mirrors/Nim
1
0
Fork 0
mirror of https://github.com/nim-lang/Nim.git synced 2026-02-13 06:43:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

fixed UTF-16 to UTF-8 conversion in widestrs.nim

Browse Source 
the source of problem for issue #3228
This commit is contained in:
jangko
2015-08-20 14:17:46 +07:00
parent 69b32637b1
commit 493dbc8932
1 changed files with 18 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

44
lib/system/widestrs.nim
View File

@@ -119,36 +119,28 @@ proc `$`*(w: WideCString, estimate: int): string =
var i = 0
while w[i].int16 != 0'i16:
var ch = w[i].int
var ch = uint32(cast[uint16](w[i]))
inc i
if ch >=% UNI_SUR_HIGH_START and ch <=% UNI_SUR_HIGH_END:
if ch >= uint32(UNI_SUR_HIGH_START) and ch <= uint32(UNI_SUR_HIGH_END):
# If the 16 bits following the high surrogate are in the source buffer...
let ch2 = w[i].int
# If it's a low surrogate, convert to UTF32:
if ch2 >=% UNI_SUR_LOW_START and ch2 <=% UNI_SUR_LOW_END:
ch = ((ch -% UNI_SUR_HIGH_START) shr halfShift) +%
(ch2 -% UNI_SUR_LOW_START) +% halfBase
inc i
if ch <=% 127:
let ch2 = uint32(cast[uint16](w[i]))
ch = (ch shl halfShift) + ch2 + halfBase
inc i
if ch < 0x80'u32:
result.add chr(ch)
elif ch <=% 0x07FF:
result.add chr((ch shr 6) or 0b110_00000)
result.add chr((ch and ones(6)) or 0b10_000000)
elif ch <=% 0xFFFF:
result.add chr(ch shr 12 or 0b1110_0000)
result.add chr(ch shr 6 and ones(6) or 0b10_0000_00)
result.add chr(ch and ones(6) or 0b10_0000_00)
elif ch <=% 0x0010FFFF:
result.add chr(ch shr 18 or 0b1111_0000)
result.add chr(ch shr 12 and ones(6) or 0b10_0000_00)
result.add chr(ch shr 6 and ones(6) or 0b10_0000_00)
result.add chr(ch and ones(6) or 0b10_0000_00)
elif ch < 0x800'u32:
result.add chr((ch shr 6) or 0xc0)
result.add chr((ch and 0x3f) or 0x80)
elif ch < 0x10000'u32:
result.add chr((ch shr 12) or 0xe0)
result.add chr(((ch shr 6) and 0x3f) or 0x80)
result.add chr((ch and 0x3f) or 0x80)
else:
# replacement char:
result.add chr(0xFFFD shr 12 or 0b1110_0000)
result.add chr(0xFFFD shr 6 and ones(6) or 0b10_0000_00)
result.add chr(0xFFFD and ones(6) or 0b10_0000_00)
result.add chr((ch shr 18) or 0xf0)
result.add chr(((ch shr 12) and 0x3f) or 0x80)
result.add chr(((ch shr 6) and 0x3f) or 0x80)
result.add chr((ch and 0x3f) or 0x80)
proc `$`*(s: WideCString): string =
result = s $ 80