test/test_parsers/test_rst/test_inline_markup.py

   1 #! /usr/bin/env python
   2
   3 # $Id$
   4 # Author: David Goodger <goodger@python.org>
   5 # Copyright: This module has been placed in the public domain.
   6
   7 """
   8 Tests for inline markup in docutils/parsers/rst/states.py.
   9 Interpreted text tests are in a separate module, test_interpreted.py.
  10 """
  11
  12 from __init__ import DocutilsTestSupport
  13
  14 def suite():
  15     s = DocutilsTestSupport.ParserTestSuite()
  16     s.generateTests(totest)
  17     return s
  18
  19 totest = {}
  20
  21 totest['emphasis'] = [
  22 ["""\
  23 *emphasis*
  24 """,
  25 """\
  26 <document source="test data">
  27     <paragraph>
  28         <emphasis>
  29             emphasis
  30 """],
  31 [u"""\
  32 l'*emphasis* and l\u2019*emphasis* with apostrophe
  33 """,
  34 u"""\
  35 <document source="test data">
  36     <paragraph>
  37         l'
  38         <emphasis>
  39             emphasis
  40          and l\u2019
  41         <emphasis>
  42             emphasis
  43          with apostrophe
  44 """],
  45 ["""\
  46 *emphasized sentence
  47 across lines*
  48 """,
  49 """\
  50 <document source="test data">
  51     <paragraph>
  52         <emphasis>
  53             emphasized sentence
  54             across lines
  55 """],
  56 ["""\
  57 *emphasis without closing asterisk
  58 """,
  59 """\
  60 <document source="test data">
  61     <paragraph>
  62         <problematic ids="id2" refid="id1">
  63             *
  64         emphasis without closing asterisk
  65     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
  66         <paragraph>
  67             Inline emphasis start-string without end-string.
  68 """],
  69 [r"""
  70 '*emphasis*' and 1/*emphasis*/2 and 3-*emphasis*-4 and 5:*emphasis*:6
  71 but not '*' or '"*"' or  x*2* or 2*x* or \*args or *
  72 or *the\* *stars\\\* *inside*
  73
  74 (however, '*args' will trigger a warning and may be problematic)
  75
  76 what about *this**?
  77 """,
  78 """\
  79 <document source="test data">
  80     <paragraph>
  81         '
  82         <emphasis>
  83             emphasis
  84         ' and 1/
  85         <emphasis>
  86             emphasis
  87         /2 and 3-
  88         <emphasis>
  89             emphasis
  90         -4 and 5:
  91         <emphasis>
  92             emphasis
  93         :6
  94         but not '*' or '"*"' or  x*2* or 2*x* or *args or *
  95         or \n\
  96         <emphasis>
  97             the* *stars\* *inside
  98     <paragraph>
  99         (however, '
 100         <problematic ids="id2" refid="id1">
 101             *
 102         args' will trigger a warning and may be problematic)
 103     <system_message backrefs="id2" ids="id1" level="2" line="6" source="test data" type="WARNING">
 104         <paragraph>
 105             Inline emphasis start-string without end-string.
 106     <paragraph>
 107         what about \n\
 108         <emphasis>
 109             this*
 110         ?
 111 """],
 112 [u"""\
 113 quoted '*emphasis*', quoted "*emphasis*",
 114 quoted \u2018*emphasis*\u2019, quoted \u201c*emphasis*\u201d,
 115 quoted \xab*emphasis*\xbb
 116 """,
 117 u"""\
 118 <document source="test data">
 119     <paragraph>
 120         quoted '
 121         <emphasis>
 122             emphasis
 123         ', quoted "
 124         <emphasis>
 125             emphasis
 126         ",
 127         quoted \u2018
 128         <emphasis>
 129             emphasis
 130         \u2019, quoted \u201c
 131         <emphasis>
 132             emphasis
 133         \u201d,
 134         quoted \xab
 135         <emphasis>
 136             emphasis
 137         \xbb
 138 """],
 139 [r"""
 140 Emphasized asterisk: *\**
 141
 142 Emphasized double asterisk: *\***
 143 """,
 144 """\
 145 <document source="test data">
 146     <paragraph>
 147         Emphasized asterisk: \n\
 148         <emphasis>
 149             *
 150     <paragraph>
 151         Emphasized double asterisk: \n\
 152         <emphasis>
 153             **
 154 """],
 155 ]
 156
 157 totest['strong'] = [
 158 ["""\
 159 **strong**
 160 """,
 161 """\
 162 <document source="test data">
 163     <paragraph>
 164         <strong>
 165             strong
 166 """],
 167 [u"""\
 168 l'**strong** and l\u2019**strong** with apostrophe
 169 """,
 170 u"""\
 171 <document source="test data">
 172     <paragraph>
 173         l'
 174         <strong>
 175             strong
 176          and l\u2019
 177         <strong>
 178             strong
 179          with apostrophe
 180 """],
 181 [u"""\
 182 quoted '**strong**', quoted "**strong**",
 183 quoted \u2018**strong**\u2019, quoted \u201c**strong**\u201d,
 184 quoted \xab**strong**\xbb
 185 """,
 186 u"""\
 187 <document source="test data">
 188     <paragraph>
 189         quoted '
 190         <strong>
 191             strong
 192         ', quoted "
 193         <strong>
 194             strong
 195         ",
 196         quoted \u2018
 197         <strong>
 198             strong
 199         \u2019, quoted \u201c
 200         <strong>
 201             strong
 202         \u201d,
 203         quoted \xab
 204         <strong>
 205             strong
 206         \xbb
 207 """],
 208 [r"""
 209 (**strong**) but not (**) or '(** ' or x**2 or \**kwargs or **
 210
 211 (however, '**kwargs' will trigger a warning and may be problematic)
 212 """,
 213 """\
 214 <document source="test data">
 215     <paragraph>
 216         (
 217         <strong>
 218             strong
 219         ) but not (**) or '(** ' or x**2 or **kwargs or **
 220     <paragraph>
 221         (however, '
 222         <problematic ids="id2" refid="id1">
 223             **
 224         kwargs' will trigger a warning and may be problematic)
 225     <system_message backrefs="id2" ids="id1" level="2" line="4" source="test data" type="WARNING">
