X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=util%2Fmarcdump.c;h=f5087e46f9ff8523fda5923fc2c64493037c061c;hp=b28a8385d135fbb6257b37c8d493096b5576f88d;hb=7d280285b7dd17c9a71b8d458171f40f35dd6682;hpb=d9ee01635f03f9095a66f71b73580560d48798e8 diff --git a/util/marcdump.c b/util/marcdump.c index b28a838..f5087e4 100644 --- a/util/marcdump.c +++ b/util/marcdump.c @@ -1,43 +1,40 @@ /* - * Copyright (c) 1995-1997, Index Data + * Copyright (C) 1995-2006, Index Data ApS * See the file LICENSE for details. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: marcdump.c,v $ - * Revision 1.9 1999-11-30 13:47:12 adam - * Improved installation. Moved header files to include/yaz. - * - * Revision 1.8 1999/05/26 07:49:35 adam - * C++ compilation. - * - * Revision 1.7 1998/02/11 11:53:36 adam - * Changed code so that it compiles as C++. - * - * Revision 1.6 1997/12/12 06:32:33 adam - * Added include of string.h. - * - * Revision 1.5 1997/09/24 13:29:40 adam - * Added verbose option -v to marcdump utility. - * - * Revision 1.4 1995/11/01 13:55:05 quinn - * Minor adjustments - * - * Revision 1.3 1995/05/16 08:51:12 quinn - * License, documentation, and memory fixes - * - * Revision 1.2 1995/05/15 11:56:56 quinn - * Debuggng & adjustments. - * - * Revision 1.1 1995/04/10 10:28:47 quinn - * Added copy of CCL and MARC display * + * $Id: marcdump.c,v 1.39 2006-06-12 16:12:58 mike Exp $ */ +#define _FILE_OFFSET_BITS 64 + +#if HAVE_CONFIG_H +#include +#endif + +#if HAVE_XML2 +#include +#include + +#include +#include + +#endif + #include #include #include #include +#include + +#if HAVE_LOCALE_H +#include +#endif +#if HAVE_LANGINFO_H +#include +#endif + #include +#include #include #include @@ -47,78 +44,301 @@ #ifndef SEEK_END #define SEEK_END 2 #endif - + + +static char *prog; + +static void usage(const char *prog) +{ + fprintf (stderr, "Usage: %s [-c cfile] [-f from] [-t to] [-x] [-X] [-e] [-I] [-v] [-s splitfname] file...\n", + prog); +} + +#if HAVE_XML2 +static void marcdump_read_xml(yaz_marc_t mt, const char *fname) +{ + xmlNodePtr ptr; + xmlDocPtr doc = xmlParseFile(fname); + if (!doc) + return; + + ptr = xmlDocGetRootElement(doc); + if (ptr) + { + int r; + WRBUF wrbuf = wrbuf_alloc(); + r = yaz_marc_read_xml(mt, ptr); + if (r) + fprintf(stderr, "yaz_marc_read_xml failed\n"); + + yaz_marc_write_mode(mt, wrbuf); + + fputs(wrbuf_buf(wrbuf), stdout); + + wrbuf_free(wrbuf, 1); + } + xmlFreeDoc(doc); +} +#endif + +static void dump(const char *fname, const char *from, const char *to, + int read_xml, int xml, + int print_offset, const char *split_fname, int verbose, + FILE *cfile) +{ + yaz_marc_t mt = yaz_marc_create(); + yaz_iconv_t cd = 0; + + if (from && to) + { + cd = yaz_iconv_open(to, from); + if (!cd) + { + fprintf(stderr, "conversion from %s to %s " + "unsupported\n", from, to); + exit(2); + } + yaz_marc_iconv(mt, cd); + } + yaz_marc_xml(mt, xml); + yaz_marc_debug(mt, verbose); + + if (read_xml) + { +#if HAVE_XML2 + marcdump_read_xml(mt, fname); +#else + return; +#endif + } + else + { + FILE *inf = fopen(fname, "rb"); + int count = 0; + int num = 1; + if (!inf) + { + fprintf (stderr, "%s: cannot open %s:%s\n", + prog, fname, strerror (errno)); + exit(1); + } + if (cfile) + fprintf (cfile, "char *marc_records[] = {\n"); + if (1) + { + int marc_no = 0; + for(;; marc_no++) + { + size_t len; + char *result = 0; + size_t rlen; + size_t r; + char buf[100001]; + + r = fread (buf, 1, 5, inf); + if (r < 5) + { + if (r && print_offset && verbose) + printf ("\n", + (long) r); + break; + } + while (*buf < '0' || *buf > '9') + { + int i; + long off = ftell(inf) - 5; + if (verbose || print_offset) + printf("\n", + *buf & 0xff, *buf & 0xff, + off, off); + for (i = 0; i<4; i++) + buf[i] = buf[i+1]; + r = fread(buf+4, 1, 1, inf); + if (r < 1) + break; + } + if (r < 1) + { + if (verbose || print_offset) + printf ("\n"); + break; + } + if (print_offset) + { + long off = ftell(inf) - 5; + printf ("\n", + num, off, off); + } + len = atoi_n(buf, 5); + if (len < 25 || len > 100000) + { + long off = ftell(inf) - 5; + printf("Bad Length %d read at offset %ld (%lx)\n", + len, (long) off, (long) off); + break; + } + rlen = len - 5; + r = fread (buf + 5, 1, rlen, inf); + if (r < rlen) + break; + if (split_fname) + { + char fname[256]; + FILE *sf; + sprintf(fname, "%.200s%07d", split_fname, marc_no); + sf = fopen(fname, "wb"); + if (!sf) + { + fprintf(stderr, "Could not open %s\n", fname); + split_fname = 0; + } + else + { + if (fwrite(buf, 1, len, sf) != len) + { + fprintf(stderr, "Could write content to %s\n", + fname); + split_fname = 0; + } + fclose(sf); + } + } + { int rlentmp = (int) rlen; + r = yaz_marc_decode_buf (mt, buf, -1, &result, &rlentmp); + rlen = (size_t) rlentmp; } + if (r > 0 && result) + { + fwrite (result, rlen, 1, stdout); + } + if (r > 0 && cfile) + { + char *p = buf; + size_t i; + if (count) + fprintf (cfile, ","); + fprintf (cfile, "\n"); + for (i = 0; i < r; i++) + { + if ((i & 15) == 0) + fprintf (cfile, " \""); + fprintf (cfile, "\\x%02X", p[i] & 255); + + if (i < r - 1 && (i & 15) == 15) + fprintf (cfile, "\"\n"); + + } + fprintf (cfile, "\"\n"); + } + num++; + if (verbose) + printf("\n"); + } + count++; + } + if (cfile) + fprintf (cfile, "};\n"); + fclose(inf); + } + if (cd) + yaz_iconv_close(cd); + yaz_marc_destroy(mt); +} + int main (int argc, char **argv) { - int ret; + int r; + int print_offset = 0; char *arg; int verbose = 0; - FILE *inf; - long file_size; - char *buf; - char *prog = *argv; - int count = 0; int no = 0; + int xml = 0; + FILE *cfile = 0; + char *from = 0, *to = 0; + int read_xml = 0; + const char *split_fname = 0; + +#if HAVE_LOCALE_H + setlocale(LC_CTYPE, ""); +#endif +#if HAVE_LANGINFO_H +#ifdef CODESET + to = nl_langinfo(CODESET); +#endif +#endif - while ((ret = options("v", argv, argc, &arg)) != -2) + prog = *argv; + while ((r = options("pvc:xOeXIf:t:s:", argv, argc, &arg)) != -2) { - no++; - switch (ret) + no++; + switch (r) { + case 'f': + from = arg; + break; + case 't': + to = arg; + break; + case 'c': + if (cfile) + fclose (cfile); + cfile = fopen(arg, "w"); + break; + case 'x': +#if HAVE_XML2 + read_xml = 1; +#else + fprintf(stderr, "%s: -x not supported." + " YAZ not compiled with Libxml2 support\n", prog); + exit(3); +#endif + break; + case 'O': + fprintf(stderr, "%s: OAI MARC no longer supported." + " Use MARCXML instead.\n", prog); + exit(1); + break; + case 'e': + xml = YAZ_MARC_XCHANGE; + break; + case 'X': + xml = YAZ_MARC_MARCXML; + break; + case 'I': + xml = YAZ_MARC_ISO2709; + break; + case 'p': + print_offset = 1; + break; + case 's': + split_fname = arg; + break; case 0: - inf = fopen (arg, "r"); - if (!inf) - { - fprintf (stderr, "%s: cannot open %s:%s\n", - prog, arg, strerror (errno)); - exit (1); - } - if (fseek (inf, 0L, SEEK_END)) - { - fprintf (stderr, "%s: cannot seek in %s:%s\n", - prog, arg, strerror (errno)); - exit (1); - } - file_size = ftell (inf); - if (fseek (inf, 0L, SEEK_SET)) - { - fprintf (stderr, "%s: cannot seek in %s:%s\n", - prog, arg, strerror (errno)); - exit (1); - } - buf = (char *)xmalloc (file_size); - if (!buf) - { - fprintf (stderr, "%s: cannot xmalloc: %s\n", - prog, strerror (errno)); - exit (1); - } - if ((long) fread (buf, 1, file_size, inf) != file_size) - { - fprintf (stderr, "%s: cannot read %s: %s\n", - prog, arg, strerror (errno)); - exit (1); - } - while ((ret = marc_display_ex (buf, stdout, verbose)) > 0) - { - buf += ret; - count++; - } - fclose (inf); - xfree (buf); + dump(arg, from, to, read_xml, xml, + print_offset, split_fname, verbose, cfile); break; case 'v': - verbose++; + verbose++; break; default: - fprintf (stderr, "Usage: %s [-v] file...\n", prog); + usage(prog); exit (1); } } + if (cfile) + fclose (cfile); if (!no) { - fprintf (stderr, "Usage: %s [-v] file...\n", prog); - exit (1); + usage(prog); + exit (1); } exit (0); } +/* + * Local variables: + * c-basic-offset: 4 + * indent-tabs-mode: nil + * End: + * vim: shiftwidth=4 tabstop=8 expandtab + */ +