icuSources/test/testdata/ssearch.xml

   1 <?xml version="1.0" encoding="UTF-8"?>
   2
   3 <!-- Copyright (C) 2016 and later: Unicode, Inc. and others. License & terms of use: http://www.unicode.org/copyright.html -->
   4 <!-- Copyright (c) 2007-2009 IBM Corporation and others. All rights reserved -->
   5
   6 <!-- Test data file for string search  -->
   7 <!DOCTYPE stringsearch-tests [
   8 <!ELEMENT stringsearch-tests (test-case+)>
   9 <!ATTLIST stringsearch-tests debug IDREF #IMPLIED >
  10 <!ELEMENT test-case (pattern, pre?, m?, post?)>
  11 <!ATTLIST test-case
  12           id ID #REQUIRED
  13           locale CDATA "en"
  14           strength (PRIMARY | SECONDARY | TERTIARY | QUATERNARY | IDENTICAL) "TERTIARY"
  15           norm (ON | OFF) "OFF"
  16           alternate_handling (NON_IGNORABLE | SHIFTED) "NON_IGNORABLE"
  17           >
  18
  19 <!ELEMENT pattern (#PCDATA)>
  20 <!ELEMENT pre  (#PCDATA)>
  21 <!ELEMENT m    (#PCDATA)>
  22 <!ELEMENT post (#PCDATA)>
  23 ]>
  24
  25 <stringsearch-tests>
  26   <!-- debug="test11"     (for copying into the above element)  -->
  27
  28     <!-- Very simple match  -->
  29     <test-case id="test01" >
  30        <pattern>abc</pattern>
  31        <pre>xxx</pre><m>abc</m><post>yyy</post>
  32     </test-case>
  33
  34     <!-- Very simple no-match  -->
  35     <test-case id="test02" >
  36        <pattern>abc</pattern>
  37        <pre>xxx</pre><post>yyy</post>
  38     </test-case>
  39
  40     <!-- Match after several near-misses. -->
  41     <test-case id="test03" >
  42        <pattern>string</pattern>
  43        <pre>silly spring stling strxng strilg strinx stri</pre><m>string</m><post> fling</post>
  44     </test-case>
  45
  46     <test-case id="test04" strength="PRIMARY" >
  47        <pattern>FUSS</pattern>
  48        <pre>abc</pre><m>fuss</m><post>sss</post>
  49     </test-case>
  50
  51     <test-case id="test05" strength="PRIMARY" >
  52        <pattern>FUSS</pattern>
  53        <pre>abc</pre><m>fuß</m><post>sss</post>
  54     </test-case>
  55
  56   <test-case id="test05.5" strength="PRIMARY" >
  57     <pattern>fuss</pattern>
  58     <pre>a </pre>
  59     <m>fuß</m>
  60     <post>ball table</post>
  61   </test-case>
  62
  63   <test-case id="test06" strength="PRIMARY" >
  64       <pattern>fuß</pattern>
  65        <pre>abc</pre><m>fuss</m><post>xyz</post>
  66     </test-case>
  67
  68     <test-case id="test07" strength="SECONDARY" >
  69       <pattern>fuß</pattern>
  70       <pre>abcfussxyz</pre>
  71     </test-case>
  72
  73     <test-case id="test08" strength="PRIMARY" >
  74       <pattern>fus</pattern>
  75       <pre>abcfuß</pre><post>xyz</post>
  76     </test-case>
  77
  78     <!-- A good match following an initial match that failed because
  79          of not ending on a character boundary -->
  80     <test-case id="test09" strength="PRIMARY">
  81       <pattern>fus</pattern>
  82       <pre>fuß  </pre><m>fus</m><post>sss</post>
  83     </test-case>
  84
  85
  86     <!-- Test cases from usrchdat.c  BREAKITERATOREXACT -->
  87
  88     <test-case id="test10" strength="TERTIARY">
