jyapayne/Nim
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

cleaned up twchartoutf8 test

Browse source
This commit is contained in:
Araq 2015-08-21 21:40:24 +02:00
commit dd2a0ec431
1 changed files with 91 additions and 87 deletions
Download patch file Download diff file Expand all files Collapse all files

94
tests/stdlib/twchartoutf8.nim
View file

@ -1,35 +1,39 @@
discard """
output: '''OK'''
"""
#assume WideCharToMultiByte always produce correct result #assume WideCharToMultiByte always produce correct result
#windows only #windows only
when not defined(windows): when not defined(windows):
{.error: "windows only".} echo "OK"
else:
{.push gcsafe.}
{.push gcsafe.} const CP_UTF8 = 65001'i32
const CP_UTF8 = 65001'i32 type
type
LPBOOL = ptr int32 LPBOOL = ptr int32
LPWCSTR = ptr uint16 LPWCSTR = ptr uint16
proc WideCharToMultiByte*(CodePage: int32, dwFlags: int32, proc WideCharToMultiByte*(CodePage: int32, dwFlags: int32,
lpWideCharStr: LPWCSTR, cchWideChar: int32, lpWideCharStr: LPWCSTR, cchWideChar: int32,
lpMultiByteStr: cstring, cchMultiByte: int32, lpMultiByteStr: cstring, cchMultiByte: int32,
lpDefaultChar: cstring, lpUsedDefaultChar: LPBOOL): int32{. lpDefaultChar: cstring, lpUsedDefaultChar: LPBOOL): int32{.
stdcall, dynlib: "kernel32", importc: "WideCharToMultiByte".} stdcall, dynlib: "kernel32", importc: "WideCharToMultiByte".}
{.pop.} {.pop.}
proc convertToUTF8(wc: WideCString, wclen: int32): string = proc convertToUTF8(wc: WideCString, wclen: int32): string =
let size = WideCharToMultiByte(CP_UTF8, 0'i32, cast[LPWCSTR](addr(wc[0])), wclen, let size = WideCharToMultiByte(CP_UTF8, 0'i32, cast[LPWCSTR](addr(wc[0])), wclen,
cstring(nil), 0'i32, cstring(nil), LPBOOL(nil)) cstring(nil), 0'i32, cstring(nil), LPBOOL(nil))
result = newString(size) result = newString(size)
let res = WideCharToMultiByte(CP_UTF8, 0'i32, cast[LPWCSTR](addr(wc[0])), wclen, let res = WideCharToMultiByte(CP_UTF8, 0'i32, cast[LPWCSTR](addr(wc[0])), wclen,
cstring(result), size, cstring(nil), LPBOOL(nil)) cstring(result), size, cstring(nil), LPBOOL(nil))
result[size] = chr(0) result[size] = chr(0)
assert size == res doAssert size == res
proc testCP(wc: WideCString, lo, hi: int) = proc testCP(wc: WideCString, lo, hi: int) =
var x = 0 var x = 0
let chunk = 1024 let chunk = 1024
for i in lo..hi: for i in lo..hi:
@ -38,12 +42,12 @@ proc testCP(wc: WideCString, lo, hi: int) =
wc[x] = Utf16Char(0) wc[x] = Utf16Char(0)
var a = convertToUTF8(wc, int32(x)) var a = convertToUTF8(wc, int32(x))
var b = wc $ chunk var b = wc $ chunk
assert a == b doAssert a == b
x = 0 x = 0
inc x inc x
proc testCP2(wc: WideCString, lo, hi: int) = proc testCP2(wc: WideCString, lo, hi: int) =
assert((lo >= 0x10000) and (hi <= 0x10FFFF)) doAssert((lo >= 0x10000) and (hi <= 0x10FFFF))
var x = 0 var x = 0
let chunk = 1024 let chunk = 1024
for i in lo..hi: for i in lo..hi:
@ -58,49 +62,49 @@ proc testCP2(wc: WideCString, lo, hi: int) =
wc[x] = Utf16Char(0) wc[x] = Utf16Char(0)
var a = convertToUTF8(wc, int32(x)) var a = convertToUTF8(wc, int32(x))
var b = wc $ chunk var b = wc $ chunk
assert a == b doAssert a == b
x = 0 x = 0
#RFC-2781 "UTF-16, an encoding of ISO 10646" #RFC-2781 "UTF-16, an encoding of ISO 10646"
var wc: WideCString var wc: WideCString
unsafeNew(wc, 1024 * 4 + 2) unsafeNew(wc, 1024 * 4 + 2)
#U+0000 to U+D7FF #U+0000 to U+D7FF
#skip the U+0000 #skip the U+0000
wc.testCP(1, 0xD7FF) wc.testCP(1, 0xD7FF)
#U+E000 to U+FFFF #U+E000 to U+FFFF
wc.testCP(0xE000, 0xFFFF) wc.testCP(0xE000, 0xFFFF)
#U+10000 to U+10FFFF #U+10000 to U+10FFFF
wc.testCP2(0x10000, 0x10FFFF) wc.testCP2(0x10000, 0x10FFFF)
#invalid UTF-16 #invalid UTF-16
const const
b = "\xEF\xBF\xBD" b = "\xEF\xBF\xBD"
c = "\xEF\xBF\xBF" c = "\xEF\xBF\xBF"
wc[0] = cast[Utf16Char](0xDC00) wc[0] = cast[Utf16Char](0xDC00)
wc[1] = Utf16Char(0) wc[1] = Utf16Char(0)
var a = $wc var a = $wc
assert a == b doAssert a == b
wc[0] = cast[Utf16Char](0xFFFF) wc[0] = cast[Utf16Char](0xFFFF)
wc[1] = cast[Utf16Char](0xDC00) wc[1] = cast[Utf16Char](0xDC00)
wc[2] = Utf16Char(0) wc[2] = Utf16Char(0)
a = $wc a = $wc
assert a == c & b doAssert a == c & b
wc[0] = cast[Utf16Char](0xD800) wc[0] = cast[Utf16Char](0xD800)
wc[1] = Utf16Char(0) wc[1] = Utf16Char(0)
a = $wc a = $wc
assert a == b doAssert a == b
wc[0] = cast[Utf16Char](0xD800) wc[0] = cast[Utf16Char](0xD800)
wc[1] = cast[Utf16Char](0xFFFF) wc[1] = cast[Utf16Char](0xFFFF)
wc[2] = Utf16Char(0) wc[2] = Utf16Char(0)
a = $wc a = $wc
assert a == b & c doAssert a == b & c
echo "OK" echo "OK"