-/*
- * Copyright (C) 1995-2007, Index Data ApS
+/* This file is part of the YAZ toolkit.
+ * Copyright (C) 1995-2012 Index Data
* See the file LICENSE for details.
- *
- * $Id: marcdump.c,v 1.51 2007-03-19 14:40:08 adam Exp $
*/
#define _FILE_OFFSET_BITS 64
#if YAZ_HAVE_XML2
#include <libxml/parser.h>
#include <libxml/tree.h>
-
#include <libxml/xpath.h>
#include <libxml/xpathInternals.h>
+/* Libxml2 version < 2.6.15. xmlreader not reliable/present */
+#if LIBXML_VERSION < 20615
+#define USE_XMLREADER 0
+#else
+#define USE_XMLREADER 1
+#endif
+
+#if USE_XMLREADER
+#include <libxml/xmlreader.h>
+#endif
+
#endif
#include <stdio.h>
static void usage(const char *prog)
{
- fprintf (stderr, "Usage: %s [-c cfile] [-f from] [-t to] "
- "[-i format] [-o format] "
- "[-n] [-l pos=value] [-v] [-C chunk] [-s splitfname] file...\n",
- prog);
-}
+ fprintf(stderr, "Usage: %s [-i format] [-o format] [-f from] [-t to] "
+ "[-l pos=value] [-c cfile] [-s prefix] [-C size] [-n] "
+ "[-p] [-v] [-V] file...\n",
+ prog);
+}
+
+static void show_version(void)
+{
+ char vstr[20], sha1_str[41];
+
+ yaz_version(vstr, sha1_str);
+ printf("YAZ version: %s %s\n", YAZ_VERSION, YAZ_VERSION_SHA1);
+ if (strcmp(sha1_str, YAZ_VERSION_SHA1))
+ printf("YAZ DLL/SO: %s %s\n", vstr, sha1_str);
+ exit(0);
+}
static int getbyte_stream(void *client_data)
{
FILE *inf = fopen(fname, "rb");
if (!inf)
{
- fprintf (stderr, "%s: cannot open %s:%s\n",
- prog, fname, strerror (errno));
+ fprintf(stderr, "%s: cannot open %s:%s\n",
+ prog, fname, strerror(errno));
exit(1);
}
-
+
while (yaz_marc_read_line(mt, getbyte_stream,
ungetbyte_stream, inf) == 0)
{
#if YAZ_HAVE_XML2
static void marcdump_read_xml(yaz_marc_t mt, const char *fname)
{
- xmlNodePtr ptr;
- xmlDocPtr doc = xmlParseFile(fname);
- if (!doc)
- return;
+ WRBUF wrbuf = wrbuf_alloc();
+#if USE_XMLREADER
+ xmlTextReaderPtr reader = xmlReaderForFile(fname, 0 /* encoding */,
+ 0 /* options */);
- ptr = xmlDocGetRootElement(doc);
- if (ptr)
+ if (reader)
{
- int r;
- WRBUF wrbuf = wrbuf_alloc();
- r = yaz_marc_read_xml(mt, ptr);
- if (r)
- fprintf(stderr, "yaz_marc_read_xml failed\n");
- else
+ int ret;
+ while ((ret = xmlTextReaderRead(reader)) == 1)
{
- yaz_marc_write_mode(mt, wrbuf);
-
- fputs(wrbuf_cstr(wrbuf), stdout);
+ int type = xmlTextReaderNodeType(reader);
+ if (type == XML_READER_TYPE_ELEMENT)
+ {
+ const char *name = (const char *)
+ xmlTextReaderLocalName(reader);
+ if (!strcmp(name, "record") || !strcmp(name, "r"))
+ {
+ xmlNodePtr ptr = xmlTextReaderExpand(reader);
+
+ int r = yaz_marc_read_xml(mt, ptr);
+ if (r)
+ fprintf(stderr, "yaz_marc_read_xml failed\n");
+ else
+ {
+ int write_rc = yaz_marc_write_mode(mt, wrbuf);
