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

fix changing case for Unicode characters which change size (#11271)

Browse source 
Some Unicode characters don't have the same byte length
in their upper- and lower-case variants.
This commit is contained in:
Miran 2019-05-20 09:18:47 +02:00 • committed by Andreas Rumpf
commit a63c2a25d0
1 changed files with 16 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

25
lib/pure/unicode.nim
View file

@ -656,13 +656,12 @@ template convertRune(s, runeProc) =
result = newString(len(s)) result = newString(len(s))
var var
i = 0 i = 0
lastIndex = 0 resultIndex = 0
rune: Rune rune: Rune
while i < len(s): while i < len(s):
lastIndex = i
fastRuneAt(s, i, rune, doInc=true) fastRuneAt(s, i, rune, doInc=true)
rune = runeProc(rune) rune = runeProc(rune)
rune.fastToUTF8Copy(result, lastIndex) fastToUTF8Copy(rune, result, resultIndex, doInc=true)
proc toUpper*(s: string): string {.noSideEffect, procvar, proc toUpper*(s: string): string {.noSideEffect, procvar,
rtl, extern: "nuc$1Str".} = rtl, extern: "nuc$1Str".} =
@ -689,17 +688,16 @@ proc swapCase*(s: string): string {.noSideEffect, procvar,
var var
i = 0 i = 0
lastIndex = 0 resultIndex = 0
rune: Rune rune: Rune
result = newString(len(s)) result = newString(len(s))
while i < len(s): while i < len(s):
lastIndex = i
fastRuneAt(s, i, rune) fastRuneAt(s, i, rune)
if rune.isUpper(): if rune.isUpper():
rune = rune.toLower() rune = rune.toLower()
elif rune.isLower(): elif rune.isLower():
rune = rune.toUpper() rune = rune.toUpper()
rune.fastToUTF8Copy(result, lastIndex) fastToUTF8Copy(rune, result, resultIndex, doInc=true)
proc capitalize*(s: string): string {.noSideEffect, procvar, proc capitalize*(s: string): string {.noSideEffect, procvar,
rtl, extern: "nuc$1".} = rtl, extern: "nuc$1".} =
@ -779,20 +777,19 @@ proc title*(s: string): string {.noSideEffect, procvar,
var var
i = 0 i = 0
lastIndex = 0 resultIndex = 0
rune: Rune rune: Rune
result = newString(len(s)) result = newString(len(s))
var firstRune = true var firstRune = true
while i < len(s): while i < len(s):
lastIndex = i
fastRuneAt(s, i, rune) fastRuneAt(s, i, rune)
if not rune.isWhiteSpace() and firstRune: if not rune.isWhiteSpace() and firstRune:
rune = rune.toUpper() rune = rune.toUpper()
firstRune = false firstRune = false
elif rune.isWhiteSpace(): elif rune.isWhiteSpace():
firstRune = true firstRune = true
rune.fastToUTF8Copy(result, lastIndex) fastToUTF8Copy(rune, result, resultIndex, doInc=true)
iterator runes*(s: string): Rune = iterator runes*(s: string): Rune =
@ -1468,3 +1465,13 @@ when isMainModule:
doAssert alignLeft("1232", 6) == "1232 " doAssert alignLeft("1232", 6) == "1232 "
doAssert alignLeft("1232", 6, '#'.Rune) == "1232##" doAssert alignLeft("1232", 6, '#'.Rune) == "1232##"
doAssert alignLeft("1232", 6, "×".asRune) == "1232××" doAssert alignLeft("1232", 6, "×".asRune) == "1232××"
block differentSizes:
# upper and lower variants have different number of bytes
doAssert toLower("AẞC") == "aßc"
doAssert toLower("ȺẞCD") == "ⱥßcd"
doAssert toUpper("ⱥbc") == "ȺBC"
doAssert toUpper("rsⱦuv") == "RSȾUV"
doAssert swapCase("ⱥbCd") == "ȺBcD"
doAssert swapCase("XyꟆaB") == "xYᶎAb"
doAssert swapCase("aᵹcᲈd") == "AꝽCꙊD"