-/* $Id: marcread.c,v 1.24 2004-06-16 22:12:30 adam Exp $
- Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
- Index Data Aps
+/* $Id: marcread.c,v 1.32 2005-12-08 11:06:49 adam Exp $
+ Copyright (C) 1995-2005
+ Index Data ApS
This file is part of the Zebra server.
#include <ctype.h>
#include <assert.h>
-#include <yaz/log.h>
#include <yaz/yaz-util.h>
#include <yaz/marcdisp.h>
-#include "grsread.h"
+#include <idzebra/recgrs.h>
#include "marcomp.h"
#include "inline.h"
#define MARC_DEBUG 0
#define MARCOMP_DEBUG 0
+struct marc_info {
+ char type[256];
+};
+
static data1_node *grs_read_iso2709 (struct grs_read_info *p, int marc_xml)
{
+ struct marc_info *mi = (struct marc_info*) p->clientData;
char buf[100000];
int entry_p;
int record_length;
int indicator_length;
int identifier_length;
int base_address;
+ int end_of_directory;
int length_data_entry;
int length_starting;
int length_implementation;
if ((*p->readf)(p->fh, buf, 5) != 5)
return NULL;
+ while (*buf < '0' || *buf > '9')
+ {
+ int i;
+
+ yaz_log(YLOG_WARN, "MARC: Skipping bad byte %d (0x%02X)",
+ *buf & 0xff, *buf & 0xff);
+ for (i = 0; i<4; i++)
+ buf[i] = buf[i+1];
+
+ if ((*p->readf)(p->fh, buf+4, 1) != 1)
+ return NULL;
+ }
record_length = atoi_n (buf, 5);
if (record_length < 25)
{
- logf (LOG_WARN, "MARC record length < 25, is %d", record_length);
+ yaz_log (YLOG_WARN, "MARC record length < 25, is %d", record_length);
return NULL;
}
/* read remaining part - attempt to read one byte furhter... */
read_bytes = (*p->readf)(p->fh, buf+5, record_length-4);
if (read_bytes < record_length-5)
{
- logf (LOG_WARN, "Couldn't read whole MARC record");
+ yaz_log (YLOG_WARN, "Couldn't read whole MARC record");
return NULL;
}
if (read_bytes == record_length - 4)
if (p->endf)
(*p->endf)(p->fh, cur_offset - 1);
}
- absynName = p->type;
+ absynName = mi->type;
res_root = data1_mk_root (p->dh, p->mem, absynName);
if (!res_root)
{
- yaz_log (LOG_WARN, "cannot read MARC without an abstract syntax");
+ yaz_log (YLOG_WARN, "cannot read MARC without an abstract syntax");
return 0;
}
if (marc_xml)
else
res_top = data1_mk_tag (p->dh, p->mem, absynName, 0, res_root);
- if ((marctab = res_root->u.root.absyn->marc))
+ if ((marctab = data1_absyn_getmarctab(p->dh, res_root->u.root.absyn)))
{
memcpy(marctab->leader, buf, 24);
memcpy(marctab->implementation_codes, buf+6, 4);
length_implementation = atoi_n (buf+22, 1);
for (entry_p = 24; buf[entry_p] != ISO2709_FS; )
- entry_p += 3+length_data_entry+length_starting;
- base_address = entry_p+1;
- for (entry_p = 24; buf[entry_p] != ISO2709_FS; )
+ {
+ int l = 3 + length_data_entry + length_starting;
+ if (entry_p + l >= record_length)
+ {
+ yaz_log(YLOG_WARN, "MARC: Directory offset %d: end of record.",
+ entry_p);
+ return 0;
+ }
+ /* check for digits in length info */
+ while (--l >= 3)
+ if (!isdigit(*(const unsigned char *) (buf + entry_p+l)))
+ break;
