Blame tools/utfdump.py

Packit 0f19cf
import codecs
Packit 0f19cf
import sys
Packit 0f19cf
Packit 0f19cf
def main():
Packit 0f19cf
Packit 0f19cf
    decode = codecs.getdecoder("utf8")
Packit 0f19cf
    encode = codecs.getencoder("latin-1")
Packit 0f19cf
Packit 0f19cf
    f = open(sys.argv[1])
Packit 0f19cf
    lineno = 0
Packit 0f19cf
    for line in f:
Packit 0f19cf
        line = decode(line)[0]
Packit 0f19cf
        lineno += 1
Packit 0f19cf
        outline = ""
Packit 0f19cf
        for uchar in line:
Packit 0f19cf
            try:
Packit 0f19cf
                o = encode(uchar)[0]
Packit 0f19cf
            except:
Packit 0f19cf
                o = "?"
Packit 0f19cf
            print "U%04X: %s" % (ord(uchar), o)
Packit 0f19cf
            outline += o
Packit 0f19cf
        print "Line %3d: %s" % (lineno, outline)
Packit 0f19cf
Packit 0f19cf
Packit 0f19cf
if __name__ == "__main__":
Packit 0f19cf
    main()