test/test_parsers/test_rst/test_enumerated_lists.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 states.py.
   9 """
  10
  11 from __init__ import DocutilsTestSupport
  12
  13 def suite():
  14     s = DocutilsTestSupport.ParserTestSuite()
  15     s.generateTests(totest)
  16     return s
  17
  18 totest = {}
  19
  20 totest['enumerated_lists'] = [
  21 ["""\
  22 1. Item one.
  23
  24 2. Item two.
  25
  26 3. Item three.
  27 """,
  28 """\
  29 <document source="test data">
  30     <enumerated_list enumtype="arabic" prefix="" suffix=".">
  31         <list_item>
  32             <paragraph>
  33                 Item one.
  34         <list_item>
  35             <paragraph>
  36                 Item two.
  37         <list_item>
  38             <paragraph>
  39                 Item three.
  40 """],
  41 ["""\
  42 No blank lines betwen items:
  43
  44 1. Item one.
  45 2. Item two.
  46 3. Item three.
  47 """,
  48 """\
  49 <document source="test data">
  50     <paragraph>
  51         No blank lines betwen items:
  52     <enumerated_list enumtype="arabic" prefix="" suffix=".">
  53         <list_item>
  54             <paragraph>
  55                 Item one.
  56         <list_item>
  57             <paragraph>
  58                 Item two.
  59         <list_item>
  60             <paragraph>
  61                 Item three.
  62 """],
  63 ["""\
  64 1.
  65 empty item above, no blank line
  66 """,
  67 """\
  68 <document source="test data">
  69     <paragraph>
  70         1.
  71         empty item above, no blank line
  72 """],
  73 ["""\
  74 Scrambled:
  75
  76 3. Item three.
  77
  78 2. Item two.
  79
  80 1. Item one.
  81
  82 3. Item three.
  83 2. Item two.
  84 1. Item one.
  85 """,
  86 """\
  87 <document source="test data">
  88     <paragraph>
  89         Scrambled:
  90     <enumerated_list enumtype="arabic" prefix="" start="3" suffix=".">
  91         <list_item>
  92             <paragraph>
  93                 Item three.
  94     <system_message level="1" line="3" source="test data" type="INFO">
  95         <paragraph>
  96             Enumerated list start value not ordinal-1: "3" (ordinal 3)
  97     <enumerated_list enumtype="arabic" prefix="" start="2" suffix=".">
  98         <list_item>
  99             <paragraph>
 100                 Item two.
 101     <system_message level="1" line="5" source="test data" type="INFO">
 102         <paragraph>
 103             Enumerated list start value not ordinal-1: "2" (ordinal 2)
 104     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 105         <list_item>
 106             <paragraph>
 107                 Item one.
 108     <paragraph>
 109         3. Item three.
 110         2. Item two.
 111         1. Item one.
 112 """],
 113 ["""\
 114 Skipping item 3:
 115
 116 1. Item 1.
 117 2. Item 2.
 118 4. Item 4.
 119 """,
 120 """\
 121 <document source="test data">
 122     <paragraph>
 123         Skipping item 3:
 124     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 125         <list_item>
 126             <paragraph>
 127                 Item 1.
 128     <system_message level="2" line="4" source="test data" type="WARNING">
 129         <paragraph>
 130             Enumerated list ends without a blank line; unexpected unindent.
 131     <paragraph>
 132         2. Item 2.
 133         4. Item 4.
 134 """],
 135 ["""\
 136 Start with non-ordinal-1:
 137
 138 0. Item zero.
 139 1. Item one.
 140 2. Item two.
 141 3. Item three.
 142
 143 And again:
 144
 145 2. Item two.
 146 3. Item three.
