_WinAPI_MultiByteToWideChar() String or Structure

[czardas]

Here is a script which removes diacritic marks. It appears to work, but I'm not entirely sure what is happening. There was a certain amount of trial and error involved in writing this, so I would like someone with more experience to look at it. It feels like a clumsy technique I'm using. Diacritics should be removed from both extended ansi and from unicode. I couldn't find a way to do this without calling _WinAPI_MultiByteToWideChar twice.

expandcollapse popup

#include <WinAPI.au3>

#region - Example
Local $sTestString = ""
;For $i = 192 To 255
;    $sTestString &= Chr($i) ; extended latin from Windows 1252 code page
;Next

For $i = 256 To 382
    $sTestString &= ChrW($i) ; Extended latin alpha characters
Next

Local $newString = _StripDiacriticMarks($sTestString)
MsgBox(0, "", $sTestString & @LF & @LF & $newString)
#endregion

Func _StripDiacriticMarks($sText)
    If Not IsString($sText) Then Return SetError(1, 0, $sText)

    Local $sCurrChar, $sSplitChar, $sElement, $sNewString = ""
    For $i = 1 To StringLen($sText)
        $sCurrChar = StringMid($sText, $i, 1)

        $sSplitChar = _WinAPI_MultiByteToWideChar($sCurrChar, 3, $MB_COMPOSITE)
        $sElement = DllStructGetData($sSplitChar, 1)
        If StringIsAlpha($sElement) Then
            $sCurrChar = $sElement

        ElseIf DllStructGetSize($sSplitChar) > 4 Then
            $sSplitChar = _WinAPI_MultiByteToWideChar($sCurrChar, 3, $MB_COMPOSITE, True)

            For $j = 1 To Stringlen($sSplitChar)
                $sElement = StringMid($sSplitChar, $j, 1)
                If StringIsAlpha($sElement) Then
                    $sCurrChar = $sElement
                    ExitLoop
                EndIf
            Next
        EndIf
        $sNewString &= $sCurrChar
    Next
    Return $sNewString
EndFunc

I'm sure I don't need all this code.

[guinness]

I personally can't see any other way, but then I won't pretend that I'm an expert in character encoding. This is where jchd and trancexx come into their element.

[czardas]

Okay thanks for the responce. I imagine I'll eventually figure out what's best. I'll use this until that time comes, unless someone can improve it. At least it's working AFAIK. I think it's pretty useful for (catch all) search features. Most people type the keys available on their keyboard, and potentially valid results are often missed because of correct (or incorrect) diacritic mark interference.

Edited March 27, 2013 by czardas

[JohnOne]

There are some charw that do not work for me, including

$sTestString = ChrW(272) and $sTestString = ChrW(338)

Edited March 27, 2013 by JohnOne

[czardas]

There are some charw that do not work for me, including

$sTestString = ChrW(272) and $sTestString = ChrW(338)

I noticed that myself. I'm not sure why though. Actually those two are okay. One is a letter in itself and the other is a ligature. I was thinking you meant some others. There are one or two others that seem strange to me:

Here's the main culprit.

Ŀ

Edited March 27, 2013 by czardas

[JohnOne]

I imagine you already done this but if you change flag in first call to _WinAPI_MultiByteToWideChar from 3 to 1, something different is returned.

[czardas]

How very true! >_

[trancexx]

You can try with something like this as alternative:

Func _StripDiacriticMarks($sText)
    Local $oMessageStream = ObjCreate("ADODB.Stream")
    If @error Then Return SetError(1, 0, $sText)
    Local $sOut
    With $oMessageStream
        .Type = 2
        .charset = "us-ascii"
        .Open()
        .WriteText($sText)
        .Flush()
        .Position = 0
        $sOut = .ReadText()
        .Close()
    EndWith
    Return $sOut
EndFunc

Edited March 27, 2013 by trancexx

[czardas]

Thanks for the code trancexx. It's a nice example (I can certainly learn from it ), but it replaces stuff it doesn't like with ? (question marks). I don't want the function to corrupt any other types of character. A quick test shows some of the characters are being corrupted:

ßÞIJijĸŋŊʼn× ==> ?????????

Don't worry about it: the function I have written seems good enough. I just didn't like calling _WinAPI_MultiByteToWideChar twice. I still need to test this on more strings to be certain there's no other corruption.

