summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--src/utf8.c124
-rw-r--r--src/utf8.h1
2 files changed, 89 insertions, 36 deletions
diff --git a/src/utf8.c b/src/utf8.c
index c798981..e144c72 100644
--- a/src/utf8.c
+++ b/src/utf8.c
@@ -190,14 +190,32 @@ void utf8proc_case_fold(strbuf *dest, const uint8_t *str, int len)
}
}
+// matches anything in the Zs class, plus LF, CR, TAB, FF.
+int utf8proc_is_space(int32_t uc)
+{
+ return (uc == 9 ||
+ uc == 10 ||
+ uc == 12 ||
+ uc == 13 ||
+ uc == 32 ||
+ uc == 160 ||
+ uc == 5760 ||
+ (uc >= 8192 && uc <= 8202) ||
+ uc == 8239 ||
+ uc == 8287 ||
+ uc == 12288);
+}
+
// matches anything in the P[cdefios] classes.
int utf8proc_is_punctuation(int32_t uc)
{
return ((uc >= 33 && uc <= 35) ||
(uc >= 37 && uc <= 42) ||
(uc >= 44 && uc <= 47) ||
- (uc >= 58 && uc <= 59) ||
- (uc >= 63 && uc <= 64) ||
+ uc == 58 ||
+ uc == 59 ||
+ uc == 63 ||
+ uc == 64 ||
(uc >= 91 && uc <= 93) ||
uc == 95 ||
uc == 123 ||
@@ -205,77 +223,98 @@ int utf8proc_is_punctuation(int32_t uc)
uc == 161 ||
uc == 167 ||
uc == 171 ||
- (uc >= 182 && uc <= 183) ||
+ uc == 182 ||
+ uc == 183 ||
uc == 187 ||
uc == 191 ||
uc == 894 ||
uc == 903 ||
(uc >= 1370 && uc <= 1375) ||
- (uc >= 1417 && uc <= 1418) ||
+ uc == 1417 ||
+ uc == 1418 ||
uc == 1470 ||
uc == 1472 ||
uc == 1475 ||
uc == 1478 ||
- (uc >= 1523 && uc <= 1524) ||
- (uc >= 1545 && uc <= 1546) ||
- (uc >= 1548 && uc <= 1549) ||
+ uc == 1523 ||
+ uc == 1524 ||
+ uc == 1545 ||
+ uc == 1546 ||
+ uc == 1548 ||
+ uc == 1549 ||
uc == 1563 ||
- (uc >= 1566 && uc <= 1567) ||
+ uc == 1566 ||
+ uc == 1567 ||
(uc >= 1642 && uc <= 1645) ||
uc == 1748 ||
(uc >= 1792 && uc <= 1805) ||
(uc >= 2039 && uc <= 2041) ||
(uc >= 2096 && uc <= 2110) ||
uc == 2142 ||
- (uc >= 2404 && uc <= 2405) ||
+ uc == 2404 ||
+ uc == 2405 ||
uc == 2416 ||
uc == 2800 ||
uc == 3572 ||
uc == 3663 ||
- (uc >= 3674 && uc <= 3675) ||
+ uc == 3674 ||
+ uc == 3675 ||
(uc >= 3844 && uc <= 3858) ||
uc == 3860 ||
(uc >= 3898 && uc <= 3901) ||
uc == 3973 ||
(uc >= 4048 && uc <= 4052) ||
- (uc >= 4057 && uc <= 4058) ||
+ uc == 4057 ||
+ uc == 4058 ||
(uc >= 4170 && uc <= 4175) ||
uc == 4347 ||
(uc >= 4960 && uc <= 4968) ||
uc == 5120 ||
- (uc >= 5741 && uc <= 5742) ||
- (uc >= 5787 && uc <= 5788) ||
+ uc == 5741 ||
+ uc == 5742 ||
+ uc == 5787 ||
+ uc == 5788 ||
(uc >= 5867 && uc <= 5869) ||
- (uc >= 5941 && uc <= 5942) ||
+ uc == 5941 ||
+ uc == 5942 ||
(uc >= 6100 && uc <= 6102) ||
(uc >= 6104 && uc <= 6106) ||
(uc >= 6144 && uc <= 6154) ||
- (uc >= 6468 && uc <= 6469) ||
- (uc >= 6686 && uc <= 6687) ||
+ uc == 6468 ||
+ uc == 6469 ||
+ uc == 6686 ||
+ uc == 6687 ||
(uc >= 6816 && uc <= 6822) ||
(uc >= 6824 && uc <= 6829) ||
(uc >= 7002 && uc <= 7008) ||
(uc >= 7164 && uc <= 7167) ||
(uc >= 7227 && uc <= 7231) ||
- (uc >= 7294 && uc <= 7295) ||
+ uc == 7294 ||
+ uc == 7295 ||
(uc >= 7360 && uc <= 7367) ||
uc == 7379 ||
(uc >= 8208 && uc <= 8231) ||
(uc >= 8240 && uc <= 8259) ||
(uc >= 8261 && uc <= 8273) ||
(uc >= 8275 && uc <= 8286) ||
- (uc >= 8317 && uc <= 8318) ||
- (uc >= 8333 && uc <= 8334) ||
+ uc == 8317 ||
