-/* $Id: icu_I18N.c,v 1.7 2007-05-07 12:52:04 marc Exp $
+/* $Id: icu_I18N.c,v 1.13 2007-05-15 15:11:42 marc Exp $
Copyright (c) 2006-2007, Index Data.
This file is part of Pazpar2.
int icu_check_status (UErrorCode status)
{
- //if(U_FAILURE(status))
- if(!U_SUCCESS(status))
+ if(U_FAILURE(status)){
yaz_log(YLOG_WARN,
"ICU: %d %s\n", status, u_errorName(status));
- return status;
+ return 0;
+ }
+ return 1;
+
}
};
+struct icu_buf_utf16 * icu_buf_utf16_copy(struct icu_buf_utf16 * dest16,
+ struct icu_buf_utf16 * src16)
+{
+ if(!dest16 || !src16
+ || dest16 == src16)
+ return 0;
+
+ if (dest16->utf16_cap < src16->utf16_len)
+ icu_buf_utf16_resize(dest16, src16->utf16_len * 2);
+
+ u_strncpy(dest16->utf16, src16->utf16, src16->utf16_len);
+
+ return dest16;
+};
+
+
void icu_buf_utf16_destroy(struct icu_buf_utf16 * buf16)
{
if (buf16){
buf8->utf8 = (uint8_t *) malloc(sizeof(uint8_t) * capacity);
else
buf8->utf8
- = (uint8_t *) realloc(buf8->utf8, sizeof(uint8_t) * capacity);
+ = (uint8_t *) realloc(buf8->utf8,
+ sizeof(uint8_t) * capacity);
buf8->utf8[0] = (uint8_t) 0;
buf8->utf8_len = 0;
buf8->utf8_cap = capacity;
};
+struct icu_buf_utf8 * icu_buf_utf8_copy(struct icu_buf_utf8 * dest8,
+ struct icu_buf_utf8 * src8)
+{
+ if(!dest8 || !src8
+ || dest8 == src8)
+ return 0;
+
+
+ if (dest8->utf8_cap < src8->utf8_len)
+ icu_buf_utf8_resize(dest8, src8->utf8_len * 2);
+
+ strncpy((char*) dest8->utf8, (char*) src8->utf8, src8->utf8_len);
+
+ return dest8;
+};
+
+
void icu_buf_utf8_destroy(struct icu_buf_utf8 * buf8)
{
// check for buffer overflow, resize and retry
if (*status == U_BUFFER_OVERFLOW_ERROR
+ && dest16 != src16 // do not resize if in-place conversion
//|| dest16_len > dest16->utf16_cap
){
icu_buf_utf16_resize(dest16, dest16_len * 2);
dest8->utf8_len = 0;
}
- return *status;
+ return sortkey_len;
+};
+
+
+
+struct icu_tokenizer * icu_tokenizer_create(const char *locale, char action,
+ UErrorCode *status)
+{
+ struct icu_tokenizer * tokenizer
+ = (struct icu_tokenizer *) malloc(sizeof(struct icu_tokenizer));
+
+ strcpy(tokenizer->locale, locale);
+ tokenizer->action = action;
+ tokenizer->bi = 0;
+ tokenizer->buf16 = 0;
+ tokenizer->token_count = 0;
+ tokenizer->token_id = 0;
+ tokenizer->token_start = 0;
+ tokenizer->token_end = 0;
+
+
+ switch(tokenizer->action) {
+ case 'l':
+ tokenizer->bi
+ = ubrk_open(UBRK_LINE, tokenizer->locale,
+ 0, 0, status);
+ break;
+ case 's':
+ tokenizer->bi
+ = ubrk_open(UBRK_SENTENCE, tokenizer->locale,
+ 0, 0, status);
+ break;
+ case 'w':
+ tokenizer->bi
+ = ubrk_open(UBRK_WORD, tokenizer->locale,
+ 0, 0, status);
+ break;
+ case 'c':
+ tokenizer->bi
+ = ubrk_open(UBRK_CHARACTER, tokenizer->locale,
+ 0, 0, status);
+ break;
+ case 't':
+ tokenizer->bi
+ = ubrk_open(UBRK_TITLE, tokenizer->locale,
+ 0, 0, status);
+ break;
+ default:
+ *status = U_UNSUPPORTED_ERROR;
+ return 0;
+ break;
+ }
+
+ // ICU error stuff is a very funny business
+ if (U_SUCCESS(*status))
+ return tokenizer;
+
+ // freeing if failed
+ icu_tokenizer_destroy(tokenizer);
+ return 0;
+};
+
+void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer)
+{
+ if (tokenizer) {
+ if (tokenizer->bi)
+ ubrk_close(tokenizer->bi);
+ free(tokenizer);
+ }
+};
+
+int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
+ struct icu_buf_utf16 * src16,
+ UErrorCode *status)
+{
+ if (!tokenizer || !tokenizer->bi || !src16)
+ return 0;
+
+
+ tokenizer->buf16 = src16;
+ tokenizer->token_count = 0;
+ tokenizer->token_id = 0;
+ tokenizer->token_start = 0;
+ tokenizer->token_end = 0;
+
+ ubrk_setText(tokenizer->bi, src16->utf16, src16->utf16_len, status);
+
+
+ if (U_FAILURE(*status))
+ return 0;
+
+ return 1;
+};
+