Edited March 27, 2013 by czardas

[jchd]

If you want a more complete covering for Unicode, try this:

#Region ;**** Directives created by AutoIt3Wrapper_GUI ****
#AutoIt3Wrapper_Run_AU3Check=n
#EndRegion ;**** Directives created by AutoIt3Wrapper_GUI ****
#include <sqlite.au3>
#include "SQLiteExtLoad.au3"

_SQLite_Startup("C:\Program Files (x86)\SQLite\sqlite3.dll")
_SQLite_LoadAutoExtension("..\bin\unifuzz.dll")
_SQLite_Open() ; use a :memory: DB


Local $sWeird = "azerty àƵĕřťƳ ʤʯΐΰӝӸḄṳἻᾀÅÇḈﻺ Å Ɽ♬♞☊◔╳∌ῴ"
Local $sNoDiacr = _Unaccent($sWeird)
MsgBox(0, "Unaccent demo", "Once diacritics are removed" & @LF & $sWeird & @LF & "becomes" & @LF & $sNoDiacr)
_SQLite_Close()
_SQLite_ShutDown()

Func _Unaccent($sStr)
Local $aRow
_SQLite_QuerySingleRow(-1, "select unaccent(" & _SQLite_FastEscape($sStr) & ");", $aRow)
Return($aRow[0])
EndFunc

Use forum search for posted includes.

Also you now have access to a whole lot of Unicode-aware string and other functions. Read the header of the C source of the extension about details.

Edited March 27, 2013 by jchd

[czardas]

Thanks a lot jchd.

Is there a reason why my function would not work with all unicode? - I'm not sure. Scite doesn't seem to like $sWeird BTW. Let me get back to you later, I have several other things to do today.

Got the includes.

Edited March 27, 2013 by czardas

[jchd]

The test string requires the file to be in UTF-8 encoding.

Yes, the reason is that StringIsAlpha only checks for A-Z and possibly ç and some other Latin-1 characters. Same for the function trancexx supplied. They won't cope with greek or cyrillic, for instance.

Append your code and let it run againts the same string to see the difference.

[czardas]

Yes I had thought that it would not work for other alphabets. In fact it was the intention, but the additional options would be good to have. I'll play around with it. Thanks again.

[czardas]

I had intended to return to this topic before now. Anyway I never managed to get your example to run jchd. I get the following error:

Return($aRow[0])

Return($aRow^ ERROR

Subscript used with non-Array variable

I have decided to adopt the brute force approach and stick to latin which is more familiar territory to me (at least for the time being). I don't know what effect a more greedy function might have on other languages, for example Chinese. Also I find I learn a lot by taking things apart piece by piece. I believe the following function removes accents from all latin characters and should also work with grapheme clusters. It's much more effective than my original attempt. The data was gathered using this online app: http://www.conlang.info/cgi/dia.cgi