+ if (l >= 3)
+ {
+ /* not all digits, so stop directory scan */
+ yaz_log(YLOG_LOG, "MARC: Bad directory");
+ break;
+ }
+ entry_p += 3 + length_data_entry + length_starting;
+ }
+ end_of_directory = entry_p;
+ if (base_address != entry_p+1)
+ {
+ yaz_log(YLOG_WARN, "MARC: Base address does not follow directory");
+ }
+ for (entry_p = 24; entry_p != end_of_directory; )
{
int data_length;
int data_offset;
i = data_offset + base_address;
end_offset = i+data_length-1;
+ if (data_length <= 0 || data_offset < 0 || end_offset >= record_length)
+ {
+ yaz_log(YLOG_WARN, "MARC: Bad offsets in data. Skipping rest");
+ break;
+ }
+
if (memcmp (tag, "00", 2) && indicator_length)
{
/* generate indicator node */
wrbuf_puts(buf, " ");
}
#if MARCOMP_DEBUG
- logf(LOG_LOG, "cat_inline_subfield(): add subfield $%s", found->name);
+ yaz_log(YLOG_LOG, "cat_inline_subfield(): add subfield $%s", found->name);
#endif
pisf = found->next;
}
int i;
if ((i=inline_parse(pif, psubf->u.tag.tag, get_data(psubf, &len)))<0)
{
- logf(LOG_WARN, "inline subfield ($%s): parse error",
+ yaz_log(YLOG_WARN, "inline subfield ($%s): parse error",
psubf->u.tag.tag);
inline_destroy_field(pif);
return;
}
else
{
- logf(LOG_WARN, "In-line field %s missed -- indicators do not match", pif->name);
+ yaz_log(YLOG_WARN, "In-line field %s missed -- indicators do not match", pif->name);
}
}
}
inline_destroy_field(pif);
}
#if MARCOMP_DEBUG
- logf(LOG_LOG, "cat_inline_field(): got buffer {%s}", buf);
+ yaz_log(YLOG_LOG, "cat_inline_field(): got buffer {%s}", buf);
#endif
}
wrbuf_puts(buf, " ");
}
#if MARCOMP_DEBUG
- logf(LOG_LOG, "cat_subfield(): add subfield $%s", found->u.tag.tag);
+ yaz_log(YLOG_LOG, "cat_subfield(): add subfield $%s", found->u.tag.tag);
#endif
subfield = found->next;
}
wrbuf_puts(buf, "");
}
#if MARCOMP_DEBUG
- logf(LOG_LOG, "cat_field(): got buffer {%s}", buf);
+ yaz_log(YLOG_LOG, "cat_field(): got buffer {%s}", buf);
#endif
return field->next;
}
))
{
#if MARCOMP_DEBUG
- logf(LOG_WARN, "Field %s missed -- does not match indicators", field->u.tag.tag);
+ yaz_log(YLOG_WARN, "Field %s missed -- does not match indicators", field->u.tag.tag);
#endif
return field->next;
}
cat_subfield(pf->list, buf, subfield);
#if MARCOMP_DEBUG
- logf(LOG_LOG, "cat_field(): got buffer {%s}", buf);
+ yaz_log(YLOG_LOG, "cat_field(): got buffer {%s}", buf);
#endif
return field->next;
for (p = s; *p; p++)
{
- if (!isspace(*p))
+ if (!isspace(*(unsigned char *)p))
return 0;
}
return 1;
static void parse_data1_tree(struct grs_read_info *p, const char *mc_stmnt,
data1_node *root)
{
- data1_marctab *marctab = root->u.root.absyn->marc;
+ data1_marctab *marctab = data1_absyn_getmarctab(p->dh, root->u.root.absyn);
data1_node *top = root->child;
data1_node *field;
mc_context *c;
}
buf = wrbuf_alloc();
#if MARCOMP_DEBUG
- logf(LOG_LOG, "parse_data1_tree(): statement -{%s}", mc_stmnt);
