util/marcdump.c

   1 /* This file is part of the YAZ toolkit.
   2  * Copyright (C) 1995-2012 Index Data
   3  * See the file LICENSE for details.
   4  */
   5
   6 #define _FILE_OFFSET_BITS 64
   7
   8 #if HAVE_CONFIG_H
   9 #include <config.h>
  10 #endif
  11
  12 #if YAZ_HAVE_XML2
  13 #include <libxml/parser.h>
  14 #include <libxml/tree.h>
  15 #include <libxml/xpath.h>
  16 #include <libxml/xpathInternals.h>
  17
  18 /* Libxml2 version < 2.6.15. xmlreader not reliable/present */
  19 #if LIBXML_VERSION < 20615
  20 #define USE_XMLREADER 0
  21 #else
  22 #define USE_XMLREADER 1
  23 #endif
  24
  25 #if USE_XMLREADER
  26 #include <libxml/xmlreader.h>
  27 #endif
  28
  29 #endif
  30
  31 #include <stdio.h>
  32 #include <stdlib.h>
  33 #include <string.h>
  34 #include <errno.h>
  35 #include <assert.h>
  36
  37 #if HAVE_LOCALE_H
  38 #include <locale.h>
  39 #endif
  40 #if HAVE_LANGINFO_H
  41 #include <langinfo.h>
  42 #endif
  43
  44 #include <yaz/marcdisp.h>
  45 #include <yaz/yaz-util.h>
  46 #include <yaz/xmalloc.h>
  47 #include <yaz/options.h>
  48
  49 #ifndef SEEK_SET
  50 #define SEEK_SET 0
  51 #endif
  52 #ifndef SEEK_END
  53 #define SEEK_END 2
  54 #endif
  55
  56
  57 static char *prog;
  58
  59 static void usage(const char *prog)
  60 {
  61     fprintf(stderr, "Usage: %s [-i format] [-o format] [-f from] [-t to] "
  62             "[-l pos=value] [-c cfile] [-s prefix] [-C size] [-n] "
  63             "[-p] [-v] [-V] file...\n",
  64             prog);
  65 }
  66
  67 static void show_version(void)
  68 {
  69     char vstr[20], sha1_str[41];
  70
  71     yaz_version(vstr, sha1_str);
  72     printf("YAZ version: %s %s\n", YAZ_VERSION, YAZ_VERSION_SHA1);
  73     if (strcmp(sha1_str, YAZ_VERSION_SHA1))
  74         printf("YAZ DLL/SO: %s %s\n", vstr, sha1_str);
  75     exit(0);
  76 }
  77
  78 static int getbyte_stream(void *client_data)
  79 {
  80     FILE *f = (FILE*) client_data;
  81
  82     int c = fgetc(f);
  83     if (c == EOF)
  84         return 0;
  85     return c;
  86 }
  87
  88 static void ungetbyte_stream(int c, void *client_data)
  89 {
  90     FILE *f = (FILE*) client_data;
  91
  92     if (c == 0)
  93         c = EOF;
  94     ungetc(c, f);
  95 }
  96
  97 static void marcdump_read_line(yaz_marc_t mt, const char *fname)
  98 {
  99     FILE *inf = fopen(fname, "rb");
 100     if (!inf)
 101     {
 102         fprintf(stderr, "%s: cannot open %s:%s\n",
 103                 prog, fname, strerror(errno));
 104         exit(1);
 105     }
 106
 107     while (yaz_marc_read_line(mt, getbyte_stream,
 108                               ungetbyte_stream, inf) == 0)
 109     {
 110         WRBUF wrbuf = wrbuf_alloc();
 111         yaz_marc_write_mode(mt, wrbuf);
 112         fputs(wrbuf_cstr(wrbuf), stdout);
 113         wrbuf_destroy(wrbuf);
 114     }
 115     fclose(inf);
 116 }
 117
 118 #if YAZ_HAVE_XML2
 119 static void marcdump_read_xml(yaz_marc_t mt, const char *fname)
 120 {
 121     WRBUF wrbuf = wrbuf_alloc();
 122 #if USE_XMLREADER
 123     xmlTextReaderPtr reader = xmlReaderForFile(fname, 0 /* encoding */,
 124                                                0 /* options */);
 125
 126     if (reader)
 127     {
 128         int ret;
 129         while ((ret = xmlTextReaderRead(reader)) == 1)
 130         {
 131             int type = xmlTextReaderNodeType(reader);
 132             if (type == XML_READER_TYPE_ELEMENT)
 133             {
 134                 const char *name = (const char *)
 135                     xmlTextReaderLocalName(reader);
 136                 if (!strcmp(name, "record") || !strcmp(name, "r"))
 137                 {
 138                     xmlNodePtr ptr = xmlTextReaderExpand(reader);
 139
 140                     int r = yaz_marc_read_xml(mt, ptr);
 141                     if (r)
 142                         fprintf(stderr, "yaz_marc_read_xml failed\n");
 143                     else
 144                     {
 145                         int write_rc = yaz_marc_write_mode(mt, wrbuf);
 146                         if (write_rc)
 147                             yaz_log(YLOG_WARN, "yaz_marc_write_mode: write error: %d", write_rc);
 148
 149                         fputs(wrbuf_cstr(wrbuf), stdout);
 150                         wrbuf_rewind(wrbuf);
 151                     }
 152                 }
 153             }
 154         }
 155     }
 156 #else
