test/test_record_conv.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2013 Index Data
   3  * See the file LICENSE for details.
   4  */
   5 #if HAVE_CONFIG_H
   6 #include <config.h>
   7 #endif
   8
   9 #include <yaz/record_conv.h>
  10 #include <yaz/test.h>
  11 #include <yaz/wrbuf.h>
  12 #include <string.h>
  13 #include <yaz/log.h>
  14 #include <yaz/proto.h>
  15 #include <yaz/prt-ext.h>
  16 #include <yaz/oid_db.h>
  17 #if YAZ_HAVE_XML2
  18
  19 #include <libxml/parser.h>
  20 #include <libxml/tree.h>
  21
  22 #if YAZ_HAVE_XSLT
  23 #include <libxslt/xslt.h>
  24 #endif
  25
  26 yaz_record_conv_t conv_configure(const char *xmlstring, WRBUF w)
  27 {
  28     xmlDocPtr doc = xmlParseMemory(xmlstring, strlen(xmlstring));
  29     if (!doc)
  30     {
  31         wrbuf_printf(w, "xmlParseMemory");
  32         return 0;
  33     }
  34     else
  35     {
  36         xmlNodePtr ptr = xmlDocGetRootElement(doc);
  37         yaz_record_conv_t p = yaz_record_conv_create();
  38
  39         if (p)
  40         {
  41             const char *srcdir = getenv("srcdir");
  42             if (srcdir)
  43                 yaz_record_conv_set_path(p, srcdir);
  44         }
  45         if (!ptr)
  46         {
  47             wrbuf_printf(w, "xmlDocGetRootElement");
  48             yaz_record_conv_destroy(p);
  49             p = 0;
  50         }
  51         else if (!p)
  52         {
  53             wrbuf_printf(w, "yaz_record_conv_create");
  54         }
  55         else
  56         {
  57
  58
  59             int r = yaz_record_conv_configure(p, ptr);
  60
  61             if (r)
  62             {
  63                 wrbuf_puts(w, yaz_record_conv_get_error(p));
  64                 yaz_record_conv_destroy(p);
  65                 p = 0;
  66             }
  67         }
  68         xmlFreeDoc(doc);
  69         return p;
  70     }
  71 }
  72
  73 int conv_configure_test(const char *xmlstring, const char *expect_error,
  74                         yaz_record_conv_t *pt)
  75 {
  76     WRBUF w = wrbuf_alloc();
  77     int ret;
  78
  79     yaz_record_conv_t p = conv_configure(xmlstring, w);
  80
  81     if (!p)
  82     {
  83         if (expect_error && !strcmp(wrbuf_cstr(w), expect_error))
  84             ret = 1;
  85         else
  86         {
  87             ret = 0;
  88             printf("%s\n", wrbuf_cstr(w));
  89         }
  90     }
  91     else
  92     {
  93         if (expect_error)
  94             ret = 0;
  95         else
  96             ret = 1;
  97     }
  98
  99     if (pt)
 100         *pt = p;
 101     else
 102         if (p)
 103             yaz_record_conv_destroy(p);
 104
 105     wrbuf_destroy(w);
 106     return ret;
 107 }
 108
 109 static void tst_configure(void)
 110 {
 111
 112
 113
 114     YAZ_CHECK(conv_configure_test("<bad", "xmlParseMemory", 0));
 115
 116
 117     YAZ_CHECK(conv_configure_test("<backend syntax='usmarc' name='F'>"
 118                                   "<bad/></backend>",
 119                                   "Element <backend>: expected <marc> or "
 120                                   "<xslt> element, got <bad>", 0));
 121
 122 #if YAZ_HAVE_XSLT
 123     YAZ_CHECK(conv_configure_test("<backend syntax='usmarc' name='F'>"
 124                                   "<xslt stylesheet=\"test_record_conv.xsl\"/>"
 125                                   "<marc"
 126                                   " inputcharset=\"marc-8\""
 127                                   " outputcharset=\"marc-8\""
 128                                   "/>"
 129                                   "</backend>",
 130                                   "Element <marc>: attribute 'inputformat' "
 131                                   "required", 0));
 132     YAZ_CHECK(conv_configure_test("<backend syntax='usmarc' name='F'>"
 133                                   "<xslt/>"
 134                                   "</backend>",
 135                                   "Element <xslt>: attribute 'stylesheet' "
 136                                   "expected", 0));
 137     YAZ_CHECK(conv_configure_test("<backend syntax='usmarc' name='F'>"
 138                                   "<marc"
 139                                   " inputcharset=\"utf-8\""
 140                                   " outputcharset=\"marc-8\""
 141                                   " inputformat=\"xml\""
 142                                   " outputformat=\"marc\""
 143                                   "/>"
 144                                   "<xslt stylesheet=\"test_record_conv.xsl\"/>"
 145                                   "</backend>",
 146                                   0, 0));
 147 #else
 148     YAZ_CHECK(conv_configure_test("<backend syntax='usmarc' name='F'>"
 149                                   "<xslt stylesheet=\"test_record_conv.xsl\"/>"
 150                                   "</backend>",
 151                                   "xslt unsupported."
