#!/usr/bin/env python
# Author:Anand babu
# Date: 26 Jun 2008
# Updates freeipmi download counter through cron
# License: GNU GPL v3 or later

import string
import os
import gzip
import glob
import md5
import pickle

OUTPUT_FILE="/home/www.gluster.org/glusterfs-download-counter.txt"
ROOT_PATH="/home/ftp/pub/gluster/glusterfs"
LOG_WEB="/var/log/apache/ftp.zresearch.com-access.log*"
LOG_FTP="/var/log/xferlog*"
## PATTERNS: glusterfs-n.n.n.tar.gz, glusterfs-CURRENT.tar.gz, glusterfs-n.n.n.tar.gz, glusterfs-server*.rpm, glusterfs-debuginfo*rpm
RELEASE_LIST_CMD='find '+ROOT_PATH+' -regex ".*glusterfs-[0-9Csd].*\.rpm" -or -regex ".*glusterfs-[a-zA-Z0-9].*\.tar\.gz"'
COUNTERDB_FILE='/var/cache/glusterfs-download-counter/counter-db.pkl'

class CounterDB:
    cdb={}
    def loadDB(self):
        try:
            fobj=open(COUNTERDB_FILE, 'rb')
            self.cdb=pickle.load(fobj)
            fobj.close()
        except IOError:
            self.cdb={}

    def saveDB(self):
        fobj=open(COUNTERDB_FILE, 'wb')
        pickle.dump(self.cdb, fobj)
        fobj.close();
        
    def set(self, hash, hits):
        self.cdb[hash]=hits

    def get(self, hash):
        try:
            return self.cdb[hash]
        except KeyError:
            return -1

class GlusterDownloadCounter:
    release_files=[]
    totalDownloads=0
    cdb=CounterDB()
    
    def __init__(self):
	fd=os.popen(RELEASE_LIST_CMD)
	_release_files=fd.read().split('\n')[:-1]
	fd.close()
        for release in _release_files:
            _release=release.split("/")[-1]
            self.release_files.append(_release)
        self.cdb.loadDB()

    def md5sumFile(self, filename):
        fobj=open(filename)
        mobj = md5.new()
        while True:
            block = fobj.read(8096)
            if not block:
                break
            mobj.update(block)
        fobj.close()
        return mobj.hexdigest()
        
    def readLog(self, filename):
        if filename.endswith(".gz"):
            fobj = gzip.GzipFile(filename, 'rb')
        else:
            fobj = open(filename, 'rb')
        pattern=fobj.read()
        fobj.close()
        return pattern.split('\n')
        
    def updateDownloads(self):
        self.totalDownloads=0
        for fname in glob.glob(LOG_FTP):
            perFileCount=0
            md5=self.md5sumFile(fname)
            count=self.cdb.get(md5)
            if count>=0:
                self.totalDownloads=self.totalDownloads+count
                continue
            lines=self.readLog(fname)
            for line in lines:
                if line.find("glusterfs-")<0:
                    continue
                for release in self.release_files:
                    if line.find(release+" ")>=0:
                        self.totalDownloads=self.totalDownloads+1
                        perFileCount=perFileCount+1
                        break
            self.cdb.set(md5, perFileCount)
            
        for fname in glob.glob(LOG_WEB):
            perFileCount=0
            md5=self.md5sumFile(fname)
            count=self.cdb.get(md5)
            if count>=0:
                self.totalDownloads=self.totalDownloads+count
                continue            
            lines=self.readLog(fname)
            for line in lines:
                if line.find("glusterfs-")<0:
                    continue
                for release in self.release_files:
                    if line.find(release+" ")>=0:
                        if line.find("\" 200 ")>=0:
                            self.totalDownloads=self.totalDownloads+1
                            perFileCount=perFileCount+1
                            break
            self.cdb.set(md5, perFileCount)
        self.cdb.saveDB()
        return self.totalDownloads

    def fileUpdate(self, filename):
        fobj=open(filename, "w")
        fobj.write(str(self.updateDownloads()))
        fobj.close()

GlusterDownloadCounter().fileUpdate(OUTPUT_FILE)