expandcollapse popup

Func _StripAccentsLatin($sText)
    If Not IsString($sText) Then Return SetError(1, 0, "")
    $sText = StringRegExpReplace($sText, "[\x{00E1}\x{0103}\x{01CE}\x{00E2}\x{00E4}\x{0227}\x{1EA1}\x{0201}\x{00E0}\x{1EA3}\x{0203}\x{0101}\x{0105}\x{1D8F}\x{1E9A}\x{00E5}\x{1E01}\x{2C65}\x{00E3}]", ChrW(0x61))
    $sText = StringRegExpReplace($sText, "[\x{1E03}\x{1E05}\x{0253}\x{1E07}\x{1D6C}\x{1D80}\x{0180}\x{0183}]", ChrW(0x62))
    $sText = StringRegExpReplace($sText, "[\x{0107}\x{010D}\x{00E7}\x{0109}\x{0255}\x{010B}\x{0188}\x{023C}]", ChrW(0x63))
    $sText = StringRegExpReplace($sText, "[\x{010F}\x{1E11}\x{1E13}\x{0221}\x{1E0B}\x{1E0D}\x{0257}\x{1E0F}\x{1D6D}\x{1D81}\x{0111}\x{0256}\x{018C}]", ChrW(0x64))
    $sText = StringRegExpReplace($sText, "[\x{00E9}\x{0115}\x{011B}\x{0229}\x{00EA}\x{1E19}\x{00EB}\x{0117}\x{1EB9}\x{0205}\x{00E8}\x{1EBB}\x{0207}\x{0113}\x{2C78}\x{0119}\x{1D92}\x{0247}\x{1EBD}\x{1E1B}]", ChrW(0x65))
    $sText = StringRegExpReplace($sText, "[\x{1E1F}\x{0192}\x{1D6E}\x{1D82}]", ChrW(0x66))
    $sText = StringRegExpReplace($sText, "[\x{01F5}\x{011F}\x{01E7}\x{0123}\x{011D}\x{0121}\x{0260}\x{1E21}\x{1D83}\x{01E5}]", ChrW(0x67))
    $sText = StringRegExpReplace($sText, "[\x{1E2B}\x{021F}\x{1E29}\x{0125}\x{2C68}\x{1E27}\x{1E23}\x{1E25}\x{0266}\x{1E96}\x{0127}]", ChrW(0x68))
    $sText = StringRegExpReplace($sText, "[\x{00ED}\x{012D}\x{01D0}\x{00EE}\x{00EF}\x{1ECB}\x{0209}\x{00EC}\x{1EC9}\x{020B}\x{012B}\x{012F}\x{1D96}\x{0268}\x{0129}\x{1E2D}]", ChrW(0x69))
    $sText = StringRegExpReplace($sText, "[\x{01F0}\x{0135}\x{029D}\x{0249}]", ChrW(0x6A))
    $sText = StringRegExpReplace($sText, "[\x{1E31}\x{01E9}\x{0137}\x{2C6A}\x{A743}\x{1E33}\x{0199}\x{1E35}\x{1D84}\x{A741}]", ChrW(0x6B))
    $sText = StringRegExpReplace($sText, "[\x{013A}\x{019A}\x{026C}\x{013E}\x{013C}\x{1E3D}\x{0234}\x{1E37}\x{2C61}\x{A749}\x{1E3B}\x{0140}\x{026B}\x{1D85}\x{026D}\x{0142}]", ChrW(0x6C))
    $sText = StringRegExpReplace($sText, "[\x{1E3F}\x{1E41}\x{1E43}\x{0271}\x{1D6F}\x{1D86}]", ChrW(0x6D))
    $sText = StringRegExpReplace($sText, "[\x{0144}\x{0148}\x{0146}\x{1E4B}\x{0235}\x{1E45}\x{1E47}\x{01F9}\x{0272}\x{1E49}\x{019E}\x{1D70}\x{1D87}\x{0273}\x{00F1}\x{0149}\x{014B}]", ChrW(0x6E))
    $sText = StringRegExpReplace($sText, "[\x{00F3}\x{014F}\x{01D2}\x{00F4}\x{00F6}\x{022F}\x{1ECD}\x{0151}\x{020D}\x{00F2}\x{1ECF}\x{01A1}\x{020F}\x{A74B}\x{A74D}\x{2C7A}\x{014D}\x{01EB}\x{00F8}\x{00F5}]", ChrW(0x6F))
    $sText = StringRegExpReplace($sText, "[\x{1E55}\x{1E57}\x{A753}\x{01A5}\x{1D71}\x{1D88}\x{A755}\x{1D7D}\x{A751}]", ChrW(0x70))
    $sText = StringRegExpReplace($sText, "[\x{A759}\x{02A0}\x{024B}\x{A757}]", ChrW(0x71))