 157     xmlDocPtr doc = xmlParseFile(fname);
 158     if (doc)
 159     {
 160         xmlNodePtr ptr = xmlDocGetRootElement(doc);
 161         for (; ptr; ptr = ptr->next)
 162         {
 163             if (ptr->type == XML_ELEMENT_NODE)
 164             {
 165                 if (!strcmp((const char *) ptr->name, "collection"))
 166                 {
 167                     ptr = ptr->children;
 168                     continue;
 169                 }
 170                 if (!strcmp((const char *) ptr->name, "record") ||
 171                     !strcmp((const char *) ptr->name, "r"))
 172                 {
 173                     int r = yaz_marc_read_xml(mt, ptr);
 174                     if (r)
 175                         fprintf(stderr, "yaz_marc_read_xml failed\n");
 176                     else
 177                     {
 178                         yaz_marc_write_mode(mt, wrbuf);
 179
 180                         fputs(wrbuf_cstr(wrbuf), stdout);
 181                         wrbuf_rewind(wrbuf);
 182                     }
 183                 }
 184             }
 185         }
 186         xmlFreeDoc(doc);
 187     }
 188 #endif
 189     fputs(wrbuf_cstr(wrbuf), stdout);
 190     wrbuf_destroy(wrbuf);
 191 }
 192 #endif
 193
 194 static void dump(const char *fname, const char *from, const char *to,
 195                  int input_format, int output_format,
 196                  int write_using_libxml2,
 197                  int print_offset, const char *split_fname, int split_chunk,
 198                  int verbose, FILE *cfile, const char *leader_spec)
 199 {
 200     yaz_marc_t mt = yaz_marc_create();
 201     yaz_iconv_t cd = 0;
 202
 203     if (yaz_marc_leader_spec(mt, leader_spec))
 204     {
 205         fprintf(stderr, "bad leader spec: %s\n", leader_spec);
 206         yaz_marc_destroy(mt);
 207         exit(2);
 208     }
 209     if (from && to)
 210     {
 211         cd = yaz_iconv_open(to, from);
 212         if (!cd)
 213         {
 214             fprintf(stderr, "conversion from %s to %s "
 215                     "unsupported\n", from, to);
 216             yaz_marc_destroy(mt);
 217             exit(2);
 218         }
 219         yaz_marc_iconv(mt, cd);
 220     }
 221     yaz_marc_enable_collection(mt);
 222     yaz_marc_xml(mt, output_format);
 223     yaz_marc_write_using_libxml2(mt, write_using_libxml2);
 224     yaz_marc_debug(mt, verbose);
 225
 226     if (input_format == YAZ_MARC_MARCXML || input_format == YAZ_MARC_TURBOMARC || input_format == YAZ_MARC_XCHANGE)
 227     {
 228 #if YAZ_HAVE_XML2
 229         marcdump_read_xml(mt, fname);
 230 #endif
 231     }
 232     else if (input_format == YAZ_MARC_LINE)
 233     {
 234         marcdump_read_line(mt, fname);
 235     }
 236     else if (input_format == YAZ_MARC_ISO2709)
 237     {
 238         FILE *inf = fopen(fname, "rb");
 239         int num = 1;
 240         int marc_no = 0;
 241         int split_file_no = -1;
 242         if (!inf)
 243         {
 244             fprintf(stderr, "%s: cannot open %s:%s\n",
 245                     prog, fname, strerror(errno));
 246             exit(1);
 247         }
 248         if (cfile)
 249             fprintf(cfile, "char *marc_records[] = {\n");
 250         for(;; marc_no++)
 251         {
 252             const char *result = 0;
 253             size_t len;
 254             size_t rlen;
 255             size_t len_result;
 256             size_t r;
 257             char buf[100001];
 258
 259             r = fread(buf, 1, 5, inf);
 260             if (r < 5)
 261             {
 262                 if (r && print_offset && verbose)
 263                     printf("<!-- Extra %ld bytes at end of file -->\n",
 264                            (long) r);
 265                 break;
 266             }
 267             while (*buf < '0' || *buf > '9')
 268             {
 269                 int i;
 270                 long off = ftell(inf) - 5;
 271                 if (verbose || print_offset)
 272                     printf("<!-- Skipping bad byte %d (0x%02X) at offset "
 273                            "%ld (0x%lx) -->\n",
 274                            *buf & 0xff, *buf & 0xff,
 275                            off, off);
 276                 for (i = 0; i<4; i++)
 277                     buf[i] = buf[i+1];
 278                 r = fread(buf+4, 1, 1, inf);
 279                 if (r < 1)
 280                     break;