 147 """,
 148 """\
 149 <document source="test data">
 150     <paragraph>
 151         Start with non-ordinal-1:
 152     <enumerated_list enumtype="arabic" prefix="" start="0" suffix=".">
 153         <list_item>
 154             <paragraph>
 155                 Item zero.
 156         <list_item>
 157             <paragraph>
 158                 Item one.
 159         <list_item>
 160             <paragraph>
 161                 Item two.
 162         <list_item>
 163             <paragraph>
 164                 Item three.
 165     <system_message level="1" line="3" source="test data" type="INFO">
 166         <paragraph>
 167             Enumerated list start value not ordinal-1: "0" (ordinal 0)
 168     <paragraph>
 169         And again:
 170     <enumerated_list enumtype="arabic" prefix="" start="2" suffix=".">
 171         <list_item>
 172             <paragraph>
 173                 Item two.
 174         <list_item>
 175             <paragraph>
 176                 Item three.
 177     <system_message level="1" line="10" source="test data" type="INFO">
 178         <paragraph>
 179             Enumerated list start value not ordinal-1: "2" (ordinal 2)
 180 """],
 181 ["""\
 182 1. Item one: line 1,
 183    line 2.
 184 2. Item two: line 1,
 185    line 2.
 186 3. Item three: paragraph 1, line 1,
 187    line 2.
 188
 189    Paragraph 2.
 190 """,
 191 """\
 192 <document source="test data">
 193     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 194         <list_item>
 195             <paragraph>
 196                 Item one: line 1,
 197                 line 2.
 198         <list_item>
 199             <paragraph>
 200                 Item two: line 1,
 201                 line 2.
 202         <list_item>
 203             <paragraph>
 204                 Item three: paragraph 1, line 1,
 205                 line 2.
 206             <paragraph>
 207                 Paragraph 2.
 208 """],
 209 ["""\
 210 Different enumeration sequences:
 211
 212 1. Item 1.
 213 2. Item 2.
 214 3. Item 3.
 215
 216 A. Item A.
 217 B. Item B.
 218 C. Item C.
 219
 220 a. Item a.
 221 b. Item b.
 222 c. Item c.
 223
 224 I. Item I.
 225 II. Item II.
 226 III. Item III.
 227
 228 i. Item i.
 229 ii. Item ii.
 230 iii. Item iii.
 231 """,
 232 """\
 233 <document source="test data">
 234     <paragraph>
 235         Different enumeration sequences:
 236     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 237         <list_item>
