#!/usr/bin/python
"""HTML Diff: http://www.aaronsw.com/2002/diff
Rough code, badly documented. Send me comments and patches."""

__author__ = 'Aaron Swartz <me@aaronsw.com>, \
Richard Henwood <rhenwood@whamcloud.com>'
__copyright__ = '(C) 2003 Aaron Swartz, (C) 2011,2012 Whamcloud, Inc \
GNU GPL 2 or 3.'
__version__ = '1.0'

import difflib
import string
import re

def isTag(x): return x[0] == "<" and x[-1] == ">"

def textDiff(a, b):
    """Takes in strings a and b and returns a human-readable HTML diff."""
    changeCounter = 0

    out = []
    a, b = html2list(a), html2list(b)
    s = difflib.SequenceMatcher(None, a, b)
    gap = 0;
    wait = 7; # fudge factor to avoid having too many changes anchors.
    for e in s.get_opcodes():
        if e[0] == "replace":
            # @@ need to do something more complicated here
            # call textDiff but not for html, but for some html... ugh
            # gonna cop-out for now
            # I've put this hack in to try accomodate id's generated by
            # docbook compilation. RJH 30/12/2011
            if ('<a id="id' in a[e[1]:e[2]][0] or
                    '<a href="#id' in a[e[1]:e[2]][0] or
                    '<a id="ftn.id' in a[e[1]:e[2]][0] or
                    '<a class="indexterm" href="#id' in a[e[1]:e[2]][0] or
                    '<a id="id' in b[e[3]:e[4]][0] or
                    '<a href="#id' in b[e[3]:e[4]][0] or
                    '<a id="ftn.id' in b[e[3]:e[4]][0] or
                    '<a class="indexterm" href="#id' in b[e[3]:e[4]][0]):
                out.append(''.join(b[e[3]:e[4]]))
            else:
                if gap > wait:
                    out.append('<a name="change_%s"></a>' % changeCounter)
                    changeCounter += 1
                withopen = addOpen(a[e[1]:e[2]])
                out.append(annotateTags(withopen, 'delete'));
                out.append(annotateTags(b[e[3]:e[4]], 'insert'));
                gap = 0
        elif e[0] == "delete":
            if gap > wait:
                out.append('<a name="change_%s"></a>' % changeCounter)
                changeCounter += 1
            out.append(annotateTags(a[e[1]:e[2]], 'delete'));
            gap = 0
        elif e[0] == "insert":
            if gap > wait:
                out.append('<a name="change_%s"></a>' % changeCounter)
                changeCounter += 1
            out.append(annotateTags(b[e[3]:e[4]], 'insert'));
            gap = 0
        elif e[0] == "equal":
            out.append(''.join(b[e[3]:e[4]]))
            gap += 1
        else:
            raise "Um, something's broken. I didn't expect a '" + `e[0]` + "'."
    return [''.join(out), changeCounter]

# this functiona finds all unmatched
# close tags in htmllist and prepends
# corresponding open tags to htmllist.
def addOpen(htmllist):
    unmatched = []
    tmpstack = []
    for item in htmllist:
        if "<" in item:
            if "</" in item:
                try:
                    if tmpstack[-1] == item:
                        tmpstack.pop()
                    else:
                        tmpstack.append(item)
                except IndexError:
                    unmatched.append(item)
                    pass
                pass
            else:
                tmpstack.append(re.sub(r'<(\w+).*', r'</\g<1>>', item, flags=re.IGNORECASE))
                pass
        else:
            pass
    unmatched = map(lambda x: x.replace('/', ''), unmatched)
    unmatched.reverse()
    return unmatched + htmllist

def annotateTags(htmllist, className):
    res = map(lambda tag: addClass(tag, className), htmllist)
    #the first element should be surrounded with a span + class
    #provided it is not a tag.
    if '<' not in htmllist[0]:
        res[0] = '<span class="%s">%s</span>' % (className, htmllist[0])
    return ''.join(res)

# this method adds a new class to a html tag. i.e.
# <a id="idp9794432"> -> <a id="idp9794432" class='test'>
# <a id="idp9794432" class='test'> -> <a id="idp9794432" class='test test2'>
def addClass (tag, newClass):
    if "<" in tag and "</" not in tag:
        if 'class' not in tag:
            # assume all tags end in '>', so chop it off:
            tag = tag[:-1]
            # add in an empty class element.
            tag = '%s class="">' % tag
        tag = re.sub(r'(.*)class=[\'"]([ \w]*)[\'"]',
                r'\g<1>class="\g<2> '+newClass+'"', tag , flags=re.IGNORECASE)
    return tag