 226         <paragraph>
 227             Inline strong start-string without end-string.
 228 """],
 229 ["""\
 230 Strong asterisk: *****
 231
 232 Strong double asterisk: ******
 233 """,
 234 """\
 235 <document source="test data">
 236     <paragraph>
 237         Strong asterisk: \n\
 238         <strong>
 239             *
 240     <paragraph>
 241         Strong double asterisk: \n\
 242         <strong>
 243             **
 244 """],
 245 ["""\
 246 **strong without closing asterisks
 247 """,
 248 """\
 249 <document source="test data">
 250     <paragraph>
 251         <problematic ids="id2" refid="id1">
 252             **
 253         strong without closing asterisks
 254     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 255         <paragraph>
 256             Inline strong start-string without end-string.
 257 """],
 258 ]
 259
 260 totest['literal'] = [
 261 ["""\
 262 ``literal``
 263 """,
 264 """\
 265 <document source="test data">
 266     <paragraph>
 267         <literal>
 268             literal
 269 """],
 270 [r"""
 271 ``\literal``
 272 """,
 273 """\
 274 <document source="test data">
 275     <paragraph>
 276         <literal>
 277             \\literal
 278 """],
 279 [r"""
 280 ``lite\ral``
 281 """,
 282 """\
 283 <document source="test data">
 284     <paragraph>
 285         <literal>
 286             lite\\ral
 287 """],
 288 [r"""
 289 ``literal\``
 290 """,
 291 """\
 292 <document source="test data">
 293     <paragraph>
 294         <literal>
 295             literal\\
 296 """],
 297 [u"""\
 298 l'``literal`` and l\u2019``literal`` with apostrophe
 299 """,
 300 u"""\
 301 <document source="test data">
 302     <paragraph>
 303         l'
 304         <literal>
 305             literal
 306          and l\u2019
 307         <literal>
 308             literal
 309          with apostrophe
 310 """],
 311 [u"""\
 312 quoted '``literal``', quoted "``literal``",
 313 quoted \u2018``literal``\u2019, quoted \u201c``literal``\u201d,
 314 quoted \xab``literal``\xbb
 315 """,
 316 u"""\
 317 <document source="test data">
 318     <paragraph>
 319         quoted '
 320         <literal>
 321             literal
 322         ', quoted "
 323         <literal>
 324             literal
 325         ",
 326         quoted \u2018
 327         <literal>
 328             literal
 329         \u2019, quoted \u201c
 330         <literal>
 331             literal
 332         \u201d,
 333         quoted \xab
 334         <literal>
 335             literal
 336         \xbb
 337 """],
 338 [u"""\
 339 ``'literal'`` with quotes, ``"literal"`` with quotes,
 340 ``\u2018literal\u2019`` with quotes, ``\u201cliteral\u201d`` with quotes,
 341 ``\xabliteral\xbb`` with quotes
 342 """,
 343 u"""\
 344 <document source="test data">
 345     <paragraph>
 346         <literal>
 347             'literal'
 348          with quotes,
 349         <literal>
 350             "literal"
 351          with quotes,
 352         <literal>
 353             \u2018literal\u2019
 354          with quotes,
 355         <literal>
 356             \u201cliteral\u201d
 357          with quotes,
 358         <literal>
 359             \xabliteral\xbb
 360          with quotes
 361 """],
 362 [r"""
 363 ``literal ``TeX quotes'' & \backslash`` but not "``" or ``
 364
 365 (however, ``standalone TeX quotes'' will trigger a warning
 366 and may be problematic)
 367 """,
 368 """\
 369 <document source="test data">
 370     <paragraph>
 371         <literal>
 372             literal ``TeX quotes'' & \\backslash
 373          but not "``" or ``
 374     <paragraph>
 375         (however, \n\
 376         <problematic ids="id2" refid="id1">
 377             ``
 378         standalone TeX quotes'' will trigger a warning
 379         and may be problematic)
 380     <system_message backrefs="id2" ids="id1" level="2" line="4" source="test data" type="WARNING">
 381         <paragraph>
 382             Inline literal start-string without end-string.
 383 """],
 384 ["""\
 385 Find the ```interpreted text``` in this paragraph!
 386 """,
 387 """\
 388 <document source="test data">
 389     <paragraph>
 390         Find the \n\
 391         <literal>
 392             `interpreted text`
 393          in this paragraph!
 394 """],
 395 ["""\
 396 ``literal without closing backquotes
 397 """,
 398 """\
 399 <document source="test data">
 400     <paragraph>
 401         <problematic ids="id2" refid="id1">
 402             ``
 403         literal without closing backquotes
 404     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 405         <paragraph>
 406             Inline literal start-string without end-string.
 407 """],
 408 [r"""
 409 Python ``list``\s use square bracket syntax.
 410 """,
 411 """\
 412 <document source="test data">
 413     <paragraph>
 414         Python \n\
 415         <literal>
 416             list
 417         s use square bracket syntax.
