-BREAK SPACE*\xa0,
1728 \u1680*OGHAM SPACE MARK*\u1680,
1729 \u2000*EN QUAD*\u2000,
1730 \u2001*EM QUAD*\u2001,
1731 \u2002*EN SPACE*\u2002,
1732 \u2003*EM SPACE*\u2003,
1733 \u2004*THREE-PER-EM SPACE*\u2004,
1734 \u2005*FOUR-PER-EM SPACE*\u2005,
1735 \u2006*SIX-PER-EM SPACE*\u2006,
1736 \u2007*FIGURE SPACE*\u2007,
1737 \u2008*PUNCTUATION SPACE*\u2008,
1738 \u2009*THIN SPACE*\u2009,
1739 \u200a*HAIR SPACE*\u200a,
1740 \u202f*NARROW NO-BREAK SPACE*\u202f,
1741 \u205f*MEDIUM MATHEMATICAL SPACE*\u205f,
1742 \u3000*IDEOGRAPHIC SPACE*\u3000,
1743 \u2028*LINE SEPARATOR*\u2028
1744 """,
1745 u"""\
1746 <document source="test data">
1747     <paragraph>
1748         text separated by
1749         <emphasis>
1750             newline
1751         \n\
1752         or \n\
1753         <emphasis>
1754             space
1755          or one of
1756         \xa0
1757         <emphasis>
1758             NO-BREAK SPACE
1759         \xa0,
1760         \u1680
1761         <emphasis>
1762             OGHAM SPACE MARK
1763         \u1680,
1764         \u2000
1765         <emphasis>
1766             EN QUAD
1767         \u2000,
1768         \u2001
1769         <emphasis>
1770             EM QUAD
1771         \u2001,
1772         \u2002
1773         <emphasis>
1774             EN SPACE
1775         \u2002,
1776         \u2003
1777         <emphasis>
1778             EM SPACE
1779         \u2003,
1780         \u2004
1781         <emphasis>
1782             THREE-PER-EM SPACE
1783         \u2004,
1784         \u2005
1785         <emphasis>
1786             FOUR-PER-EM SPACE
1787         \u2005,
1788         \u2006
1789         <emphasis>
1790             SIX-PER-EM SPACE
1791         \u2006,
1792         \u2007
1793         <emphasis>
1794             FIGURE SPACE
1795         \u2007,
1796         \u2008
1797         <emphasis>
1798             PUNCTUATION SPACE
1799         \u2008,
1800         \u2009
1801         <emphasis>
1802             THIN SPACE
1803         \u2009,
1804         \u200a
1805         <emphasis>
1806             HAIR SPACE
1807         \u200a,
1808         \u202f
1809         <emphasis>
1810             NARROW NO-BREAK SPACE
1811         \u202f,
1812         \u205f
1813         <emphasis>
1814             MEDIUM MATHEMATICAL SPACE
1815         \u205f,
1816         \u3000
1817         <emphasis>
1818             IDEOGRAPHIC SPACE
1819         \u3000,
1820     <paragraph>
1821         <emphasis>
1822             LINE SEPARATOR
1823 """],
1824 [u"""\
1825 inline markup separated by non-ASCII whitespace
1826 \xa0**NO-BREAK SPACE**\xa0, \xa0``NO-BREAK SPACE``\xa0, \xa0`NO-BREAK SPACE`\xa0,
1827 \u2000**EN QUAD**\u2000, \u2000``EN QUAD``\u2000, \u2000`EN QUAD`\u2000,
1828 \u202f**NARROW NBSP**\u202f, \u202f``NARROW NBSP``\u202f, \u202f`NARROW NBSP`\u202f,
1829 """,
1830 u"""\
1831 <document source="test data">
1832     <paragraph>
1833         inline markup separated by non-ASCII whitespace
1834         \xa0
1835         <strong>
1836             NO-BREAK SPACE
1837         \xa0, \xa0
1838         <literal>
1839             NO-BREAK SPACE
1840         \xa0, \xa0
1841         <title_reference>
1842             NO-BREAK SPACE
1843         \xa0,
1844         \u2000
1845         <strong>
1846             EN QUAD
1847         \u2000, \u2000
1848         <literal>
1849             EN QUAD
1850         \u2000, \u2000
1851         <title_reference>
1852             EN QUAD
1853         \u2000,
1854         \u202f
1855         <strong>
1856             NARROW NBSP
1857         \u202f, \u202f
1858         <literal>
1859             NARROW NBSP
1860         \u202f, \u202f
1861         <title_reference>
1862             NARROW NBSP
1863         \u202f,
1864 """],
1865 [u"""\