    $sText = StringRegExpReplace($sText, "[\x{0155}\x{0159}\x{0157}\x{1E59}\x{1E5B}\x{0211}\x{027E}\x{0213}\x{1E5F}\x{027C}\x{1D72}\x{1D89}\x{024D}\x{027D}]", ChrW(0x72))
    $sText = StringRegExpReplace($sText, "[\x{015B}\x{0161}\x{015F}\x{015D}\x{0219}\x{1E61}\x{1E63}\x{0282}\x{1D74}\x{1D8A}\x{023F}]", ChrW(0x73))
    $sText = StringRegExpReplace($sText, "[\x{0165}\x{0163}\x{1E71}\x{021B}\x{0236}\x{1E97}\x{2C66}\x{1E6B}\x{1E6D}\x{01AD}\x{1E6F}\x{1D75}\x{01AB}\x{0288}\x{0167}]", ChrW(0x74))
    $sText = StringRegExpReplace($sText, "[\x{00FA}\x{016D}\x{01D4}\x{00FB}\x{1E77}\x{00FC}\x{1E73}\x{1EE5}\x{0171}\x{0215}\x{00F9}\x{1EE7}\x{01B0}\x{0217}\x{016B}\x{0173}\x{1D99}\x{016F}\x{0169}\x{1E75}]", ChrW(0x75))
    $sText = StringRegExpReplace($sText, "[\x{2C74}\x{A75F}\x{1E7F}\x{028B}\x{1D8C}\x{2C71}\x{1E7D}]", ChrW(0x76))
    $sText = StringRegExpReplace($sText, "[\x{1E83}\x{0175}\x{1E85}\x{1E87}\x{1E89}\x{1E81}\x{2C73}\x{1E98}]", ChrW(0x77))
    $sText = StringRegExpReplace($sText, "[\x{1E8D}\x{1E8B}\x{1D8D}]", ChrW(0x78))
    $sText = StringRegExpReplace($sText, "[\x{00FD}\x{0177}\x{00FF}\x{1E8F}\x{1EF5}\x{1EF3}\x{01B4}\x{1EF7}\x{1EFF}\x{0233}\x{1E99}\x{024F}\x{1EF9}]", ChrW(0x79))
    $sText = StringRegExpReplace($sText, "[\x{017A}\x{017E}\x{1E91}\x{0291}\x{2C6C}\x{017C}\x{1E93}\x{0225}\x{1E95}\x{1D76}\x{1D8E}\x{0290}\x{01B6}\x{0240}]", ChrW(0x7A))
    $sText = StringRegExpReplace($sText, "[\x{00C1}\x{0102}\x{01CD}\x{00C2}\x{00C4}\x{0226}\x{1EA0}\x{0200}\x{00C0}\x{1EA2}\x{0202}\x{0100}\x{0104}\x{00C5}\x{1E00}\x{023A}\x{00C3}]", ChrW(0x41))
    $sText = StringRegExpReplace($sText, "[\x{1E02}\x{1E04}\x{0181}\x{1E06}\x{0243}\x{0182}]", ChrW(0x42))
    $sText = StringRegExpReplace($sText, "[\x{0106}\x{010C}\x{00C7}\x{0108}\x{010A}\x{0187}\x{023B}]", ChrW(0x43))
    $sText = StringRegExpReplace($sText, "[\x{010E}\x{1E10}\x{1E12}\x{1E0A}\x{1E0C}\x{018A}\x{1E0E}\x{0110}\x{018B}]", ChrW(0x44))
    $sText = StringRegExpReplace($sText, "[\x{00C9}\x{0114}\x{011A}\x{0228}\x{00CA}\x{1E18}\x{00CB}\x{0116}\x{1EB8}\x{0204}\x{00C8}\x{1EBA}\x{0206}\x{0112}\x{0118}\x{0246}\x{1EBC}\x{1E1A}]", ChrW(0x45))
    $sText = StringRegExpReplace($sText, "[\x{1E1E}\x{0191}]", ChrW(0x46))
    $sText = StringRegExpReplace($sText, "[\x{01F4}\x{011E}\x{01E6}\x{0122}\x{011C}\x{0120}\x{0193}\x{1E20}\x{01E4}]", ChrW(0x47))
    $sText = StringRegExpReplace($sText, "[\x{1E2A}\x{021E}\x{1E28}\x{0124}\x{2C67}\x{1E26}\x{1E22}\x{1E24}\x{0126}]", ChrW(0x48))
    $sText = StringRegExpReplace($sText, "[\x{00CD}\x{012C}\x{01CF}\x{00CE}\x{00CF}\x{0130}\x{1ECA}\x{0208}\x{00CC}\x{1EC8}\x{020A}\x{012A}\x{012E}\x{0197}\x{0128}\x{1E2C}]", ChrW(0x49))
    $sText = StringRegExpReplace($sText, "[\x{0134}\x{0248}]", ChrW(0x4A))
