diff options
| author | Xavier Roche <xroche@users.noreply.github.com> | 2014-05-27 16:35:57 +0000 |
|---|---|---|
| committer | Xavier Roche <xroche@users.noreply.github.com> | 2014-05-27 16:35:57 +0000 |
| commit | fdcff762f661ba37e48604039f9e17c510ad18d6 (patch) | |
| tree | 3e779847225750c26560f1629c8b32c4a5a9dddd | |
| parent | c3987aca717d438aff9b9d9378a3bc69748400d1 (diff) | |
Rewritten UTF8 writer to avoir spurious GCC 4.8.3 warnings.
| -rw-r--r-- | src/htscharset.c | 77 |
1 files changed, 35 insertions, 42 deletions
diff --git a/src/htscharset.c b/src/htscharset.c index 4137d65..50be2d2 100644 --- a/src/htscharset.c +++ b/src/htscharset.c @@ -723,43 +723,21 @@ static unsigned int nlz8(unsigned char x) { 26 U+200000 U+3FFFFFF 5 111110xx 31 U+4000000 U+7FFFFFFF 6 1111110x */ -#define ADD_SEQ(UC, BITS, EMITTER) do { \ - /* number of data bits in first octet */ \ - const unsigned int bits = BITS % 6; \ - /* shift for first octet */ \ - const unsigned int shift0 = BITS - bits; \ - /* first octet */ \ - const unsigned char lead = \ - /* leading bits */ \ - ( 0xff ^ ( ( 1 << ( bits + 1 ) ) - 1 ) ) \ - /* encoded bits */ \ - | ( ( (UC) >> shift0 ) & ( ( 1 << ( bits + 1 ) ) - 1 ) ) \ - ; \ - /* further bytes are encoding 6 bits */ \ - const unsigned char second = \ - 0x80 | ( ( (UC) >> ( shift0 - 6 ) ) & 0x3f ); \ - EMITTER(lead); \ - EMITTER(second); \ - if (BITS > 6*2) { \ - const unsigned char next = \ - 0x80 | ( ( (UC) >> ( shift0 - 6*2 ) ) & 0x3f ); \ - EMITTER(next); \ - if (BITS > 6*3) { \ - const unsigned char next = \ - 0x80 | ( ( (UC) >> ( shift0 - 6*3 ) ) & 0x3f ); \ - EMITTER(next); \ - if (BITS > 6*4) { \ - const unsigned char next = \ - 0x80 | ( ( (UC) >> ( shift0 - 6*4 ) ) & 0x3f ); \ - EMITTER(next); \ - if (BITS > 6*5) { \ - const unsigned char next = \ - 0x80 | ( ( (UC) >> ( shift0 - 6*5 ) ) & 0x3f ); \ - EMITTER(next); \ - } \ - } \ - } \ - } \ +#define ADD_FIRST_SEQ(UC, LEN, EMITTER) do { \ + /* first octet */ \ + const unsigned char lead = \ + /* leading bits: LEN "1" bits */ \ + ~ ( ( 1 << (unsigned) ( 8 - LEN ) ) - 1 ) \ + /* encoded bits */ \ + | ( (UC) >> (unsigned) ( ( LEN - 1 ) * 6 ) ); \ + EMITTER(lead); \ + } while(0) + +#define ADD_NEXT_SEQ(UC, SHIFT, EMITTER) do { \ + /* further bytes are encoding 6 bits */ \ + const unsigned char next = \ + 0x80 | ( ( (UC) >> SHIFT ) & 0x3f ); \ + EMITTER(next); \ } while(0) /* UC is a constant. EMITTER is a macro function taking an unsigned int. */ @@ -767,15 +745,30 @@ static unsigned int nlz8(unsigned char x) { if ((UC) < 0x80) { \ EMITTER(((unsigned char) (UC))); \ } else if ((UC) < 0x0800) { \ - ADD_SEQ(UC, 11, EMITTER); \ + ADD_FIRST_SEQ(UC, 2, EMITTER); \ + ADD_NEXT_SEQ(UC, 0, EMITTER); \ } else if ((UC) < 0x10000) { \ - ADD_SEQ(UC, 16, EMITTER); \ + ADD_FIRST_SEQ(UC, 3, EMITTER); \ + ADD_NEXT_SEQ(UC, 6, EMITTER); \ + ADD_NEXT_SEQ(UC, 0, EMITTER); \ } else if ((UC) < 0x200000) { \ - ADD_SEQ(UC, 21, EMITTER); \ + ADD_FIRST_SEQ(UC, 4, EMITTER); \ + ADD_NEXT_SEQ(UC, 12, EMITTER); \ + ADD_NEXT_SEQ(UC, 6, EMITTER); \ + ADD_NEXT_SEQ(UC, 0, EMITTER); \ } else if ((UC) < 0x4000000) { \ - ADD_SEQ(UC, 26, EMITTER); \ + ADD_FIRST_SEQ(UC, 5, EMITTER); \ + ADD_NEXT_SEQ(UC, 18, EMITTER); \ + ADD_NEXT_SEQ(UC, 12, EMITTER); \ + ADD_NEXT_SEQ(UC, 6, EMITTER); \ + ADD_NEXT_SEQ(UC, 0, EMITTER); \ } else { \ - ADD_SEQ(UC, 31, EMITTER); \ + ADD_FIRST_SEQ(UC, 6, EMITTER); \ + ADD_NEXT_SEQ(UC, 24, EMITTER); \ + ADD_NEXT_SEQ(UC, 18, EMITTER); \ + ADD_NEXT_SEQ(UC, 12, EMITTER); \ + ADD_NEXT_SEQ(UC, 6, EMITTER); \ + ADD_NEXT_SEQ(UC, 0, EMITTER); \ } \ } while(0) |