-/* $Id: zebramap.c,v 1.68 2007-11-08 08:18:37 adam Exp $
+/* $Id: zebramap.c,v 1.78 2008-01-24 16:13:12 adam Exp $
Copyright (C) 1995-2007
Index Data ApS
#include <yaz/yaz-util.h>
#if YAZ_HAVE_ICU
-#include <yaz/icu_I18N.h>
+#include <yaz/icu.h>
#endif
#include <zebramap.h>
int first_in_field;
int type;
int use_chain;
+ int debug;
union {
struct {
int entry_size;
} u;
chrmaptab maptab;
const char *maptab_name;
- const char *locale;
zebra_maps_t zebra_maps;
#if YAZ_HAVE_XML2
xmlDocPtr doc;
#if YAZ_HAVE_ICU
struct icu_chain *icu_chain;
#endif
- WRBUF simple_buf;
+ WRBUF input_str;
+ WRBUF print_str;
size_t simple_off;
struct zebra_map *next;
};
#if YAZ_HAVE_XML2
xmlFreeDoc(zm->doc);
#endif
- wrbuf_destroy(zm->simple_buf);
+ wrbuf_destroy(zm->input_str);
+ wrbuf_destroy(zm->print_str);
zm = zm->next;
}
wrbuf_destroy(zms->wrbuf_1);
zm->id = nmem_strdup(zms->nmem, index_type);
zm->maptab_name = 0;
zm->use_chain = 0;
- zm->locale = 0;
+ zm->debug = 0;
zm->maptab = 0;
zm->type = map_type;
zm->completeness = 0;
#if YAZ_HAVE_XML2
zm->doc = 0;
#endif
- zm->simple_buf = wrbuf_alloc();
+ zm->input_str = wrbuf_alloc();
+ zm->print_str = wrbuf_alloc();
return zm;
}
return -1;
}
}
- else if (!yaz_matchstr(argv[0], "locale"))
- {
- zm->locale = nmem_strdup(zms->nmem, argv[1]);
- }
else if (!yaz_matchstr(argv[0], "simplechain"))
{
zm->use_chain = 1;
}
else if (!yaz_matchstr(argv[0], "icuchain"))
{
-#if YAZ_HAVE_XML2
- if (!zm->locale)
+ char full_path[1024];
+ if (!yaz_filepath_resolve(argv[1], zms->tabpath, zms->tabroot,
+ full_path))
{
- yaz_log(YLOG_WARN, "%s:%d: locale required before icuchain",
- fname, lineno);
+ yaz_log(YLOG_WARN, "%s:%d: Could not locate icuchain config '%s'",
+ fname, lineno, argv[1]);
return -1;
}
- zm->doc = xmlParseFile(argv[1]);
+#if YAZ_HAVE_XML2
+ zm->doc = xmlParseFile(full_path);
if (!zm->doc)
{
yaz_log(YLOG_WARN, "%s:%d: Could not load icuchain config '%s'",
UErrorCode status;
xmlNode *xml_node = xmlDocGetRootElement(zm->doc);
zm->icu_chain =
- icu_chain_xml_config(xml_node, zm->locale,
+ icu_chain_xml_config(xml_node,
/* not sure about sort for this function yet.. */
#if 1
1,
return -1;
#endif
}
+ else if (!yaz_matchstr(argv[0], "debug") && argc == 2)
+ {
+ zm->debug = atoi(argv[1]);
+ }
else
{
yaz_log(YLOG_WARN, "%s:%d: Unrecognized directive '%s'",
static int tokenize_simple(zebra_map_t zm,
const char **result_buf, size_t *result_len)
{
- char *buf = wrbuf_buf(zm->simple_buf);
- size_t len = wrbuf_len(zm->simple_buf);
+ char *buf = wrbuf_buf(zm->input_str);
+ size_t len = wrbuf_len(zm->input_str);
size_t i = zm->simple_off;
size_t start;
return 0;
}
-int zebra_map_tokenize(zebra_map_t zm,
- const char *buf, size_t len,
- const char **result_buf, size_t *result_len)
+
+int zebra_map_tokenize_next(zebra_map_t zm,
+ const char **result_buf, size_t *result_len,
+ const char **display_buf, size_t *display_len)
{
assert(zm->use_chain);
- if (buf)
- {
- wrbuf_rewind(zm->simple_buf);
- wrbuf_write(zm->simple_buf, buf, len);
- zm->simple_off = 0;
- }
-
#if YAZ_HAVE_ICU
if (!zm->icu_chain)
return tokenize_simple(zm, result_buf, result_len);
else
{
UErrorCode status;
- if (buf)
- {
- yaz_log(YLOG_LOG, "assicn_cstr %s", wrbuf_cstr(zm->simple_buf));
- icu_chain_assign_cstr(zm->icu_chain,
- wrbuf_cstr(zm->simple_buf),
- &status);
- assert(U_SUCCESS(status));
- }
while (icu_chain_next_token(zm->icu_chain, &status))
{
- assert(U_SUCCESS(status));
- *result_buf = icu_chain_token_norm(zm->icu_chain);
+ if (!U_SUCCESS(status))
+ return 0;
+ *result_buf = icu_chain_token_sortkey(zm->icu_chain);
assert(*result_buf);
- yaz_log(YLOG_LOG, "got result %s", *result_buf);
+
*result_len = strlen(*result_buf);
+
+ if (display_buf)
+ {
+ *display_buf = icu_chain_token_display(zm->icu_chain);
+ if (display_len)
+ *display_len = strlen(*display_buf);
+ }
+ if (zm->debug)
+ {
+ wrbuf_rewind(zm->print_str);
+ wrbuf_write_escaped(zm->print_str, *result_buf, *result_len);
+ yaz_log(YLOG_LOG, "output %s", wrbuf_cstr(zm->print_str));
+ }
+
if (**result_buf != '\0')
return 1;
}
- assert(U_SUCCESS(status));
}
return 0;
#else
#endif
}
+int zebra_map_tokenize_start(zebra_map_t zm,
+ const char *buf, size_t len)
+{
+ assert(zm->use_chain);
+
+ wrbuf_rewind(zm->input_str);
+ wrbuf_write(zm->input_str, buf, len);
+ zm->simple_off = 0;
+#if YAZ_HAVE_ICU
+ if (zm->icu_chain)
+ {
+ UErrorCode status;
+ if (zm->debug)
+ {
+ wrbuf_rewind(zm->print_str);
+ wrbuf_write_escaped(zm->print_str, wrbuf_buf(zm->input_str),
+ wrbuf_len(zm->input_str));
+
+ yaz_log(YLOG_LOG, "input %s",
+ wrbuf_cstr(zm->print_str));
+ }
+ icu_chain_assign_cstr(zm->icu_chain,
+ wrbuf_cstr(zm->input_str),
+ &status);
+ if (!U_SUCCESS(status))
+ {
+ if (zm->debug)
+ {
+ yaz_log(YLOG_WARN, "bad encoding for input");
+ }
+ return -1;
+ }
+ }
+#endif
+ return 0;
+}
+
int zebra_maps_is_icu(zebra_map_t zm)
{
+ assert(zm);
#if YAZ_HAVE_ICU
return zm->use_chain;
#else