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 1.34 2005-12-17 20:22:01 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     const char *split_fname = 0;
 119
 120 #if HAVE_LOCALE_H
 121     setlocale(LC_CTYPE, "");
 122 #endif
 123 #if HAVE_LANGINFO_H
 124 #ifdef CODESET
 125     to = nl_langinfo(CODESET);
 126 #endif
 127 #endif
 128
 129     while ((r = options("pvc:xOeXIf:t:2s:", argv, argc, &arg)) != -2)
 130     {
 131         int count;
 132         no++;
 133         switch (r)
 134         {
 135         case 'f':
 136             from = arg;
 137             break;
 138         case 't':
 139             to = arg;
 140             break;
 141         case 'c':
 142             if (cfile)
 143                 fclose (cfile);
 144             cfile = fopen(arg, "w");
 145             break;
 146         case 'x':
 147             xml = YAZ_MARC_SIMPLEXML;
 148             break;
 149         case 'O':
 150             xml = YAZ_MARC_OAIMARC;
 151             break;
 152         case 'e':
 153             xml = YAZ_MARC_XCHANGE;
 154             break;
 155         case 'X':
 156             xml = YAZ_MARC_MARCXML;
 157             break;
 158         case 'I':
 159             xml = YAZ_MARC_ISO2709;
 160             break;
 161         case 'p':
 162             print_offset = 1;
 163             break;
 164         case '2':
 165             libxml_dom_test = 1;
 166             break;
 167         case 's':
 168             split_fname = arg;
 169             break;
 170         case 0:
 171             inf = fopen(arg, "rb");
 172             count = 0;
 173             if (!inf)
 174             {
 175                 fprintf (stderr, "%s: cannot open %s:%s\n",
 176                          prog, arg, strerror (errno));
 177                 exit(1);
 178             }
 179             if (cfile)
 180                 fprintf (cfile, "char *marc_records[] = {\n");
 181             if (1)
 182             {
 183                 yaz_marc_t mt = yaz_marc_create();
 184                 yaz_iconv_t cd = 0;
 185                 int marc_no = 0;
 186
 187                 if (from && to)
 188                 {
 189                     cd = yaz_iconv_open(to, from);
 190                     if (!cd)
 191                     {
 192                         fprintf(stderr, "conversion from %s to %s "
 193                                 "unsupported\n", from, to);
 194                         exit(2);
 195                     }
 196                     yaz_marc_iconv(mt, cd);
 197                 }
 198                 yaz_marc_xml(mt, xml);
 199                 yaz_marc_debug(mt, verbose);
 200                 for(;; marc_no++)
 201                 {
 202                     int len;
 203                     char *result = 0;
 204                     int rlen;
 205
 206                     r = fread (buf, 1, 5, inf);
 207                     if (r < 5)
 208                     {
 209                         if (r && print_offset && verbose)
 210                             printf ("<!-- Extra %d bytes at end of file -->\n", r);
 211                         break;
 212                     }
 213                     while (*buf < '0' || *buf > '9')
 214                     {
 215                         int i;
 216                         long off = ftell(inf) - 5;
 217                         if (verbose || print_offset)
 218                             printf("<!-- Skipping bad byte %d (0x%02X) at offset "
 219                                    "%ld (0x%lx) -->\n",
 220                                    *buf & 0xff, *buf & 0xff,
 221                                    off, off);
 222                         for (i = 0; i<4; i++)
 223                             buf[i] = buf[i+1];
 224                         r = fread(buf+4, 1, 1, inf);
 225                         if (r < 1)
 226                             break;
 227                     }
 228                     if (r < 1)
 229                     {
 230                         if (verbose || print_offset)
 231                             printf ("<!-- End of file with data -->\n");
 232                         break;
 233                     }
 234                     if (print_offset)
 235                     {
 236                         long off = ftell(inf) - 5;
 237                         printf ("<!-- Record %d offset %ld (0x%lx) -->\n",
 238                                 num, off, off);
 239                     }
 240                     len = atoi_n(buf, 5);
 241                     if (len < 25 || len > 100000)
 242                     {
 243                         long off = ftell(inf) - 5;
 244                         printf("Bad Length %d read at offset %ld (%lx)\n",
 245                                len, (long) off, (long) off);
 246                         break;
 247                     }
 248                     rlen = len - 5;
 249                     r = fread (buf + 5, 1, rlen, inf);
