view harvestToPurl.py @ 0:08a476cdcfea

first
author dwinter
date Wed, 31 Oct 2012 14:57:53 +0100
parents
children caeede0c9464
line wrap: on
line source

'''
Created on 31.10.2012

@author: dwinter
'''

import manageIndexMetaPURLs



import os
from os.path import join, getsize
import sys
import re

def harvestIndexMeta(path,user,delpath="",replacepath=""):
    
    md = manageIndexMetaPURLs.IndexMetaPURLManager()
    
    for root, dirs, files in os.walk(path):
    
    
        for name in files:
            if name.endswith(".meta"):
                
                fl=join(root, name)
                fl=re.sub("^"+delpath,replacepath,fl) #loesche den teil vom path der mir delpath beginnt
                
                val,purl = md.register(fl, True, "", user)
                if val==manageIndexMetaPURLs.ALREADY_EXISTS:
                    print "found %s -> %s"%(fl,purl)
                else:
                    print "added %s -> %s"%(fl,purl)
        if 'pageimg' in dirs:
            dirs.remove('pageimg')  # don't visit pageimf
        for dir in dirs:
            if  dir== "pageimg":
                dirs.remove('pageimg')
            if  dir.startswith("."):
                dirs.remove(dir)




if __name__ == '__main__':
    args = sys.argv[1:]
    if not (len(args)==2 or len(args)==3 or len(args)==4):
        print "USAGE: python harvestToPurl.py path user (optional)pathPrefixToDelete (optional)replacedeleted"
        sys.exit(2)
    path=args[0]
    user=args[1]
    
    delpath=""
    replacepath=""
    
    if len(args)==3:
        delpath=args[2]
    elif len(args)==4:
        delpath=args[2]
        replacepath=args[3]
       
        
    if not os.path.exists(path):
        print "ERROR: path %s does not exist!"%path
        sys.exit(2)
        
    harvestIndexMeta(path,user,delpath=delpath,replacepath=replacepath)