 281             }
 282             if (r < 1)
 283             {
 284                 if (verbose || print_offset)
 285                     printf("<!-- End of file with data -->\n");
 286                 break;
 287             }
 288             if (print_offset)
 289             {
 290                 long off = ftell(inf) - 5;
 291                 printf("<!-- Record %d offset %ld (0x%lx) -->\n",
 292                        num, off, off);
 293             }
 294             len = atoi_n(buf, 5);
 295             if (len < 25 || len > 100000)
 296             {
 297                 long off = ftell(inf) - 5;
 298                 printf("Bad Length %ld read at offset %ld (%lx)\n",
 299                        (long)len, (long) off, (long) off);
 300                 break;
 301             }
 302             rlen = len - 5;
 303             r = fread(buf + 5, 1, rlen, inf);
 304             if (r < rlen)
 305                 break;
 306             while (buf[len-1] != ISO2709_RS)
 307             {
 308                 if (len > sizeof(buf)-2)
 309                     break;
 310                 r = fread(buf + len, 1, 1, inf);
 311                 if (r != 1)
 312                     break;
 313                 len++;
 314             }
 315             if (split_fname)
 316             {
 317                 char fname[256];
 318                 const char *mode = 0;
 319                 FILE *sf;
 320                 if ((marc_no % split_chunk) == 0)
 321                 {
 322                     mode = "wb";
 323                     split_file_no++;
 324                 }
 325                 else
 326                     mode = "ab";
 327                 sprintf(fname, "%.200s%07d", split_fname, split_file_no);
 328                 sf = fopen(fname, mode);
 329                 if (!sf)
 330                 {
 331                     fprintf(stderr, "Could not open %s\n", fname);
 332                     split_fname = 0;
 333                 }
 334                 else
 335                 {
 336                     if (fwrite(buf, 1, len, sf) != len)
 337                     {
 338                         fprintf(stderr, "Could write content to %s\n",
 339                                 fname);
 340                         split_fname = 0;
 341                     }
 342                     fclose(sf);
 343                 }
 344             }
 345             len_result = rlen;
 346             r = yaz_marc_decode_buf(mt, buf, -1, &result, &len_result);
 347             if (r > 0 && result && len_result)
 348             {
 349                 if (fwrite(result, len_result, 1, stdout) != 1)
 350                 {
 351                     fprintf(stderr, "Write to stdout failed\n");
 352                     break;
 353                 }
 354             }
 355             if (r > 0 && cfile)
 356             {
 357                 char *p = buf;
 358                 size_t i;
 359                 if (marc_no)
 360                     fprintf(cfile, ",");
 361                 fprintf(cfile, "\n");
 362                 for (i = 0; i < r; i++)
 363                 {
 364                     if ((i & 15) == 0)
 365                         fprintf(cfile, "  \"");
 366                     fprintf(cfile, "\\x%02X", p[i] & 255);
 367
 368                     if (i < r - 1 && (i & 15) == 15)
 369                         fprintf(cfile, "\"\n");
 370
 371                 }
 372                 fprintf(cfile, "\"\n");
 373             }
 374             num++;
 375             if (verbose)
 376                 printf("\n");
 377         }
 378         if (cfile)
 379             fprintf(cfile, "};\n");
 380         fclose(inf);
 381     }
 382     {
 383         WRBUF wrbuf = wrbuf_alloc();
 384         yaz_marc_write_trailer(mt, wrbuf);
 385         fputs(wrbuf_cstr(wrbuf), stdout);
 386         wrbuf_destroy(wrbuf);
 387     }
 388     if (cd)
 389         yaz_iconv_close(cd);
 390     yaz_marc_destroy(mt);
 391 }
 392
 393 int main (int argc, char **argv)
 394 {
 395     int r;
 396     int print_offset = 0;
 397     char *arg;
 398     int verbose = 0;
 399     int no = 0;
 400     int output_format = YAZ_MARC_LINE;
 401     FILE *cfile = 0;
 402     char *from = 0, *to = 0;
 403     int input_format = YAZ_MARC_ISO2709;
 404     int split_chunk = 1;
 405     const char *split_fname = 0;
 406     const char *leader_spec = 0;
 407     int write_using_libxml2 = 0;
