Comments 6-5

some file

import os.path

from hashlib import md5
from pprint import pprint

# spacing is a bit wacked, should be 4 spaces per tab

def get_files():
        # let's make these generators
        for f in os.listdir('.'):
                if os.path.isfile(f) and f != "manifest.txt":
                        yield f
 
def get_files_recursive():
        # this too
        for (path, dirs, files) in os.walk('.'):
                for f in files:
                        if f != "manifest.txt":
                                yield (os.path.join(path,f))
        
  
def calc_hash(fname):
        m = md5()
        data = "tmp"
        with open(fname,'rb') as f:
                while len(data) > 0:
                        data = f.read(512000)
                        m.update(data)
        return m.hexdigest()
 
def output(data):
        x=0
        with open('manifest.txt','w') as f:
            # don't use hash -- its reserved
                for fn in sorted(data.keys()):
                        f.write("%s     %s\n" % (fn,data[fn]))
                        x+=1 # could use enumerate here if you wanted
 
def read_manifest():
        if not os.path.isfile('manifest.txt'):
            # let's make this an exception
                raise Exception("manifest.txt does not exist.")
                
        flist = {}
        with open('manifest.txt','r') as f:
                for line in f:
                    # let's use rpartition
                    # that way names can have spaces
                        fn, _, h = line.rpartition(' ')
                        flist[fn.strip()] = h.strip()
        return flist
 
def check_manifest():
    # a little strange since this only looks at the files that then existed
    # and not any adds to the directory,
    # but I guess that is what you wanted?
        changed = {}
        # changed to dict
        flist = read_manifest()
        for fn in flist.keys():
                oldh = flist[fn]
                newh = calc_hash(fn)
                if newh != oldh:
                        changed[fn] = newh
        # i'd return changed and let the calling code figure it out
        return changed

 
def build_manifest(recursive = False):
    # let's use a dict
    # that way we don't have to worry about it being matchy matchy
        hlist = {}
        # hacked up since I am not running with args,
        # but I would do the parsing in main anyway
        if recursive:
            # see what I do here
                flist = get_files_recursive
        else:
                flist = get_files
        for fn in flist():
            hlist[fn] = calc_hash(fn)
        output(hlist)
 
def main():
    # put parsing in here maybe
    pprint(check_manifest())

if __name__ == '__main__':
    main()