 418 """],
 419 ]
 420
 421 totest['references'] = [
 422 ["""\
 423 ref_
 424 """,
 425 """\
 426 <document source="test data">
 427     <paragraph>
 428         <reference name="ref" refname="ref">
 429             ref
 430 """],
 431 [u"""\
 432 l'ref_ and l\u2019ref_ with apostrophe
 433 """,
 434 u"""\
 435 <document source="test data">
 436     <paragraph>
 437         l'
 438         <reference name="ref" refname="ref">
 439             ref
 440          and l\u2019
 441         <reference name="ref" refname="ref">
 442             ref
 443          with apostrophe
 444 """],
 445 [u"""\
 446 quoted 'ref_', quoted "ref_",
 447 quoted \u2018ref_\u2019, quoted \u201cref_\u201d,
 448 quoted \xabref_\xbb,
 449 but not 'ref ref'_, "ref ref"_, \u2018ref ref\u2019_,
 450 \u201cref ref\u201d_, or \xabref ref\xbb_
 451 """,
 452 u"""\
 453 <document source="test data">
 454     <paragraph>
 455         quoted '
 456         <reference name="ref" refname="ref">
 457             ref
 458         ', quoted "
 459         <reference name="ref" refname="ref">
 460             ref
 461         ",
 462         quoted \u2018
 463         <reference name="ref" refname="ref">
 464             ref
 465         \u2019, quoted \u201c
 466         <reference name="ref" refname="ref">
 467             ref
 468         \u201d,
 469         quoted \xab
 470         <reference name="ref" refname="ref">
 471             ref
 472         \xbb,
 473         but not 'ref ref'_, "ref ref"_, \u2018ref ref\u2019_,
 474         \u201cref ref\u201d_, or \xabref ref\xbb_
 475 """],
 476 ["""\
 477 ref__
 478 """,
 479 """\
 480 <document source="test data">
 481     <paragraph>
 482         <reference anonymous="1" name="ref">
 483             ref
 484 """],
 485 [u"""\
 486 l'ref__ and l\u2019ref__ with apostrophe
 487 """,
 488 u"""\
 489 <document source="test data">
 490     <paragraph>
 491         l'
 492         <reference anonymous="1" name="ref">
 493             ref
 494          and l\u2019
 495         <reference anonymous="1" name="ref">
 496             ref
 497          with apostrophe
 498 """],
 499 [u"""\
 500 quoted 'ref__', quoted "ref__",
 501 quoted \u2018ref__\u2019, quoted \u201cref__\u201d,
 502 quoted \xabref__\xbb,
 503 but not 'ref ref'__, "ref ref"__, \u2018ref ref\u2019__,
 504 \u201cref ref\u201d__, or \xabref ref\xbb__
 505 """,
 506 u"""\
 507 <document source="test data">
 508     <paragraph>
 509         quoted '
 510         <reference anonymous="1" name="ref">
 511             ref
 512         ', quoted "
 513         <reference anonymous="1" name="ref">
 514             ref
 515         ",
 516         quoted \u2018
 517         <reference anonymous="1" name="ref">
 518             ref
 519         \u2019, quoted \u201c
 520         <reference anonymous="1" name="ref">
 521             ref
 522         \u201d,
 523         quoted \xab
 524         <reference anonymous="1" name="ref">
 525             ref
 526         \xbb,
 527         but not 'ref ref'__, "ref ref"__, \u2018ref ref\u2019__,
 528         \u201cref ref\u201d__, or \xabref ref\xbb__
 529 """],
 530 ["""\
 531 ref_, r_, r_e-f_, -ref_, and anonymousref__,
 532 but not _ref_ or __attr__ or object.__attr__
 533 """,
 534 """\
 535 <document source="test data">
 536     <paragraph>
 537         <reference name="ref" refname="ref">
 538             ref
 539         , \n\
 540         <reference name="r" refname="r">
 541             r
 542         , \n\
 543         <reference name="r_e-f" refname="r_e-f">
 544             r_e-f
 545         , -
 546         <reference name="ref" refname="ref">
 547             ref
 548         , and \n\
 549         <reference anonymous="1" name="anonymousref">
 550             anonymousref
 551         ,
 552         but not _ref_ or __attr__ or object.__attr__
 553 """],
 554 ]
 555
 556 totest['phrase_references'] = [
 557 ["""\
 558 `phrase reference`_
 559 """,
 560 """\
 561 <document source="test data">
 562     <paragraph>
 563         <reference name="phrase reference" refname="phrase reference">
 564             phrase reference
 565 """],
 566 [u"""\
 567 l'`phrase reference`_ and l\u2019`phrase reference`_ with apostrophe
 568 """,
 569 u"""\
 570 <document source="test data">
 571     <paragraph>
 572         l'
 573         <reference name="phrase reference" refname="phrase reference">
 574             phrase reference
 575          and l\u2019
 576         <reference name="phrase reference" refname="phrase reference">
 577             phrase reference
 578          with apostrophe
 579 """],
 580 [u"""\
 581 quoted '`phrase reference`_', quoted "`phrase reference`_",
 582 quoted \u2018`phrase reference`_\u2019,
 583 quoted \u201c`phrase reference`_\u201d,
 584 quoted \xab`phrase reference`_\xbb
 585 """,
 586 u"""\
 587 <document source="test data">
 588     <paragraph>
 589         quoted '
 590         <reference name="phrase reference" refname="phrase reference">
 591             phrase reference
 592         ', quoted "
 593         <reference name="phrase reference" refname="phrase reference">
 594             phrase reference
 595         ",
 596         quoted \u2018
 597         <reference name="phrase reference" refname="phrase reference">
 598             phrase reference
 599         \u2019,
 600         quoted \u201c
 601         <reference name="phrase reference" refname="phrase reference">
 602             phrase reference
 603         \u201d,
 604         quoted \xab
 605         <reference name="phrase reference" refname="phrase reference">
 606             phrase reference
 607         \xbb
 608 """],
 609 [u"""\
 610 `'phrase reference'`_ with quotes, `"phrase reference"`_ with quotes,
 611 `\u2018phrase reference\u2019`_ with quotes,
