#!/usr/bin/python24

"""This app searches through a folder containing files recovered
    when an fsck found errors on the hard drive and put some files
    into a lost+found folder.  Since there are thousands of files
    in the folder, I need an easy way to sort them so I can make 
    some decisions about whether I need to keep them or not.
    
    This app can find all mbox-type mail files, .png picture files, 
    and anything else. 
    """
    
import os, os.path, string, shutil, sys, re
try:
    import Tkinter, Pmw
except ImportError:
    print "can't find Tkinter files"

#user whose files we want to check
user = "stewart"
#debug flag
dbg=0
print "Starting up.."
#get user's home
home = os.environ.get('HOME',None)
#if root is running this script, force the answer
if (home == '/root') or (home is None):
    home = "/home/%s" % user
if dbg: print "%s is running this script" % home
#where are the files to investigate?
#workdir = os.path.join(home,'.thunderbird','default','d8j853iq.slt','Mail','Local Folders','Messages')
workdir = os.path.join(home,'lost+found')
os.chdir(workdir)

#create a dictionary to deal with all of the types of files we will try to find and process
fileDict = {}
"""elements in fileDict have the following properties:
    label   - label for this type of file (this is the key)
    destdir - the subdirectory where files of this type will be placed
    counter - counter for how many files of this type we've found
    extend  - filename extension to add to files of this type (or None)
    check   - check-string that will identify this type
    """
fileDict["mail1"] = ['mail',0,None,'Return-Path']
fileDict["mail2"] = ['mail',0,None,'Return-path']
fileDict["mail3"] = ['mail',0,None,'From ']
fileDict["python"] = ['python',0,'.py','import ']
fileDict["xml"] = ['xml',0,'.xml','<?xml ']
fileDict["png"] = ['pics',0,'.png','PNG ']
fileDict["html"] = ['html',0,'.html','<html']
fileDict["xpm"] = ['xpm',0,'.xpm','XPM']
fileDict["cpp"] = ['cpp',0,'.h','\/*']

folderList = []
checkList  = []
for key in fileDict.keys():
    destdir, counter, extend, check = fileDict[key]
    folderList.append(destdir)
    checkList.append(len(check))
    destdir = os.path.join(workdir, destdir)
    if not os.path.exists(destdir):
        os.mkdir(destdir)
        if dbg: print "created folder %s" % destdir

listdir = os.listdir(os.getcwd())
print "There are %i files to process" % len(listdir)
if dbg: print "The first file to open is %s" % listdir[0]

#list of all non-mail files' first lines
firstLines = []
#file containing all first lines
firstfile = "firstlines.txt"

#now step through all files and process them
#some files and folders can't be opened, so ignore them
initialLength = len(listdir)
removeList = []
for lis in listdir:
    try:
        fid = open(lis,'rb')
        fid.close()
        if (lis == firstfile) or (lis in folderList):
            try:
                removeList.append(lis)
                listdir.remove(lis)
            except ValueError:
                pass
    except IOError:
        removeList.append(lis)
        listdir.remove(lis)
print "We have %i files to process, of which %i are accessible" % (initialLength, len(listdir))
print "Ignoring the following folders, and inaccessible files:"
for rem in removeList:
    print " - %s" % rem

#we only need to read from the each file the length of the longest check-string:
checkList.sort(); checkList.reverse(); long = int(checkList[0])
for lis in listdir:
    haveone = 0
    try:
        fid = open(lis,'r')
        """elements in fileDict have the following properties:
            label   - label for this type of file (this is the key)
            destdir - the subdirectory where files of this type will be placed
            counter - counter for how many files of this type we've found
            extend  - filename extension to add to files of this type (or None)
            check   - check-string that will identify this type
            """
        data = fid.read(long)
        fid.close()
        for key in fileDict.keys():
            destdir, counter, extend, check = fileDict[key]
            if listdir.index(lis) <10:
                #just print out detail on the first 10 files, not on all of them
                if dbg: print "looking for %s in %s" % (check, data)
            if check in data:                   # if string.find(data,check) != -1
                #then we have a file of interest
                haveone += 1
                destfile = os.path.join(destdir,lis)
                if extend is None:
                    shutil.copyfile(lis, destfile)
                else:
                    shutil.copyfile(lis, destfile+extend)
                if haveone == 1: counter += 1
                print "we are at index", listdir.index(lis)
                if listdir.index(lis) <10:
                    #just print out detail on the first 10 files, not on all of them
                    if dbg: print "looking for %s in %s" % (check, data)
                    if dbg: print "trying to copy %s to %s" % (lis, destfile)
                    print "counter is now", counter
                if counter != 0: fileDict[key][1] = counter
                os.remove(lis)
                if dbg: print "moved %s" % lis
                break
        if not haveone:
            #then we have something else; grab its 1st line
            firstLines.append("%s %s" % (lis,data) )
    except IOError:
        pass
        
print "All done classifying files"
for key in fileDict.keys():
    destdir, counter, extend, check = fileDict[key]
    print "Copied %i files of type %s" % (counter, key)
    
print "there are %i non-mail files" % len(firstLines)
#put all first lines in a a file for viewing
if dbg: print "1st lines of all non-mail files are in %s" % firstfile
fout = open(firstfile,'w+')

#display first lines
root = Tkinter.Tk()
text = Pmw.ScrolledText(root)
text.insert('end', 'UNCLASSIFIED FILES: (Name, Initial data)\n\n')
for first in firstLines:
    fout.write(first+'\n')
    text.insert('end', first+'\n')
text.pack()
but = Tkinter.Button(root,text="Quit",command=sys.exit)
but.pack()
fout.close()
root.mainloop()