python_html_diff.py

def html_diff(val1, val2):
    from difflib import SequenceMatcher
    if val1 is None or val2 is None:
        return (val1, val2)
    diff = SequenceMatcher(None, val1, val2)
    op_codes = diff.get_opcodes()
    offset1 = 0
    offset2 = 0

    for op_code in op_codes:
        op, i1, i2, j1, j2 = op_code
        if op != 'equal':
            if i1 != i2:
                to_insert = '<span class="diff diff-{code}">'.format(code=op)
                val1 = val1[:i1 + offset1] + to_insert + val1[i1 + offset1:]
                offset1 += len(to_insert)

                to_insert = '</span>'
                val1 = val1[:i2 + offset1] + to_insert + val1[i2 + offset1:]
                offset1 += len(to_insert)
            if j1 != j2:
                to_insert = '<span class="diff diff-{code}">'.format(code=op)
                val2 = val2[:j1 + offset2] + to_insert + val2[j1 + offset2:]
                offset2 += len(to_insert)

                to_insert = '</span>'
                val2 = val2[:j2 + offset2] + to_insert + val2[j2 + offset2:]
                offset2 += len(to_insert)

    return (val1, val2)