def html2list(x, b=0):
    mode = 'char'
    cur = ''
    out = []
    for c in x:
        if mode == 'tag':
            if c == '>':
                if b: cur += ']'
                else: cur += c
                out.append(cur); cur = ''; mode = 'char'
            else: cur += c
        elif mode == 'char':
            if c == '<':
                out.append(cur)
                if b: cur = '['
                else: cur = c
                mode = 'tag'
            elif c in string.whitespace: out.append(cur+c); cur = ''
            else: cur += c
    out.append(cur)
    return filter(lambda x: x is not '', out)

def test ():
    print "\nrunning tests..."
    test1res = addClass('<a id="idp9794432">', 'test')
    test1suc = '<a id="idp9794432" class=" test">'
    if test1res == test1suc:
        print "SUCCESS: %s == %s" % (test1res, test1suc)
    else:
        print "FAIL: %s != %s" % (test1res, test1suc)
    test2res = addClass('<a id="idp9794432" class="test">', 'test2')
    test2suc = '<a id="idp9794432" class="test test2">'
    if test2res == test2suc:
        print "SUCCESS: %s == %s" % (test2res, test2suc)
    else:
        print "FAIL: %s != %s" % (test2res, test2suc)

    test3res = addOpen(['hello','</div>', '</p>'])
    test3suc = ['<p>', '<div>', 'hello','</div>', '</p>']
    if test3res == test3suc:
        print "SUCCESS: %s == %s" % (test3res, test3suc)
    else:
        print "FAIL: %s != %s" % (test3res, test3suc)

    test4res = addOpen(['hello','</div>', '<p class="test">', 'test', '</p>', '</p>'])
    test4suc = ['<p>', '<div>', 'hello','</div>', '<p class="test">', 'test', '</p>', '</p>']
    if test4res == test4suc:
        print "SUCCESS: %s == %s" % (test4res, test4suc)
    else:
        print "FAIL: %s != %s" % (test4res, test4suc)

    test5res = addOpen(['0','</p>', '</div>', '</div>', '<div>', 'test', '</div>'])
    test5suc = ['<div>', '<div>', '<p>', '0','</p>', '</div>', '</div>', '<div>', 'test', '</div>']
    if test5res == test5suc:
        print "SUCCESS: %s == %s" % (test5res, test5suc)
    else:
        print "FAIL: %s != %s" % (test5res, test5suc)

    test6res = annotateTags(['<div class="hello">', '0', '</div>'], 'test')
    test6suc = '<div class="hello test">0</div>'
    if test6res == test6suc:
        print "SUCCESS: %s == %s" % (test6res, test6suc)
    else:
        print "FAIL: %s != %s" % (test6res, test6suc)

    test7res = addOpen(['0','</p>', '</div>', '</div>', '<div>', '<div hello="world">', 'test', '</div>', '</div>', '</div>'])
    test7suc = ['<div>', '<div>', '<div>', '<p>', '0','</p>', '</div>', '</div>', '<div>', '<div hello="world">', 'test', '</div>', '</div>', '</div>']
    if test7res == test7suc:
        print "SUCCESS: %s == %s" % (test7res, test7suc)
    else:
        print "FAIL: %s != %s" % (test7res, test7suc)

    pass

def jsHeader (changeCounter):
    jscript = """
<script type='text/javascript'>
    var changes = %s;
    function showChanges()
    {
        var diffList, link, warn;
        diffList=document.getElementById('changeDiffs');
        for (var i = 0; i < changes; i += 1) {
            if (i == 0) {
                diffList.innerHTML = "";
            }
            link = document.createElement('a');
            link.setAttribute('class', 'change_link');
            link.setAttribute('href', '#change_'+i);
            link.innerHTML = ' change #'+i;
            diffList.appendChild(link);
        }
        warn = document.createTextNode("NOTE: these changes are automatically detected and may be inaccurate.");
        diffList.appendChild(warn);
    }
</script>
    """ % changeCounter
    return jscript;

def cssHeader ():

    css = """<style type="text/css">
    .delete {
    background-color: Pink; text-decoration: line-through;
    }

    .insert {
    background-color: PaleGreen;
    }

    .change_link {
    padding:5px;
    }

    </style>
    """
    return css


if __name__ == '__main__':
    import sys
    try:
        a, b = sys.argv[1:3]
    except ValueError:
        print "htmldiff: highlight the differences between two html files"
        print "usage: " + sys.argv[0] + " a b"
        test()
        sys.exit(1)
    changeCounter = 0
    (diffxhtml, changeCounter) = textDiff(open(a).read(), open(b).read())
    # it is, regrettably, possible that the result of textDiff is not
    # valid xhtml. I have noticed issues with nested <p>'s for example.
    # Because of this, search and replace to insert our css, js and dom
    # elements:
    diffxhtml = diffxhtml.replace('</head><body>', cssHeader() + jsHeader(changeCounter) +
            '</head><body onload="showChanges()"><div id="changeDiffs">No changes detected.</div>')
    print diffxhtml