/* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2012 Index Data
+ * Copyright (C) 1995-2013 Index Data
* See the file LICENSE for details.
*/
collection_first,
collection_second
};
-
+
/** \brief node types for yaz_marc_node */
enum YAZ_MARC_NODE_TYPE
-{
+{
YAZ_MARC_DATAFIELD,
YAZ_MARC_CONTROLFIELD,
YAZ_MARC_COMMENT,
size_t size);
#if YAZ_HAVE_XML2
static int yaz_marc_write_xml_turbo_xml(yaz_marc_t mt, xmlNode **root_ptr,
- const char *ns,
+ const char *ns,
const char *format,
const char *type);
#endif
}
/** \brief adds a attribute value to the element name if it is plain chars
-
+
If not, and if the attribute name is not null, it will append a
attribute element with the value if attribute name is null it will
return a non-zero value meaning it couldnt handle the value.
}
}
+static void check_ascii(yaz_marc_t mt, char *leader, int offset,
+ int ch_default)
+{
+ if (leader[offset] < ' ' || leader[offset] > 127)
+ {
+ yaz_marc_cprintf(mt,
+ "Leader character at offset %d is non-ASCII. "
+ "Setting value to '%c'", offset, ch_default);
+ leader[offset] = ch_default;
+ }
+}
+
void yaz_marc_set_leader(yaz_marc_t mt, const char *leader_c,
int *indicator_length,
int *identifier_length,
memcpy(leader, leader_c, 24);
+ check_ascii(mt, leader, 5, 'a');
+ check_ascii(mt, leader, 6, 'a');
+ check_ascii(mt, leader, 7, 'a');
+ check_ascii(mt, leader, 8, '#');
+ check_ascii(mt, leader, 9, '#');
if (!atoi_n_check(leader+10, 1, indicator_length))
{
- yaz_marc_cprintf(mt,
+ yaz_marc_cprintf(mt,
"Indicator length at offset 10 should hold a digit."
" Assuming 2");
leader[10] = '2';
}
if (!atoi_n_check(leader+11, 1, identifier_length))
{
- yaz_marc_cprintf(mt,
+ yaz_marc_cprintf(mt,
"Identifier length at offset 11 should hold a digit."
" Assuming 2");
leader[11] = '2';
}
if (!atoi_n_check(leader+12, 5, base_address))
{
- yaz_marc_cprintf(mt,
+ yaz_marc_cprintf(mt,
"Base address at offsets 12..16 should hold a number."
" Assuming 0");
*base_address = 0;
}
+ check_ascii(mt, leader, 17, '#');
+ check_ascii(mt, leader, 18, '#');
+ check_ascii(mt, leader, 19, '#');
if (!atoi_n_check(leader+20, 1, length_data_entry))
{
- yaz_marc_cprintf(mt,
+ yaz_marc_cprintf(mt,
"Length data entry at offset 20 should hold a digit."
" Assuming 4");
*length_data_entry = 4;
}
if (!atoi_n_check(leader+22, 1, length_implementation))
{
- yaz_marc_cprintf(mt,
+ yaz_marc_cprintf(mt,
"Length implementation at offset 22 should hold a digit."
" Assuming 0");
*length_implementation = 0;
leader[22] = '0';
}
+ check_ascii(mt, leader, 23, '0');
if (mt->debug)
{
size_t inbytesleft = i;
size_t r = yaz_iconv(mt->iconv_cd, (char**) &inp, &inbytesleft,
&outp, &outbytesleft);
+ yaz_iconv(mt->iconv_cd, 0, 0, &outp, &outbytesleft);
if (r != (size_t) (-1))
return i; /* got a complete sequence */
}
}
return 1; /* we don't know */
}
-
+
void yaz_marc_reset(yaz_marc_t mt)
{
nmem_reset(mt->nmem);
leader = n->u.leader;
break;
}
-
+
if (!leader)
return -1;
if (!atoi_n_check(leader+11, 1, &identifier_length))
switch(n->which)
{
case YAZ_MARC_COMMENT:
- wrbuf_iconv_write(wr, mt->iconv_cd,
+ wrbuf_iconv_write(wr, mt->iconv_cd,
n->u.comment, strlen(n->u.comment));
wrbuf_puts(wr, "\n");
break;
leader = n->u.leader;
break;
}
-
+
if (!leader)
return -1;
if (!atoi_n_check(leader+11, 1, &identifier_length))
{
size_t using_code_len = get_subfield_len(mt, s->code_data,
identifier_length);
-
- wrbuf_puts (wr, mt->subfield_str);
- wrbuf_iconv_write(wr, mt->iconv_cd, s->code_data,
+
+ wrbuf_puts (wr, mt->subfield_str);
+ wrbuf_iconv_write(wr, mt->iconv_cd, s->code_data,
using_code_len);
wrbuf_iconv_puts(wr, mt->iconv_cd, " ");
- wrbuf_iconv_puts(wr, mt->iconv_cd,
+ wrbuf_iconv_puts(wr, mt->iconv_cd,
s->code_data + using_code_len);
marc_iconv_reset(mt, wr);
}
break;
case YAZ_MARC_COMMENT:
wrbuf_puts(wr, "(");
- wrbuf_iconv_write(wr, mt->iconv_cd,
+ wrbuf_iconv_write(wr, mt->iconv_cd,
n->u.comment, strlen(n->u.comment));
marc_iconv_reset(mt, wr);
wrbuf_puts(wr, ")\n");
