#!/usr/bin/python # secSorter.py import os,re # Variables theDir=os.listdir(".") comFile="sectionThreads.txt" comFileTmp=comFile+".tmp" # html to lines def getLines(): wComFileTmp=open(comFileTmp,"w") for uFile in theDir: if uFile.endswith(".html"): rData=open(uFile,"r") for line in rData: if "whoposted" in line: wComFileTmp.write(line) print uFile,"- Done" wComFileTmp.close() # Lines to thread ids def slimLines(): rComFileTmp=open(comFileTmp,"r") wComFile=open(comFile,"w") subPat="" patt1="\" onclick.*$" patt2="[^\d\n]" for line in rComFileTmp: skim_1=re.sub(patt1,subPat,line) skim_2=re.sub(patt2,subPat,skim_1) wComFile.write(skim_2) skim_2 wComFile.close() rComFileTmp.close() os.remove(comFileTmp) print " :: Finished OK :: " ##### Run The Stuff ###### # Run Functions getLines() slimLines()
Categories