 612 `\u201cphrase reference\u201d`_ with quotes,
 613 `\xabphrase reference\xbb`_ with quotes
 614 """,
 615 u"""\
 616 <document source="test data">
 617     <paragraph>
 618         <reference name="'phrase reference'" refname="'phrase reference'">
 619             'phrase reference'
 620          with quotes,
 621         <reference name=""phrase reference"" refname=""phrase reference"">
 622             "phrase reference"
 623          with quotes,
 624         <reference name="\u2018phrase reference\u2019" refname="\u2018phrase reference\u2019">
 625             \u2018phrase reference\u2019
 626          with quotes,
 627         <reference name="\u201cphrase reference\u201d" refname="\u201cphrase reference\u201d">
 628             \u201cphrase reference\u201d
 629          with quotes,
 630         <reference name="\xabphrase reference\xbb" refname="\xabphrase reference\xbb">
 631             \xabphrase reference\xbb
 632          with quotes
 633 """],
 634 ["""\
 635 `anonymous reference`__
 636 """,
 637 """\
 638 <document source="test data">
 639     <paragraph>
 640         <reference anonymous="1" name="anonymous reference">
 641             anonymous reference
 642 """],
 643 [u"""\
 644 l'`anonymous reference`__ and l\u2019`anonymous reference`__ with apostrophe
 645 """,
 646 u"""\
 647 <document source="test data">
 648     <paragraph>
 649         l'
 650         <reference anonymous="1" name="anonymous reference">
 651             anonymous reference
 652          and l\u2019
 653         <reference anonymous="1" name="anonymous reference">
 654             anonymous reference
 655          with apostrophe
 656 """],
 657 [u"""\
 658 quoted '`anonymous reference`__', quoted "`anonymous reference`__",
 659 quoted \u2018`anonymous reference`__\u2019,
 660 quoted \u201c`anonymous reference`__\u201d,
 661 quoted \xab`anonymous reference`__\xbb
 662 """,
 663 u"""\
 664 <document source="test data">
 665     <paragraph>
 666         quoted '
 667         <reference anonymous="1" name="anonymous reference">
 668             anonymous reference
 669         ', quoted "
 670         <reference anonymous="1" name="anonymous reference">
 671             anonymous reference
 672         ",
 673         quoted \u2018
 674         <reference anonymous="1" name="anonymous reference">
 675             anonymous reference
 676         \u2019,
 677         quoted \u201c
 678         <reference anonymous="1" name="anonymous reference">
 679             anonymous reference
 680         \u201d,
 681         quoted \xab
 682         <reference anonymous="1" name="anonymous reference">
 683             anonymous reference
 684         \xbb
 685 """],
 686 [u"""\
 687 `'anonymous reference'`__ with quotes, `"anonymous reference"`__ with quotes,
 688 `\u2018anonymous reference\u2019`__ with quotes,
 689 `\u201canonymous reference\u201d`__ with quotes,
 690 `\xabanonymous reference\xbb`__ with quotes
 691 """,
 692 u"""\
 693 <document source="test data">
 694     <paragraph>
 695         <reference anonymous="1" name="'anonymous reference'">
 696             'anonymous reference'
 697          with quotes,
 698         <reference anonymous="1" name=""anonymous reference"">
 699             "anonymous reference"
 700          with quotes,
 701         <reference anonymous="1" name="\u2018anonymous reference\u2019">
 702             \u2018anonymous reference\u2019
 703          with quotes,
 704         <reference anonymous="1" name="\u201canonymous reference\u201d">
 705             \u201canonymous reference\u201d
 706          with quotes,
 707         <reference anonymous="1" name="\xabanonymous reference\xbb">
 708             \xabanonymous reference\xbb
 709          with quotes
 710 """],
 711 ["""\
 712 `phrase reference
 713 across lines`_
 714 """,
 715 """\
 716 <document source="test data">
 717     <paragraph>
 718         <reference name="phrase reference across lines" refname="phrase reference across lines">
 719             phrase reference
 720             across lines
 721 """],
 722 ["""\
 723 `phrase\`_ reference`_
 724 """,
 725 """\
 726 <document source="test data">
 727     <paragraph>
 728         <reference name="phrase`_ reference" refname="phrase`_ reference">
 729             phrase`_ reference
 730 """],
 731 ["""\
 732 Invalid phrase reference:
 733
 734 :role:`phrase reference`_
 735 """,
 736 """\
 737 <document source="test data">
 738     <paragraph>
 739         Invalid phrase reference:
 740     <paragraph>
 741         <problematic ids="id2" refid="id1">
 742             :role:`phrase reference`_
 743     <system_message backrefs="id2" ids="id1" level="2" line="3" source="test data" type="WARNING">
 744         <paragraph>
 745             Mismatch: both interpreted text role prefix and reference suffix.
 746 """],
 747 ["""\
 748 Invalid phrase reference:
 749
 750 `phrase reference`:role:_
 751 """,
 752 """\
 753 <document source="test data">
 754     <paragraph>
 755         Invalid phrase reference:
 756     <paragraph>
 757         <problematic ids="id2" refid="id1">
 758             `phrase reference`:role:_
 759     <system_message backrefs="id2" ids="id1" level="2" line="3" source="test data" type="WARNING">
 760         <paragraph>
 761             Mismatch: both interpreted text role suffix and reference suffix.
 762 """],
 763 ["""\
 764 `phrase reference_ without closing backquote
 765 """,
 766 """\
 767 <document source="test data">
 768     <paragraph>
 769         <problematic ids="id2" refid="id1">
 770             `
 771         phrase \n\
 772         <reference name="reference" refname="reference">
 773             reference
 774          without closing backquote
 775     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 776         <paragraph>