 250                     if (r < rlen)
 251                         break;
 252                     if (split_fname)
 253                     {
 254                         char fname[256];
 255                         sprintf(fname, "%.200s%07d", split_fname, marc_no);
 256                         FILE *sf = fopen(fname, "wb");
 257                         if (!sf)
 258                         {
 259                             fprintf(stderr, "Could not open %s\n", fname);
 260                             split_fname = 0;
 261                         }
 262                         else
 263                         {
 264                             if (fwrite(buf, 1, len, sf) != len)
 265                             {
 266                                 fprintf(stderr, "Could write content to %s\n",
 267                                         fname);
 268                                 split_fname = 0;
 269                             }
 270                             fclose(sf);
 271                         }
 272                     }
 273                     r = yaz_marc_decode_buf (mt, buf, -1, &result, &rlen);
 274                     if (result)
 275                         fwrite (result, rlen, 1, stdout);
 276 #if HAVE_XML2
 277                     if (r > 0 && libxml_dom_test)
 278                     {
 279                         xmlDocPtr doc = xmlParseMemory(result, rlen);
 280                         if (!doc)
 281                             fprintf(stderr, "xmLParseMemory failed\n");
 282                         else
 283                         {
 284                             int i;
 285                             xmlXPathContextPtr xpathCtx;
 286                             xmlXPathObjectPtr xpathObj;
 287                             static const char *xpathExpr[] = {
 288                                 "/record/datafield[@tag='245']/subfield[@code='a']",
 289                                 "/record/datafield[@tag='100']/subfield",
 290                                 "/record/datafield[@tag='245']/subfield[@code='a']",
 291                                 "/record/datafield[@tag='650']/subfield",
 292                                 "/record/datafield[@tag='650']",
 293                                 0};
 294
 295                             xpathCtx = xmlXPathNewContext(doc);
 296
 297                             for (i = 0; xpathExpr[i]; i++) {
 298                                 xpathObj = xmlXPathEvalExpression(BAD_CAST xpathExpr[i], xpathCtx);
 299                                 if(xpathObj == NULL) {
 300                                     fprintf(stderr,"Error: unable to evaluate xpath expression \"%s\"\n", xpathExpr[i]);
 301                                 }
 302                                 else
 303                                 {
 304                                     print_xpath_nodes(xpathObj->nodesetval, stdout);
 305                                     xmlXPathFreeObject(xpathObj);
 306                                 }
 307                             }
 308                             xmlXPathFreeContext(xpathCtx);
 309                             xmlFreeDoc(doc);
 310                         }
 311                     }
 312 #endif
 313                     if (r > 0 && cfile)
 314                     {
 315                         char *p = buf;
 316                         int i;
 317                         if (count)
 318                             fprintf (cfile, ",");
 319                         fprintf (cfile, "\n");
 320                         for (i = 0; i < r; i++)
 321                         {
 322                             if ((i & 15) == 0)
 323                                 fprintf (cfile, "  \"");
 324                             fprintf (cfile, "\\x%02X", p[i] & 255);
 325
 326                             if (i < r - 1 && (i & 15) == 15)
 327                                 fprintf (cfile, "\"\n");
 328
 329                         }
 330                         fprintf (cfile, "\"\n");
 331                     }
 332                     num++;
 333                     if (verbose)
 334                         printf("\n");
 335                 }
 336                 count++;
 337                 if (cd)
 338                     yaz_iconv_close(cd);
 339                 yaz_marc_destroy(mt);
 340             }
 341             if (cfile)
 342                 fprintf (cfile, "};\n");
 343             fclose(inf);
 344             break;
 345         case 'v':
 346             verbose++;
 347             break;
 348         default:
 349             usage(prog);
 350             exit (1);
 351         }
 352     }
 353     if (cfile)
 354         fclose (cfile);
 355     if (!no)
 356     {
 357         usage(prog);
 358         exit (1);
 359     }
 360     exit (0);
 361 }
 362 /*
 363  * Local variables:
 364  * c-basic-offset: 4
 365  * indent-tabs-mode: nil
 366  * End:
 367  * vim: shiftwidth=4 tabstop=8 expandtab
 368  */
 369