    $sText = StringRegExpReplace($sText, "[\x{1E30}\x{01E8}\x{0136}\x{2C69}\x{A742}\x{1E32}\x{0198}\x{1E34}\x{A740}]", ChrW(0x4B))
    $sText = StringRegExpReplace($sText, "[\x{0139}\x{023D}\x{013D}\x{013B}\x{1E3C}\x{1E36}\x{2C60}\x{A748}\x{1E3A}\x{013F}\x{2C62}\x{0141}]", ChrW(0x4C))
    $sText = StringRegExpReplace($sText, "[\x{1E3E}\x{1E40}\x{1E42}\x{2C6E}]", ChrW(0x4D))
    $sText = StringRegExpReplace($sText, "[\x{0143}\x{0147}\x{0145}\x{1E4A}\x{1E44}\x{1E46}\x{01F8}\x{019D}\x{1E48}\x{0220}\x{00D1}\x{014A}]", ChrW(0x4E))
    $sText = StringRegExpReplace($sText, "[\x{00D3}\x{014E}\x{01D1}\x{00D4}\x{00D6}\x{022E}\x{1ECC}\x{0150}\x{020C}\x{00D2}\x{1ECE}\x{01A0}\x{020E}\x{A74A}\x{A74C}\x{014C}\x{019F}\x{01EA}\x{00D8}\x{00D5}]", ChrW(0x4F))
    $sText = StringRegExpReplace($sText, "[\x{1E54}\x{1E56}\x{A752}\x{01A4}\x{A754}\x{2C63}\x{A750}]", ChrW(0x50))
    $sText = StringRegExpReplace($sText, "[\x{A758}\x{A756}]", ChrW(0x51))
    $sText = StringRegExpReplace($sText, "[\x{0154}\x{0158}\x{0156}\x{1E58}\x{1E5A}\x{0210}\x{0212}\x{1E5E}\x{024C}\x{2C64}]", ChrW(0x52))
    $sText = StringRegExpReplace($sText, "[\x{015A}\x{0160}\x{015E}\x{015C}\x{0218}\x{1E60}\x{1E62}]", ChrW(0x53))
    $sText = StringRegExpReplace($sText, "[\x{0164}\x{0162}\x{1E70}\x{021A}\x{023E}\x{1E6A}\x{1E6C}\x{01AC}\x{1E6E}\x{01AE}\x{0166}]", ChrW(0x54))
    $sText = StringRegExpReplace($sText, "[\x{00DA}\x{016C}\x{01D3}\x{00DB}\x{1E76}\x{00DC}\x{1E72}\x{1EE4}\x{0170}\x{0214}\x{00D9}\x{1EE6}\x{01AF}\x{0216}\x{016A}\x{0172}\x{016E}\x{0168}\x{1E74}]", ChrW(0x55))
    $sText = StringRegExpReplace($sText, "[\x{A75E}\x{1E7E}\x{01B2}\x{1E7C}]", ChrW(0x56))
    $sText = StringRegExpReplace($sText, "[\x{1E82}\x{0174}\x{1E84}\x{1E86}\x{1E88}\x{1E80}\x{2C72}]", ChrW(0x57))
    $sText = StringRegExpReplace($sText, "[\x{1E8C}\x{1E8A}]", ChrW(0x58))
    $sText = StringRegExpReplace($sText, "[\x{00DD}\x{0176}\x{0178}\x{1E8E}\x{1EF4}\x{1EF2}\x{01B3}\x{1EF6}\x{1EFE}\x{0232}\x{024E}\x{1EF8}]", ChrW(0x59))
    $sText = StringRegExpReplace($sText, "[\x{0179}\x{017D}\x{1E90}\x{2C6B}\x{017B}\x{1E92}\x{0224}\x{1E94}\x{01B5}]", ChrW(0x5A))
    $sText = StringRegExpReplace($sText, "[\x{01FD}\x{01E3}]", ChrW(0xE6))
    $sText = StringRegExpReplace($sText, "[\x{01FC}\x{01E2}]", ChrW(0xC6))
    $sText = StringRegExpReplace($sText, "[\x{1D90}]", ChrW(0x251))
    $sText = StringRegExpReplace($sText, "[\x{A73B}]", ChrW(0xA739))
    $sText = StringRegExpReplace($sText, "[\x{A73A}]", ChrW(0xA738))
    $sText = StringRegExpReplace($sText, "[\x{A72F}]", ChrW(0xA72D))
    $sText = StringRegExpReplace($sText, "[\x{A72E}]", ChrW(0xA72C))
    $sText = StringRegExpReplace($sText, "[\x{01C6}]", ChrW(0x01F3))
    $sText = StringRegExpReplace($sText, "[\x{01C4}]", ChrW(0x1F1))