 777             Inline interpreted text or phrase reference start-string without end-string.
 778 """],
 779 ["""\
 780 `anonymous phrase reference__ without closing backquote
 781 """,
 782 """\
 783 <document source="test data">
 784     <paragraph>
 785         <problematic ids="id2" refid="id1">
 786             `
 787         anonymous phrase \n\
 788         <reference anonymous="1" name="reference">
 789             reference
 790          without closing backquote
 791     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
 792         <paragraph>
 793             Inline interpreted text or phrase reference start-string without end-string.
 794 """],
 795 ]
 796
 797 totest['embedded_URIs'] = [
 798 ["""\
 799 `phrase reference <http://example.com>`_
 800 """,
 801 """\
 802 <document source="test data">
 803     <paragraph>
 804         <reference name="phrase reference" refuri="http://example.com">
 805             phrase reference
 806         <target ids="phrase-reference" names="phrase\ reference" refuri="http://example.com">
 807 """],
 808 ["""\
 809 `anonymous reference <http://example.com>`__
 810 """,
 811 """\
 812 <document source="test data">
 813     <paragraph>
 814         <reference name="anonymous reference" refuri="http://example.com">
 815             anonymous reference
 816 """],
 817 ["""\
 818 `embedded URI on next line
 819 <http://example.com>`__
 820 """,
 821 """\
 822 <document source="test data">
 823     <paragraph>
 824         <reference name="embedded URI on next line" refuri="http://example.com">
 825             embedded URI on next line
 826 """],
 827 ["""\
 828 `embedded URI across lines <http://example.com/
 829 long/path>`__
 830 """,
 831 """\
 832 <document source="test data">
 833     <paragraph>
 834         <reference name="embedded URI across lines" refuri="http://example.com/long/path">
 835             embedded URI across lines
 836 """],
 837 ["""\
 838 `embedded URI with whitespace <http://example.com/
 839 long/path /and  /whitespace>`__
 840 """,
 841 """\
 842 <document source="test data">
 843     <paragraph>
 844         <reference name="embedded URI with whitespace" refuri="http://example.com/long/path/and/whitespace">
 845             embedded URI with whitespace
 846 """],
 847 ["""\
 848 `embedded email address <jdoe@example.com>`__
 849
 850 `embedded email address broken across lines <jdoe
 851 @example.com>`__
 852 """,
 853 """\
 854 <document source="test data">
 855     <paragraph>
 856         <reference name="embedded email address" refuri="mailto:jdoe@example.com">
 857             embedded email address
 858     <paragraph>
 859         <reference name="embedded email address broken across lines" refuri="mailto:jdoe@example.com">
 860             embedded email address broken across lines
 861 """],
 862 [r"""
 863 `embedded URI with too much whitespace < http://example.com/
 864 long/path /and  /whitespace >`__
 865
 866 `embedded URI with too much whitespace at end <http://example.com/
 867 long/path /and  /whitespace >`__
 868
 869 `embedded URI with no preceding whitespace<http://example.com>`__
 870
 871 `escaped URI \<http://example.com>`__
 872
 873 See `HTML Anchors: \<a>`_.
 874 """,
 875 """\
 876 <document source="test data">
 877     <paragraph>
 878         <reference anonymous="1" name="embedded URI with too much whitespace < http://example.com/ long/path /and /whitespace >">
 879             embedded URI with too much whitespace < http://example.com/
 880             long/path /and  /whitespace >
 881     <paragraph>
 882         <reference anonymous="1" name="embedded URI with too much whitespace at end <http://example.com/ long/path /and /whitespace >">
 883             embedded URI with too much whitespace at end <http://example.com/
 884             long/path /and  /whitespace >
 885     <paragraph>
 886         <reference anonymous="1" name="embedded URI with no preceding whitespace<http://example.com>">
 887             embedded URI with no preceding whitespace<http://example.com>
 888     <paragraph>
 889         <reference anonymous="1" name="escaped URI <http://example.com>">
 890             escaped URI <http://example.com>
 891     <paragraph>
 892         See \n\
 893         <reference name="HTML Anchors: <a>" refname="html anchors: <a>">
 894             HTML Anchors: <a>
 895         .
 896 """],
 897 ["""\
 898 Relative URIs' reference text can be omitted:
 899
 900 `<reference>`_
 901
 902 `<anonymous>`__
 903 """,
 904 """\
 905 <document source="test data">
 906     <paragraph>
 907         Relative URIs' reference text can be omitted:
 908     <paragraph>
 909         <reference name="reference" refuri="reference">
 910             reference
 911         <target ids="reference" names="reference" refuri="reference">
 912     <paragraph>
 913         <reference name="anonymous" refuri="anonymous">
 914             anonymous
 915 """],
 916 ]
 917
 918 totest['inline_targets'] = [
 919 ["""\
 920 _`target`
 921
 922 Here is _`another target` in some text. And _`yet
 923 another target`, spanning lines.
 924
 925 _`Here is  a    TaRgeT` with case and spacial difficulties.
 926 """,
 927 """\
 928 <document source="test data">
 929     <paragraph>
 930         <target ids="target" names="target">
 931             target
 932     <paragraph>
 933         Here is \n\
 934         <target ids="another-target" names="another\ target">
 935             another target
 936          in some text. And \n\
 937         <target ids="yet-another-target" names="yet\ another\ target">
 938             yet
 939             another target
 940         , spanning lines.
 941     <paragraph>
 942         <target ids="here-is-a-target" names="here\ is\ a\ target">
 943             Here is  a    TaRgeT
 944          with case and spacial difficulties.
