#!/usr/bin/python
# Stephen Skory
# stephenskory@yahoo.com
# Part of a hack to make Gallery2 work with Amazon's S3 service

# uncomment this below if you want to do some debugging.
#print "Content-Type: text/plain"
#print

# User edited values
AWSBase = 'http://bucketname.s3.amazonaws.com/' # yes trailing slash
albumsAWSBase = AWSBase + 'albums/' # yes trailing slash
derivativeAWSBase = AWSBase + 'derivative/' # yes trailing slash
s3fsBase = '/path/to/bucket/' # yes trailing slash
galleryBase = '/gallery2/' # your web location for gallery2, minus your domain name
                           # if your gallery is at http://mysite.com/gallery2/main.php
                           # put /gallery2/ here, with the trailing slash
table_prefix = 'g2_'
column_prefix = 'g_'
refreshTime = '180' # days
myHost = 'localhost' # 99% of the time
myUser = 'database_username'
myPasswd = 'database_password'
myDB = 'database_name'

# --- EDIT ABOVE --- #

# get the needed modules
import cgi, time, sys, re, MySQLdb, urllib2
#import cgitb; cgitb.enable() # for debugging
from path import path # for 'touch'ing files, requries path.py: http://www.jorendorff.com/articles/python/path/

# if the image is a full sized image, get the pieces of its pathname
def fullSizedImage(numbers):
    path = []
    # loop over the numbers in its location
    for number in numbers:
        if (number != '7'): # 7 is the top album
            db.query("""SELECT %spathComponent from %sFileSystemEntity where g_id=%s""" \
            % (column_prefix,table_prefix,number))
            r = db.store_result()
            # store the results in path
            path.append(r.fetch_row()[0][0])
    # get the file name
    db.query("""SELECT %spathComponent from %sFileSystemEntity where g_id=%s""" \
    % (column_prefix,table_prefix,line))
    r = db.store_result()
    for row in r.fetch_row(1):
        path.append(row[0])
    
    # make the path to the full S3 version
    string = ''
    string = "/".join("%s" % part for part in path)
    string = albumsAWSBase + string
    return string

# if the id is for a reduced size image, it's easy to make the S3 location
def reducedSizeImage(line):
    if (int(line) <= 100): # the id=100 is a weird case I'm not 100% on...
        string = derivativeAWSBase + '0/%s/' % line[0]
        string = string + line.strip() + '.dat'
    else:
        string = derivativeAWSBase + '%s/%s/' % (line[0], line[1])
        string = string  + line.strip() + '.dat'
    return string

# we call this after we've formed a S3 location to see if it's there already.
# if it is and it's new enough, we return 0 and preserve the S3 version
# otherwise we return 1 and will create a new local location
def chooseRefreshedVersion(Location):
    # try to get the file off of S3
    req = urllib2.Request(Location)
    try:
        url_handle = urllib2.urlopen(req)
    # if it's not there, return 1
    except urllib2.HTTPError:
        return 1
    # otherwise, get the file headers
    headers = url_handle.info()
    # get the date of last change for the S3 version
    last_modified = headers.getheader("Last-Modified")
    time_mod = time.mktime(time.strptime(last_modified, "%a, %d %b %Y %H:%M:%S %Z"))
    # now
    time_now = time.mktime(time.localtime())
    # calculate the difference
    timeDiff = time_now - time_mod
    # calculate the refresh time in seconds
    rTimeSecs = 60*60*24*refreshTime
    # if the S3 file is too old, return 1
    # but also we'll 'touch' it to 'refresh' it. If it needs regeneration, Gallery will do it
    # because we're going through the 'official' channel, so an extra touch won't hurt it
    if (timeDiff > rTimeSecs):
        # take off the s3 address part
        localFileName = sb.subn('',Location)[0]
        # add the local stuff
        localFileName = s3fsBase + localFileName
        # create f under path framework
        f = path(localFileName)
        # touch it
        f.touch()
        return 1
    # if the S3 version is not too old, or if it exists on S3, return 0
    return 0

# Establish a database connection
# yeah, yeah, it's database specific. You get what you pay for!
db = MySQLdb.connection(host=myHost,
                user=myUser,
                passwd=myPasswd,
                db=myDB)

# Make some regular expressions
# to find the numbers from parentSequence
p = re.compile('\d+')
# to strip off stuff that's not a number
D = re.compile('\D')
# for updating refresh time
sb = re.compile(AWSBase)

# get the values from the request
inputValue = cgi.FieldStorage()

# get the ID
if(inputValue.has_key('id')):
    id = inputValue['id'].value
    line = inputValue['id'].value

# get the serial number
if(inputValue.has_key('sn')):
    sn = inputValue['sn'].value

# get the file name
if(inputValue.has_key('fn')):
    fn = inputValue['fn'].value

# clean the line of anything but digits, i.e. SQL injection queries
line = D.subn('',line)[0]

# if line is too long, or there's nothing, let's just exit 'cause something's wrong
# 20 is an arbitrary number
if (len(line) > 20 or len(line) == 0):
    sys.exit()

# Make the SQL query
db.query("""SELECT %sparentSequence FROM %sItemAttributesMap WHERE g_itemId=%s""" \
% (column_prefix,table_prefix,line))
r = db.store_result()
string = ''
# if the id queried is in ItemAttributesMap, it's a full sized image,
# on error (as in nothing returned), we know it's a resized version, and make the path
try:
    numbers = p.findall(r.fetch_row()[0][0])
except IndexError:
    string = reducedSizeImage(line)
# here, if we haven't filled 'string' it's a full sized image
if (len(string) == 0):
    string = fullSizedImage(numbers)

# after making the S3 version, we test to see if it's fresh or not, or even on S3
if (chooseRefreshedVersion(string)):
    string = '%smain.php?g2_view=core.DownloadItem&g2_itemId=%s&g2_serialNumber=%s&g2_fileName=%s' \
    % (galleryBase,id,sn,fn)

# after all that, we forward the client to the correct address
print 'Location: %s' % string
print