+ uc == 8318 ||
+ uc == 8333 ||
+ uc == 8334 ||
(uc >= 8968 && uc <= 8971) ||
- (uc >= 9001 && uc <= 9002) ||
+ uc == 9001 ||
+ uc == 9002 ||
(uc >= 10088 && uc <= 10101) ||
- (uc >= 10181 && uc <= 10182) ||
+ uc == 10181 ||
+ uc == 10182 ||
(uc >= 10214 && uc <= 10223) ||
(uc >= 10627 && uc <= 10648) ||
(uc >= 10712 && uc <= 10715) ||
- (uc >= 10748 && uc <= 10749) ||
+ uc == 10748 ||
+ uc == 10749 ||
(uc >= 11513 && uc <= 11516) ||
- (uc >= 11518 && uc <= 11519) ||
+ uc == 11518 ||
+ uc == 11519 ||
uc == 11632 ||
(uc >= 11776 && uc <= 11822) ||
(uc >= 11824 && uc <= 11842) ||
@@ -286,34 +325,44 @@ int utf8proc_is_punctuation(int32_t uc)
uc == 12349 ||
uc == 12448 ||
uc == 12539 ||
- (uc >= 42238 && uc <= 42239) ||
+ uc == 42238 ||
+ uc == 42239 ||
(uc >= 42509 && uc <= 42511) ||
uc == 42611 ||
uc == 42622 ||
(uc >= 42738 && uc <= 42743) ||
(uc >= 43124 && uc <= 43127) ||
- (uc >= 43214 && uc <= 43215) ||
+ uc == 43214 ||
+ uc == 43215 ||
(uc >= 43256 && uc <= 43258) ||
- (uc >= 43310 && uc <= 43311) ||
+ uc == 43310 ||
+ uc == 43311 ||
uc == 43359 ||
(uc >= 43457 && uc <= 43469) ||
- (uc >= 43486 && uc <= 43487) ||
+ uc == 43486 ||
+ uc == 43487 ||
(uc >= 43612 && uc <= 43615) ||
- (uc >= 43742 && uc <= 43743) ||
- (uc >= 43760 && uc <= 43761) ||
+ uc == 43742 ||
+ uc == 43743 ||
+ uc == 43760 ||
+ uc == 43761 ||
uc == 44011 ||
- (uc >= 64830 && uc <= 64831) ||
+ uc == 64830 ||
+ uc == 64831 ||
(uc >= 65040 && uc <= 65049) ||
(uc >= 65072 && uc <= 65106) ||
(uc >= 65108 && uc <= 65121) ||
uc == 65123 ||
uc == 65128 ||
- (uc >= 65130 && uc <= 65131) ||
+ uc == 65130 ||
+ uc == 65131 ||
(uc >= 65281 && uc <= 65283) ||
(uc >= 65285 && uc <= 65290) ||
(uc >= 65292 && uc <= 65295) ||
- (uc >= 65306 && uc <= 65307) ||
- (uc >= 65311 && uc <= 65312) ||
+ uc == 65306 ||
+ uc == 65307 ||
+ uc == 65311 ||
+ uc == 65312 ||
(uc >= 65339 && uc <= 65341) ||
uc == 65343 ||
uc == 65371 ||
@@ -332,10 +381,12 @@ int utf8proc_is_punctuation(int32_t uc)
(uc >= 68409 && uc <= 68415) ||
(uc >= 68505 && uc <= 68508) ||
(uc >= 69703 && uc <= 69709) ||
- (uc >= 69819 && uc <= 69820) ||
+ uc == 69819 ||
+ uc == 69820 ||
(uc >= 69822 && uc <= 69825) ||
(uc >= 69952 && uc <= 69955) ||
- (uc >= 70004 && uc <= 70005) ||
+ uc == 70004 ||
+ uc == 70005 ||
(uc >= 70085 && uc <= 70088) ||
uc == 70093 ||
(uc >= 70200 && uc <= 70205) ||
@@ -343,7 +394,8 @@ int utf8proc_is_punctuation(int32_t uc)
(uc >= 71105 && uc <= 71113) ||
(uc >= 71233 && uc <= 71235) ||
(uc >= 74864 && uc <= 74868) ||
- (uc >= 92782 && uc <= 92783) ||
+ uc == 92782 ||
+ uc == 92783 ||
uc == 92917 ||
(uc >= 92983 && uc <= 92987) ||
uc == 92996 ||
diff --git a/src/utf8.h b/src/utf8.h
index 2474e07..319e39a 100644
--- a/src/utf8.h
+++ b/src/utf8.h
@@ -13,6 +13,7 @@ void utf8proc_encode_char(int32_t uc, cmark_strbuf *buf);
int utf8proc_iterate(const uint8_t *str, int str_len, int32_t *dst);
int utf8proc_charlen(const uint8_t *str, int str_len);
void utf8proc_detab(cmark_strbuf *dest, const uint8_t *line, size_t size);
+int utf8proc_is_space(int32_t uc);
int utf8proc_is_punctuation(int32_t uc);
#ifdef __cplusplus