test/tst_record_conv.c

   1 /*
   2  * Copyright (C) 2005-2007, Index Data ApS
   3  * See the file LICENSE for details.
   4  *
   5  * $Id: tst_record_conv.c,v 1.13 2007-01-03 08:42:16 adam Exp $
   6  *
   7  */
   8 #include <yaz/record_conv.h>
   9 #include <yaz/test.h>
  10 #include <yaz/wrbuf.h>
  11 #include <string.h>
  12 #include <yaz/log.h>
  13 #include <yaz/libxml2_error.h>
  14
  15 #if HAVE_CONFIG_H
  16 #include <config.h>
  17 #endif
  18
  19 #if YAZ_HAVE_XML2
  20
  21 #include <libxml/parser.h>
  22 #include <libxml/tree.h>
  23
  24 yaz_record_conv_t conv_configure(const char *xmlstring, WRBUF w)
  25 {
  26     xmlDocPtr doc = xmlParseMemory(xmlstring, strlen(xmlstring));
  27     if (!doc)
  28     {
  29         wrbuf_printf(w, "xmlParseMemory");
  30         return 0;
  31     }
  32     else
  33     {
  34         xmlNodePtr ptr = xmlDocGetRootElement(doc);
  35         yaz_record_conv_t p = yaz_record_conv_create();
  36
  37         if (p)
  38         {
  39             const char *srcdir = getenv("srcdir");
  40             if (srcdir)
  41                 yaz_record_conv_set_path(p, srcdir);
  42         }
  43         if (!ptr)
  44         {
  45             wrbuf_printf(w, "xmlDocGetRootElement");
  46             yaz_record_conv_destroy(p);
  47             p = 0;
  48         }
  49         else if (!p)
  50         {
  51             wrbuf_printf(w, "yaz_record_conv_create");
  52         }
  53         else
  54         {
  55
  56
  57             int r = yaz_record_conv_configure(p, ptr);
  58
  59             if (r)
  60             {
  61                 wrbuf_puts(w, yaz_record_conv_get_error(p));
  62                 yaz_record_conv_destroy(p);
  63                 p = 0;
  64             }
  65         }
  66         xmlFreeDoc(doc);
  67         return p;
  68     }
  69 }
  70
  71 int conv_configure_test(const char *xmlstring, const char *expect_error,
  72                         yaz_record_conv_t *pt)
  73 {
  74     WRBUF w = wrbuf_alloc();
  75     int ret;
  76
  77     yaz_record_conv_t p = conv_configure(xmlstring, w);
  78
  79     if (!p)
  80     {
  81         if (expect_error && !strcmp(wrbuf_buf(w), expect_error))
  82             ret = 1;
  83         else
  84         {
  85             ret = 0;
  86             printf("%.*s\n", wrbuf_len(w), wrbuf_buf(w));
  87         }
  88     }
  89     else
  90     {
  91         if (expect_error)
  92             ret = 0;
  93         else
  94             ret = 1;
  95     }
  96
  97     if (pt)
  98         *pt = p;
  99     else
 100         if (p)
 101             yaz_record_conv_destroy(p);
 102
 103     wrbuf_free(w, 1);
 104     return ret;
 105 }
 106
 107 static void tst_configure(void)
 108 {
 109
 110
 111
 112     YAZ_CHECK(conv_configure_test("<bad", "xmlParseMemory", 0));
 113
 114
 115     YAZ_CHECK(conv_configure_test("<backend syntax='usmarc' name='F'>"
 116                                   "<bad/></backend>",
 117                                   "Element <backend>: expected <marc> or "
 118                                   "<xslt> element, got <bad>", 0));
 119
 120 #if YAZ_HAVE_XSLT
 121     YAZ_CHECK(conv_configure_test("<backend syntax='usmarc' name='F'>"
 122                                   "<xslt stylesheet=\"tst_record_conv.xsl\"/>"
 123                                   "<marc"
 124                                   " inputcharset=\"marc-8\""
 125                                   " outputcharset=\"marc-8\""
 126                                   "/>"
 127                                   "</backend>",
 128                                   "Element <marc>: attribute 'inputformat' "
 129                                   "required", 0));
 130     YAZ_CHECK(conv_configure_test("<backend syntax='usmarc' name='F'>"
 131                                   "<xslt/>"
 132                                   "</backend>",
 133                                   "Element <xslt>: attribute 'stylesheet' "
 134                                   "expected", 0));
 135     YAZ_CHECK(conv_configure_test("<backend syntax='usmarc' name='F'>"
 136                                   "<marc"
 137                                   " inputcharset=\"utf-8\""
 138                                   " outputcharset=\"marc-8\""
 139                                   " inputformat=\"xml\""
 140                                   " outputformat=\"marc\""
 141                                   "/>"
 142                                   "<xslt stylesheet=\"tst_record_conv.xsl\"/>"
 143                                   "</backend>",
 144                                   0, 0));
 145 #else
 146     YAZ_CHECK(conv_configure_test("<backend syntax='usmarc' name='F'>"
 147                                   "<xslt stylesheet=\"tst_record_conv.xsl\"/>"
 148                                   "</backend>",
 149                                   "xslt unsupported."
