int icu_chain_assign_cstr(struct icu_chain *chain, const char *src8cstr,
UErrorCode *status)
{
if (chain->iter)
icu_iter_destroy(chain->iter);
chain->iter = icu_iter_create(chain);
icu_iter_first(chain->iter, src8cstr);
return 1;
}
pp2_charset_token_t pp2_charset_tokenize(pp2_charset_t pct)
{
pp2_charset_token_t prt = xmalloc(sizeof(*prt));
assert(pct);
prt->norm_str = wrbuf_alloc();
prt->sort_str = wrbuf_alloc();
prt->cp = 0;
prt->last_cp = 0;
prt->pct = pct;
#if YAZ_HAVE_ICU
prt->iter = 0;
if (pct->icu_chn)
prt->iter = icu_iter_create(pct->icu_chn);
#endif
return prt;
}
static void check_icu_iter1(void)
{
UErrorCode status = U_ZERO_ERROR;
struct icu_chain *chain = 0;
xmlNode *xml_node;
yaz_icu_iter_t iter;
const char *xml_str = "<icu locale=\"en\">"
"<tokenize rule=\"w\"/>"
"<transform rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>"
"</icu>";
xmlDoc *doc = xmlParseMemory(xml_str, strlen(xml_str));
YAZ_CHECK(doc);
if (!doc)
return;
xml_node = xmlDocGetRootElement(doc);
YAZ_CHECK(xml_node);
if (!xml_node)
return ;
chain = icu_chain_xml_config(xml_node, 1, &status);
xmlFreeDoc(doc);
YAZ_CHECK(chain);
iter = icu_iter_create(chain);
icu_iter_first(iter, "a string with 15 tokens and 8 displays");
YAZ_CHECK(iter);
if (!iter)
return;
while (icu_iter_next(iter))
{
yaz_log(YLOG_LOG, "[%s]", icu_iter_get_norm(iter));
}
icu_iter_destroy(iter);
icu_chain_destroy(chain);
}
static void check_norm(void)
{
UErrorCode status = U_ZERO_ERROR;
struct icu_chain *chain = 0;
xmlNode *xml_node;
yaz_icu_iter_t it;
const char *xml_str =
" <icu_chain id=\"relevance\" locale=\"en\">"
" <transform rule=\"[:Control:] Any-Remove\"/>"
" <tokenize rule=\"l\"/>"
" <transform rule=\"[[:WhiteSpace:][:Punctuation:]`] Remove\"/>"
" <casemap rule=\"l\"/>"
" </icu_chain>";
xmlDoc *doc = xmlParseMemory(xml_str, strlen(xml_str));
YAZ_CHECK(doc);
if (!doc)
return;
xml_node = xmlDocGetRootElement(doc);
YAZ_CHECK(xml_node);
if (!xml_node)
return ;
chain = icu_chain_xml_config(xml_node, 1, &status);
it = icu_iter_create(chain);
if (it)
{
icu_iter_first(it, " y😄");
while (icu_iter_next(it))
{
const char *norm_str = icu_iter_get_norm(it);
size_t start, len;
YAZ_CHECK(norm_str);
if (norm_str)
yaz_log(YLOG_LOG, "norm_str len=%ld=%s",
(long) strlen(norm_str), norm_str);
icu_iter_get_org_info(it, &start, &len);
YAZ_CHECK(start <= 1000);
YAZ_CHECK(len <= 1000);
}
icu_iter_first(it, "\n y😄");
while (icu_iter_next(it))
{
const char *norm_str = icu_iter_get_norm(it);
size_t start, len;
YAZ_CHECK(norm_str);
if (norm_str)
yaz_log(YLOG_LOG, "norm_str len=%ld=%s",
(long) strlen(norm_str), norm_str);
icu_iter_get_org_info(it, &start, &len);
YAZ_CHECK(start <= 1000);
YAZ_CHECK(len <= 1000);
}
}
icu_iter_destroy(it);
icu_chain_destroy(chain);
xmlFreeDoc(doc);
}
static int test_iter(struct icu_chain *chain, const char *input,
const char *expected)
{
yaz_icu_iter_t iter = icu_iter_create(chain);
WRBUF result, second, sort_result;
int success = 1;
if (!iter)
{
yaz_log(YLOG_WARN, "test_iter: input=%s !iter", input);
return 0;
}
if (icu_iter_next(iter))
{
yaz_log(YLOG_WARN, "test_iter: expecting 0 before icu_iter_first");
return 0;
}
sort_result = wrbuf_alloc();
result = wrbuf_alloc();
icu_iter_first(iter, input);
while (icu_iter_next(iter))
{
const char *sort_str = icu_iter_get_sortkey(iter);
if (sort_str)
{
wrbuf_puts(sort_result, "[");
wrbuf_puts_escaped(sort_result, sort_str);
wrbuf_puts(sort_result, "]");
}
else
{
wrbuf_puts(sort_result, "[NULL]");
}
wrbuf_puts(result, "[");
wrbuf_puts(result, icu_iter_get_norm(iter));
wrbuf_puts(result, "]");
}
yaz_log(YLOG_LOG, "sortkey=%s", wrbuf_cstr(sort_result));
second = wrbuf_alloc();
icu_iter_first(iter, input);
while (icu_iter_next(iter))
{
wrbuf_puts(second, "[");
wrbuf_puts(second, icu_iter_get_norm(iter));
wrbuf_puts(second, "]");
}
icu_iter_destroy(iter);
if (strcmp(expected, wrbuf_cstr(result)))
{
yaz_log(YLOG_WARN, "test_iter: input=%s expected=%s got=%s",
input, expected, wrbuf_cstr(result));
success = 0;
}
if (strcmp(expected, wrbuf_cstr(second)))
{
yaz_log(YLOG_WARN, "test_iter: input=%s expected=%s got=%s (2nd)",
input, expected, wrbuf_cstr(second));
success = 0;
}
wrbuf_destroy(result);
wrbuf_destroy(second);
wrbuf_destroy(sort_result);
return success;
}