 152                                   " YAZ compiled without XSLT support", 0));
 153 #endif
 154 }
 155
 156 static int conv_convert_test(yaz_record_conv_t p,
 157                              const char *input_record,
 158                              const char *output_expect_record)
 159 {
 160     int ret = 0;
 161     if (!p)
 162     {
 163         YAZ_CHECK(ret);
 164     }
 165     else
 166     {
 167         WRBUF output_record = wrbuf_alloc();
 168         int r = yaz_record_conv_record(p, input_record, strlen(input_record),
 169                                        output_record);
 170         if (r)
 171         {
 172             if (output_expect_record)
 173             {
 174                 printf("yaz_record_conv error=%s\n",
 175                        yaz_record_conv_get_error(p));
 176                 ret = 0;
 177             }
 178             else
 179                 ret = 1;
 180         }
 181         else
 182         {
 183             if (!output_expect_record)
 184             {
 185                 ret = 0;
 186             }
 187             else if (strcmp(output_expect_record, wrbuf_cstr(output_record)))
 188             {
 189                 ret = 0;
 190                 printf("got-output_record len=%ld: %s\n",
 191                        (long) wrbuf_len(output_record),
 192                        wrbuf_cstr(output_record));
 193                 printf("output_expect_record len=%ld %s\n",
 194                        (long) strlen(output_expect_record),
 195                        output_expect_record);
 196             }
 197             else
 198             {
 199                 ret = 1;
 200             }
 201         }
 202         wrbuf_destroy(output_record);
 203     }
 204     return ret;
 205 }
 206
 207 static int conv_convert_test_iter(yaz_record_conv_t p,
 208                                   const char *input_record,
 209                                   const char *output_expect_record,
 210                                   int num_iter)
 211 {
 212     int i;
 213     int ret;
 214     for (i = 0; i < num_iter; i++)
 215     {
 216         ret = conv_convert_test(p, input_record, output_expect_record);
 217         if (!ret)
 218             break;
 219     }
 220     return ret;
 221 }
 222
 223 static void tst_convert1(void)
 224 {
 225     yaz_record_conv_t p = 0;
 226     const char *marcxml_rec =
 227         "<record xmlns=\"http://www.loc.gov/MARC21/slim\">\n"
 228         "  <leader>00080nam a22000498a 4500</leader>\n"
 229         "  <controlfield tag=\"001\">   11224466 </controlfield>\n"
 230         "  <datafield tag=\"010\" ind1=\" \" ind2=\" \">\n"
 231         "    <subfield code=\"a\">   11224466 </subfield>\n"
 232         "  </datafield>\n"
 233         "</record>\n";
 234     const char *tmarcxml_rec =
 235         "<r xmlns=\"http://www.indexdata.com/MARC21/turboxml\">\n"
 236         "  <l>00080nam a22000498a 4500</l>\n"
 237         "  <c001>   11224466 </c001>\n"
 238         "  <d010 i1=\" \" i2=\" \">\n"
 239         "    <sa>   11224466 </sa>\n"
 240         "  </d010>\n"
 241         "</r>\n";
 242     const char *iso2709_rec =
 243         "\x30\x30\x30\x38\x30\x6E\x61\x6D\x20\x61\x32\x32\x30\x30\x30\x34"
 244         "\x39\x38\x61\x20\x34\x35\x30\x30\x30\x30\x31\x30\x30\x31\x33\x30"
 245         "\x30\x30\x30\x30\x30\x31\x30\x30\x30\x31\x37\x30\x30\x30\x31\x33"
 246         "\x1E\x20\x20\x20\x31\x31\x32\x32\x34\x34\x36\x36\x20\x1E\x20\x20"