 238             <paragraph>
 239                 Item 1.
 240         <list_item>
 241             <paragraph>
 242                 Item 2.
 243         <list_item>
 244             <paragraph>
 245                 Item 3.
 246     <enumerated_list enumtype="upperalpha" prefix="" suffix=".">
 247         <list_item>
 248             <paragraph>
 249                 Item A.
 250         <list_item>
 251             <paragraph>
 252                 Item B.
 253         <list_item>
 254             <paragraph>
 255                 Item C.
 256     <enumerated_list enumtype="loweralpha" prefix="" suffix=".">
 257         <list_item>
 258             <paragraph>
 259                 Item a.
 260         <list_item>
 261             <paragraph>
 262                 Item b.
 263         <list_item>
 264             <paragraph>
 265                 Item c.
 266     <enumerated_list enumtype="upperroman" prefix="" suffix=".">
 267         <list_item>
 268             <paragraph>
 269                 Item I.
 270         <list_item>
 271             <paragraph>
 272                 Item II.
 273         <list_item>
 274             <paragraph>
 275                 Item III.
 276     <enumerated_list enumtype="lowerroman" prefix="" suffix=".">
 277         <list_item>
 278             <paragraph>
 279                 Item i.
 280         <list_item>
 281             <paragraph>
 282                 Item ii.
 283         <list_item>
 284             <paragraph>
 285                 Item iii.
 286 """],
 287 ["""\
 288 Bad Roman numerals:
 289
 290 i. i
 291
 292 ii. ii
 293
 294 iii. iii
 295
 296 iiii. iiii
 297       second line
 298
 299 (LCD) is an acronym made up of Roman numerals
 300
 301 (livid) is a word made up of Roman numerals
 302
 303 (CIVIL) is another such word
 304
 305 (I) I
 306
 307 (IVXLCDM) IVXLCDM
 308 """,
 309 """\
 310 <document source="test data">
 311     <paragraph>
 312         Bad Roman numerals:
 313     <enumerated_list enumtype="lowerroman" prefix="" suffix=".">
 314         <list_item>
 315             <paragraph>
 316                 i
 317         <list_item>
 318             <paragraph>
 319                 ii
 320         <list_item>
 321             <paragraph>
 322                 iii
 323     <definition_list>
 324         <definition_list_item>
 325             <term>
 326                 iiii. iiii
 327             <definition>
 328                 <paragraph>
 329                     second line
 330     <paragraph>
 331         (LCD) is an acronym made up of Roman numerals
 332     <paragraph>
 333         (livid) is a word made up of Roman numerals
 334     <paragraph>
 335         (CIVIL) is another such word
 336     <enumerated_list enumtype="upperroman" prefix="(" suffix=")">
 337         <list_item>
 338             <paragraph>
 339                 I
 340     <paragraph>
 341         (IVXLCDM) IVXLCDM
 342 """],
 343 ["""\
 344 Potentially ambiguous cases:
 345
 346 A. Item A.
 347 B. Item B.
 348 C. Item C.
 349
 350 I. Item I.
 351 II. Item II.
 352 III. Item III.
 353
 354 a. Item a.
 355 b. Item b.
 356 c. Item c.
 357
 358 i. Item i.
 359 ii. Item ii.
 360 iii. Item iii.
 361
 362 Phew! Safe!
 363 """,
 364 """\
 365 <document source="test data">
 366     <paragraph>
 367         Potentially ambiguous cases:
 368     <enumerated_list enumtype="upperalpha" prefix="" suffix=".">
 369         <list_item>
 370             <paragraph>
 371                 Item A.
 372         <list_item>
 373             <paragraph>
 374                 Item B.
 375         <list_item>
 376             <paragraph>
 377                 Item C.
 378     <enumerated_list enumtype="upperroman" prefix="" suffix=".">
 379         <list_item>
 380             <paragraph>
 381                 Item I.
 382         <list_item>
 383             <paragraph>
 384                 Item II.
 385         <list_item>
 386             <paragraph>
 387                 Item III.
 388     <enumerated_list enumtype="loweralpha" prefix="" suffix=".">
 389         <list_item>
 390             <paragraph>
 391                 Item a.
 392         <list_item>
 393             <paragraph>
 394                 Item b.
 395         <list_item>
 396             <paragraph>
 397                 Item c.
 398     <enumerated_list enumtype="lowerroman" prefix="" suffix=".">
 399         <list_item>
 400             <paragraph>
 401                 Item i.
 402         <list_item>
 403             <paragraph>
 404                 Item ii.
 405         <list_item>
 406             <paragraph>
 407                 Item iii.
 408     <paragraph>
 409         Phew! Safe!
 410 """],
 411 ["""\
 412 Definitely ambiguous:
 413
 414 A. Item A.
 415 B. Item B.
 416 C. Item C.
 417 D. Item D.
 418 E. Item E.
 419 F. Item F.
 420 G. Item G.
 421 H. Item H.
 422 I. Item I.
 423 II. Item II.
 424 III. Item III.
 425
 426 a. Item a.
 427 b. Item b.
 428 c. Item c.
 429 d. Item d.
 430 e. Item e.
 431 f. Item f.
 432 g. Item g.
 433 h. Item h.
 434 i. Item i.
 435 ii. Item ii.
 436 iii. Item iii.
