diff options
| -rw-r--r-- | Makefile | 6 | ||||
| -rw-r--r-- | convert-to-joined.c | 4 | ||||
| -rw-r--r-- | libcharconv.h | 5 | ||||
| -rw-r--r-- | libcharconv_joined.c | 68 | ||||
| -rw-r--r-- | libcharconv_latin.c | 6 |
5 files changed, 87 insertions, 2 deletions
@@ -62,7 +62,8 @@ BIN =\ convert-to-yijing-hexagrams\ convert-to-vulgar-fractions\ convert-to-flipped\ - convert-to-overlaid + convert-to-overlaid\ + convert-to-joined LIBOBJ =\ libcharconv_decode_utf8_.o\ @@ -113,7 +114,8 @@ LIBOBJ =\ libcharconv_yijing_hexagrams.o\ libcharconv_vulgar_fractions.o\ libcharconv_flipped.o\ - libcharconv_overlaid.o + libcharconv_overlaid.o\ + libcharconv_joined.o LOBJ = $(LIBOBJ:.o=.lo) diff --git a/convert-to-joined.c b/convert-to-joined.c new file mode 100644 index 0000000..71295bf --- /dev/null +++ b/convert-to-joined.c @@ -0,0 +1,4 @@ +/* See LICENSE file for copyright and license details. */ +#include "common.h" + +SIMPLE(libcharconv_joined) diff --git a/libcharconv.h b/libcharconv.h index 289f9e8..731c7ce 100644 --- a/libcharconv.h +++ b/libcharconv.h @@ -338,6 +338,11 @@ LIBCHARCONV_FUNC_(libcharconv_flipped); */ LIBCHARCONV_FUNC_(libcharconv_overlaid); +/** + * Join characters horizontally + */ +LIBCHARCONV_FUNC_(libcharconv_joined); + #undef LIBCHARCONV_FUNC_ #endif diff --git a/libcharconv_joined.c b/libcharconv_joined.c new file mode 100644 index 0000000..72377a2 --- /dev/null +++ b/libcharconv_joined.c @@ -0,0 +1,68 @@ +/* See LICENSE file for copyright and license details. */ +#include "lib-common.h" + + +static struct { + uint_least32_t a; + uint_least32_t b; + uint_least32_t to; +} pairs[] = { + {(uint_least32_t)'?', (uint_least32_t)'!', 0x2048}, + {(uint_least32_t)'?', (uint_least32_t)'?', 0x2047}, + {(uint_least32_t)'!', (uint_least32_t)'!', 0x203C}, + {(uint_least32_t)'!', (uint_least32_t)'?', 0x2049} +}; + + +enum libcharconv_result +libcharconv_joined(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp) +{ + uint_least32_t a, b; + size_t i, alen, blen; + *n = 0; + while (slen) { + alen = libcharconv_decode_utf8_(s, slen, &a); + if (alen > slen) { + if (*n) + goto no_conv; + return LIBCHARCONV_INDETERMINATE; + } + if (!alen) { + *n += 1u; + slen -= 1u; + s = &s[1]; + continue; + } + + for (i = 0u; i < sizeof(pairs) / sizeof(*pairs); i++) { + if (a != pairs[i].a) + continue; + if (*n) + goto no_conv; + if (slen == alen) + return LIBCHARCONV_INDETERMINATE; + blen = libcharconv_decode_utf8_(&s[alen], slen - alen, &b); + if (blen > slen) + return LIBCHARCONV_INDETERMINATE; + if (!blen) + goto no_conv; + if (b == pairs[i].b) + goto conv; + } + + *n += alen; + s = &s[alen]; + slen -= alen; + } +no_conv: + return LIBCHARCONV_NO_CONVERT; + +conv: + if (*n) + goto no_conv; + if (*ncp) + *cp = pairs[i].to; + *n += alen + blen; + *ncp = 1u; + return LIBCHARCONV_CONVERTED; +} diff --git a/libcharconv_latin.c b/libcharconv_latin.c index ee975e1..a4b9059 100644 --- a/libcharconv_latin.c +++ b/libcharconv_latin.c @@ -630,6 +630,12 @@ libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, siz case UINT32_C(0x203D): c1 = '!'; c2 = '?'; goto conv2; case UINT32_C(0x2E18): cp1 = UINT32_C(0xBF); cp2 = UINT32_C(0xA1); goto conv2cp; + /* joined */ + case UINT32_C(0x2048): c1 = '?'; c2 = '!'; goto conv2; + case UINT32_C(0x2047): c1 = '?'; c2 = '?'; goto conv2; + case UINT32_C(0x203C): c1 = '!'; c2 = '!'; goto conv2; + case UINT32_C(0x2049): c1 = '!'; c2 = '?'; goto conv2; + default: no_match: *n += clen; |