文字エンコーディングを自動判別してファイル読み込み v1.2.1

Posted on 2月 6, 2018 by Takaaki Naganoya

AppleScript名：文字エンコーディングを自動判別してファイル読み込み v1.2.1

— Created 2014-12-28 by Takaaki Naganoya
— Modified 2014-12-29 by Shane Stanley
— Modified 2015-10-03 by Takaaki Naganoya
use AppleScript version "2.5"
use scripting additions
use framework "Foundation"

set aPath to POSIX path of (choose file)
set aRes to readJapanesTextFileWithGuessingEncoding(aPath) of me
set bRes to aRes as string

–Read Japanese text with detecting its text encoding
on readJapanesTextFileWithGuessingEncoding(aPOSIXpath as string)
　　
　　–ISO2022JP check
　　set aNSData to current application’s NSData’s dataWithContentsOfFile:aPOSIXpath
　　set aDataLength to aNSData’s |length|()
　　if aDataLength > 1024 then set aDataLength to 1024
　　
　　–0x1B check
　　set anNSString to current application’s NSString’s stringWithString:(character id 27) — 0x1B
　　set theData to anNSString’s dataUsingEncoding:(current application’s NSUTF8StringEncoding)
　　set theRange to aNSData’s rangeOfData:theData options:0 range:(current application’s NSMakeRange(0, aDataLength))
　　
　　–found 0x1B in aNSData
　　if |length| of theRange = 1 and location of theRange < aDataLength then
　　　　set aStr to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSISO2022JPStringEncoding)) –21
　　　　if aStr is not equal to missing value then return (aStr as text) — ISO2022JP
　　end if
　　
　　–EUC
　　set resValue to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSJapaneseEUCStringEncoding))
　　–log resValue
　　if resValue is not equal to missing value then return (resValue as text)
　　–UTF-8
　　set resValue to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSUTF8StringEncoding))
　　–log resValue
　　if resValue is not equal to missing value then return (resValue as text)
　　–SHift JIS
　　set resValue to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSShiftJISStringEncoding))
　　–log resValue
　　if resValue is not equal to missing value then return (resValue as text)
　　
　　
　　–UTF-16BE/LE/無印Unicodeは多数決を取る
　　set resValue1 to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSUTF16BigEndianStringEncoding)) as text
　　–log resValue1
　　set sample1 to getTextSample(resValue1) of me
　　set lang1 to specifyLanguageOfText(sample1) of me
　　set para1 to length of (paragraphs of sample1)
　　set words1 to length of (words of sample1)
　　
　　set resValue2 to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSUTF16LittleEndianStringEncoding)) as text
　　–log resValue2
　　set sample2 to getTextSample(resValue2) of me
　　set lang2 to specifyLanguageOfText(sample2) of me
　　set para2 to length of (paragraphs of sample2)
　　set words2 to length of (words of sample2)
　　
　　set resValue3 to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSUnicodeStringEncoding)) as text
　　–log resValue3
　　set sample3 to getTextSample(resValue3) of me
　　set lang3 to specifyLanguageOfText(sample3) of me
　　set para3 to length of (paragraphs of sample3)
　　set words3 to length of (words of sample3)
　　
　　–文字および文法的に見て「日本語」ならそれを返す
　　if lang1 = "ja" then return resValue1
　　if lang2 = "ja" then return resValue2
　　if lang3 = "ja" then return resValue2
　　
　　
　　–文字化けしたときには、日本語の「Word」として認識されづらく、Paragraphも少ない（１とか）なので条件で除外する
　　if para1 is not equal to 1 then
　　　　if (words1 ≤ words2) or (words1 ≤ words3) then
　　　　　　return resValue1
　　　　end if
　　end if
　　
　　if para2 is not equal to 1 then
　　　　if (words2 ≤ words1) or (words2 ≤ words3) then
　　　　　　return resValue2
　　　　end if
　　end if
　　
　　if para3 is not equal to 1 then
　　　　if (words3 ≤ words1) or (words3 ≤ words2) then
　　　　　　return resValue3
　　　　end if
　　end if
　　
　　return false
　　
　　(*
　　–おまけ（未確認）
　　set resValue to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSWindowsCP1251StringEncoding))
　　if resValue is not equal to missing value then return resValue
　　
　　set resValue to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSWindowsCP1252StringEncoding))
　　if resValue is not equal to missing value then return resValue
　　
　　set resValue to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSWindowsCP1253StringEncoding))
　　if resValue is not equal to missing value then return resValue
　　
　　set resValue to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSWindowsCP1254StringEncoding))
　　if resValue is not equal to missing value then return resValue
　　
　　set resValue to (current application’s NSString’s alloc()’s initWithData:aNSData encoding:(current application’s NSWindowsCP1250StringEncoding))
　　if resValue is not equal to missing value then return resValue
　　
　　return false
　　*)
end readJapanesTextFileWithGuessingEncoding

on specifyLanguageOfText(aStr)
　　set aNSstring to current application’s NSString’s stringWithString:aStr
　　set tagSchemes to current application’s NSArray’s arrayWithObjects:(current application’s NSLinguisticTagSchemeLanguage)
　　set tagger to current application’s NSLinguisticTagger’s alloc()’s initWithTagSchemes:tagSchemes options:0
　　tagger’s setString:aNSstring
　　set aLanguage to tagger’s tagAtIndex:0 |scheme|:(current application’s NSLinguisticTagSchemeLanguage) tokenRange:(missing value) sentenceRange:(missing value)
　　return aLanguage as text
end specifyLanguageOfText

on getTextSample(aText)
　　set aLen to length of aText
　　if aLen < 1024 then
　　　　set bLen to aLen
　　else
　　　　set bLen to 1024
　　end if
　　return (text 1 thru bLen of aText)
end getTextSample