+int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
+ struct icu_buf_utf16 * tkn16,
+ UErrorCode *status)
+{
+ int32_t tkn_start = 0;
+ int32_t tkn_end = 0;
+ int32_t tkn_len = 0;
+
+
+ if (!tokenizer || !tokenizer->bi
+ || !tokenizer->buf16 || !tokenizer->buf16->utf16_len)
+ return 0;
+
+ // never change tokenizer->buf16 and keep always invariant
+ // 0 <= tokenizer->token_start
+ // <= tokenizer->token_end
+ // <= tokenizer->buf16->utf16_len
+ // returns length of token
+
+ if (0 == tokenizer->token_end) // first call
+ tkn_start = ubrk_first(tokenizer->bi);
+ else //successive calls
+ tkn_start = tokenizer->token_end;
+
+ // get next position
+ tkn_end = ubrk_next(tokenizer->bi);
+
+ // repairing invariant at end of ubrk, which is UBRK_DONE = -1
+ if (UBRK_DONE == tkn_end)
+ tkn_end = tokenizer->buf16->utf16_len;
+
+ // copy out if everything is well
+ if(U_FAILURE(*status))
+ return 0;
+
+ // everything OK, now update internal state
+ tkn_len = tkn_end - tkn_start;
+
+ if (0 < tkn_len){
+ tokenizer->token_count++;
+ tokenizer->token_id++;
+ } else {
+ tokenizer->token_id = 0;
+ }
+ tokenizer->token_start = tkn_start;
+ tokenizer->token_end = tkn_end;
+
+
+ // copying into token buffer if it exists
+ if (tkn16){
+ if (tkn16->utf16_cap < tkn_len)
+ icu_buf_utf16_resize(tkn16, (size_t) tkn_len * 2);
+
+ u_strncpy(tkn16->utf16, &(tokenizer->buf16->utf16)[tkn_start],
+ tkn_len);
+
+ tkn16->utf16_len = tkn_len;
+ }
+
+ return tkn_len;
+}
+
+
+int32_t icu_tokenizer_token_id(struct icu_tokenizer * tokenizer)
+{
+ return tokenizer->token_id;
+};
+
+int32_t icu_tokenizer_token_start(struct icu_tokenizer * tokenizer)
+{
+ return tokenizer->token_start;
+};
+
+int32_t icu_tokenizer_token_end(struct icu_tokenizer * tokenizer)
+{
+ return tokenizer->token_end;
+};
+
+int32_t icu_tokenizer_token_length(struct icu_tokenizer * tokenizer)
+{
+ return (tokenizer->token_end - tokenizer->token_start);
+};
+
+int32_t icu_tokenizer_token_count(struct icu_tokenizer * tokenizer)
+{
+ return tokenizer->token_count;
+};
+
+
+
+//struct icu_normalizer
+//{
+// char action;
+// struct icu_buf_utf16 * rules16;
+// UParseError parse_error[256];
+// UTransliterator * trans;
+//};
+
+
+struct icu_normalizer * icu_normalizer_create(const char *rules, char action,
+ UErrorCode *status)
+{
+
+ struct icu_normalizer * normalizer
+ = (struct icu_normalizer *) malloc(sizeof(struct icu_normalizer));
+
+ normalizer->action = action;
+ normalizer->trans = 0;
+ normalizer->rules16 = icu_buf_utf16_create(0);
+ icu_utf16_from_utf8_cstr(normalizer->rules16, rules, status);
+
+ switch(normalizer->action) {
+ case 'f':
+ normalizer->trans
+ = utrans_openU(normalizer->rules16->utf16,
+ normalizer->rules16->utf16_len,
+ UTRANS_FORWARD,
+ 0, 0,
+ normalizer->parse_error, status);
+ break;
+ case 'r':
+ normalizer->trans
+ = utrans_openU(normalizer->rules16->utf16,
+ normalizer->rules16->utf16_len,
+ UTRANS_REVERSE ,
+ 0, 0,
+ normalizer->parse_error, status);
+ break;
+ default:
+ *status = U_UNSUPPORTED_ERROR;
+ return 0;
+ break;
+ }
+
+ if (U_SUCCESS(*status))
+ return normalizer;
+
+ // freeing if failed
+ icu_normalizer_destroy(normalizer);
+ return 0;
+};
+
+
+void icu_normalizer_destroy(struct icu_normalizer * normalizer){
+ if (normalizer) {
+ if (normalizer->rules16)
+ icu_buf_utf16_destroy(normalizer->rules16);
+ if (normalizer->trans)
+ utrans_close(normalizer->trans);
+ free(normalizer);
+ }
};
+int icu_normalizer_normalize(struct icu_normalizer * normalizer,
+ struct icu_buf_utf16 * dest16,
+ struct icu_buf_utf16 * src16,
+ UErrorCode *status)
+{
+ if (!normalizer || !normalizer->trans || !src16 || !dest16)
+ return 0;
+
+ if (!icu_buf_utf16_copy(dest16, src16))
+ return 0;