 247         "\x1F\x61\x20\x20\x20\x31\x31\x32\x32\x34\x34\x36\x36\x20\x1E\x1D";
 248
 249     YAZ_CHECK(conv_configure_test("<backend>"
 250                                   "<marc"
 251                                   " inputcharset=\"utf-8\""
 252                                   " outputcharset=\"marc-8\""
 253                                   " inputformat=\"xml\""
 254                                   " outputformat=\"marc\""
 255                                   "/>"
 256                                   "</backend>",
 257                                   0, &p));
 258     YAZ_CHECK(conv_convert_test(p, marcxml_rec, iso2709_rec));
 259     YAZ_CHECK(conv_convert_test(p, tmarcxml_rec, iso2709_rec));
 260     yaz_record_conv_destroy(p);
 261
 262     YAZ_CHECK(conv_configure_test("<backend>"
 263                                   "<marc"
 264                                   " outputcharset=\"utf-8\""
 265                                   " inputcharset=\"marc-8\""
 266                                   " outputformat=\"marcxml\""
 267                                   " inputformat=\"marc\""
 268                                   "/>"
 269                                   "</backend>",
 270                                   0, &p));
 271     YAZ_CHECK(conv_convert_test(p, iso2709_rec, marcxml_rec));
 272     yaz_record_conv_destroy(p);
 273
 274
 275     YAZ_CHECK(conv_configure_test("<backend>"
 276                                   "<xslt stylesheet=\"test_record_conv.xsl\"/>"
 277                                   "<xslt stylesheet=\"test_record_conv.xsl\"/>"
 278                                   "<marc"
 279                                   " inputcharset=\"utf-8\""
 280                                   " outputcharset=\"marc-8\""
 281                                   " inputformat=\"xml\""
 282                                   " outputformat=\"marc\""
 283                                   "/>"
 284                                   "<marc"
 285                                   " outputcharset=\"utf-8\""
 286                                   " inputcharset=\"marc-8\""
 287                                   " outputformat=\"marcxml\""
 288                                   " inputformat=\"marc\""
 289                                   "/>"
 290                                   "</backend>",
 291                                   0, &p));
 292     YAZ_CHECK(conv_convert_test(p, marcxml_rec, marcxml_rec));
 293     yaz_record_conv_destroy(p);
 294
 295
 296     YAZ_CHECK(conv_configure_test("<backend>"
 297                                   "<xslt stylesheet=\"test_record_conv.xsl\"/>"
 298                                   "<xslt stylesheet=\"test_record_conv.xsl\"/>"
 299                                   "<marc"
 300                                   " outputcharset=\"marc-8\""
 301                                   " inputformat=\"xml\""
 302                                   " outputformat=\"marc\""
 303                                   "/>"
 304                                   "<marc"
 305                                   " inputcharset=\"marc-8\""
 306                                   " outputformat=\"marcxml\""
 307                                   " inputformat=\"marc\""
 308                                   "/>"
 309                                   "</backend>",
 310                                   0, &p));
 311     YAZ_CHECK(conv_convert_test(p, marcxml_rec, marcxml_rec));
 312     yaz_record_conv_destroy(p);
 313 }
 314
 315 static void tst_convert2(void)