+ if (write_rc)
+ yaz_log(YLOG_WARN, "yaz_marc_write_mode: write error: %d", write_rc);
+
+ fputs(wrbuf_cstr(wrbuf), stdout);
+ wrbuf_rewind(wrbuf);
+ }
+ }
+ }
}
- wrbuf_destroy(wrbuf);
}
- xmlFreeDoc(doc);
+#else
+ xmlDocPtr doc = xmlParseFile(fname);
+ if (doc)
+ {
+ xmlNodePtr ptr = xmlDocGetRootElement(doc);
+ for (; ptr; ptr = ptr->next)
+ {
+ if (ptr->type == XML_ELEMENT_NODE)
+ {
+ if (!strcmp((const char *) ptr->name, "collection"))
+ {
+ ptr = ptr->children;
+ continue;
+ }
+ if (!strcmp((const char *) ptr->name, "record") ||
+ !strcmp((const char *) ptr->name, "r"))
+ {
+ int r = yaz_marc_read_xml(mt, ptr);
+ if (r)
+ fprintf(stderr, "yaz_marc_read_xml failed\n");
+ else
+ {
+ yaz_marc_write_mode(mt, wrbuf);
+
+ fputs(wrbuf_cstr(wrbuf), stdout);
+ wrbuf_rewind(wrbuf);
+ }
+ }
+ }
+ }
+ xmlFreeDoc(doc);
+ }
+#endif
+ fputs(wrbuf_cstr(wrbuf), stdout);
+ wrbuf_destroy(wrbuf);
}
#endif
}
yaz_marc_iconv(mt, cd);
}
+ yaz_marc_enable_collection(mt);
yaz_marc_xml(mt, output_format);
yaz_marc_write_using_libxml2(mt, write_using_libxml2);
yaz_marc_debug(mt, verbose);
- if (input_format == YAZ_MARC_MARCXML || input_format == YAZ_MARC_XCHANGE)
+ if (input_format == YAZ_MARC_MARCXML || input_format == YAZ_MARC_TURBOMARC || input_format == YAZ_MARC_XCHANGE)
{
#if YAZ_HAVE_XML2
marcdump_read_xml(mt, fname);
int split_file_no = -1;
if (!inf)
{
- fprintf (stderr, "%s: cannot open %s:%s\n",
- prog, fname, strerror (errno));
+ fprintf(stderr, "%s: cannot open %s:%s\n",
+ prog, fname, strerror(errno));
exit(1);
}
if (cfile)
- fprintf (cfile, "char *marc_records[] = {\n");
+ fprintf(cfile, "char *marc_records[] = {\n");
for(;; marc_no++)
{
const char *result = 0;
size_t len_result;
size_t r;
char buf[100001];
-
- r = fread (buf, 1, 5, inf);
+
+ r = fread(buf, 1, 5, inf);
if (r < 5)
{
if (r && print_offset && verbose)
- printf ("<!-- Extra %ld bytes at end of file -->\n",
- (long) r);
+ printf("<!-- Extra %ld bytes at end of file -->\n",
+ (long) r);
break;
}
while (*buf < '0' || *buf > '9')
long off = ftell(inf) - 5;
if (verbose || print_offset)
printf("<!-- Skipping bad byte %d (0x%02X) at offset "
- "%ld (0x%lx) -->\n",
+ "%ld (0x%lx) -->\n",
*buf & 0xff, *buf & 0xff,
off, off);
for (i = 0; i<4; i++)
if (r < 1)
{
if (verbose || print_offset)
- printf ("<!-- End of file with data -->\n");
+ printf("<!-- End of file with data -->\n");
break;
}
if (print_offset)
{
long off = ftell(inf) - 5;
- printf ("<!-- Record %d offset %ld (0x%lx) -->\n",
- num, off, off);
+ printf("<!-- Record %d offset %ld (0x%lx) -->\n",
+ num, off, off);
}
len = atoi_n(buf, 5);
if (len < 25 || len > 100000)
break;
}
rlen = len - 5;
- r = fread (buf + 5, 1, rlen, inf);