1866 no inline markup due to whitespace inside and behind: *
1867 newline
1868 *
1869 * space * or one of
1870 *\xa0NO-BREAK SPACE\xa0*
1871 *\u1680OGHAM SPACE MARK\u1680*
1872 *\u2000EN QUAD\u2000*
1873 *\u2001EM QUAD\u2001*
1874 *\u2002EN SPACE\u2002*
1875 *\u2003EM SPACE\u2003*
1876 *\u2004THREE-PER-EM SPACE\u2004*
1877 *\u2005FOUR-PER-EM SPACE\u2005*
1878 *\u2006SIX-PER-EM SPACE\u2006*
1879 *\u2007FIGURE SPACE\u2007*
1880 *\u2008PUNCTUATION SPACE\u2008*
1881 *\u2009THIN SPACE\u2009*
1882 *\u200aHAIR SPACE\u200a*
1883 *\u202fNARROW NO-BREAK SPACE\u202f*
1884 *\u205fMEDIUM MATHEMATICAL SPACE\u205f*
1885 *\u3000IDEOGRAPHIC SPACE\u3000*
1886 *\u2028LINE SEPARATOR\u2028*
1887 """,
1888 u"""\
1889 <document source="test data">
1890     <paragraph>
1891         no inline markup due to whitespace inside and behind: *
1892         newline
1893         *
1894         * space * or one of
1895         *\xa0NO-BREAK SPACE\xa0*
1896         *\u1680OGHAM SPACE MARK\u1680*
1897         *\u2000EN QUAD\u2000*
1898         *\u2001EM QUAD\u2001*
1899         *\u2002EN SPACE\u2002*
1900         *\u2003EM SPACE\u2003*
1901         *\u2004THREE-PER-EM SPACE\u2004*
1902         *\u2005FOUR-PER-EM SPACE\u2005*
1903         *\u2006SIX-PER-EM SPACE\u2006*
1904         *\u2007FIGURE SPACE\u2007*
1905         *\u2008PUNCTUATION SPACE\u2008*
1906         *\u2009THIN SPACE\u2009*
1907         *\u200aHAIR SPACE\u200a*
1908         *\u202fNARROW NO-BREAK SPACE\u202f*
1909         *\u205fMEDIUM MATHEMATICAL SPACE\u205f*
1910         *\u3000IDEOGRAPHIC SPACE\u3000*
1911         *
1912         LINE SEPARATOR
1913         *"""],
1914 [u"""\
1915 no inline markup because of non-ASCII whitespace following /preceding the markup
1916 **\xa0NO-BREAK SPACE\xa0** ``\xa0NO-BREAK SPACE\xa0`` `\xa0NO-BREAK SPACE\xa0`
1917 **\u2000EN QUAD\u2000** ``\u2000EN QUAD\u2000`` `\u2000EN QUAD\u2000`
1918 **\u202fNARROW NBSP\u202f** ``\u202fNARROW NBSP\u202f`` `\u202fNARROW NBSP\u202f`
1919 """,
1920 u"""\
1921 <document source="test data">
1922     <paragraph>
1923         no inline markup because of non-ASCII whitespace following /preceding the markup
1924         **\xa0NO-BREAK SPACE\xa0** ``\xa0NO-BREAK SPACE\xa0`` `\xa0NO-BREAK SPACE\xa0`
1925         **\u2000EN QUAD\u2000** ``\u2000EN QUAD\u2000`` `\u2000EN QUAD\u2000`
1926         **\u202fNARROW NBSP\u202f** ``\u202fNARROW NBSP\u202f`` `\u202fNARROW NBSP\u202f`\
1927 """],
1928 # « * » ‹ * › « * » ‹ * › « * » ‹ * › French,
1929 [u"""\
1930 "Quoted" markup start-string (matched openers & closers) -> no markup:
1931
1932 '*' "*" (*) <*> [*] {*}
1933 ⁅*⁆
1934
1935 Some international quoting styles:
1936 ‘*’ “*” English, ...,
1937 „*“ ‚*‘ »*« ›*‹ German, Czech, ...,
1938 „*” «*» Romanian,
1939 “*„ ‘*‚ Greek,
1940 「*」 『*』traditional Chinese,
1941 ”*” ’*’ »*» ›*› Swedish, Finnish,
1942 „*” ‚*’ Polish,
1943 „*” »*« ’*’ Hungarian,
1944
1945 But this is „*’ emphasized »*‹.
1946 """,
1947 u"""\
1948 <document source="test data">
1949     <paragraph>
1950         "Quoted" markup start-string (matched openers & closers) -> no markup:
1951     <paragraph>
1952         '*' "*" (*) <*> [*] {*}
1953         ⁅*⁆
1954     <paragraph>
1955         Some international quoting styles:
1956         ‘*’ “*” English, ...,
1957         „*“ ‚*‘ »*« ›*‹ German, Czech, ...,
1958         „*” «*» Romanian,
1959         “*„ ‘*‚ Greek,
1960         「*」 『*』traditional Chinese,
1961         ”*” ’*’ »*» ›*› Swedish, Finnish,
1962         „*” ‚*’ Polish,
1963         „*” »*« ’*’ Hungarian,
1964     <paragraph>
1965         But this is „
1966         <emphasis>
1967             ’ emphasized »
1968         ‹.
1969 """],
1970 ]
1971
1972
1973 if __name__ == '__main__':
1974     import unittest
1975     unittest.main(defaultTest='suite')