util/marcdump.c

   1 /*
   2  * Copyright (C) 1995-2005, Index Data ApS
   3  * See the file LICENSE for details.
   4  *
   5  * $Id: marcdump.c,v 1.31 2005-06-09 18:47:07 adam Exp $
   6  */
   7
   8 #define _FILE_OFFSET_BITS 64
   9
  10 #if HAVE_CONFIG_H
  11 #include <config.h>
  12 #endif
  13
  14 #if HAVE_XML2
  15 #include <libxml/parser.h>
  16 #include <libxml/tree.h>
  17
  18 #include <libxml/xpath.h>
  19 #include <libxml/xpathInternals.h>
  20
  21 #endif
  22
  23 #include <stdio.h>
  24 #include <stdlib.h>
  25 #include <string.h>
  26 #include <errno.h>
  27 #include <assert.h>
  28
  29 #if HAVE_LOCALE_H
  30 #include <locale.h>
  31 #endif
  32 #if HAVE_LANGINFO_H
  33 #include <langinfo.h>
  34 #endif
  35
  36 #include <yaz/marcdisp.h>
  37 #include <yaz/yaz-util.h>
  38 #include <yaz/xmalloc.h>
  39 #include <yaz/options.h>
  40
  41 #ifndef SEEK_SET
  42 #define SEEK_SET 0
  43 #endif
  44 #ifndef SEEK_END
  45 #define SEEK_END 2
  46 #endif
  47
  48 static void usage(const char *prog)
  49 {
  50     fprintf (stderr, "Usage: %s [-c cfile] [-f from] [-t to] [-x] [-O] [-X] [-e] [-I] [-v] file...\n",
  51              prog);
  52 }
  53
  54 #if HAVE_XML2
  55 void print_xpath_nodes(xmlNodeSetPtr nodes, FILE* output) {
  56     xmlNodePtr cur;
  57     int size;
  58     int i;
  59
  60     assert(output);
  61     size = (nodes) ? nodes->nodeNr : 0;
  62
  63     fprintf(output, "Result (%d nodes):\n", size);
  64     for(i = 0; i < size; ++i) {
  65         assert(nodes->nodeTab[i]);
  66
  67         if(nodes->nodeTab[i]->type == XML_NAMESPACE_DECL)
  68         {
  69             xmlNsPtr ns;
  70
  71             ns = (xmlNsPtr)nodes->nodeTab[i];
  72             cur = (xmlNodePtr)ns->next;
  73             if(cur->ns) {
  74                 fprintf(output, "= namespace \"%s\"=\"%s\" for node %s:%s\n",
  75                     ns->prefix, ns->href, cur->ns->href, cur->name);
  76             } else {
  77                 fprintf(output, "= namespace \"%s\"=\"%s\" for node %s\n",
  78                     ns->prefix, ns->href, cur->name);
  79             }
  80         }
  81         else if(nodes->nodeTab[i]->type == XML_ELEMENT_NODE)
  82         {
  83             cur = nodes->nodeTab[i];
  84             if(cur->ns) {
  85                 fprintf(output, "= element node \"%s:%s\"\n",
  86                     cur->ns->href, cur->name);
  87             }
  88             else
  89             {
  90                 fprintf(output, "= element node \"%s\"\n",
  91                     cur->name);
  92             }
  93         }
  94         else
  95         {
  96             cur = nodes->nodeTab[i];
  97             fprintf(output, "= node \"%s\": type %d\n", cur->name, cur->type);
  98         }
  99     }
 100 }
 101 #endif
 102
 103 int main (int argc, char **argv)
 104 {
 105     int r;
 106     int libxml_dom_test = 0;
 107     int print_offset = 0;
 108     char *arg;
 109     int verbose = 0;
 110     FILE *inf;
 111     char buf[100001];
 112     char *prog = *argv;
 113     int no = 0;
 114     int xml = 0;
 115     FILE *cfile = 0;
 116     char *from = 0, *to = 0;
 117     int num = 1;
 118
 119 #if HAVE_LOCALE_H
 120     setlocale(LC_CTYPE, "");
 121 #endif
 122 #if HAVE_LANGINFO_H
 123 #ifdef CODESET
 124     to = nl_langinfo(CODESET);
 125 #endif
 126 #endif
 127
 128     while ((r = options("pvc:xOeXIf:t:2", argv, argc, &arg)) != -2)
 129     {
 130         int count;
 131         no++;
 132         switch (r)
 133         {
 134         case 'f':
 135             from = arg;
 136             break;
 137         case 't':
 138             to = arg;
 139             break;
 140         case 'c':
 141             if (cfile)
 142                 fclose (cfile);
 143             cfile = fopen(arg, "w");
 144             break;
 145         case 'x':
 146             xml = YAZ_MARC_SIMPLEXML;
 147             break;
 148         case 'O':
 149             xml = YAZ_MARC_OAIMARC;
 150             break;
 151         case 'e':
 152             xml = YAZ_MARC_XCHANGE;
 153             break;
 154         case 'X':
 155             xml = YAZ_MARC_MARCXML;
 156             break;
 157         case 'I':
 158             xml = YAZ_MARC_ISO2709;
 159             break;
 160         case 'p':
 161             print_offset = 1;
 162             break;
 163         case '2':
 164             libxml_dom_test = 1;
 165             break;
 166         case 0:
 167             inf = fopen(arg, "rb");
 168             count = 0;
 169             if (!inf)
 170             {
 171                 fprintf (stderr, "%s: cannot open %s:%s\n",
 172                          prog, arg, strerror (errno));
 173                 exit(1);
 174             }
 175             if (cfile)
 176                 fprintf (cfile, "char *marc_records[] = {\n");
 177             if (1)
 178             {
 179                 yaz_marc_t mt = yaz_marc_create();
 180                 yaz_iconv_t cd = 0;
 181
 182                 if (from && to)
 183                 {
 184                     cd = yaz_iconv_open(to, from);
 185                     if (!cd)
 186                     {
 187                         fprintf(stderr, "conversion from %s to %s "
 188                                 "unsupported\n", from, to);
 189                         exit(2);
 190                     }
 191                     yaz_marc_iconv(mt, cd);
 192                 }
 193                 yaz_marc_xml(mt, xml);
 194                 yaz_marc_debug(mt, verbose);
 195                 while (1)
 196                 {
 197                     int len;
 198                     char *result = 0;
 199                     int rlen;
 200
 201                     r = fread (buf, 1, 5, inf);
 202                     if (r < 5)
 203                     {