 316 {
 317     yaz_record_conv_t p = 0;
 318     const char *marcxml_rec =
 319         "<record xmlns=\"http://www.loc.gov/MARC21/slim\">\n"
 320         "  <leader>00080nam a22000498a 4500</leader>\n"
 321         "  <controlfield tag=\"001\">   11224466 </controlfield>\n"
 322         "  <datafield tag=\"010\" ind1=\" \" ind2=\" \">\n"
 323         "    <subfield code=\"a\">k&#xf8;benhavn</subfield>\n"
 324         "  </datafield>\n"
 325         "</record>\n";
 326     const char *iso2709_rec =
 327         "\x30\x30\x30\x37\x37\x6E\x61\x6D\x20\x61\x32\x32\x30\x30\x30\x34"
 328         "\x39\x38\x61\x20\x34\x35\x30\x30\x30\x30\x31\x30\x30\x31\x33\x30"
 329         "\x30\x30\x30\x30\x30\x31\x30\x30\x30\x31\x34\x30\x30\x30\x31\x33"
 330         "\x1E\x20\x20\x20\x31\x31\x32\x32\x34\x34\x36\x36\x20\x1E\x20\x20"
 331         "\x1F\x61\x6b\xb2\x62\x65\x6e\x68\x61\x76\x6e\x1E\x1D";
 332
 333     YAZ_CHECK(conv_configure_test("<backend>"
 334                                   "<marc"
 335                                   " inputcharset=\"utf-8\""
 336                                   " outputcharset=\"marc-8\""
 337                                   " inputformat=\"xml\""
 338                                   " outputformat=\"marc\""
 339                                   "/>"
 340                                   "</backend>",
 341                                   0, &p));
 342     YAZ_CHECK(conv_convert_test_iter(p, marcxml_rec, iso2709_rec, 100));
 343     yaz_record_conv_destroy(p);
 344 }
 345
 346 static void tst_convert3(void)
 347 {
 348     NMEM nmem = nmem_create();
 349     int ret;
 350     yaz_record_conv_t p = 0;
 351
 352     const char *iso2709_rec =
 353         "\x30\x30\x30\x37\x37\x6E\x61\x6D\x20\x61\x32\x32\x30\x30\x30\x34"
 354         "\x39\x38\x61\x20\x34\x35\x30\x30\x30\x30\x31\x30\x30\x31\x33\x30"
 355         "\x30\x30\x30\x30\x30\x31\x30\x30\x30\x31\x34\x30\x30\x30\x31\x33"
 356         "\x1E\x20\x20\x20\x31\x31\x32\x32\x34\x34\x36\x36\x20\x1E\x20\x20"
 357         "\x1F\x61\x6b\xb2\x62\x65\x6e\x68\x61\x76\x6e\x1E\x1D";
 358
 359     const char *opacxml_rec =
 360         "<opacRecord>\n"
 361         "  <bibliographicRecord>\n"
 362         "<record xmlns=\"http://www.loc.gov/MARC21/slim\">\n"
 363         "  <leader>00077nam a22000498a 4500</leader>\n"
 364         "  <controlfield tag=\"001\">   11224466 </controlfield>\n"
 365         "  <datafield tag=\"010\" ind1=\" \" ind2=\" \">\n"
 366         "    <subfield code=\"a\">k" "\xc3" "\xb8" /* oslash in UTF_8 */
 367         "benhavn</subfield>\n"
 368         "  </datafield>\n"
 369         "</record>\n"
 370         "  </bibliographicRecord>\n"
 371         "<holdings>\n"
 372         " <holding>\n"
 373         "  <typeOfRecord>u</typeOfRecord>\n"
 374         "  <encodingLevel>U</encodingLevel>\n"
 375         "  <receiptAcqStatus>0</receiptAcqStatus>\n"
 376         "  <dateOfReport>000000</dateOfReport>\n"
 377         "  <nucCode>s-FM/GC</nucCode>\n"
 378         "  <localLocation>Main or Science/Business Reading Rms - STORED OFFSITE</localLocation>\n"
 379         "  <callNumber>MLCM 89/00602 (N)</callNumber>\n"
 380         "  <shelvingData>FT MEADE</shelvingData>\n"
 381         "  <copyNumber>Copy 1</copyNumber>\n"
 382         "  <volumes>\n"
 383         "   <volume>\n"
 384         "    <enumeration>1</enumeration>\n"
 385         "    <chronology>2</chronology>\n"
 386         "    <enumAndChron>3</enumAndChron>\n"
 387         "   </volume>\n"
