try:
    import json
except:
    import simplejson as json
import os
import hashlib
import difflib
import urllib2
import sys

def hashfile(afile, hasher, blocksize=65536):
    buf = afile.read(blocksize)
    while len(buf) > 0:
        hasher.update(buf)
        buf = afile.read(blocksize)
    return hasher.hexdigest()

class hashtable():
    def __init__(self):
        self.files = dict()
    
    def addhash(self, filename, hash, version):
        if filename not in self.files:
            self.files[filename] = dict()
        if hash not in self.files[filename]:
            self.files[filename][hash] = []
        self.files[filename][hash].append(version)
    
    def dump(self):
        print self.files

def replace_current_site():
    searchandreplace = dict()
    for i in open('/tmp/wpv.rpl').read().split('====\n'):
        if "FROM:" in i:
            (fromline,toline,junk) = i.split('\n')
            print fromline
            print toline
            searchandreplace[fromline[6:].replace('\n','').replace('\r','')] = toline[4:].replace('\n','').replace('\r','')
    for root, dir, files in os.walk('.'):
        if len(root) > 2 and root[:2] == './':
            root = root[2:]
        else:
            root = ""
        for i in files:
            try:
                if i[-4:] == ".php" or i[-3:] == ".js":
                    if root != "":
                        current = root + '/' + i
                    else:
                        current = i
                    rewrite = False
                    lines = open(current).read().split('\n')
                    for i in searchandreplace:
                        j = 0
                        while j < len(lines):
                            while i in lines[j]:
                                lines[j] = lines[j].replace(i, searchandreplace[i])
                                rewrite = True
                    if rewrite:
                        outfile = open(current, 'w')
                        outfile.write("\n".join(lines))
                        outfile.close()               
            except:
                pass

def check_current_site(returnbadlines = False, rootpath = ''):
    for root, dir, files in os.walk('.'):
        if len(root) > 2 and root[:2] == './':
            root = root[2:]
        else:
            root = ""
        for i in files:
            if i[-4:] == ".php" or i[-3:] == ".js":
                if root != "":
                    current = root + '/' + i
                else:
                    current = i
                if current in hashes.files:
                    currenthash = hashfile(open(current, 'rb'), hashlib.sha256())
                    if currenthash not in hashes.files[current]:
                        print "Bad File: " + rootpath + current,
                        answer = 0
                        closestresult = "NONE"
                        for j in hashes.files[current]:
                            k = max(hashes.files[current][j])
                            checkfile = urllib2.urlopen('http://wpv.as22384.net/wordpress/' + k + '/' + current).read()
                            diffratio = difflib.SequenceMatcher(None, open(current).read(), checkfile).ratio()
                            if diffratio >= answer:
                                answer = diffratio
                                closestresult = ", ".join(sorted(hashes.files[current][j]))
                                closestresultfordiff = k
                        print "  Closest Match (" + str(answer) + "): " + str(closestresult)
                        for j in difflib.unified_diff(open(current).read().split('\n'), urllib2.urlopen('http://wpv.as22384.net/wordpress/' + closestresultfordiff + '/' + current).read().split('\n')):
                            print j
                            if j[0] == '-':
                                badlines.add(j[1:].replace("\n","").replace("\r",""))
    if returnbadlines:
        return badlines
    else:
        outfile = open('/tmp/wpv.rpl', 'a')
        for i in sorted(badlines):
            outfile.write('FROM: ')
            outfile.write(i.replace("\n",'').replace("\r",''))
            outfile.write("\n")
            outfile.write("TO: \n====\n")
        outfile.close()


if __name__ == "__main__":
    hashes = hashtable()
    # Workaround for python 2.5
    temp = json.JsonReader()
    hashes.files = temp.read(urllib2.urlopen('http://wpv.as22384.net/hashes.json').read())
    badlines = set()
    if '-x' in sys.argv:
        replace_current_site()
    else:
        if '-a' in sys.argv:
            badlines = set()
            for root, dir, files in os.walk('/'):
                for i in files:
                    if i == 'wp-config.php':
                        os.chdir(root)
                        for j in check_current_site(returnbadlines = True, rootpath = root + '/'):
                            badlines.add(j)
            outfile = open('/tmp/wpv.rpl', 'a')
            for i in sorted(badlines):
                outfile.write('FROM: ')
                outfile.write(i.replace("\n",'').replace("\r",''))
                outfile.write("\n")
                outfile.write("TO: \n====\n")
            outfile.close()
                                                                                    
        else:
            check_current_site()