+
+ utrans_transUChars (normalizer->trans,
+ dest16->utf16, &(dest16->utf16_len),
+ dest16->utf16_cap,
+ 0, &(src16->utf16_len), status);
+
+ if (U_FAILURE(*status)){
+ dest16->utf16[0] = (UChar) 0;
+ dest16->utf16_len = 0;
+ }
+
+ return dest16->utf16_len;
+}
+
+
+
+
+struct icu_chain_step * icu_chain_step_create(struct icu_chain * chain,
+ enum icu_chain_step_type type,
+ const uint8_t * rule,
+ struct icu_buf_utf16 * src16,
+ UErrorCode *status)
+{
+ struct icu_chain_step * step = 0;
+
+ if(!chain || !type || !rule)
+ return 0;
+
+ step = (struct icu_chain_step *) malloc(sizeof(struct icu_chain_step));
+
+ // create auxilary objects
+ switch(step->type) {
+ case ICU_chain_step_type_display:
+ break;
+ case ICU_chain_step_type_norm:
+ break;
+ case ICU_chain_step_type_sort:
+ break;
+ case ICU_chain_step_type_charmap:
+ break;
+ case ICU_chain_step_type_normalize:
+ step->u.normalizer = icu_normalizer_create((char *) rule, 'f', status);
+ break;
+ case ICU_chain_step_type_tokenize:
+ step->u.tokenizer = icu_tokenizer_create((char *) chain->locale,
+ (char) rule[0], status);
+ break;
+ default:
+ break;
+ }
+
+ if (src16)
+ step->src16 = src16;
+
+
+ return step;
+};
+
+
+void icu_chain_step_destroy(struct icu_chain_step * step){
+
+ if (!step)
+ return;
+
+ if (step->previous)
+ icu_chain_step_destroy(step->previous);
+
+ if (step->src16)
+ icu_buf_utf16_destroy(step->src16);
+
+ // destroy last living icu_chain_step
+
+ switch(step->type) {
+ case ICU_chain_step_type_display:
+ break;
+ case ICU_chain_step_type_norm:
+ break;
+ case ICU_chain_step_type_sort:
+ break;
+ case ICU_chain_step_type_charmap:
+ break;
+ case ICU_chain_step_type_normalize:
+ icu_normalizer_destroy(step->u.normalizer);
+ break;
+ case ICU_chain_step_type_tokenize:
+ icu_tokenizer_destroy(step->u.tokenizer);
+ break;
+ default:
+ break;
+ }
+
+
+};
+
+
+
+struct icu_chain * icu_chain_create(const uint8_t * identifier,
+ const uint8_t * locale)
+{
+
+ struct icu_chain * chain
+ = (struct icu_chain *) malloc(sizeof(struct icu_chain));
+
+ strncpy((char *) chain->identifier, (const char *) identifier, 128);
+ chain->identifier[128 - 1] = '\0';
+ strncpy((char *) chain->locale, (const char *) locale, 16);
+ chain->locale[16 - 1] = '\0';
+
+ chain->token_count = 0;
+
+ chain->display8 = icu_buf_utf8_create(0);
+ chain->norm8 = icu_buf_utf8_create(0);
+ chain->sort8 = icu_buf_utf8_create(0);
+
+ chain->src16 = icu_buf_utf16_create(0);
+
+ chain->steps = 0;
+
+ return chain;
+};
+
+
+void icu_chain_destroy(struct icu_chain * chain)
+{
+ icu_buf_utf8_destroy(chain->display8);
+ icu_buf_utf8_destroy(chain->norm8);
+ icu_buf_utf8_destroy(chain->sort8);
+
+ icu_buf_utf16_destroy(chain->src16);
+
+ icu_chain_step_destroy(chain->steps);
+};
+
+
+struct icu_chain_step * icu_chain_insert_step(struct icu_chain * chain,
+ enum icu_chain_step_type type,
+ const uint8_t * rule,
+ UErrorCode *status)
+{
+ struct icu_chain_step * step = 0;
+ struct icu_buf_utf16 * src16 = 0;
+
+ if (!chain || !type || !rule)
+ return 0;
+
+ //if(chain->steps && chain->steps->src16)
+
+ // assign utf16 src buffers as needed
+ switch(step->type) {
+ case ICU_chain_step_type_display:
+ break;
+ case ICU_chain_step_type_norm:
+ break;
+ case ICU_chain_step_type_sort:
+ break;
+ case ICU_chain_step_type_charmap:
+ break;
+ case ICU_chain_step_type_normalize:
+ break;
+ case ICU_chain_step_type_tokenize:
+ break;
+ default:
+ break;
+ }
+
+ // create actual chain step with this buffer
+ // leave zero for implicit buffer creation
+ step = icu_chain_step_create(chain, type, rule, src16, status);
+
+ step->previous = chain->steps;
+ chain->steps = step;
+
+ return step;
+};
+
#endif // HAVE_ICU