 204                         if (r && print_offset && verbose)
 205                             printf ("<!-- Extra %d bytes at end of file -->\n", r);
 206                         break;
 207                     }
 208                     while (*buf < '0' || *buf > '9')
 209                     {
 210                         int i;
 211                         long off = ftell(inf) - 5;
 212                         if (verbose || print_offset)
 213                             printf("<!-- Skipping bad byte %d (0x%02X) at offset "
 214                                    "%ld (0x%lx) -->\n",
 215                                    *buf & 0xff, *buf & 0xff,
 216                                    off, off);
 217                         for (i = 0; i<4; i++)
 218                             buf[i] = buf[i+1];
 219                         r = fread(buf+4, 1, 1, inf);
 220                         if (r < 1)
 221                             break;
 222                     }
 223                     if (r < 1)
 224                     {
 225                         if (verbose || print_offset)
 226                             printf ("<!-- End of file with data -->\n");
 227                         break;
 228                     }
 229                     if (print_offset)
 230                     {
 231                         long off = ftell(inf) - 5;
 232                         printf ("<!-- Record %d offset %ld (0x%lx) -->\n",
 233                                 num, off, off);
 234                     }
 235                     len = atoi_n(buf, 5);
 236                     if (len < 25 || len > 100000)
 237                     {
 238                         long off = ftell(inf) - 5;
 239                         printf("Bad Length %d read at offset %ld (%lx)\n",
 240                                len, (long) off, (long) off);
 241                         break;
 242                     }
 243                     len = len - 5;
 244                     r = fread (buf + 5, 1, len, inf);
 245                     if (r < len)
 246                         break;
 247                     r = yaz_marc_decode_buf (mt, buf, -1, &result, &rlen);
 248                     if (result)
 249                         fwrite (result, rlen, 1, stdout);
 250 #if HAVE_XML2
 251                     if (r > 0 && libxml_dom_test)
 252                     {
 253                         xmlDocPtr doc = xmlParseMemory(result, rlen);
 254                         if (!doc)
 255                             fprintf(stderr, "xmLParseMemory failed\n");
 256                         else
 257                         {
 258                             int i;
 259                             xmlXPathContextPtr xpathCtx;
 260                             xmlXPathObjectPtr xpathObj;
 261                             static const char *xpathExpr[] = {
 262                                 "/record/datafield[@tag='245']/subfield[@code='a']",
 263                                 "/record/datafield[@tag='100']/subfield",
 264                                 "/record/datafield[@tag='245']/subfield[@code='a']",
 265                                 "/record/datafield[@tag='650']/subfield",
 266                                 "/record/datafield[@tag='650']",
 267                                 0};
 268
 269                             xpathCtx = xmlXPathNewContext(doc);
 270
 271                             for (i = 0; xpathExpr[i]; i++) {
 272                                 xpathObj = xmlXPathEvalExpression(xpathExpr[i], xpathCtx);
 273                                 if(xpathObj == NULL) {
 274                                     fprintf(stderr,"Error: unable to evaluate xpath expression \"%s\"\n", xpathExpr[i]);
 275                                 }
 276                                 else
 277                                 {
 278                                     print_xpath_nodes(xpathObj->nodesetval, stdout);
 279                                     xmlXPathFreeObject(xpathObj);
 280                                 }
 281                             }
 282                             xmlXPathFreeContext(xpathCtx);
 283                             xmlFreeDoc(doc);
 284                         }
 285                     }
 286 #endif
 287                     if (r > 0 && cfile)
 288                     {
 289                         char *p = buf;
 290                         int i;
 291                         if (count)
 292                             fprintf (cfile, ",");
 293                         fprintf (cfile, "\n");
 294                         for (i = 0; i < r; i++)
 295                         {
 296                             if ((i & 15) == 0)
 297                                 fprintf (cfile, "  \"");
 298                             fprintf (cfile, "\\x%02X", p[i] & 255);
 299
 300                             if (i < r - 1 && (i & 15) == 15)
 301                                 fprintf (cfile, "\"\n");
 302
 303                         }
 304                         fprintf (cfile, "\"\n");
 305                     }
 306                     num++;
 307                     if (verbose)
 308                         printf("\n");
 309                 }
 310                 count++;
 311                 if (cd)
 312                     yaz_iconv_close(cd);
 313                 yaz_marc_destroy(mt);
 314             }
 315             if (cfile)
 316                 fprintf (cfile, "};\n");
 317             fclose(inf);
 318             break;
 319         case 'v':
 320             verbose++;
 321             break;
 322         default:
 323             usage(prog);
 324             exit (1);
 325         }
 326     }
 327     if (cfile)
 328         fclose (cfile);
 329     if (!no)
 330     {
 331         usage(prog);
 332         exit (1);
 333     }
 334     exit (0);
 335 }