-/* $Id: icu_I18N.h,v 1.11 2007-05-11 10:38:42 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 */
-// forward declarations
-//struct UBreakIterator;
-
-
-
-
// declared structs and functions
-
int icu_check_status (UErrorCode status);
struct icu_buf_utf16
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,
UErrorCode *status);
+#if 0
+struct icu_token
+{
+ int32_t token_id;
+ uint8_t * display8;
+ uint8_t * norm8;
+ uint8_t * sort8;
+}
+#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_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
+};
+
+
+
+struct icu_chain_step
+{
+ // 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 * buf16;
+ struct icu_chain_step * previous;
+ int more_tokens;
+ int need_new_token;
+};
+
+
+struct icu_chain;
+
+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 * buf16,
+ UErrorCode *status);
+
+
+void icu_chain_step_destroy(struct icu_chain_step * step);
+
+
+struct icu_chain
+{
+ uint8_t identifier[128];
+ uint8_t locale[16];
+
+ // number of tokens returned so far
+ int32_t token_count;
+
+ // utf8 output buffers
+ struct icu_buf_utf8 * display8;
+ struct icu_buf_utf8 * norm8;
+ struct icu_buf_utf8 * sort8;
+
+ // utf16 source buffer
+ struct icu_buf_utf16 * src16;
+
+ // linked list of chain steps
+ struct icu_chain_step * steps;
+};
+
+struct icu_chain * icu_chain_create(const uint8_t * identifier,
+ const uint8_t * locale);
+
+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