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 1.29 2005-04-20 13:04:04 adam Exp $
   6  */
   7
   8 #if HAVE_CONFIG_H
   9 #include <config.h>
  10 #endif
  11
  12 #if HAVE_XML2
  13 #include <libxml/parser.h>
  14 #include <libxml/tree.h>
  15
  16 #include <libxml/xpath.h>
  17 #include <libxml/xpathInternals.h>
  18
  19 #endif
  20
  21 #include <stdio.h>
  22 #include <stdlib.h>
  23 #include <string.h>
  24 #include <errno.h>
  25 #include <assert.h>
  26
  27 #if HAVE_LOCALE_H
  28 #include <locale.h>
  29 #endif
  30 #if HAVE_LANGINFO_H
  31 #include <langinfo.h>
  32 #endif
  33
  34 #include <yaz/marcdisp.h>
  35 #include <yaz/yaz-util.h>
  36 #include <yaz/xmalloc.h>
  37 #include <yaz/options.h>
  38
  39 #ifndef SEEK_SET
  40 #define SEEK_SET 0
  41 #endif
  42 #ifndef SEEK_END
  43 #define SEEK_END 2
  44 #endif
  45
  46 static void usage(const char *prog)
  47 {
  48     fprintf (stderr, "Usage: %s [-c cfile] [-f from] [-t to] [-x] [-O] [-X] [-e] [-I] [-v] file...\n",
  49              prog);
  50 }
  51
  52 #if HAVE_XML2
  53 void print_xpath_nodes(xmlNodeSetPtr nodes, FILE* output) {
  54     xmlNodePtr cur;
  55     int size;
  56     int i;
  57
  58     assert(output);
  59     size = (nodes) ? nodes->nodeNr : 0;
  60
  61     fprintf(output, "Result (%d nodes):\n", size);
  62     for(i = 0; i < size; ++i) {
  63         assert(nodes->nodeTab[i]);
  64
  65         if(nodes->nodeTab[i]->type == XML_NAMESPACE_DECL)
  66         {
  67             xmlNsPtr ns;
  68
  69             ns = (xmlNsPtr)nodes->nodeTab[i];
  70             cur = (xmlNodePtr)ns->next;
  71             if(cur->ns) {
  72                 fprintf(output, "= namespace \"%s\"=\"%s\" for node %s:%s\n",
  73                     ns->prefix, ns->href, cur->ns->href, cur->name);
  74             } else {
  75                 fprintf(output, "= namespace \"%s\"=\"%s\" for node %s\n",
  76                     ns->prefix, ns->href, cur->name);
  77             }
  78         }
  79         else if(nodes->nodeTab[i]->type == XML_ELEMENT_NODE)
  80         {
  81             cur = nodes->nodeTab[i];
  82             if(cur->ns) {
  83                 fprintf(output, "= element node \"%s:%s\"\n",
  84                     cur->ns->href, cur->name);
  85             }
  86             else
  87             {
  88                 fprintf(output, "= element node \"%s\"\n",
  89                     cur->name);
  90             }
  91         }
  92         else
  93         {
  94             cur = nodes->nodeTab[i];
  95             fprintf(output, "= node \"%s\": type %d\n", cur->name, cur->type);
  96         }
  97     }
  98 }
  99 #endif
 100
 101 int main (int argc, char **argv)
 102 {
 103     int r;
 104     int libxml_dom_test = 0;
 105     int print_offset = 0;
 106     char *arg;
 107     int verbose = 0;
 108     FILE *inf;
 109     char buf[100001];
 110     char *prog = *argv;
 111     int no = 0;
 112     int xml = 0;
 113     FILE *cfile = 0;
 114     char *from = 0, *to = 0;
 115     int num = 1;
 116
 117 #if HAVE_LOCALE_H
 118     setlocale(LC_CTYPE, "");
 119 #endif
 120 #if HAVE_LANGINFO_H
 121 #ifdef CODESET
 122     to = nl_langinfo(CODESET);
 123 #endif
 124 #endif
 125
 126     while ((r = options("pvc:xOeXIf:t:2", argv, argc, &arg)) != -2)
 127     {
 128         int count;
 129         no++;
 130         switch (r)
 131         {
 132         case 'f':
 133             from = arg;
 134             break;
 135         case 't':
 136             to = arg;
 137             break;
 138         case 'c':
 139             if (cfile)
 140                 fclose (cfile);
 141             cfile = fopen (arg, "w");
 142             break;
 143         case 'x':
 144             xml = YAZ_MARC_SIMPLEXML;
 145             break;
 146         case 'O':
 147             xml = YAZ_MARC_OAIMARC;
 148             break;
 149         case 'e':
 150             xml = YAZ_MARC_XCHANGE;
 151             break;
 152         case 'X':
 153             xml = YAZ_MARC_MARCXML;
 154             break;
 155         case 'I':
 156             xml = YAZ_MARC_ISO2709;
 157             break;
 158         case 'p':
 159             print_offset = 1;
 160             break;
 161         case '2':
 162             libxml_dom_test = 1;
 163             break;
 164         case 0:
 165             inf = fopen (arg, "rb");
 166             count = 0;
 167             if (!inf)
 168             {
 169                 fprintf (stderr, "%s: cannot open %s:%s\n",
 170                          prog, arg, strerror (errno));
 171                 exit(1);
 172             }
 173             if (cfile)
 174                 fprintf (cfile, "char *marc_records[] = {\n");
 175             if (1)
 176             {
 177                 yaz_marc_t mt = yaz_marc_create();
 178                 yaz_iconv_t cd = 0;
 179
 180                 if (from && to)
 181                 {
 182                     cd = yaz_iconv_open(to, from);
 183                     if (!cd)
 184                     {
 185                         fprintf(stderr, "conversion from %s to %s "
 186                                 "unsupported\n", from, to);
 187                         exit(2);
 188                     }
 189                     yaz_marc_iconv(mt, cd);
 190                 }
 191                 yaz_marc_xml(mt, xml);
 192                 yaz_marc_debug(mt, verbose);
 193                 while (1)
 194                 {
 195                     int len;
 196                     char *result = 0;
 197                     int rlen;
 198
 199                     r = fread (buf, 1, 5, inf);
 200                     if (r < 5)
 201                     {
