-/*
- * Copyright (C) 1995-2007, Index Data ApS
+/* This file is part of the YAZ toolkit.
+ * Copyright (C) 1995-2008 Index Data
* See the file LICENSE for details.
- *
- * $Id: marcdump.c,v 1.48 2007-02-17 10:53:06 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>
{
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",
+ "[-n] [-l pos=value] [-v] [-C chunk] [-s splitfname] [-p] file...\n",
prog);
}
{
WRBUF wrbuf = wrbuf_alloc();
yaz_marc_write_mode(mt, wrbuf);
- fputs(wrbuf_buf(wrbuf), stdout);
- wrbuf_free(wrbuf, 1);
+ fputs(wrbuf_cstr(wrbuf), stdout);
+ wrbuf_destroy(wrbuf);
}
fclose(inf);
}
#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_buf(wrbuf), stdout);
+ int type = xmlTextReaderNodeType(reader);
+ if (type == XML_READER_TYPE_ELEMENT)
+ {
+ const char *name = (const char *)
+ xmlTextReaderLocalName(reader);
+ if (!strcmp(name, "record"))
+ {
+ xmlNodePtr ptr = xmlTextReaderExpand(reader);
+
+ 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);
+ }
+ }
+ }
+ }
+ }
+#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"))
+ {
+ 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);
+ }
+ }
+ }
}
- wrbuf_free(wrbuf, 1);
+ xmlFreeDoc(doc);
}
- xmlFreeDoc(doc);
+#endif
+ fputs(wrbuf_cstr(wrbuf), stdout);
+ wrbuf_destroy(wrbuf);
}
#endif
yaz_marc_iconv(mt, cd);
}
yaz_marc_xml(mt, output_format);
+ yaz_marc_enable_collection(mt);
yaz_marc_write_using_libxml2(mt, write_using_libxml2);
yaz_marc_debug(mt, verbose);
fprintf (cfile, "char *marc_records[] = {\n");
for(;; marc_no++)
{
- char *result = 0;
+ const char *result = 0;
size_t len;
size_t rlen;
- int len_result;
+ size_t len_result;
size_t r;
char buf[100001];
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);
+ if (r != 1)
+ break;
+ len++;
+ }
if (split_fname)
{
char fname[256];
fclose(sf);
}
}
- len_result = (int) rlen;
+ len_result = rlen;
r = yaz_marc_decode_buf(mt, buf, -1, &result, &len_result);
if (r > 0 && result)
{
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 *split_fname = 0;
const char *leader_spec = 0;
int write_using_libxml2 = 0;
-
+
#if HAVE_LOCALE_H
setlocale(LC_CTYPE, "");
#endif