 437 """,
 438 """\
 439 <document source="test data">
 440     <paragraph>
 441         Definitely ambiguous:
 442     <enumerated_list enumtype="upperalpha" prefix="" suffix=".">
 443         <list_item>
 444             <paragraph>
 445                 Item A.
 446         <list_item>
 447             <paragraph>
 448                 Item B.
 449         <list_item>
 450             <paragraph>
 451                 Item C.
 452         <list_item>
 453             <paragraph>
 454                 Item D.
 455         <list_item>
 456             <paragraph>
 457                 Item E.
 458         <list_item>
 459             <paragraph>
 460                 Item F.
 461         <list_item>
 462             <paragraph>
 463                 Item G.
 464         <list_item>
 465             <paragraph>
 466                 Item H.
 467     <system_message level="2" line="11" source="test data" type="WARNING">
 468         <paragraph>
 469             Enumerated list ends without a blank line; unexpected unindent.
 470     <enumerated_list enumtype="upperroman" prefix="" suffix=".">
 471         <list_item>
 472             <paragraph>
 473                 Item I.
 474         <list_item>
 475             <paragraph>
 476                 Item II.
 477         <list_item>
 478             <paragraph>
 479                 Item III.
 480     <enumerated_list enumtype="loweralpha" prefix="" suffix=".">
 481         <list_item>
 482             <paragraph>
 483                 Item a.
 484         <list_item>
 485             <paragraph>
 486                 Item b.
 487         <list_item>
 488             <paragraph>
 489                 Item c.
 490         <list_item>
 491             <paragraph>
 492                 Item d.
 493         <list_item>
 494             <paragraph>
 495                 Item e.
 496         <list_item>
 497             <paragraph>
 498                 Item f.
 499         <list_item>
 500             <paragraph>
 501                 Item g.
 502         <list_item>
 503             <paragraph>
 504                 Item h.
 505     <system_message level="2" line="23" source="test data" type="WARNING">
 506         <paragraph>
 507             Enumerated list ends without a blank line; unexpected unindent.
 508     <enumerated_list enumtype="lowerroman" prefix="" suffix=".">
 509         <list_item>
 510             <paragraph>
 511                 Item i.
 512         <list_item>
 513             <paragraph>
 514                 Item ii.
 515         <list_item>
 516             <paragraph>
 517                 Item iii.
 518 """],
 519 ["""\
 520 Different enumeration formats:
 521
 522 1. Item 1.
 523 2. Item 2.
 524 3. Item 3.
 525
 526 1) Item 1).
 527 2) Item 2).
 528 3) Item 3).
 529
 530 (1) Item (1).
 531 (2) Item (2).
 532 (3) Item (3).
 533 """,
 534 """\
 535 <document source="test data">
 536     <paragraph>
 537         Different enumeration formats:
 538     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 539         <list_item>
 540             <paragraph>
 541                 Item 1.
 542         <list_item>
 543             <paragraph>
 544                 Item 2.
 545         <list_item>
 546             <paragraph>
 547                 Item 3.
 548     <enumerated_list enumtype="arabic" prefix="" suffix=")">
 549         <list_item>
 550             <paragraph>
 551                 Item 1).
 552         <list_item>
 553             <paragraph>
 554                 Item 2).
 555         <list_item>
 556             <paragraph>
 557                 Item 3).
 558     <enumerated_list enumtype="arabic" prefix="(" suffix=")">
 559         <list_item>
 560             <paragraph>
 561                 Item (1).
 562         <list_item>
 563             <paragraph>
 564                 Item (2).
 565         <list_item>
 566             <paragraph>
 567                 Item (3).
 568 """],
 569 ["""\
 570 Nested enumerated lists:
 571
 572 1. Item 1.
 573
 574    A) Item A).
 575    B) Item B).
 576    C) Item C).
 577
 578 2. Item 2.
 579
 580    (a) Item (a).
 581
 582        I) Item I).
 583        II) Item II).
 584        III) Item III).
 585
 586    (b) Item (b).
 587
 588    (c) Item (c).
 589
 590        (i) Item (i).
 591        (ii) Item (ii).
 592        (iii) Item (iii).
 593
 594 3. Item 3.