  89       <pattern>fox</pattern>
  90       <m>fox</m><post>y fox</post>
  91     </test-case>
  92
  93     <test-case id="test11" strength="PRIMARY" locale="de_DE@collation=phonebook">
  94       <pattern>toe</pattern>
  95       <pre>This is a </pre><m>Tö</m><post>ne</post>
  96     </test-case>
  97
  98     <test-case id="test11a" strength="SECONDARY" locale="de_DE@collation=phonebook">
  99       <pattern>toe</pattern>
 100       <pre>This is a </pre><post>Töne</post>
 101     </test-case>
 102
 103     <test-case id="test12" strength="TERTIARY">
 104       <pattern>e</pattern>
 105       <pre>tésting that é doés not match </pre><m>e</m><post></post>
 106     </test-case>
 107
 108     <test-case id="test13" strength="PRIMARY" locale="fr">
 109       <pattern>e</pattern>
 110       <pre></pre><m>É</m><post>É</post>
 111     </test-case>
 112
 113     <test-case id="test14" strength="PRIMARY" locale="fr">
 114       <pattern>O</pattern>
 115       <pre>C</pre><m>O\u0302</m><post>TÉ</post>
 116     </test-case>
 117
 118
 119     <!-- Test cases from usrchdat.c  STRENGTH -->
 120
 121
 122     <test-case id="test15" strength="PRIMARY" locale="en">
 123       <pattern>fox</pattern>
 124       <pre>The quick brown </pre><m>fox</m><post> jumps over the lazy foxes</post>
 125     </test-case>
 126
 127     <test-case id="test16" strength="PRIMARY" locale="fr">
 128       <pattern>peche</pattern>
 129       <pre>blackbirds pat </pre><m>p\u00E9ch\u00E9</m><post> </post>
 130     </test-case>
 131
 132     <test-case id="test17" strength="PRIMARY" locale="fr">
 133       <pattern>peche</pattern>
 134       <pre>blackbirds pat </pre><m>p\u00EAche</m><post> </post>
 135     </test-case>
 136
 137     <test-case id="test18" strength="PRIMARY" locale="fr">
 138       <pattern>peche</pattern>
 139       <pre>blackbirds pat </pre><m>p\u00E9che</m><post>r </post>
 140     </test-case>
 141
 142     <test-case id="test19" strength="PRIMARY" locale="fr">
 143       <pattern>peche</pattern>
 144       <pre>blackbirds pat </pre><m>p\u00EAche</m><post>r </post>
 145     </test-case>
 146
 147     <test-case id="test20" strength="PRIMARY" locale="es">
 148       <pattern>channel</pattern>
 149       <pre>A </pre><m>channel</m><post>, </post>
 150     </test-case>
 151
 152     <test-case id="test21" strength="PRIMARY" locale="es">
 153       <pattern>channel</pattern>
 154       <pre>A </pre><m>CHANNEL</m><post>, </post>
 155     </test-case>
 156
 157     <test-case id="test22" strength="PRIMARY" locale="es">
 158       <pattern>channel</pattern>
 159       <pre>A </pre><m>Channel</m><post>s, </post>
 160     </test-case>
 161
 162     <test-case id="test23" strength="PRIMARY" locale="es">
 163       <pattern>channel</pattern>
 164       <pre>A </pre><m>channel</m><post>... </post>
 165     </test-case>
 166
 167     <test-case id="test24" strength="TERTIARY" locale="en">
 168       <pattern>A\u0300</pattern>
 169       <pre>A miss, and then </pre><m>\u00c0</m><post> should match but not A"</post>
 170     </test-case>
 171
 172     <!-- TODO:  In the original test data, this test matched at IDENTICAL strength.