 388         "   <volume>\n"
 389         "    <enumeration>1</enumeration>\n"
 390         "    <chronology>2</chronology>\n"
 391         "    <enumAndChron>3</enumAndChron>\n"
 392         "   </volume>\n"
 393         "  </volumes>\n"
 394         "  <circulations>\n"
 395         "   <circulation>\n"
 396         "    <availableNow value=\"1\"/>\n"
 397         "    <availabilityDate>20130129</availabilityDate>\n"
 398         "    <itemId>1226176</itemId>\n"
 399         "    <renewable value=\"0\"/>\n"
 400         "    <onHold value=\"0\"/>\n"
 401         "   </circulation>\n"
 402         "  </circulations>\n"
 403         " </holding>\n"
 404         "</holdings>\n"
 405         "</opacRecord>\n";
 406
 407     Z_OPACRecord *z_opac = nmem_malloc(nmem, sizeof(*z_opac));
 408     Z_HoldingsAndCircData *h;
 409     Z_CircRecord *circ;
 410
 411     z_opac->bibliographicRecord =
 412         z_ext_record_oid_nmem(nmem, yaz_oid_recsyn_usmarc,
 413                               iso2709_rec, strlen(iso2709_rec));
 414     z_opac->num_holdingsData = 1;
 415     z_opac->holdingsData = (Z_HoldingsRecord **)
 416         nmem_malloc(nmem, sizeof(Z_HoldingsRecord *) * 1);
 417     z_opac->holdingsData[0] = (Z_HoldingsRecord *)
 418         nmem_malloc(nmem, sizeof(Z_HoldingsRecord));
 419     z_opac->holdingsData[0]->which = Z_HoldingsRecord_holdingsAndCirc;
 420     h = z_opac->holdingsData[0]->u.holdingsAndCirc = (Z_HoldingsAndCircData *)
 421          nmem_malloc(nmem, sizeof(*h));
 422     h->typeOfRecord = nmem_strdup(nmem, "u");
 423     h->encodingLevel = nmem_strdup(nmem, "U");
 424     h->format = 0;
 425     h->receiptAcqStatus = nmem_strdup(nmem, "0");
 426     h->generalRetention = 0;
 427     h->completeness = 0;
 428     h->dateOfReport = nmem_strdup(nmem, "000000");
 429     h->nucCode = nmem_strdup(nmem, "s-FM/GC");
 430     h->localLocation = nmem_strdup(nmem,
 431                                    "Main or Science/Business Reading "
 432                                    "Rms - STORED OFFSITE");
 433     h->shelvingLocation = 0;
 434     h->callNumber = nmem_strdup(nmem, "MLCM 89/00602 (N)");
 435     h->shelvingData = nmem_strdup(nmem, "FT MEADE");
 436     h->copyNumber = nmem_strdup(nmem, "Copy 1");
 437     h->publicNote = 0;
 438     h->reproductionNote = 0;
 439     h->termsUseRepro = 0;
 440     h->enumAndChron = 0;
 441     h->num_volumes = 2;
 442     h->volumes = 0;
 443
 444     h->volumes = (Z_Volume **)
 445         nmem_malloc(nmem, 2 * sizeof(Z_Volume *));
 446
 447     h->volumes[0] = (Z_Volume *)
 448         nmem_malloc(nmem, sizeof(Z_Volume));
 449     h->volumes[1] = h->volumes[0];
 450
 451     h->volumes[0]->enumeration = nmem_strdup(nmem, "1");
 452     h->volumes[0]->chronology = nmem_strdup(nmem, "2");
 453     h->volumes[0]->enumAndChron = nmem_strdup(nmem, "3");
 454
 455     h->num_circulationData = 1;
 456     h->circulationData = (Z_CircRecord **)
 457         nmem_malloc(nmem, 1 * sizeof(Z_CircRecord *));
 458     circ = h->circulationData[0] = (Z_CircRecord *)
 459         nmem_malloc(nmem, sizeof(Z_CircRecord));
 460     circ->availableNow = nmem_booldup(nmem, 1);
 461     circ->availablityDate = nmem_strdup(nmem, "20130129");
 462     circ->availableThru = 0;
 463     circ->restrictions = 0;
 464     circ->itemId = nmem_strdup(nmem, "1226176");
 465     circ->renewable = nmem_booldup(nmem, 0);
 466     circ->onHold = nmem_booldup(nmem, 0);
 467     circ->enumAndChron = 0;
 468     circ->midspine = 0;
 469     circ->temporaryLocation = 0;