\retval -1 failure
*/
static int yaz_marc_write_marcxml_wrbuf(yaz_marc_t mt, WRBUF wr,
- const char *ns,
+ const char *ns,
const char *format,
const char *type,
int turbo)
leader = n->u.leader;
break;
}
-
+
if (!leader)
return -1;
if (!atoi_n_check(leader+11, 1, &identifier_length))
return -1;
-
+
if (mt->enable_collection != no_collection)
{
if (mt->enable_collection == collection_first)
wrbuf_iconv_write_cdata(wr, mt->iconv_cd,
s->code_data, using_code_len);
wrbuf_iconv_puts(wr, mt->iconv_cd, "\">");
- }
+ }
else
{
element_name_append_attribute_value(mt, wr, "code", s->code_data, using_code_len);
return 0;
}
-static void sanitise_leader_for_utf8(yaz_marc_t mt)
-{
- /* the leader MUST be ASCII for UTF-8 output (XML) */
- struct yaz_marc_node *n;
- for (n = mt->nodes; n; n = n->next)
- if (n->which == YAZ_MARC_LEADER)
- {
- size_t i;
- for (i = 0; n->u.leader[i]; i++)
- if (n->u.leader[i] < ' ' || n->u.leader[i] > 126)
- {
- n->u.leader[i] = ' ';
- yaz_marc_cprintf(mt, "Fixing leader char at offset %d",
- (int) (i+1));
- }
- }
-}
-
static int yaz_marc_write_marcxml_ns(yaz_marc_t mt, WRBUF wr,
- const char *ns,
+ const char *ns,
const char *format,
const char *type,
int turbo)
{
- sanitise_leader_for_utf8(mt);
if (mt->write_using_libxml2)
{
#if YAZ_HAVE_XML2
{
char ind_str[6];
char ind_val[2];
-
+
ind_val[0] = n->u.datafield.indicator[i];
ind_val[1] = '\0';
sprintf(ind_str, "%s%d", indicator_name[1], i+1);
wrbuf_rewind(wr_cdata);
wrbuf_iconv_puts(wr_cdata, mt->iconv_cd, s->code_data + using_code_len);
marc_iconv_reset(mt, wr_cdata);
-
+
wrbuf_rewind(subfield_name);
wrbuf_puts(subfield_name, "s");
not_written = element_name_append_attribute_value(mt, subfield_name, 0, s->code_data, using_code_len) != 0;
}
static int yaz_marc_write_xml_turbo_xml(yaz_marc_t mt, xmlNode **root_ptr,
- const char *ns,
+ const char *ns,
const char *format,
const char *type)
{
leader = n->u.leader;
break;
}
-
+
if (!leader)
return -1;
if (!atoi_n_check(leader+11, 1, &identifier_length))
char field[10];
field[0] = 'c';
field[4] = '\0';
-
+
switch(n->which)
{
case YAZ_MARC_DATAFIELD:
wrbuf_rewind(wr_cdata);
wrbuf_iconv_puts(wr_cdata, mt->iconv_cd, n->u.controlfield.data);
marc_iconv_reset(mt, wr_cdata);
-
+
strncpy(field + 1, n->u.controlfield.tag, 3);
ptr = xmlNewTextChild(record_ptr, ns_record,
BAD_CAST field,
int yaz_marc_write_xml(yaz_marc_t mt, xmlNode **root_ptr,
- const char *ns,
+ const char *ns,
const char *format,
const char *type)
{
leader = n->u.leader;
break;
}
-
+
if (!leader)
return -1;
if (!atoi_n_check(leader+11, 1, &identifier_length))
wrbuf_rewind(wr_cdata);
wrbuf_iconv_puts(wr_cdata, mt->iconv_cd, n->u.controlfield.data);
marc_iconv_reset(mt, wr_cdata);
-
+
ptr = xmlNewTextChild(record_ptr, ns_record,
BAD_CAST "controlfield",
BAD_CAST wrbuf_cstr(wr_cdata));
-
+
xmlNewProp(ptr, BAD_CAST "tag", BAD_CAST n->u.controlfield.tag);
break;
case YAZ_MARC_COMMENT:
const char *leader = 0;
WRBUF wr_dir, wr_head, wr_data_tmp;
int base_address;
-
+
for (n = mt->nodes; n; n = n->next)
if (n->which == YAZ_MARC_LEADER)
leader = n->u.leader;
-
+
if (!leader)
return -1;
if (!atoi_n_check(leader+10, 1, &indicator_length))
wrbuf_printf(wr_dir, "%.3s", n->u.controlfield.tag);
wrbuf_rewind(wr_data_tmp);
- wrbuf_iconv_puts(wr_data_tmp, mt->iconv_cd,
+ wrbuf_iconv_puts(wr_data_tmp, mt->iconv_cd,
n->u.controlfield.data);
marc_iconv_reset(mt, wr_data_tmp);
wrbuf_iconv_putchar(wr_data_tmp, mt->iconv_cd, ' ');/* field sep */
wrbuf_printf(wr_head, "%05d", base_address);
/* from "original" leader */
wrbuf_write(wr_head, leader+17, 7);
-
+
wrbuf_write(wr, wrbuf_buf(wr_head), 24);
wrbuf_write(wr, wrbuf_buf(wr_dir), wrbuf_len(wr_dir));
wrbuf_destroy(wr_head);
{
const char *vp = strchr(val+1, '\'');
size_t len;
-
+
if (!vp)
return -1;
len = vp-val-1;
int yaz_marc_decode_formatstr(const char *arg)
{
- int mode = -1;
+ int mode = -1;
if (!strcmp(arg, "marc"))
mode = YAZ_MARC_ISO2709;
if (!strcmp(arg, "marcxml"))