 173                 Doesn't seem right.  The characters are different.
 174                 -->
 175     <test-case id="test24a" strength="IDENTICAL" locale="en">
 176       <pattern>A\u0300</pattern>
 177       <pre>At IDENTICAL, shoud this match?  </pre><m>\u00c0</m><post></post>
 178     </test-case>
 179
 180   <test-case id="test24b" strength="IDENTICAL" alternate_handling="SHIFTED" locale="en">
 181     <pattern>A\u0300</pattern>
 182     <pre>At IDENTICAL, shoud this match?  </pre>
 183     <m>\u00c0</m>
 184     <post></post>
 185   </test-case>
 186
 187   <test-case id="test25" strength="SECONDARY" locale="en">
 188       <pattern>Ű</pattern>
 189       <pre>12</pre><m>ű</m><post> Ű</post>
 190     </test-case>
 191
 192     <test-case id="test26" strength="SECONDARY" locale="en">
 193       <pattern>A</pattern>
 194       <pre>12</pre><m>a</m><post>...</post>
 195     </test-case>
 196
 197
 198     <!--  Test Cases from usrchdat.c,  VARIABLE -->
 199     <test-case id="test27" strength="TERTIARY" locale="en">
 200       <pattern>blackbird</pattern>
 201       <pre>black-bird </pre><m>blackbird</m><post>...</post>
 202     </test-case>
 203
 204     <test-case id="test28" strength="TERTIARY" locale="en">
 205       <pattern>go</pattern>
 206       <pre> on</pre>
 207     </test-case>
 208
 209     <!-- TODO:  this gives an U_ILLEGAL_ARGUMENT error when opening
 210                 the UStringSearch.  How did the orignal test run? -->
 211     <!--
 212     <test-case id="test29" strength="PRIMARY" locale="en">
 213       <pattern>  </pattern>
 214       <pre></pre><m></m><post>abc</post>
 215     </test-case>
 216     -->
 217
 218     <test-case id="test30" strength="SECONDARY" locale="en">
 219       <pattern>abc</pattern>
 220       <pre>  a bc   ab c    a  bc     ab  c"</pre>
 221     </test-case>
 222
 223     <test-case id="test31" strength="SECONDARY" locale="en">
 224       <pattern>abc</pattern>
 225       <pre>           ---------------</pre>
 226     </test-case>
 227
 228
 229     <!--  Normalization test cases from usrchdat.c  -->
 230     <test-case id="test32" strength="TERTIARY"  norm="ON">
 231       <pattern>a\u0325\u0300</pattern>
 232       <pre></pre><m>a\u0300\u0325</m>
 233     </test-case>
 234
 235
 236     <test-case id="test32a" strength="TERTIARY"  norm="OFF">
 237       <pattern>a\u0325\u0300</pattern>
 238       <pre>a\u0300\u0325</pre>
 239     </test-case>
 240
 241
 242     <!-- COMPOSITEBOUNDARIES from usrchdat.c
 243          Boundaries are not identical to orignal test data because
 244          of matching only full combining sequences
 245     -->
 246     <test-case id="test40" strength="TERTIARY">
 247       <pattern>A</pattern>
 248       <pre>À</pre>   <!-- \u00C0 -->
 249     </test-case>
 250
 251     <test-case id="test41" strength="TERTIARY">
 252       <pattern>A</pattern>
 253       <pre>À</pre><m>A</m><post>C</post>
 254     </test-case>
 255
 256     <test-case id="test42" strength="TERTIARY">
 257       <pattern>A\u030A</pattern>
 258       <pre>À\u01FA</pre>
 259     </test-case>
 260
 261
 262
 263     <!-- SUPPLEMENTARYCANONICAL from usrchdat.c  -->
 264     <test-case id="test50" strength="TERTIARY">
 265       <pattern>\uD800\uDC00</pattern>
 266       <pre>abc \uD802\uDC00 \uD800\uDC01 \uD801\uDC00 </pre><m>\uD800\uDC00</m>