    $sText = StringRegExpReplace($sText, "[\x{0286}\x{1D8B}\x{1D98}]", ChrW(0x283))
    $sText = StringRegExpReplace($sText, "[\x{01EF}\x{0293}\x{1D9A}\x{01BA}]", ChrW(0x292))
    $sText = StringRegExpReplace($sText, "[\x{01EE}]", ChrW(0x1B7))
    $sText = StringRegExpReplace($sText, "[\x{029B}]", ChrW(0x262))
    $sText = StringRegExpReplace($sText, "[\x{0267}]", ChrW(0xA727))
    $sText = StringRegExpReplace($sText, "[\x{1D7B}]", ChrW(0x26A))
    $sText = StringRegExpReplace($sText, "[\x{1D7C}]", ChrW(0x269))
    $sText = StringRegExpReplace($sText, "[\x{1D0C}]", ChrW(0x29F))
    $sText = StringRegExpReplace($sText, "[\x{019B}]", ChrW(0x3BB))
    $sText = StringRegExpReplace($sText, "[\x{025A}\x{1D95}]", ChrW(0x259))
    $sText = StringRegExpReplace($sText, "[\x{A765}\x{A767}]", ChrW(0xFE))
    $sText = StringRegExpReplace($sText, "[\x{A764}\x{A766}]", ChrW(0xDE))
    $sText = StringRegExpReplace($sText, "[\x{1D7E}]", ChrW(0x1D1C))
    $sText = StringRegExpReplace($sText, "[\x{1D7F}]", ChrW(0x28A))
    $sText = StringRegExpReplace($sText, "[\x{0300}\x{0301}\x{0302}\x{0303}\x{0304}\x{0305}\x{0306}\x{0307}\x{0308}\x{0309}\x{030A}\x{030B}\x{030C}\x{030D}\x{030E}\x{030F}\x{0310}\x{0311}\x{0312}\x{0313}\x{0314}\x{0315}\x{0316}\x{0317}\x{0318}\x{0319}\x{031A}\x{031B}\x{031C}\x{031D}\x{031E}\x{031F}\x{0320}\x{0321}\x{0322}\x{0323}\x{0324}\x{0325}\x{0326}\x{0327}\x{0328}\x{0329}\x{032A}\x{032B}\x{032C}\x{032D}\x{032E}\x{032F}\x{0330}\x{0331}\x{0332}\x{0333}\x{0334}\x{0335}\x{0336}\x{0337}\x{0338}\x{0339}\x{033A}\x{033B}\x{033C}\x{033D}\x{033E}\x{033F}\x{0340}\x{0341}\x{0342}\x{0343}\x{0344}\x{0345}\x{0346}\x{0347}\x{0348}\x{0349}\x{034A}\x{034B}\x{034C}\x{034D}\x{034E}\x{034F}\x{0350}\x{0351}\x{0352}\x{0353}\x{0354}\x{0355}\x{0356}\x{0357}\x{0358}\x{0359}\x{035A}\x{035B}\x{035C}\x{035D}\x{035E}\x{035F}\x{0360}\x{0361}\x{0362}\x{0363}\x{0364}\x{0365}\x{0366}\x{0367}\x{0368}\x{0369}\x{036A}\x{036B}\x{036C}\x{036D}\x{036E}\x{036F}]", "")
    Return $sText
EndFunc ;==> _StripAccentsLatin

Duh, code was missing a comma.

Edited April 22, 2013 by czardas

[jchd]

Are you sure no error occurs, e.g. the SQLite extension isn't found or something similar?

[czardas]

Yes I'm sure, but it looks like I'm missing unifuzz.dll

[jchd]

Umm, that's what I suspected.

Here you are.

[czardas]

Thanks jchd. Looking at the c source. It certainly looks interesting. I think it should work now, but I'll test it in the morning.

[czardas]

I just got around to trying this again and I'm still having no joy with it. I have tried modifying the path to full script path (where I put the dll) and still get the same error.

Edited April 23, 2013 by czardas

[jchd]

Which error are you getting and when?