Fix (R) and (TM) mojibake

fixRegAndTm.groovy

//
// Fix (R) and (TM) mojibake
//

String replace(String str){

	def log = { text,i->
		def out = ''<<''
		for(int j=Math.max(0,i-10); j<i; j++){
			out << text.charAt(j)
		}
		out.toString()
	}

	def sb = ''<<''

	for( int i=0; i<str.length(); i++){
		
		boolean found = false

		if( i>0 ){
			def m0 = ( str.charAt(i-1) =~ /[a-zA-Z]/ )
			def m1 = ( str.charAt(i-1) =~ />/ )
			if( m0.find() || m1.find() ){
				String hexstr = Integer.toHexString( (int)str.charAt(i) )
				if( hexstr == '30a3' ){
					println "found (R) -> (${log(str,i)})"
					sb << '&reg;'
					found = true
				}
				else if( hexstr == '30a7' ){
					println "found TM -> (${log(str,i)})"
					sb << '&trade;'
				}
			}
		}

		if( found==false ){
			sb << str.charAt(i)
		}
	}
	sb.toString()
}


if( args.length<1 )
	System.exit(0)


def ENC = 'MS932'

def inf  = new File(args[0])
def outf = inf

println "--- ${inf.name} ---"

def text = inf.getText(ENC)

def w    = outf.newWriter(ENC)
w.print( replace(text) )
w.close()