diff options
author | Nicholas Noll <nbnoll@eml.cc> | 2021-10-26 21:32:55 -0700 |
---|---|---|
committer | Nicholas Noll <nbnoll@eml.cc> | 2021-10-26 21:51:49 -0700 |
commit | 189b9e23edfe60b7e82c4c7b6071a3f98799653a (patch) | |
tree | 77c1f32726446deb298a0d2e6389358bfe16bd32 /sys/libunicode | |
parent | 29b56ef4e4113bcd091b19d6926f18814162ca53 (diff) |
fix(unicode): renamed functions to be easier to understand their functions
Diffstat (limited to 'sys/libunicode')
-rw-r--r-- | sys/libunicode/canfit.c | 20 | ||||
-rw-r--r-- | sys/libunicode/decode.c (renamed from sys/libunicode/bytetorune.c) | 2 | ||||
-rw-r--r-- | sys/libunicode/encode.c (renamed from sys/libunicode/runetobyte.c) | 2 | ||||
-rw-r--r-- | sys/libunicode/find.c | 30 | ||||
-rw-r--r-- | sys/libunicode/findlast.c (renamed from sys/libunicode/findlastrune.c) | 15 | ||||
-rw-r--r-- | sys/libunicode/findrune.c | 26 | ||||
-rw-r--r-- | sys/libunicode/fullrune.c | 18 | ||||
-rw-r--r-- | sys/libunicode/internal.h | 12 | ||||
-rw-r--r-- | sys/libunicode/len.c | 21 | ||||
-rw-r--r-- | sys/libunicode/rules.mk | 14 | ||||
-rw-r--r-- | sys/libunicode/runelen.c | 2 |
11 files changed, 101 insertions, 61 deletions
diff --git a/sys/libunicode/canfit.c b/sys/libunicode/canfit.c new file mode 100644 index 0000000..d44c9e6 --- /dev/null +++ b/sys/libunicode/canfit.c @@ -0,0 +1,20 @@ +#include "internal.h" + +/* returns 1 if string of length n is long enough to be decoded */ +int +utf8·canfit(byte* s, int n) +{ + int i; + rune c; + + if(n <= 0) return 0; + c = *(ubyte*) s; + if(c < Tx) return 1; + + for(i = 3; i < UTFmax + 1; i++){ + if(c < Tbyte(i)) + return n >= i - 1; + } + + return n >= UTFmax; +} diff --git a/sys/libunicode/bytetorune.c b/sys/libunicode/decode.c index fe41e7a..79271f2 100644 --- a/sys/libunicode/bytetorune.c +++ b/sys/libunicode/decode.c @@ -1,7 +1,7 @@ #include "internal.h" int -utf8·bytetorune(rune* r, byte* s) +utf8·decode(byte *s, rune* r) { int c[UTFmax], i; rune l; diff --git a/sys/libunicode/runetobyte.c b/sys/libunicode/encode.c index 27f252b..8f4d212 100644 --- a/sys/libunicode/runetobyte.c +++ b/sys/libunicode/encode.c @@ -1,7 +1,7 @@ #include "internal.h" int -utf8·runetobyte(byte* s, rune* r) +utf8·encode(rune* r, byte* s) { int i, j; rune c; diff --git a/sys/libunicode/find.c b/sys/libunicode/find.c new file mode 100644 index 0000000..659ab5b --- /dev/null +++ b/sys/libunicode/find.c @@ -0,0 +1,30 @@ +#include "internal.h" + +byte* +utf8·find(byte* s, rune c) +{ + long c1; + rune r; + int n; + + if(c < RuneSync) + return strchr(s, c); + + for(;;){ + c1 = *(ubyte*)s; + if(c1 < RuneSelf) { + if(c1 == 0) return nil; + if(c1 == c) return s; + s++; + continue; + } + n = utf8·decode(s, &r); + + if(r == c) + return s; + + s += n; + } + + return nil; +} diff --git a/sys/libunicode/findlastrune.c b/sys/libunicode/findlast.c index 0dc5032..3a4ed38 100644 --- a/sys/libunicode/findlastrune.c +++ b/sys/libunicode/findlast.c @@ -1,27 +1,28 @@ #include "internal.h" byte* -utf8·findlastrune(byte* s, long c) +utf8·findlast(byte* s, rune c) { long c1; rune r; byte *l; - if (c < RuneSync) + if(c < RuneSync) return strrchr(s, c); l = nil; for(;;){ c1 = *(ubyte*)s; - if (c1 < RuneSelf) { - if (c1 == 0) return l; - if (c1 == c) l = s; + if(c1 < RuneSelf) { + if(c1 == 0) return l; + if(c1 == c) l = s; s++; continue; } - c1 = utf8·bytetorune(&r, s); - if (r == c) + c1 = utf8·decode(s, &r); + if(r == c) l = s; + s += c1; } diff --git a/sys/libunicode/findrune.c b/sys/libunicode/findrune.c deleted file mode 100644 index 97edc3c..0000000 --- a/sys/libunicode/findrune.c +++ /dev/null @@ -1,26 +0,0 @@ -#include "internal.h" - -byte* -utf8·findrune(byte* s, long c) -{ - long c1; - rune r; - int n; - - if (c < RuneSync) return strchr(s, c); - - for (;;) { - c1 = *(ubyte*)s; - if (c1 < RuneSelf) { - if (c1 == 0) return nil; - if (c1 == c) return s; - s++; - continue; - } - n = utf8·bytetorune(&r, s); - if (r == c) return s; - s += n; - } - - return nil; -} diff --git a/sys/libunicode/fullrune.c b/sys/libunicode/fullrune.c deleted file mode 100644 index e5cf314..0000000 --- a/sys/libunicode/fullrune.c +++ /dev/null @@ -1,18 +0,0 @@ -#include "internal.h" - -int -utf8·fullrune(byte* s, int n) -{ - int i; - rune c; - - if (n <= 0) return 0; - c = *(ubyte*) s; - if (c < Tx) return 1; - - for (i = 3; i < UTFmax + 1; i++) { - if (c < Tbyte(i)) return n >= i - 1; - } - - return n >= UTFmax; -} diff --git a/sys/libunicode/internal.h b/sys/libunicode/internal.h index bf376b2..31ed2ae 100644 --- a/sys/libunicode/internal.h +++ b/sys/libunicode/internal.h @@ -4,8 +4,18 @@ #include <base.h> #include <libunicode.h> +/* UTF-8 code + * 1 byte: + * 0xxxxxxx + * 2 byte: + * 110xxxxx 10xxxxxx + * 3 byte: + * 1110xxxx 10xxxxxx 10xxxxxx + * 4 byte: + * 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx + */ #define Bit(i) (7-(i)) -/* N 0's preceded by i 1's e.g. T(Bit(2)) is 1100 0000 */ +/* 0's preceded by i 1's e.g. T(Bit(2)) is 1100 0000 */ #define Tbyte(i) (((1 << (Bit(i)+1))-1) ^ 0xFF) /* 0000 0000 0000 0111 1111 1111 */ #define RuneX(i) ((1 << (Bit(i) + ((i)-1)*Bitx))-1) diff --git a/sys/libunicode/len.c b/sys/libunicode/len.c new file mode 100644 index 0000000..c461bf3 --- /dev/null +++ b/sys/libunicode/len.c @@ -0,0 +1,21 @@ +#include "internal.h" + +int +utf8·len(char *s) +{ + int c; + long n; + rune r; + + n = 0; + for(;;){ + c = *(uchar*)s; + if(c < RuneSelf) { + if(c == 0) + return n; + s++; + } else + s += utf8·decode(s, &r); + n++; + } +} diff --git a/sys/libunicode/rules.mk b/sys/libunicode/rules.mk index fb83819..100697b 100644 --- a/sys/libunicode/rules.mk +++ b/sys/libunicode/rules.mk @@ -3,12 +3,13 @@ include share/push.mk UNICODE = 14.0.0 SRCS_$(d) := \ - $(d)/runetobyte.c \ - $(d)/bytetorune.c \ - $(d)/findrune.c \ - $(d)/findlastrune.c \ - $(d)/fullrune.c \ + $(d)/encode.c \ + $(d)/decode.c \ + $(d)/find.c \ + $(d)/findlast.c \ + $(d)/canfit.c \ $(d)/runelen.c \ + $(d)/len.c \ $(d)/runetype-$(UNICODE).c LIBS_$(d) := $(d)/libunicode.a @@ -16,13 +17,14 @@ LIBS_$(d) := $(d)/libunicode.a include share/paths.mk $(d)/vendor/UnicodeData-$(UNICODE).txt: + @echo "GET UnicodeData.txt";\ curl https://www.unicode.org/Public/$(UNICODE)/ucd/UnicodeData.txt > $@ $(d)/vendor/mkrunetype: $(d)/vendor/mkrunetype.c $(OBJ_DIR)/sys/base/base.a $(COMPLINK) $(d)/runetype-$(UNICODE).c: $(d)/vendor/UnicodeData-$(UNICODE).txt $(d)/vendor/mkrunetype - $(dir $@)/vendor/mkrunetype $< > $@ + @$(dir $@)vendor/mkrunetype $< > $@ GENS += $(d)/vendor/mkrunetype diff --git a/sys/libunicode/runelen.c b/sys/libunicode/runelen.c index 682400c..dac7f15 100644 --- a/sys/libunicode/runelen.c +++ b/sys/libunicode/runelen.c @@ -4,5 +4,5 @@ int utf8·runelen(rune r) { byte s[10]; - return utf8·runetobyte(s, &r); + return utf8·encode(&r, s); } |