 408
 409 #if HAVE_LOCALE_H
 410     setlocale(LC_CTYPE, "");
 411 #endif
 412 #if HAVE_LANGINFO_H
 413 #ifdef CODESET
 414     to = nl_langinfo(CODESET);
 415 #endif
 416 #endif
 417
 418     prog = *argv;
 419     while ((r = options("i:o:C:npc:xOeXIf:t:s:l:Vv", argv, argc, &arg)) != -2)
 420     {
 421         no++;
 422         switch (r)
 423         {
 424         case 'i':
 425             input_format = yaz_marc_decode_formatstr(arg);
 426             if (input_format == -1)
 427             {
 428                 fprintf(stderr, "%s: bad input format: %s\n", prog, arg);
 429                 exit(1);
 430             }
 431 #if YAZ_HAVE_XML2
 432 #else
 433             if (input_format == YAZ_MARC_MARCXML
 434                 || input_format == YAZ_MARC_XCHANGE)
 435             {
 436                 fprintf(stderr, "%s: Libxml2 support not enabled\n", prog);
 437                 exit(3);
 438             }
 439 #endif
 440             break;
 441         case 'o':
 442             /* dirty hack so we can make Libxml2 do the writing ..
 443                rather than WRBUF */
 444             if (strlen(arg) > 4 && strncmp(arg, "xml,", 4) == 0)
 445             {
 446                 /* Only supported for Libxml2 2.6.0 or later */
 447 #if LIBXML_VERSION >= 20600
 448                 arg = arg + 4;
 449                 write_using_libxml2 = 1;
 450 #else
 451                 fprintf(stderr, "%s: output using Libxml2 unsupported\n", prog);
 452                 exit(4);
 453 #endif
 454             }
 455             output_format = yaz_marc_decode_formatstr(arg);
 456             if (output_format == -1)
 457             {
 458                 fprintf(stderr, "%s: bad output format: %s\n", prog, arg);
 459                 exit(1);
 460             }
 461             break;
 462         case 'l':
 463             leader_spec = arg;
 464             break;
 465         case 'f':
 466             from = arg;
 467             break;
 468         case 't':
 469             to = arg;
 470             break;
 471         case 'c':
 472             if (cfile)
 473                 fclose(cfile);
 474             cfile = fopen(arg, "w");
 475             break;
 476         case 'x':
 477             fprintf(stderr, "%s: -x no longer supported. "
 478                     "Use -i marcxml instead\n", prog);
 479             exit(1);
 480             break;
 481         case 'O':
 482             fprintf(stderr, "%s: OAI MARC no longer supported."
 483                     " Use MARCXML instead.\n", prog);
 484             exit(1);
 485             break;
 486         case 'e':
 487             fprintf(stderr, "%s: -e no longer supported. "
 488                     "Use -o marcxchange instead\n", prog);
 489             exit(1);
 490             break;
 491         case 'X':
 492             fprintf(stderr, "%s: -X no longer supported. "
 493                     "Use -o marcxml instead\n", prog);
 494             exit(1);
 495             break;
 496         case 'I':
 497             fprintf(stderr, "%s: -I no longer supported. "
 498                     "Use -o marc instead\n", prog);
 499             exit(1);
 500             break;
 501         case 'n':
 502             output_format = YAZ_MARC_CHECK;
 503             break;
 504         case 'p':
 505             print_offset = 1;
 506             break;
 507         case 's':
 508             split_fname = arg;
 509             break;
 510         case 'C':
 511             split_chunk = atoi(arg);
 512             break;
 513         case 0:
 514             dump(arg, from, to, input_format, output_format,
 515                  write_using_libxml2,
 516                  print_offset, split_fname, split_chunk,
 517                  verbose, cfile, leader_spec);
 518             break;
 519         case 'v':
 520             verbose++;
 521             break;
 522         case 'V':
 523             show_version();
 524             break;
 525         default:
 526             usage(prog);
 527             exit(1);
 528         }
 529     }
 530     if (cfile)
 531         fclose(cfile);
 532     if (!no)
 533     {
 534         usage(prog);
 535         exit(1);
 536     }
 537     exit(0);
 538 }
 539 /*
 540  * Local variables:
 541  * c-basic-offset: 4
 542  * c-file-style: "Stroustrup"
 543  * indent-tabs-mode: nil
 544  * End:
 545  * vim: shiftwidth=4 tabstop=8 expandtab
 546  */
 547