 202                         if (r && print_offset && verbose)
 203                             printf ("Extra %d bytes at end of file", r);
 204                         break;
 205                     }
 206                     while (*buf < '0' || *buf > '9')
 207                     {
 208                         int i;
 209                         long off = ftell(inf) - 5;
 210                         if (verbose || print_offset)
 211                             printf("Skipping bad byte %d (0x%02X) at offset "
 212                                    "%ld (0x%lx)\n",
 213                                    *buf & 0xff, *buf & 0xff,
 214                                    off, off);
 215                         for (i = 0; i<4; i++)
 216                             buf[i] = buf[i+1];
 217                         r = fread(buf+4, 1, 1, inf);
 218                         if (r < 1)
 219                             break;
 220                     }
 221                     if (r < 1)
 222                     {
 223                         if (verbose || print_offset)
 224                             printf ("End of file with extra garbage\n");
 225                         break;
 226                     }
 227                     if (print_offset)
 228                     {
 229                         long off = ftell(inf) - 5;
 230                         printf ("Record %d offset %ld (0x%lx)\n", num,
 231                                 off, off);
 232                     }
 233                     len = atoi_n(buf, 5);
 234                     if (len < 25 || len > 100000)
 235                     {
 236                         long off = ftell(inf) - 5;
 237                         printf("Bad Length %d read at offset %ld (%lx)\n",
 238                                len, (long) off, (long) off);
 239                         break;
 240                     }
 241                     len = len - 5;
 242                     r = fread (buf + 5, 1, len, inf);
 243                     if (r < len)
 244                         break;
 245                     r = yaz_marc_decode_buf (mt, buf, -1, &result, &rlen);
 246                     if (result)
 247                         fwrite (result, rlen, 1, stdout);
 248 #if HAVE_XML2
 249                     if (r > 0 && libxml_dom_test)
 250                     {
 251                         xmlDocPtr doc = xmlParseMemory(result, rlen);
 252                         if (!doc)
 253                             fprintf(stderr, "xmLParseMemory failed\n");
 254                         else
 255                         {
 256                             int i;
 257                             xmlXPathContextPtr xpathCtx;
 258                             xmlXPathObjectPtr xpathObj;
 259                             static const char *xpathExpr[] = {
 260                                 "/record/datafield[@tag='245']/subfield[@code='a']",
 261                                 "/record/datafield[@tag='100']/subfield",
 262                                 "/record/datafield[@tag='245']/subfield[@code='a']",
 263                                 "/record/datafield[@tag='650']/subfield",
 264                                 "/record/datafield[@tag='650']",
 265                                 0};
 266
 267                             xpathCtx = xmlXPathNewContext(doc);
 268
 269                             for (i = 0; xpathExpr[i]; i++) {
 270                                 xpathObj = xmlXPathEvalExpression(xpathExpr[i], xpathCtx);
 271                                 if(xpathObj == NULL) {
 272                                     fprintf(stderr,"Error: unable to evaluate xpath expression \"%s\"\n", xpathExpr[i]);
 273                                 }
 274                                 else
 275                                 {
 276                                     print_xpath_nodes(xpathObj->nodesetval, stdout);
 277                                     xmlXPathFreeObject(xpathObj);
 278                                 }
 279                             }
 280                             xmlXPathFreeContext(xpathCtx);
 281                             xmlFreeDoc(doc);
 282                         }
 283                     }
 284 #endif
 285                     if (r > 0 && cfile)
 286                     {
 287                         char *p = buf;
 288                         int i;
 289                         if (count)
 290                             fprintf (cfile, ",");
 291                         fprintf (cfile, "\n");
 292                         for (i = 0; i < r; i++)
 293                         {
 294                             if ((i & 15) == 0)
 295                                 fprintf (cfile, "  \"");
 296                             fprintf (cfile, "\\x%02X", p[i] & 255);
 297
 298                             if (i < r - 1 && (i & 15) == 15)
 299                                 fprintf (cfile, "\"\n");
 300
 301                         }
 302                         fprintf (cfile, "\"\n");
 303                     }
 304                     num++;
 305                 }
 306                 count++;
 307                 if (cd)
 308                     yaz_iconv_close(cd);
 309                 yaz_marc_destroy(mt);
 310             }
 311             if (cfile)
 312                 fprintf (cfile, "};\n");
 313             fclose(inf);
 314             break;
 315         case 'v':
 316             verbose++;
 317             break;
 318         default:
 319             usage(prog);
 320             exit (1);
 321         }
 322     }
 323     if (cfile)
 324         fclose (cfile);
 325     if (!no)
 326     {
 327         usage(prog);
 328         exit (1);
 329     }
 330     exit (0);
 331 }