-/* $Id: icu_I18N.h,v 1.13 2007-05-15 15:11:42 marc Exp $
+/* $Id: icu_I18N.h,v 1.17 2007-05-25 13:27:21 marc Exp $
Copyright (c) 2006-2007, Index Data.
This file is part of Pazpar2.
#include <yaz/nmem.h>
+#include <libxml/parser.h>
+#include <libxml/tree.h>
#include <unicode/utypes.h> /* Basic ICU data types */
#include <unicode/uchar.h> /* char names */
struct icu_buf_utf16 * src16,
UErrorCode * status);
+struct icu_casemap
+{
+ char locale[16];
+ char action;
+};
+
+struct icu_casemap * icu_casemap_create(const char *locale, char action,
+ UErrorCode *status);
+
+void icu_casemap_destroy(struct icu_casemap * casemap);
+
+int icu_casemap_casemap(struct icu_casemap * casemap,
+ struct icu_buf_utf16 * dest16,
+ struct icu_buf_utf16 * src16,
+ UErrorCode *status);
+
int icu_utf16_casemap(struct icu_buf_utf16 * dest16,
struct icu_buf_utf16 * src16,
const char *locale, char action,
}
#endif
+
enum icu_chain_step_type {
ICU_chain_step_type_none, //
ICU_chain_step_type_display, // convert to utf8 display format
- ICU_chain_step_type_norm, // convert to utf8 norm format
- ICU_chain_step_type_sort, // convert to utf8 sort format
- ICU_chain_step_type_charmap, // apply utf16 charmap
+ ICU_chain_step_type_index, // convert to utf8 index format
+ ICU_chain_step_type_sortkey, // convert to utf8 sortkey format
+ ICU_chain_step_type_casemap, // apply utf16 charmap
ICU_chain_step_type_normalize, // apply utf16 normalization
ICU_chain_step_type_tokenize // apply utf16 tokenization
};
// type and action object
enum icu_chain_step_type type;
union {
+ struct icu_casemap * casemap;
struct icu_normalizer * normalizer;
struct icu_tokenizer * tokenizer;
} u;
// temprary post-action utf16 buffer
- struct icu_buf_utf16 * src16;
+ struct icu_buf_utf16 * buf16;
struct icu_chain_step * previous;
- int end_of_tokens;
+ int more_tokens;
+ int need_new_token;
};
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,
+ struct icu_buf_utf16 * buf16,
UErrorCode *status);
void icu_chain_destroy(struct icu_chain * chain);
+struct icu_chain * icu_chain_xml_config(xmlNode *xml_node,
+ UErrorCode * status);
+
+
struct icu_chain_step * icu_chain_insert_step(struct icu_chain * chain,
enum icu_chain_step_type type,
const uint8_t * rule,
UErrorCode *status);
+int icu_chain_step_next_token(struct icu_chain * chain,
+ struct icu_chain_step * step,
+ UErrorCode *status);
+
+int icu_chain_assign_cstr(struct icu_chain * chain,
+ const char * src8cstr,
+ UErrorCode *status);
+
+int icu_chain_next_token(struct icu_chain * chain,
+ UErrorCode *status);
+
+int icu_chain_get_token_count(struct icu_chain * chain);
+
+const char * icu_chain_get_display(struct icu_chain * chain);
+
+const char * icu_chain_get_norm(struct icu_chain * chain);
+
+const char * icu_chain_get_sort(struct icu_chain * chain);
+
+
+
#endif // HAVE_ICU