migurski/bytes2utf8.py

## bytes2utf8.py
#!/usr/bin/env python
''' Convert binary data to UTF8.

If you want binary data in Javascript over XMLHttpRequest but don't
have ArrayBuffer available, one option is to read strings of text and
use xhr.responseText.charCodeAt() to read them. Because this expects to
work with text, it will return the unicode of the character at that
position. This script converts raw binary bytes to unicode text that
evaluates to the same 8-bit unsigned int at each position. The overall
length will increase slightly.
'''

from struct import pack, unpack
from sys import stdin, stderr, stdout
from optparse import OptionParser

parser = OptionParser(usage='%(prog)s <input> <output>')

defaults = dict(order='little', verbose=False)

parser.set_defaults(**defaults)

parser.add_option('-o', '--order', choices=('little', 'big'),
                  dest='order', help='Byte order (endianness), default %(order)s.' % defaults)

parser.add_option('-v', '--verbose', action='store_true',
                  dest='verbose', help='Optionally output size report.')

def byte2uint16_little(byte):
    return pack('<H', ord(byte))

def byte2uint16_big(byte):
    return pack('>H', ord(byte))

if __name__ == '__main__':

    opts, (input, output) = parser.parse_args()

    bytes = open(input).read()

    if opts.order == 'little':
        utf16 = ''.join(map(byte2uint16_little, bytes))
        utf8 = utf16.decode('utf-16-le').encode('utf-8')

    else:
        utf16 = ''.join(map(byte2uint16_big, bytes))
        utf8 = utf16.decode('utf-16-be').encode('utf-8')

    if opts.verbose:
        print >> stderr, 'Converted', len(bytes), 'bytes binary to', len(utf8), 'bytes utf-8'

    out = open(output, 'w')
    out.write(utf8)
    out.close()

## sample.html
<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="utf-8" />
	<title>floats.html</title>
</head>
<body>

<script type="application/javascript">
<!--

    var x = new XMLHttpRequest(),
        t = undefined;

    x.onreadystatechange = function()
    {
        if(x.readyState != 4)
        {
            return;
        }

        var ab = new ArrayBuffer(x.responseText.length),
            u = new Uint8Array(ab), f = new Float32Array(ab);

        console.log(u.length, f.length);

        for(var i = 0; i < u.length; i+=1)
        {
            u[i] = x.responseText.charCodeAt(i);
        }

        console.log(f[0], f[1], f[2], f[3]);
    }

    x.open('GET', 'floats.utf8', true);
    x.send();

//-->
</script>

</body>
</html>
	#!/usr/bin/env python
	''' Convert binary data to UTF8.

	If you want binary data in Javascript over XMLHttpRequest but don't
	have ArrayBuffer available, one option is to read strings of text and
	use xhr.responseText.charCodeAt() to read them. Because this expects to
	work with text, it will return the unicode of the character at that
	position. This script converts raw binary bytes to unicode text that
	evaluates to the same 8-bit unsigned int at each position. The overall
	length will increase slightly.
	'''

	from struct import pack, unpack
	from sys import stdin, stderr, stdout
	from optparse import OptionParser

	parser = OptionParser(usage='%(prog)s <input> <output>')

	defaults = dict(order='little', verbose=False)

	parser.set_defaults(**defaults)

	parser.add_option('-o', '--order', choices=('little', 'big'),
	dest='order', help='Byte order (endianness), default %(order)s.' % defaults)

	parser.add_option('-v', '--verbose', action='store_true',
	dest='verbose', help='Optionally output size report.')

	def byte2uint16_little(byte):
	return pack('<H', ord(byte))

	def byte2uint16_big(byte):
	return pack('>H', ord(byte))

	if __name__ == '__main__':

	opts, (input, output) = parser.parse_args()

	bytes = open(input).read()

	if opts.order == 'little':
	utf16 = ''.join(map(byte2uint16_little, bytes))
	utf8 = utf16.decode('utf-16-le').encode('utf-8')

	else:
	utf16 = ''.join(map(byte2uint16_big, bytes))
	utf8 = utf16.decode('utf-16-be').encode('utf-8')

	if opts.verbose:
	print >> stderr, 'Converted', len(bytes), 'bytes binary to', len(utf8), 'bytes utf-8'

	out = open(output, 'w')
	out.write(utf8)
	out.close()
	<!DOCTYPE html>
	<html lang="en">
	<head>
	<meta charset="utf-8" />
	<title>floats.html</title>
	</head>
	<body>

	<script type="application/javascript">
	<!--

	var x = new XMLHttpRequest(),
	t = undefined;

	x.onreadystatechange = function()
	{
	if(x.readyState != 4)
	{
	return;
	}

	var ab = new ArrayBuffer(x.responseText.length),
	u = new Uint8Array(ab), f = new Float32Array(ab);

	console.log(u.length, f.length);

	for(var i = 0; i < u.length; i+=1)
	{
	u[i] = x.responseText.charCodeAt(i);
	}

	console.log(f[0], f[1], f[2], f[3]);
	}

	x.open('GET', 'floats.utf8', true);
	x.send();

	//-->
	</script>

	</body>
	</html>