X-Git-Url: https://git.hcoop.net/hcoop/debian/courier-authlib.git/blobdiff_plain/781cfcb8fd50934c470e0dabf79d32ab333dec68..f77892e671e1261ec26e5133f86b8a52635f3dd2:/libs/unicode/unicode_graphemebreak.c diff --git a/libs/unicode/unicode_graphemebreak.c b/libs/unicode/unicode_graphemebreak.c new file mode 100644 index 0000000..5939dc8 --- /dev/null +++ b/libs/unicode/unicode_graphemebreak.c @@ -0,0 +1,95 @@ +/* +** Copyright 2011 Double Precision, Inc. +** See COPYING for distribution information. +** +*/ + +#include "unicode_config.h" +#include "unicode.h" +#include +#include +#include + +#define UNICODE_GRAPHEMEBREAK_ANY 0x00 +#define UNICODE_GRAPHEMEBREAK_CR 0x01 +#define UNICODE_GRAPHEMEBREAK_LF 0x02 +#define UNICODE_GRAPHEMEBREAK_Control 0x03 +#define UNICODE_GRAPHEMEBREAK_Extend 0x04 +#define UNICODE_GRAPHEMEBREAK_Prepend 0x05 +#define UNICODE_GRAPHEMEBREAK_SpacingMark 0x06 +#define UNICODE_GRAPHEMEBREAK_L 0x07 +#define UNICODE_GRAPHEMEBREAK_V 0x08 +#define UNICODE_GRAPHEMEBREAK_T 0x09 +#define UNICODE_GRAPHEMEBREAK_LV 0x0A +#define UNICODE_GRAPHEMEBREAK_LVT 0x0B + +#include "graphemebreaktab.h" + +int unicode_grapheme_break(unicode_char a, unicode_char b) +{ + uint8_t ac=unicode_tab_lookup(a, unicode_indextab, + sizeof(unicode_indextab)/sizeof(unicode_indextab[0]), + unicode_rangetab, + unicode_classtab, + UNICODE_GRAPHEMEBREAK_ANY), + bc=unicode_tab_lookup(b, unicode_indextab, + sizeof(unicode_indextab)/sizeof(unicode_indextab[0]), + unicode_rangetab, + unicode_classtab, + UNICODE_GRAPHEMEBREAK_ANY); + + /* GB1 and GB2 are implied */ + + if (ac == UNICODE_GRAPHEMEBREAK_CR && bc == UNICODE_GRAPHEMEBREAK_LF) + return 0; /* GB3 */ + + + switch (ac) { + case UNICODE_GRAPHEMEBREAK_CR: + case UNICODE_GRAPHEMEBREAK_LF: + case UNICODE_GRAPHEMEBREAK_Control: + return 1; /* GB4 */ + default: + break; + } + + switch (bc) { + case UNICODE_GRAPHEMEBREAK_CR: + case UNICODE_GRAPHEMEBREAK_LF: + case UNICODE_GRAPHEMEBREAK_Control: + return 1; /* GB5 */ + default: + break; + } + + if (ac == UNICODE_GRAPHEMEBREAK_L) + switch (bc) { + case UNICODE_GRAPHEMEBREAK_L: + case UNICODE_GRAPHEMEBREAK_V: + case UNICODE_GRAPHEMEBREAK_LV: + case UNICODE_GRAPHEMEBREAK_LVT: + return 0; /* GB6 */ + } + + if ((ac == UNICODE_GRAPHEMEBREAK_LV || + ac == UNICODE_GRAPHEMEBREAK_V) && + (bc == UNICODE_GRAPHEMEBREAK_V || + bc == UNICODE_GRAPHEMEBREAK_T)) + return 0; /* GB7 */ + + if ((ac == UNICODE_GRAPHEMEBREAK_LVT || + ac == UNICODE_GRAPHEMEBREAK_T) && + bc == UNICODE_GRAPHEMEBREAK_T) + return 0; /* GB8 */ + + if (bc == UNICODE_GRAPHEMEBREAK_Extend) + return 0; /* GB9 */ + + if (bc == UNICODE_GRAPHEMEBREAK_SpacingMark) + return 0; /* GB9a */ + + if (ac == UNICODE_GRAPHEMEBREAK_Prepend) + return 0; /* GB9b */ + + return 1; /* GB10 */ +}