+ yaz_log(YLOG_LOG, "parse_data1_tree(): statement -{%s}", mc_stmnt);
#endif
if (!yaz_matchstr(pf->name, "ldr"))
{
data1_node *new;
#if MARCOMP_DEBUG
- logf(LOG_LOG,"parse_data1_tree(): try LEADER from {%d} to {%d} positions",
+ yaz_log(YLOG_LOG,"parse_data1_tree(): try LEADER from {%d} to {%d} positions",
pf->interval.start, pf->interval.end);
#endif
new = data1_mk_tag_n(p->dh, p->mem, mc_stmnt, strlen(mc_stmnt), 0, top);
data1_node *new;
char *pb;
#if MARCOMP_DEBUG
- logf(LOG_LOG, "parse_data1_tree(): try field {%s}", field->u.tag.tag);
+ yaz_log(YLOG_LOG, "parse_data1_tree(): try field {%s}", field->u.tag.tag);
#endif
wrbuf_rewind(buf);
wrbuf_puts(buf, "");
if (!root)
return 0;
- for (e=root->u.root.absyn->main_elements; e; e=e->next)
+ for (e = data1_absyn_getelements(p->dh, root->u.root.absyn); e; e=e->next)
{
data1_tag *tag = e->tag;
if (!root)
return 0;
- for (e=root->u.root.absyn->main_elements; e; e=e->next)
+ for (e = data1_absyn_getelements(p->dh, root->u.root.absyn); e; e=e->next)
{
data1_tag *tag = e->tag;
return root;
}
-static void *grs_init_marc(void)
+static void *init_marc(Res res, RecType rt)
{
- return 0;
+ struct marc_info *p = xmalloc(sizeof(*p));
+ strcpy(p->type, "");
+ return p;
+}
+
+static void config_marc(void *clientData, Res res, const char *args)
+{
+ struct marc_info *p = (struct marc_info*) clientData;
+ if (strlen(args) < sizeof(p->type))
+ strcpy(p->type, args);
+}
+
+static void destroy_marc(void *clientData)
+{
+ struct marc_info *p = (struct marc_info*) clientData;
+ xfree (p);
+}
+
+
+static int extract_marc(void *clientData, struct recExtractCtrl *ctrl)
+{
+ return zebra_grs_extract(clientData, ctrl, grs_read_marc);
}
-static void grs_destroy_marc(void *clientData)
+static int retrieve_marc(void *clientData, struct recRetrieveCtrl *ctrl)
{
+ return zebra_grs_retrieve(clientData, ctrl, grs_read_marc);
}
-static struct recTypeGrs marc_type = {
- "marc",
- grs_init_marc,
- grs_destroy_marc,
- grs_read_marc
+static struct recType marc_type = {
+ 0,
+ "grs.marc",
+ init_marc,
+ config_marc,
+ destroy_marc,
+ extract_marc,
+ retrieve_marc,
};
-RecTypeGrs recTypeGrs_marc = &marc_type;
+static int extract_marcxml(void *clientData, struct recExtractCtrl *ctrl)
+{
+ return zebra_grs_extract(clientData, ctrl, grs_read_marcxml);
+}
+
+static int retrieve_marcxml(void *clientData, struct recRetrieveCtrl *ctrl)
+{
+ return zebra_grs_retrieve(clientData, ctrl, grs_read_marcxml);
+}
-static struct recTypeGrs marcxml_type = {
- "marcxml",
- grs_init_marc,
- grs_destroy_marc,
- grs_read_marcxml
+static struct recType marcxml_type = {
+ 0,
+ "grs.marcxml",
+ init_marc,
+ config_marc,
+ destroy_marc,
+ extract_marcxml,
+ retrieve_marcxml,
};
-RecTypeGrs recTypeGrs_marcxml = &marcxml_type;
+RecType
+#ifdef IDZEBRA_STATIC_GRS_MARC
+idzebra_filter_grs_marc
+#else
+idzebra_filter
+#endif
+
+[] = {
+ &marc_type,
+ &marcxml_type,
+ 0,
+};
+