-/* $Id: zebramap.c,v 1.72 2007-11-15 08:53:26 adam Exp $
+/* $Id: zebramap.c,v 1.78 2008-01-24 16:13:12 adam Exp $
Copyright (C) 1995-2007
Index Data ApS
}
else if (!yaz_matchstr(argv[0], "icuchain"))
{
+ char full_path[1024];
+ if (!yaz_filepath_resolve(argv[1], zms->tabpath, zms->tabroot,
+ full_path))
+ {
+ yaz_log(YLOG_WARN, "%s:%d: Could not locate icuchain config '%s'",
+ fname, lineno, argv[1]);
+ return -1;
+ }
#if YAZ_HAVE_XML2
- zm->doc = xmlParseFile(argv[1]);
+ zm->doc = xmlParseFile(full_path);
if (!zm->doc)
{
yaz_log(YLOG_WARN, "%s:%d: Could not load icuchain config '%s'",
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->input_str);
- wrbuf_write(zm->input_str, 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)
- {
- 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);
- assert(U_SUCCESS(status));
- }
while (icu_chain_next_token(zm->icu_chain, &status))
{
- assert(U_SUCCESS(status));
+ if (!U_SUCCESS(status))
+ return 0;
*result_buf = icu_chain_token_sortkey(zm->icu_chain);
assert(*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);
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