#!/usr/bin/env python

import tarfile
import zipfile
import os
import sys
import datetime

try:
    import hashlib
    md5_cons = hashlib.md5
    sha1_cons = hashlib.sha1
except:
    import md5
    import sha
    md5_cons = md5.new
    sha1_cons = sha.new


def hash_file(fileobj, *hashobjects):
    """ RAM efficient hashing implementation for stream-based file objects. """
    data = fileobj.read(1024*1024)
    while data:
        for ho in hashobjects:
            ho.update(data)
        data = fileobj.read(1024*1024)

    return (ho.hexdigest() for ho in hashobjects)


class ArchiveInfo:
    def __init__(self, filename):
        self.filename = filename
        self.filesize = os.path.getsize(filename) # this can throw an exception
        self.mtime = os.path.getmtime(filename)
        self.basename = os.path.basename(filename)


    def write_info(self, outfiledir):
        outfilename = "%s.%s.INDEX" % (self.basename, self.mtime)
        self.out_fd = open(outfilename, "w")
        self.write_header()
        header_end = self.out_fd.tell()
        try:
            self.write_tar_content()
        except:
            self.out_fd.seek(header_end)
            try:
                self.write_zip_content()
            except:
                self.out_fd.seek(header_end)
        self.out_fd.truncate()
        self.out_fd.close()


    def write_header(self):
        self.out_fd.write("File-dist-name: %s\n" % (self.basename))
        self.out_fd.write("File-dist-size: %s\n" % (self.filesize))
        self.out_fd.write("File-dist-isdistfile: 1\n")
        #self.out_fd.write("File-dist-origin: %s\n" % (self.url))
        #self.out_fd.write("File-dist-cpv: %s\n" % (self.cpv))
        #self.out_fd.write("File-dist-cat: %s\n" % (self.cat))
        #self.out_fd.write("File-dist-pn: %s\n" % (self.pn))
        #self.out_fd.write("File-dist-pv: %s\n" % (self.pv))
        #self.out_fd.write("File-dist-pr: %s\n" % (self.pr))
        #self.out_fd.write("File-dist-pf: %s\n" % (self.pf))
        #self.out_fd.write("File-dist-cpv: %s\n" % (self.cpv))
        self.out_fd.write("File-dist-mtime: %s\n" %(self.mtime))

        distfile = open(self.filename)
        (md5, sha1) = hash_file(distfile, md5_cons(), sha1_cons())
        self.out_fd.write("File-dist-md5: %s\n" % (md5))
        self.out_fd.write("File-dist-sha1: %s\n" % (sha1))
        distfile.close()


    def write_tar_content(self):
        tar = tarfile.TarFile.open(self.filename, 'r')

        file_info = tar.next()
        filecount = 0
        while file_info != None:
            if file_info.isfile():
                file_stream = tar.extractfile(file_info)
                
                if file_stream == None:
                    print "Filestream empty on %s" % (file_info.name)
                    pass
                else:
                    self.out_fd.write("File-%05d-name: %s\n" %(filecount, file_info.name))
                    self.out_fd.write("File-%05d-size: %s\n" %(filecount, file_info.size))
                    self.out_fd.write("File-%05d-mtime: %s\n" %(filecount, file_info.mtime))
                    (md5, sha1) = hash_file(file_stream, md5_cons(), sha1_cons())
                    self.out_fd.write("File-%05d-md5: %s\n" %(filecount, md5))
                    self.out_fd.write("File-%05d-sha1: %s\n" %(filecount, sha1))
                    filecount += 1
            file_info = tar.next()
        tar.close()

    def write_zip_content(self):
        zip = zipfile.ZipFile(self.filename, 'r')

        filecount = 0
        for name in zip.namelist():
            file_info = zip.getinfo(name)
            if os.path.basename(file_info.filename) == "":
                # this is a directory
                continue
            file_content = zip.read(name) # Doesn't work stream based :-/
            file_time = datetime.datetime(*file_info.date_time)
            self.out_fd.write("File-%05d-name: %s\n" %(filecount, file_info.filename))
            self.out_fd.write("File-%05d-size: %s\n" %(filecount, file_info.file_size))
            self.out_fd.write("File-%05d-mtime: %s\n" %(filecount, file_time.strftime("%s")))

            md5 = md5_cons(file_content).hexdigest()
            sha1 = sha1_cons(file_content).hexdigest()
            self.out_fd.write("File-%05d-md5: %s\n" %(filecount, md5))
            self.out_fd.write("File-%05d-sha1: %s\n" %(filecount, sha1))

            filecount += 1

        zip.close()




def main():
    import getopt
    try:
        optlist, files = getopt.getopt(sys.argv[1:],
                                  'P:hu:')
    except getopt.GetoptError:
        usage(sys.argv[0])
        sys.exit(2)

    outdir = "."
    for opt, arg in optlist:
        if opt == '-h':
            usage(sys.argv[0])
            sys.exit(0)
        if opt == '-P':
            outdir = arg

    if len(files) == 0:
        print "Please specify a filename."
    else:
        for infilename in files:
            try:
                a = ArchiveInfo(infilename)
                filename = a.write_info(outdir)
            except Error, e:
                print infilename, " could not be opened: %s" % str(e)

def usage(programname):
    """ Print usage information """
    print "Usage: %s [-h] [-P <dir>] <file> [<file> ..]" % (programname)
    print '''
This script opens the file(s) specified, and writes the index to the directory specified by -P.

Parameters:
 -h        Display this help
 -f file   Path to the file to index, must be present
 -P path   Directory to create the index file in (default: .)
'''


if __name__ == "__main__":
    try:
        main()
    except KeyboardInterrupt:
        print '\n ! Exiting.'