 470
 471     YAZ_CHECK(conv_configure_test("<backend>"
 472                                   "<marc"
 473                                   " inputcharset=\"marc-8\""
 474                                   " outputcharset=\"utf-8\""
 475                                   " inputformat=\"marc\""
 476                                   " outputformat=\"marcxml\""
 477                                   "/>"
 478                                   "</backend>",
 479                                   0, &p));
 480
 481     if (p)
 482     {
 483         WRBUF output_record = wrbuf_alloc();
 484         ret = yaz_record_conv_opac_record(p, z_opac, output_record);
 485         YAZ_CHECK(ret == 0);
 486         if (ret == 0)
 487         {
 488             ret = strcmp(wrbuf_cstr(output_record), opacxml_rec);
 489             YAZ_CHECK(ret == 0);
 490             if (ret)
 491             {
 492                 printf("got-output_record len=%ld: %s\n",
 493                        (long) wrbuf_len(output_record),
 494                        wrbuf_cstr(output_record));
 495                 printf("output_expect_record len=%ld %s\n",
 496                        (long) strlen(opacxml_rec),
 497                        opacxml_rec);
 498             }
 499         }
 500         yaz_record_conv_destroy(p);
 501         wrbuf_destroy(output_record);
 502     }
 503     {
 504         Z_OPACRecord *opac = 0;
 505         yaz_marc_t mt =  yaz_marc_create();
 506         ret = yaz_xml_to_opac(mt, opacxml_rec, strlen(opacxml_rec),
 507                               &opac, 0 /* iconv */, nmem);
 508         YAZ_CHECK(ret);
 509         YAZ_CHECK(opac);
 510
 511         if (opac)
 512         {
 513             WRBUF output_record = wrbuf_alloc();
 514             char *p;
 515
 516             yaz_marc_xml(mt, YAZ_MARC_MARCXML);
 517             yaz_opac_decode_wrbuf(mt, opac, output_record);
 518
 519             /* change MARC size to 00077 from 00078, due to
 520                encoding of the aring (two bytes in UTF-8) */
 521             p = strstr(wrbuf_buf(output_record), "00078");
 522             YAZ_CHECK(p);
 523             if (p)
 524                 p[4] = '7';
 525
 526             ret = strcmp(wrbuf_cstr(output_record), opacxml_rec);
 527             YAZ_CHECK(ret == 0);
 528             if (ret)
 529             {
 530                 printf("got-output_record len=%ld: %s\n",
 531                        (long) wrbuf_len(output_record),
 532                        wrbuf_cstr(output_record));
 533                 printf("output_expect_record len=%ld %s\n",
 534                        (long) strlen(opacxml_rec),
 535                        opacxml_rec);
 536             }
 537             wrbuf_destroy(output_record);
 538         }
 539         yaz_marc_destroy(mt);
 540     }
 541     nmem_destroy(nmem);
 542 }
 543
 544 #endif
 545
 546 int main(int argc, char **argv)
 547 {
 548     YAZ_CHECK_INIT(argc, argv);
 549     yaz_log_xml_errors(0, 0 /* disable log */);
 550 #if YAZ_HAVE_XML2
 551     tst_configure();
 552 #endif
 553 #if YAZ_HAVE_XSLT
 554     tst_convert1();
 555     tst_convert2();
 556     tst_convert3();
 557     xsltCleanupGlobals();
 558 #endif
 559 #if YAZ_HAVE_XML2
 560     xmlCleanupParser();
 561 #endif
 562     YAZ_CHECK_TERM;
 563 }
 564
 565 /*
 566  * Local variables:
 567  * c-basic-offset: 4
 568  * c-file-style: "Stroustrup"
 569  * indent-tabs-mode: nil
 570  * End:
 571  * vim: shiftwidth=4 tabstop=8 expandtab
 572  */
 573