 150                                   " YAZ compiled without XSLT support", 0));
 151 #endif
 152 }
 153
 154 static int conv_convert_test(yaz_record_conv_t p,
 155                              const char *input_record,
 156                              const char *output_expect_record)
 157 {
 158     int ret = 0;
 159     if (!p)
 160     {
 161         YAZ_CHECK(ret);
 162     }
 163     else
 164     {
 165         WRBUF output_record = wrbuf_alloc();
 166         int r = yaz_record_conv_record(p, input_record, strlen(input_record),
 167                                        output_record);
 168         if (r)
 169         {
 170             if (output_expect_record)
 171             {
 172                 printf("yaz_record_conv error=%s\n",
 173                        yaz_record_conv_get_error(p));
 174                 ret = 0;
 175             }
 176             else
 177                 ret = 1;
 178         }
 179         else
 180         {
 181             if (!output_expect_record)
 182             {
 183                 ret = 0;
 184             }
 185             else if (strlen(output_expect_record) != wrbuf_len(output_record))
 186             {
 187                 int expect_len = strlen(output_expect_record);
 188                 ret = 0;
 189                 printf("output_record expect-len=%d got-len=%d\n", expect_len,
 190                        wrbuf_len(output_record));
 191                 printf("got-output_record = %.*s\n",
 192                        wrbuf_len(output_record), wrbuf_buf(output_record));
 193                 printf("output_expect_record = %s\n",
 194                        output_expect_record);
 195             }
 196             else if (memcmp(output_expect_record, wrbuf_buf(output_record),
 197                             strlen(output_expect_record)))
 198             {
 199                 ret = 0;
 200                 printf("got-output_record = %.*s\n",
 201                        wrbuf_len(output_record), wrbuf_buf(output_record));
 202                 printf("output_expect_record = %s\n",
 203                        output_expect_record);
 204             }
 205             else
 206             {
 207                 ret = 1;
 208             }
 209         }
 210         wrbuf_free(output_record, 1);
 211     }
 212     return ret;
 213 }
 214
 215 static void tst_convert1(void)
 216 {
 217     yaz_record_conv_t p = 0;
 218     const char *marcxml_rec =
 219         "<record xmlns=\"http://www.loc.gov/MARC21/slim\">\n"
 220         "  <leader>00080nam a22000498a 4500</leader>\n"
 221         "  <controlfield tag=\"001\">   11224466 </controlfield>\n"
 222         "  <datafield tag=\"010\" ind1=\" \" ind2=\" \">\n"
 223         "    <subfield code=\"a\">   11224466 </subfield>\n"
 224         "  </datafield>\n"
 225         "</record>\n";
 226     const char *iso2709_rec =
 227         "\x30\x30\x30\x38\x30\x6E\x61\x6D\x20\x61\x32\x32\x30\x30\x30\x34"
 228         "\x39\x38\x61\x20\x34\x35\x30\x30\x30\x30\x31\x30\x30\x31\x33\x30"
 229         "\x30\x30\x30\x30\x30\x31\x30\x30\x30\x31\x37\x30\x30\x30\x31\x33"
 230         "\x1E\x20\x20\x20\x31\x31\x32\x32\x34\x34\x36\x36\x20\x1E\x20\x20"
 231         "\x1F\x61\x20\x20\x20\x31\x31\x32\x32\x34\x34\x36\x36\x20\x1E\x1D";
 232
 233     YAZ_CHECK(conv_configure_test("<backend>"
 234                                   "<marc"
 235                                   " inputcharset=\"utf-8\""
 236                                   " outputcharset=\"marc-8\""
 237                                   " inputformat=\"xml\""
 238                                   " outputformat=\"marc\""
 239                                   "/>"
 240                                   "</backend>",
 241                                   0, &p));
 242     YAZ_CHECK(conv_convert_test(p, marcxml_rec, iso2709_rec));
 243     yaz_record_conv_destroy(p);
 244
 245     YAZ_CHECK(conv_configure_test("<backend>"
 246                                   "<marc"
 247                                   " outputcharset=\"utf-8\""
 248                                   " inputcharset=\"marc-8\""
 249                                   " outputformat=\"marcxml\""
 250                                   " inputformat=\"marc\""
 251                                   "/>"
 252                                   "</backend>",