 267       <post>abc abc\uD800\uDC00 \uD800\uD800\uDC00 \uD800\uDC00\uDC00</post>
 268     </test-case>
 269
 270     <test-case id="test51" strength="TERTIARY">
 271       <pattern>\\uD834\\uDDB9</pattern>
 272       <pre>and</pre><m>\\uD834\\uDDB9</m><post>this sentence</post>
 273     </test-case>
 274
 275     <test-case id="test52" strength="TERTIARY">
 276       <pattern> \\uD834\\uDDB9 </pattern>
 277       <pre>and</pre><m> \\uD834\\uDDB9 </m><post>this sentence</post>
 278     </test-case>
 279
 280     <test-case id="test53" strength="TERTIARY">
 281       <pattern>-\\uD834\\uDDB9-</pattern>
 282       <pre>and</pre><m>-\\uD834\\uDDB9-</m><post>this sentence</post>
 283     </test-case>
 284
 285     <test-case id="test54" strength="TERTIARY">
 286       <pattern>,\\uD834\\uDDB9,</pattern>
 287       <pre>and</pre><m>,\\uD834\\uDDB9,</m><post>this sentence</post>
 288     </test-case>
 289
 290     <test-case id="test55" strength="TERTIARY">
 291       <pattern>?\\uD834\\uDDB9?</pattern>
 292       <pre>and</pre><m>?\\uD834\\uDDB9?</m><post>this sentence</post>
 293     </test-case>
 294
 295
 296     <!-- Long combining sequences  -->
 297     <!-- Backwards search fails because patterns ends w/ ignorables
 298     <test-case id="test60" strength="PRIMARY">
 299       <pattern>A\u0301\u0301\u0301\u0301</pattern>
 300       <m>A\u0301\u0301\u0301\u0301\u0301</m>
 301     </test-case>
 302     -->
 303
 304     <test-case id="test61" strength="TERTIARY">
 305       <pattern>A\u0301\u0301\u0301\u0301</pattern>
 306           <pre>A\u0301\u0301\u0301\u0301\u0301</pre>
 307     </test-case>
 308
 309     <test-case id="test62" strength="TERTIARY">
 310       <pattern>A\u0301\u0301\u0301\u0301</pattern>
 311             <m>A\u0301\u0301\u0301\u0301</m>
 312     </test-case>
 313
 314     <!-- stand-alone combining marks don't match attached marks  -->
 315     <test-case id="test63" strength="TERTIARY">
 316       <pattern>\u0301</pattern>
 317       <pre>A\u0301\u0301\u0301\u0301</pre>
 318     </test-case>
 319
 320     <test-case id="test64" strength="TERTIARY">
 321       <pattern>\u0301</pattern>
 322       <post>\u0301\u0301\u0301\u0301</post>
 323     </test-case>
 324
 325   <!-- stand-alone combining mark does match an un-attached combining mark -->
 326     <test-case id="test65" strength="TERTIARY">
 327        <pattern>\u0301</pattern>
 328        <m>\u0301</m><post>A\u0301\u0301</post>
 329     </test-case>
 330
 331     <test-case id="test66" strength="TERTIARY">
 332        <pattern>\u0301</pattern>
 333        <m>\u0301</m>
 334     </test-case>
 335
 336     <!-- stand-alone combining marks at end of the target text -->
 337     <test-case id="test67" strength="TERTIARY">
 338        <pattern>\u0301</pattern>
 339        <pre>abcd\r</pre><m>\u0301</m>
 340     </test-case>
 341
 342       <!-- attached combining marks at end of the target text, no match -->
 343     <test-case id="test68" strength="TERTIARY">
 344        <pattern>\u0301</pattern>
 345        <pre>abcd\u0301</pre>
 346     </test-case>
 347
 348
 349
 350    <!-- no match within expansions at the start -->
 351     <test-case id="test70" strength="PRIMARY">
 352       <pattern>Eligature</pattern>
 353       <pre>Æligature</pre>
 354     </test-case>