 945 """],
 946 [u"""\
 947 l'_`target1` and l\u2019_`target2` with apostrophe
 948 """,
 949 u"""\
 950 <document source="test data">
 951     <paragraph>
 952         l'
 953         <target ids="target1" names="target1">
 954             target1
 955          and l\u2019
 956         <target ids="target2" names="target2">
 957             target2
 958          with apostrophe
 959 """],
 960 [u"""\
 961 quoted '_`target1`', quoted "_`target2`",
 962 quoted \u2018_`target3`\u2019, quoted \u201c_`target4`\u201d,
 963 quoted \xab_`target5`\xbb
 964 """,
 965 u"""\
 966 <document source="test data">
 967     <paragraph>
 968         quoted '
 969         <target ids="target1" names="target1">
 970             target1
 971         ', quoted "
 972         <target ids="target2" names="target2">
 973             target2
 974         ",
 975         quoted \u2018
 976         <target ids="target3" names="target3">
 977             target3
 978         \u2019, quoted \u201c
 979         <target ids="target4" names="target4">
 980             target4
 981         \u201d,
 982         quoted \xab
 983         <target ids="target5" names="target5">
 984             target5
 985         \xbb
 986 """],
 987 [u"""\
 988 _`'target1'` with quotes, _`"target2"` with quotes,
 989 _`\u2018target3\u2019` with quotes, _`\u201ctarget4\u201d` with quotes,
 990 _`\xabtarget5\xbb` with quotes
 991 """,
 992 u"""\
 993 <document source="test data">
 994     <paragraph>
 995         <target ids="target1" names="'target1'">
 996             'target1'
 997          with quotes,
 998         <target ids="target2" names=""target2"">
 999             "target2"
1000          with quotes,
1001         <target ids="target3" names="\u2018target3\u2019">
1002             \u2018target3\u2019
1003          with quotes,
1004         <target ids="target4" names="\u201ctarget4\u201d">
1005             \u201ctarget4\u201d
1006          with quotes,
1007         <target ids="target5" names="\xabtarget5\xbb">
1008             \xabtarget5\xbb
1009          with quotes
1010 """],
1011 ["""\
1012 But this isn't a _target; targets require backquotes.
1013
1014 And _`this`_ is just plain confusing.
1015 """,
1016 """\
1017 <document source="test data">
1018     <paragraph>
1019         But this isn't a _target; targets require backquotes.
1020     <paragraph>
1021         And \n\
1022         <problematic ids="id2" refid="id1">
1023             _`
1024         this`_ is just plain confusing.
1025     <system_message backrefs="id2" ids="id1" level="2" line="3" source="test data" type="WARNING">
1026         <paragraph>
1027             Inline target start-string without end-string.
1028 """],
1029 ["""\
1030 _`inline target without closing backquote
1031 """,
1032 """\
1033 <document source="test data">
1034     <paragraph>
1035         <problematic ids="id2" refid="id1">
1036             _`
1037         inline target without closing backquote
1038     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
1039         <paragraph>
1040             Inline target start-string without end-string.
1041 """],
1042 ]
1043
1044 totest['footnote_reference'] = [
1045 ["""\
1046 [1]_
1047 """,
1048 """\
1049 <document source="test data">
1050     <paragraph>
1051         <footnote_reference ids="id1" refname="1">
1052             1
1053 """],
1054 ["""\
1055 [#]_
1056 """,
1057 """\
1058 <document source="test data">
1059     <paragraph>
1060         <footnote_reference auto="1" ids="id1">
1061 """],
1062 ["""\
1063 [#label]_
1064 """,
1065 """\
1066 <document source="test data">
1067     <paragraph>
1068         <footnote_reference auto="1" ids="id1" refname="label">
1069 """],
1070 ["""\
1071 [*]_
1072 """,
1073 """\
1074 <document source="test data">
1075     <paragraph>
1076         <footnote_reference auto="*" ids="id1">
1077 """],
1078 ["""\
1079 Adjacent footnote refs are not possible: [*]_[#label]_ [#]_[2]_ [1]_[*]_
1080 """,
1081 """\
1082 <document source="test data">
1083     <paragraph>
1084         Adjacent footnote refs are not possible: [*]_[#label]_ [#]_[2]_ [1]_[*]_
1085 """],
1086 ]
1087
1088 totest['citation_reference'] = [
1089 ["""\
1090 [citation]_
1091 """,
1092 """\
1093 <document source="test data">
1094     <paragraph>
1095         <citation_reference ids="id1" refname="citation">
1096             citation
1097 """],
1098 ["""\
1099 [citation]_ and [cit-ation]_ and [cit.ation]_ and [CIT1]_ but not [CIT 1]_
1100 """,
1101 """\
1102 <document source="test data">
1103     <paragraph>
1104         <citation_reference ids="id1" refname="citation">
1105             citation
1106          and \n\
1107         <citation_reference ids="id2" refname="cit-ation">
1108             cit-ation
1109          and \n\
1110         <citation_reference ids="id3" refname="cit.ation">
1111             cit.ation
1112          and \n\
1113         <citation_reference ids="id4" refname="cit1">
1114             CIT1
1115          but not [CIT 1]_
1116 """],
1117 ["""\
1118 Adjacent citation refs are not possible: [citation]_[CIT1]_
1119 """,
1120 """\
1121 <document source="test data">
1122     <paragraph>
1123         Adjacent citation refs are not possible: [citation]_[CIT1]_
1124 """],
1125 ]
1126
1127 totest['substitution_references'] = [
1128 ["""\
1129 |subref|
1130 """,
1131 """\
1132 <document source="test data">
1133     <paragraph>
1134         <substitution_reference refname="subref">
1135             subref
1136 """],
1137 ["""\
1138 |subref|_ and |subref|__
1139 """,
1140 """\
1141 <document source="test data">
1142     <paragraph>
1143         <reference refname="subref">
1144             <substitution_reference refname="subref">
1145                 subref
1146          and \n\
1147         <reference anonymous="1">
1148             <substitution_reference refname="subref">
1149                 subref
1150 """],
1151 ["""\
1152 |substitution reference|
1153 """,
1154 """\
1155 <document source="test data">
1156     <paragraph>
1157         <substitution_reference refname="substitution reference">
1158             substitution reference
1159 """],
1160 ["""\
1161 |substitution
1162 reference|
1163 """,
1164 """\
1165 <document source="test data">
1166     <paragraph>
1167         <substitution_reference refname="substitution reference">
1168             substitution
1169             reference
1170 """],
1171 ["""\
1172 |substitution reference without closing verbar
1173 """,
1174 """\
1175 <document source="test data">
1176     <paragraph>
1177         <problematic ids="id2" refid="id1">
1178             |
1179         substitution reference without closing verbar
1180     <system_message backrefs="id2" ids="id1" level="2" line="1" source="test data" type="WARNING">