 595 """,
 596 """\
 597 <document source="test data">
 598     <paragraph>
 599         Nested enumerated lists:
 600     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 601         <list_item>
 602             <paragraph>
 603                 Item 1.
 604             <enumerated_list enumtype="upperalpha" prefix="" suffix=")">
 605                 <list_item>
 606                     <paragraph>
 607                         Item A).
 608                 <list_item>
 609                     <paragraph>
 610                         Item B).
 611                 <list_item>
 612                     <paragraph>
 613                         Item C).
 614         <list_item>
 615             <paragraph>
 616                 Item 2.
 617             <enumerated_list enumtype="loweralpha" prefix="(" suffix=")">
 618                 <list_item>
 619                     <paragraph>
 620                         Item (a).
 621                     <enumerated_list enumtype="upperroman" prefix="" suffix=")">
 622                         <list_item>
 623                             <paragraph>
 624                                 Item I).
 625                         <list_item>
 626                             <paragraph>
 627                                 Item II).
 628                         <list_item>
 629                             <paragraph>
 630                                 Item III).
 631                 <list_item>
 632                     <paragraph>
 633                         Item (b).
 634                 <list_item>
 635                     <paragraph>
 636                         Item (c).
 637                     <enumerated_list enumtype="lowerroman" prefix="(" suffix=")">
 638                         <list_item>
 639                             <paragraph>
 640                                 Item (i).
 641                         <list_item>
 642                             <paragraph>
 643                                 Item (ii).
 644                         <list_item>
 645                             <paragraph>
 646                                 Item (iii).
 647         <list_item>
 648             <paragraph>
 649                 Item 3.
 650 """],
 651 [u"""\
 652 A. Einstein was a great influence on
 653 B. Physicist, who was a colleague of
 654 C. Chemist.  They all worked in
 655 Princeton, NJ.
 656
 657 Using a non-breaking space as a workaround:
 658
 659 A.\u00a0Einstein was a great influence on
 660 B. Physicist, who was a colleague of
 661 C. Chemist.  They all worked in
 662 Princeton, NJ.
 663 """,
 664 """\
 665 <document source="test data">
 666     <enumerated_list enumtype="upperalpha" prefix="" suffix=".">
 667         <list_item>
 668             <paragraph>
 669                 Einstein was a great influence on
 670         <list_item>
 671             <paragraph>
 672                 Physicist, who was a colleague of
 673     <system_message level="2" line="3" source="test data" type="WARNING">
 674         <paragraph>
 675             Enumerated list ends without a blank line; unexpected unindent.
 676     <paragraph>
 677         C. Chemist.  They all worked in
 678         Princeton, NJ.
 679     <paragraph>
 680         Using a non-breaking space as a workaround:
 681     <paragraph>
 682         A.\xa0Einstein was a great influence on
 683         B. Physicist, who was a colleague of
 684         C. Chemist.  They all worked in
 685         Princeton, NJ.
 686 """],
 687 ["""\
 688 1. Item one: line 1,
 689    line 2.
 690 2. Item two: line 1,
 691   line 2.
 692 3. Item three: paragraph 1, line 1,
 693  line 2.
 694
 695    Paragraph 2.
 696 """,
 697 """\
 698 <document source="test data">
 699     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 700         <list_item>
 701             <paragraph>
 702                 Item one: line 1,
 703                 line 2.
 704         <list_item>
 705             <paragraph>
 706                 Item two: line 1,
 707     <system_message level="2" line="4" source="test data" type="WARNING">
 708         <paragraph>
 709             Enumerated list ends without a blank line; unexpected unindent.
 710     <block_quote>
 711         <paragraph>
 712             line 2.
 713     <system_message level="2" line="5" source="test data" type="WARNING">
 714         <paragraph>
 715             Block quote ends without a blank line; unexpected unindent.
 716     <enumerated_list enumtype="arabic" prefix="" start="3" suffix=".">
 717         <list_item>
 718             <paragraph>
 719                 Item three: paragraph 1, line 1,
 720     <system_message level="1" line="5" source="test data" type="INFO">
 721         <paragraph>
 722             Enumerated list start value not ordinal-1: "3" (ordinal 3)
 723     <system_message level="2" line="6" source="test data" type="WARNING">
 724         <paragraph>
 725             Enumerated list ends without a blank line; unexpected unindent.
 726     <block_quote>
 727         <paragraph>
 728             line 2.