 355
 356     <test-case id="test71" strength="PRIMARY">
 357       <pattern>AEligature</pattern>
 358       <m>Æligature</m>
 359     </test-case>
 360
 361     <test-case id="test72" strength="PRIMARY">
 362         <pattern>AEligature</pattern>
 363         <m>Æligature</m>
 364     </test-case>
 365
 366     <!-- unattached combining Tilde will not match a Tilde that is
 367          part of a composed Ñ  (\u00D1)  -->
 368     <test-case id="test73" strength="SECONDARY">
 369         <pattern>\u0303</pattern>  <!-- combining tilde -->
 370         <pre>Ñ&#x0d;</pre><m>\u0303</m>
 371     </test-case>
 372
 373     <test-case id="test74" strength="SECONDARY">
 374         <pattern>\u0303</pattern>  <!-- combining tilde -->
 375         <pre>Ñ &#x0d;</pre><m>\u0303</m><post>a</post>
 376     </test-case>
 377
 378   <test-case id="test75" strength="TERTIARY" locale="fr">
 379     <pattern>\u00EA</pattern>
 380     <pre>p</pre><m>\u00EA</m><post>che</post>
 381   </test-case>
 382
 383   <test-case id="test76" strength="TERTIARY" locale="fr">
 384     <pattern>\u00EA</pattern>
 385     <pre>p</pre><m>e\u0302</m><post>che</post>
 386   </test-case>
 387
 388   <test-case id="test77" strength="TERTIARY" locale="fr">
 389     <pattern>e\u0302</pattern>
 390     <pre>p</pre><m>\u00EA</m><post>che</post>
 391   </test-case>
 392
 393   <!-- Test cases from ticket:5382 -->
 394   <test-case id="test78" strength="SECONDARY" locale="hu_HU">
 395     <pattern>\u0170</pattern>
 396     <m>\u0171</m>
 397     <post>12</post>
 398   </test-case>
 399
 400   <test-case id="test79" strength="SECONDARY" locale="hu_HU">
 401     <pattern>\u0170</pattern>
 402     <pre>1</pre>
 403     <m>\u0171</m>
 404     <post>2</post>
 405   </test-case>
 406
 407   <test-case id="test80" strength="SECONDARY" locale="hu_HU">
 408     <pattern>\u0170</pattern>
 409     <pre>12</pre>
 410     <m>\u0171</m>
 411   </test-case>
 412
 413   <!-- Test cases from ticket:5959 -->
 414   <test-case id="test81" strength="SECONDARY">
 415     <pattern>\u2166</pattern>
 416     <m>VII</m>
 417   </test-case>
 418
 419   <test-case id="test82" strength="SECONDARY">
 420     <pattern>VII</pattern>
 421     <m>\u2166</m>
 422   </test-case>
 423
 424   <test-case id="test83" strength="IDENTICAL" alternate_handling="SHIFTED" locale="en">
 425     <pattern>Universal Declaration of Human Rights</pattern>
 426     <pre>Proclaims this </pre><m>Universal Declaration of Human Rights</m><post> as a common standard of achievement for all peoples and all nations</post>
 427   </test-case>
 428
 429   <test-case id="test83b" strength="TERTIARY" alternate_handling="SHIFTED" locale="en">
 430     <pattern>Universal Declaration of Human Rights</pattern>
 431     <pre>Proclaims this </pre>
 432     <m>Universal-Declaration-of-Human-Rights</m>
 433     <post> as a common standard of achievement for all peoples and all nations</post>
 434   </test-case>
 435
 436   <test-case id="test84" strength="TERTIARY" locale="en">
 437     <pattern>\u05E9\u0591\u05E9</pattern>
 438     <m>\u05E9\u0592\u05E9</m>
 439   </test-case>
 440
 441   <test-case id="test84b" strength="IDENTICAL" locale="en">
 442     <pattern>\u05E9\u0591\u05E9</pattern>
 443     <pre>\u05E9\u0592\u05E9</pre>
 444   </test-case>
 445 </stringsearch-tests>
 446