1181         <paragraph>
1182             Inline substitution_reference start-string without end-string.
1183 """],
1184 ["""\
1185 first | then || and finally |||
1186 """,
1187 """\
1188 <document source="test data">
1189     <paragraph>
1190         first | then || and finally |||
1191 """],
1192 ]
1193
1194 totest['standalone_hyperlink'] = [
1195 ["""\
1196 http://www.standalone.hyperlink.com
1197
1198 http:/one-slash-only.absolute.path
1199
1200 [http://example.com]
1201
1202 (http://example.com)
1203
1204 <http://example.com>
1205
1206 http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html
1207
1208 http://[3ffe:2a00:100:7031::1] (the final "]" is ambiguous in text)
1209
1210 http://[3ffe:2a00:100:7031::1]/
1211
1212 mailto:someone@somewhere.com
1213
1214 news:comp.lang.python
1215
1216 An email address in a sentence: someone@somewhere.com.
1217
1218 ftp://ends.with.a.period.
1219
1220 (a.question.mark@end?)
1221 """,
1222 """\
1223 <document source="test data">
1224     <paragraph>
1225         <reference refuri="http://www.standalone.hyperlink.com">
1226             http://www.standalone.hyperlink.com
1227     <paragraph>
1228         <reference refuri="http:/one-slash-only.absolute.path">
1229             http:/one-slash-only.absolute.path
1230     <paragraph>
1231         [
1232         <reference refuri="http://example.com">
1233             http://example.com
1234         ]
1235     <paragraph>
1236         (
1237         <reference refuri="http://example.com">
1238             http://example.com
1239         )
1240     <paragraph>
1241         <
1242         <reference refuri="http://example.com">
1243             http://example.com
1244         >
1245     <paragraph>
1246         <reference refuri="http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html">
1247             http://[1080:0:0:0:8:800:200C:417A]/IPv6address.html
1248     <paragraph>
1249         <reference refuri="http://[3ffe:2a00:100:7031::1">
1250             http://[3ffe:2a00:100:7031::1
1251         ] (the final "]" is ambiguous in text)
1252     <paragraph>
1253         <reference refuri="http://[3ffe:2a00:100:7031::1]/">
1254             http://[3ffe:2a00:100:7031::1]/
1255     <paragraph>
1256         <reference refuri="mailto:someone@somewhere.com">
1257             mailto:someone@somewhere.com
1258     <paragraph>
1259         <reference refuri="news:comp.lang.python">
1260             news:comp.lang.python
1261     <paragraph>
1262         An email address in a sentence: \n\
1263         <reference refuri="mailto:someone@somewhere.com">
1264             someone@somewhere.com
1265         .
1266     <paragraph>
1267         <reference refuri="ftp://ends.with.a.period">
1268             ftp://ends.with.a.period
1269         .
1270     <paragraph>
1271         (
1272         <reference refuri="mailto:a.question.mark@end">
1273             a.question.mark@end
1274         ?)
1275 """],
1276 [r"""
1277 Valid URLs with escaped markup characters:
1278
1279 http://example.com/\*content\*/whatever
1280
1281 http://example.com/\*content*/whatever
1282 """,
1283 """\
1284 <document source="test data">
1285     <paragraph>
1286         Valid URLs with escaped markup characters:
1287     <paragraph>
1288         <reference refuri="http://example.com/*content*/whatever">
1289             http://example.com/*content*/whatever
1290     <paragraph>
1291         <reference refuri="http://example.com/*content*/whatever">
1292             http://example.com/*content*/whatever
1293 """],
1294 ["""\
1295 Valid URLs may end with punctuation inside "<>":
1296
1297 <http://example.org/ends-with-dot.>
1298 """,
1299 """\
1300 <document source="test data">
1301     <paragraph>
1302         Valid URLs may end with punctuation inside "<>":
1303     <paragraph>
1304         <
1305         <reference refuri="http://example.org/ends-with-dot.">
1306             http://example.org/ends-with-dot.
1307         >
1308 """],
1309 ["""\
1310 Valid URLs with interesting endings:
1311
1312 http://example.org/ends-with-pluses++
1313 """,
1314 """\
1315 <document source="test data">
1316     <paragraph>
1317         Valid URLs with interesting endings:
1318     <paragraph>
1319         <reference refuri="http://example.org/ends-with-pluses++">
1320             http://example.org/ends-with-pluses++
1321 """],
1322 ["""\
1323 None of these are standalone hyperlinks (their "schemes"
1324 are not recognized): signal:noise, a:b.
1325 """,
1326 """\
1327 <document source="test data">
1328     <paragraph>
1329         None of these are standalone hyperlinks (their "schemes"
1330         are not recognized): signal:noise, a:b.
1331 """],
1332 ["""\
1333 Escaped email addresses are not recognized: test\@example.org
1334 """,
1335 """\
1336 <document source="test data">
1337     <paragraph>
1338         Escaped email addresses are not recognized: test@example.org
1339 """],
1340 ]
1341
1342 totest['markup recognition rules'] = [
1343 ["""\
1344 __This__ should be left alone.