 253                                   0, &p));
 254     YAZ_CHECK(conv_convert_test(p, iso2709_rec, marcxml_rec));
 255     yaz_record_conv_destroy(p);
 256
 257
 258     YAZ_CHECK(conv_configure_test("<backend>"
 259                                   "<xslt stylesheet=\"tst_record_conv.xsl\"/>"
 260                                   "<xslt stylesheet=\"tst_record_conv.xsl\"/>"
 261                                   "<marc"
 262                                   " inputcharset=\"utf-8\""
 263                                   " outputcharset=\"marc-8\""
 264                                   " inputformat=\"xml\""
 265                                   " outputformat=\"marc\""
 266                                   "/>"
 267                                   "<marc"
 268                                   " outputcharset=\"utf-8\""
 269                                   " inputcharset=\"marc-8\""
 270                                   " outputformat=\"marcxml\""
 271                                   " inputformat=\"marc\""
 272                                   "/>"
 273                                   "</backend>",
 274                                   0, &p));
 275     YAZ_CHECK(conv_convert_test(p, marcxml_rec, marcxml_rec));
 276     yaz_record_conv_destroy(p);
 277
 278
 279     YAZ_CHECK(conv_configure_test("<backend>"
 280                                   "<xslt stylesheet=\"tst_record_conv.xsl\"/>"
 281                                   "<xslt stylesheet=\"tst_record_conv.xsl\"/>"
 282                                   "<marc"
 283                                   " outputcharset=\"marc-8\""
 284                                   " inputformat=\"xml\""
 285                                   " outputformat=\"marc\""
 286                                   "/>"
 287                                   "<marc"
 288                                   " inputcharset=\"marc-8\""
 289                                   " outputformat=\"marcxml\""
 290                                   " inputformat=\"marc\""
 291                                   "/>"
 292                                   "</backend>",
 293                                   0, &p));
 294     YAZ_CHECK(conv_convert_test(p, marcxml_rec, marcxml_rec));
 295     yaz_record_conv_destroy(p);
 296 }
 297
 298 static void tst_convert2(void)
 299 {
 300     yaz_record_conv_t p = 0;
 301     const char *marcxml_rec =
 302         "<record xmlns=\"http://www.loc.gov/MARC21/slim\">\n"
 303         "  <leader>00080nam a22000498a 4500</leader>\n"
 304         "  <controlfield tag=\"001\">   11224466 </controlfield>\n"
 305         "  <datafield tag=\"010\" ind1=\" \" ind2=\" \">\n"
 306         "    <subfield code=\"a\">k&#xf8;benhavn</subfield>\n"
 307         "  </datafield>\n"
 308         "</record>\n";
 309     const char *iso2709_rec =
 310         "\x30\x30\x30\x37\x37\x6E\x61\x6D\x20\x61\x32\x32\x30\x30\x30\x34"
 311         "\x39\x38\x61\x20\x34\x35\x30\x30\x30\x30\x31\x30\x30\x31\x33\x30"
 312         "\x30\x30\x30\x30\x30\x31\x30\x30\x30\x31\x34\x30\x30\x30\x31\x33"
 313         "\x1E\x20\x20\x20\x31\x31\x32\x32\x34\x34\x36\x36\x20\x1E\x20\x20"
 314         "\x1F\x61\x6b\xb2\x62\x65\x6e\x68\x61\x76\x6e\x1E\x1D";
 315
 316     YAZ_CHECK(conv_configure_test("<backend>"
 317                                   "<marc"
 318                                   " inputcharset=\"utf-8\""
 319                                   " outputcharset=\"marc-8\""
 320                                   " inputformat=\"xml\""
 321                                   " outputformat=\"marc\""
 322                                   "/>"
 323                                   "</backend>",
 324                                   0, &p));
 325     YAZ_CHECK(conv_convert_test(p, marcxml_rec, iso2709_rec));
 326     yaz_record_conv_destroy(p);
 327 }
 328
 329 #endif
 330
 331 int main(int argc, char **argv)
 332 {
 333     YAZ_CHECK_INIT(argc, argv);
 334     libxml2_error_to_yazlog(0 /* disable log */, 0);
 335 #if YAZ_HAVE_XML2
 336     tst_configure();
 337 #endif
 338 #if  YAZ_HAVE_XSLT
 339     tst_convert1();
 340     tst_convert2();
 341 #endif
 342     YAZ_CHECK_TERM;
 343 }
 344
 345 /*
 346  * Local variables:
 347  * c-basic-offset: 4
 348  * indent-tabs-mode: nil
 349  * End:
 350  * vim: shiftwidth=4 tabstop=8 expandtab
 351  */
 352