From 51e92f0c8e7b50c684287bea1a55edbde128053f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Marc=20Andr=C3=A9=20Tanner?= Date: Mon, 23 Nov 2015 11:10:38 +0100 Subject: text: introduce functions to iterate over graphemes They currently consider any character for which wcwidth(3) return 0 as a combining character. --- text.c | 57 +++++++++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 55 insertions(+), 2 deletions(-) (limited to 'text.c') diff --git a/text.c b/text.c index 1e1c968..2de96b8 100644 --- a/text.c +++ b/text.c @@ -20,6 +20,7 @@ #include #include #include +#include #include #include #include @@ -1333,7 +1334,7 @@ bool text_iterator_byte_prev(Iterator *it, char *b) { return true; } -bool text_iterator_char_next(Iterator *it, char *c) { +bool text_iterator_codepoint_next(Iterator *it, char *c) { while (text_iterator_byte_next(it, NULL)) { if (ISUTF8(*it->text)) { if (c) @@ -1344,7 +1345,7 @@ bool text_iterator_char_next(Iterator *it, char *c) { return false; } -bool text_iterator_char_prev(Iterator *it, char *c) { +bool text_iterator_codepoint_prev(Iterator *it, char *c) { while (text_iterator_byte_prev(it, NULL)) { if (ISUTF8(*it->text)) { if (c) @@ -1355,6 +1356,58 @@ bool text_iterator_char_prev(Iterator *it, char *c) { return false; } +bool text_iterator_char_next(Iterator *it, char *c) { + if (!text_iterator_codepoint_next(it, c)) + return false; + mbstate_t ps = { 0 }; + for (;;) { + char buf[MB_CUR_MAX]; + size_t len = text_bytes_get(it->piece->text, it->pos, sizeof buf, buf); + wchar_t wc; + size_t wclen = mbrtowc(&wc, buf, len, &ps); + if (wclen == (size_t)-1 && errno == EILSEQ) { + return true; + } else if (wclen == (size_t)-2) { + return false; + } else if (wclen == 0) { + return true; + } else { + int width = wcwidth(wc); + if (width != 0) + return true; + if (!text_iterator_codepoint_next(it, c)) + return false; + } + } + return true; +} + +bool text_iterator_char_prev(Iterator *it, char *c) { + if (!text_iterator_codepoint_prev(it, c)) + return false; + for (;;) { + char buf[MB_CUR_MAX]; + size_t len = text_bytes_get(it->piece->text, it->pos, sizeof buf, buf); + wchar_t wc; + mbstate_t ps = { 0 }; + size_t wclen = mbrtowc(&wc, buf, len, &ps); + if (wclen == (size_t)-1 && errno == EILSEQ) { + return true; + } else if (wclen == (size_t)-2) { + return false; + } else if (wclen == 0) { + return true; + } else { + int width = wcwidth(wc); + if (width != 0) + return true; + if (!text_iterator_codepoint_prev(it, c)) + return false; + } + } + return true; +} + bool text_byte_get(Text *txt, size_t pos, char *buf) { return text_bytes_get(txt, pos, 1, buf); } -- cgit v1.2.3