+ r = fread(buf + 5, 1, rlen, inf);
if (r < rlen)
break;
while (buf[len-1] != ISO2709_RS)
{
if (len > sizeof(buf)-2)
break;
- r = fread (buf + len, 1, 1, inf);
+ r = fread(buf + len, 1, 1, inf);
if (r != 1)
break;
len++;
}
len_result = rlen;
r = yaz_marc_decode_buf(mt, buf, -1, &result, &len_result);
- if (r > 0 && result)
+ if (r > 0 && result && len_result)
{
- fwrite (result, len_result, 1, stdout);
+ if (fwrite(result, len_result, 1, stdout) != 1)
+ {
+ fprintf(stderr, "Write to stdout failed\n");
+ break;
+ }
}
if (r > 0 && cfile)
{
char *p = buf;
size_t i;
if (marc_no)
- fprintf (cfile, ",");
- fprintf (cfile, "\n");
+ fprintf(cfile, ",");
+ fprintf(cfile, "\n");
for (i = 0; i < r; i++)
{
if ((i & 15) == 0)
- fprintf (cfile, " \"");
- fprintf (cfile, "\\x%02X", p[i] & 255);
-
+ fprintf(cfile, " \"");
+ fprintf(cfile, "\\x%02X", p[i] & 255);
+
if (i < r - 1 && (i & 15) == 15)
- fprintf (cfile, "\"\n");
-
+ fprintf(cfile, "\"\n");
+
}
- fprintf (cfile, "\"\n");
+ fprintf(cfile, "\"\n");
}
num++;
if (verbose)
printf("\n");
}
if (cfile)
- fprintf (cfile, "};\n");
+ fprintf(cfile, "};\n");
fclose(inf);
}
+ {
+ WRBUF wrbuf = wrbuf_alloc();
+ yaz_marc_write_trailer(mt, wrbuf);
+ fputs(wrbuf_cstr(wrbuf), stdout);
+ wrbuf_destroy(wrbuf);
+ }
if (cd)
yaz_iconv_close(cd);
yaz_marc_destroy(mt);
const char *leader_spec = 0;
int write_using_libxml2 = 0;
- nmem_init();
#if HAVE_LOCALE_H
setlocale(LC_CTYPE, "");
#endif
#endif
prog = *argv;
- while ((r = options("i:o:C:npvc:xOeXIf:t:s:l:", argv, argc, &arg)) != -2)
+ while ((r = options("i:o:C:npc:xOeXIf:t:s:l:Vv", argv, argc, &arg)) != -2)
{
no++;
switch (r)
}
#if YAZ_HAVE_XML2
#else
- if (input_format == YAZ_MARC_MARCXML
+ if (input_format == YAZ_MARC_MARCXML
|| input_format == YAZ_MARC_XCHANGE)
{
fprintf(stderr, "%s: Libxml2 support not enabled\n", prog);
rather than WRBUF */
if (strlen(arg) > 4 && strncmp(arg, "xml,", 4) == 0)
{
+ /* Only supported for Libxml2 2.6.0 or later */
+#if LIBXML_VERSION >= 20600
arg = arg + 4;
write_using_libxml2 = 1;
+#else
+ fprintf(stderr, "%s: output using Libxml2 unsupported\n", prog);
+ exit(4);
+#endif
}
output_format = yaz_marc_decode_formatstr(arg);
if (output_format == -1)
break;
case 'c':
if (cfile)
- fclose (cfile);
+ fclose(cfile);
cfile = fopen(arg, "w");
break;
case 'x':
case 'v':
verbose++;
break;
+ case 'V':
+ show_version();
+ break;
default:
usage(prog);
exit(1);
}
}
if (cfile)
- fclose (cfile);
- nmem_exit();
+ fclose(cfile);
if (!no)
{
usage(prog);
- exit (1);
+ exit(1);
}
- exit (0);
+ exit(0);
}
/*
* Local variables:
* c-basic-offset: 4
+ * c-file-style: "Stroustrup"
* indent-tabs-mode: nil
* End:
* vim: shiftwidth=4 tabstop=8 expandtab