 729         <block_quote>
 730             <paragraph>
 731                 Paragraph 2.
 732 """],
 733 ["""\
 734 1. Item one.
 735
 736 #. Item two.
 737
 738 #. Item three.
 739 """,
 740 """\
 741 <document source="test data">
 742     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 743         <list_item>
 744             <paragraph>
 745                 Item one.
 746         <list_item>
 747             <paragraph>
 748                 Item two.
 749         <list_item>
 750             <paragraph>
 751                 Item three.
 752 """],
 753 ["""\
 754 a. Item one.
 755 #. Item two.
 756 #. Item three.
 757 """,
 758 """\
 759 <document source="test data">
 760     <enumerated_list enumtype="loweralpha" prefix="" suffix=".">
 761         <list_item>
 762             <paragraph>
 763                 Item one.
 764         <list_item>
 765             <paragraph>
 766                 Item two.
 767         <list_item>
 768             <paragraph>
 769                 Item three.
 770 """],
 771 ["""\
 772 i. Item one.
 773 ii. Item two.
 774 #. Item three.
 775 """,
 776 """\
 777 <document source="test data">
 778     <enumerated_list enumtype="lowerroman" prefix="" suffix=".">
 779         <list_item>
 780             <paragraph>
 781                 Item one.
 782         <list_item>
 783             <paragraph>
 784                 Item two.
 785         <list_item>
 786             <paragraph>
 787                 Item three.
 788 """],
 789 ["""\
 790 #. Item one.
 791 #. Item two.
 792 #. Item three.
 793 """,
 794 """\
 795 <document source="test data">
 796     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 797         <list_item>
 798             <paragraph>
 799                 Item one.
 800         <list_item>
 801             <paragraph>
 802                 Item two.
 803         <list_item>
 804             <paragraph>
 805                 Item three.
 806 """],
 807 ["""\
 808 1. Item one.
 809 #. Item two.
 810 3. Item three.
 811 """,
 812 """\
 813 <document source="test data">
 814     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 815         <list_item>
 816             <paragraph>
 817                 Item one.
 818     <system_message level="2" line="2" source="test data" type="WARNING">
 819         <paragraph>
 820             Enumerated list ends without a blank line; unexpected unindent.
 821     <paragraph>
 822         #. Item two.
 823         3. Item three.
 824 """],
 825 ["""\
 826 z.
 827 x
 828 """,
 829 """\
 830 <document source="test data">
 831     <paragraph>
 832         z.
 833         x
 834 """],
 835 ["""\
 836 3-space indent, with a trailing space:
 837
 838 1. \n\
 839    foo
 840
 841 3-space indent, no trailing space:
 842
 843 1.
 844    foo
 845
 846 2-space indent:
 847
 848 1.
 849   foo
 850
 851 1-space indent:
 852
 853 1.
 854  foo
 855
 856 0-space indent, not a list item:
 857
 858 1.
 859 foo
 860
 861 No item content:
 862
 863 1.
 864 """,
 865 """\
 866 <document source="test data">
 867     <paragraph>
 868         3-space indent, with a trailing space:
 869     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 870         <list_item>
 871             <paragraph>
 872                 foo
 873     <paragraph>
 874         3-space indent, no trailing space:
 875     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 876         <list_item>
 877             <paragraph>
 878                 foo
 879     <paragraph>
 880         2-space indent:
 881     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 882         <list_item>
 883             <paragraph>
 884                 foo
 885     <paragraph>
 886         1-space indent:
 887     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 888         <list_item>
 889             <paragraph>
 890                 foo
 891     <paragraph>
 892         0-space indent, not a list item:
 893     <paragraph>
 894         1.
 895         foo
 896     <paragraph>
 897         No item content:
 898     <enumerated_list enumtype="arabic" prefix="" suffix=".">
 899         <list_item>
 900 """],
 901 ]
 902
 903
 904 if __name__ == '__main__':
 905     import unittest
 906     unittest.main(defaultTest='suite')