1345 """,
1346 """\
1347 <document source="test data">
1348     <paragraph>
1349         __This__ should be left alone.
1350 """],
1351 [r"""
1352 Character-level m\ *a*\ **r**\ ``k``\ `u`:title:\p
1353 with backslash-escaped whitespace, including new\
1354 lines.
1355 """,
1356 """\
1357 <document source="test data">
1358     <paragraph>
1359         Character-level m
1360         <emphasis>
1361             a
1362         <strong>
1363             r
1364         <literal>
1365             k
1366         <title_reference>
1367             u
1368         p
1369         with backslash-escaped whitespace, including newlines.
1370 """],
1371 [u"""\
1372 text-*separated*\u2010*by*\u2011*various*\u2012*dashes*\u2013*and*\u2014*hyphens*.
1373 \u00bf*punctuation*? \u00a1*examples*!\u00a0*\u00a0no-break-space\u00a0*.
1374 """,
1375 u"""\
1376 <document source="test data">
1377     <paragraph>
1378         text-
1379         <emphasis>
1380             separated
1381         \u2010
1382         <emphasis>
1383             by
1384         \u2011
1385         <emphasis>
1386             various
1387         \u2012
1388         <emphasis>
1389             dashes
1390         \u2013
1391         <emphasis>
1392             and
1393         \u2014
1394         <emphasis>
1395             hyphens
1396         .
1397         \xbf
1398         <emphasis>
1399             punctuation
1400         ? \xa1
1401         <emphasis>
1402             examples
1403         !\xa0
1404         <emphasis>
1405             \u00a0no-break-space\u00a0
1406         .
1407 """],
1408 # Whitespace characters:
1409 #  \u180e*MONGOLIAN VOWEL SEPARATOR*\u180e,   fails in Python 2.4
1410 [u"""\
1411 text separated by
1412 *newline*
1413 or *space* or one of
1414 \xa0*NO-BREAK SPACE*\xa0,
1415 \u1680*OGHAM SPACE MARK*\u1680,
1416 \u2000*EN QUAD*\u2000,
1417 \u2001*EM QUAD*\u2001,
1418 \u2002*EN SPACE*\u2002,
1419 \u2003*EM SPACE*\u2003,
1420 \u2004*THREE-PER-EM SPACE*\u2004,
1421 \u2005*FOUR-PER-EM SPACE*\u2005,
1422 \u2006*SIX-PER-EM SPACE*\u2006,
1423 \u2007*FIGURE SPACE*\u2007,
1424 \u2008*PUNCTUATION SPACE*\u2008,
1425 \u2009*THIN SPACE*\u2009,
1426 \u200a*HAIR SPACE*\u200a,
1427 \u202f*NARROW NO-BREAK SPACE*\u202f,
1428 \u205f*MEDIUM MATHEMATICAL SPACE*\u205f,
1429 \u3000*IDEOGRAPHIC SPACE*\u3000,
1430 \u2028*LINE SEPARATOR*\u2028
1431 """,
1432 u"""\
1433 <document source="test data">
1434     <paragraph>
1435         text separated by
1436         <emphasis>
1437             newline
1438         \n\
1439         or \n\
1440         <emphasis>
1441             space
1442          or one of
1443         \xa0
1444         <emphasis>
1445             NO-BREAK SPACE
1446         \xa0,
1447         \u1680
1448         <emphasis>
1449             OGHAM SPACE MARK
1450         \u1680,
1451         \u2000
1452         <emphasis>
1453             EN QUAD
1454         \u2000,
1455         \u2001
1456         <emphasis>
1457             EM QUAD
1458         \u2001,
1459         \u2002
1460         <emphasis>
1461             EN SPACE
1462         \u2002,
1463         \u2003
1464         <emphasis>
1465             EM SPACE
1466         \u2003,
1467         \u2004
1468         <emphasis>
1469             THREE-PER-EM SPACE
1470         \u2004,
1471         \u2005
1472         <emphasis>
1473             FOUR-PER-EM SPACE
1474         \u2005,
1475         \u2006
1476         <emphasis>
1477             SIX-PER-EM SPACE
1478         \u2006,
1479         \u2007
1480         <emphasis>
1481             FIGURE SPACE
1482         \u2007,
1483         \u2008
1484         <emphasis>
1485             PUNCTUATION SPACE
1486         \u2008,
1487         \u2009
1488         <emphasis>
1489             THIN SPACE
1490         \u2009,
1491         \u200a
1492         <emphasis>
1493             HAIR SPACE
1494         \u200a,
1495         \u202f
1496         <emphasis>
1497             NARROW NO-BREAK SPACE
1498         \u202f,
1499         \u205f
1500         <emphasis>
1501             MEDIUM MATHEMATICAL SPACE
1502         \u205f,
1503         \u3000
1504         <emphasis>
1505             IDEOGRAPHIC SPACE
1506         \u3000,
1507     <paragraph>
1508         <emphasis>
1509             LINE SEPARATOR
1510 """],
1511 [u"""\
1512 None of these should be markup (matched openers & closers):
1513
1514 \u2018*\u2019 \u201c*\u201d \xab*\xbb \u00bf*? \u00a1*!
1515
1516 But this should:
1517
1518 l\u2019*exception*.
1519 """,
1520 u"""\
1521 <document source="test data">
1522     <paragraph>
1523         None of these should be markup (matched openers & closers):
1524     <paragraph>
1525         \u2018*\u2019 \u201c*\u201d \xab*\xbb \xbf*? \xa1*!
1526     <paragraph>
1527         But this should:
1528     <paragraph>
1529         l\u2019
1530         <emphasis>
1531             exception
1532         .
1533 """],
1534 ]
1535
1536
1537 if __name__ == '__main__':
1538     import unittest
1539     unittest.main(defaultTest='suite')