summaryrefslogtreecommitdiff
path: root/indexing/upmendex/source/mendex.h
diff options
context:
space:
mode:
Diffstat (limited to 'indexing/upmendex/source/mendex.h')
-rw-r--r--indexing/upmendex/source/mendex.h13
1 files changed, 10 insertions, 3 deletions
diff --git a/indexing/upmendex/source/mendex.h b/indexing/upmendex/source/mendex.h
index c6703761fc..091862d61e 100644
--- a/indexing/upmendex/source/mendex.h
+++ b/indexing/upmendex/source/mendex.h
@@ -11,10 +11,12 @@
#include <unicode/uchar.h>
#include <unicode/ucol.h>
+#define PAGE_COMPOSIT_DEPTH 10
+
struct page {
char *page;
char *enc;
- int attr[3];
+ int attr[PAGE_COMPOSIT_DEPTH];
};
struct index {
@@ -47,7 +49,6 @@ int lastpage(const char *filename);
/* sort.c */
void wsort(struct index *ind, int num);
void pagesort(struct index *ind, int num);
-int is_alphanumeric(UChar *c);
int is_latin(UChar *c);
int is_numeric(UChar *c);
int is_jpn_kana(UChar *c);
@@ -58,7 +59,10 @@ int is_cyrillic(UChar *c);
int is_greek(UChar *c);
int is_devanagari(UChar *c);
int is_thai(UChar *c);
-int is_comb_diacritical_mark(UChar *c);
+int is_arabic(UChar *c);
+int is_hebrew(UChar *c);
+int is_type_mark_or_punct(UChar *c);
+int is_type_symbol(UChar *c);
int chkcontinue(struct page *p, int num);
int ss_comp(UChar *s1, UChar *s2);
@@ -71,8 +75,11 @@ int ss_comp(UChar *s1, UChar *s2);
#define CH_HANZI 6
#define CH_DEVANAGARI 7
#define CH_THAI 8
+#define CH_ARABIC 9
+#define CH_HEBREW 10
#define CH_SYMBOL 0x100
#define CH_NUMERIC 0x101
+#define is_any_script(a) ((CH_LATIN<=(a) && (a)<=CH_HEBREW))
/* sort.c */
int charset(UChar *c);