#!/usr/bin/python

# Python script for generating sitemap.xml and sitemapindex.xml files.
#
# Based on protocol specification here:
# http://www.sitemaps.org/protocol.php
#
# Directory walking classes adapted from recipe at ActiveState.com
# http://code.activestate.com/recipes/528904/
#
# Hugh Brown, 14 Jan 2009
# http://www.iwebthereforeiam.com/iwebthereforeiam/

import os, os.path

class Walker(object):
	def __init__(self, dir, executeHook=None, excludeDirs=[]):
		self.dir = dir
		self.executeHook = executeHook
		self.excludeDirs = excludeDirs

	def isValidFile(self, fileName):
		return True

	def isValidDir(self, dir):
		# Remove self.dir prefix (exclusions don't apply to the supplied root dir for walking)
		if dir.startswith( self.dir ):
			dir = dir[len(self.dir):]
		# Check all sub-directories in the path of the file
		subdirs = dir.split('/')
		return not any(subdir in self.excludeDirs for subdir in subdirs)

	def executeFile(self, path, args):
		if self.executeHook and self.isValidFile(path) :
			self.executeHook(self, path, args)
		# else subclass Walker and override executeFile

	def execute(self, args):
		for root, dirs, fileNames in os.walk(self.dir):
			for exdir in self.excludeDirs:
				if exdir in dirs:
					dirs.remove(exdir)
			for fileName in fileNames:
				path = os.path.join(root, fileName)
				self.executeFile(path, args)

class ReWalker(Walker):
	def __init__(self, prefix, dir, fileMatchRe, executeHook=None, excludeDirs=[]):
		import re
		Walker.__init__(self, dir, executeHook, excludeDirs)
		self.fileMatchPat = re.compile(fileMatchRe)
		self.prefix = prefix

	def isValidFile(self, fileName):
		return self.fileMatchPat.match(fileName)

# Utility functions
def urlQuote(path) :
	import urllib
	return urllib.quote(path)
def lastModTime(path) :
	import time
	ctime = os.path.getmtime(path)
	gmtime = time.gmtime(ctime)
	return time.strftime("%Y-%m-%dT%H:%M:%S+00:00", gmtime)

def generateUrl(walker, path, args):
	fileOffset = path[ len(walker.dir) : ].replace("\\", "/")
	htmlPath = os.path.join(walker.prefix, fileOffset)

	s = '\t<url>\n'
	s += "\t\t<loc>%s</loc>\n" % urlQuote(htmlPath)
	s += "\t\t<lastmod>%s</lastmod>\n" % lastModTime(path)
	##s += "\t\t<changefreq>monthly</changefreq>\n"
	##s += "\t\t<priority>0.5</priority>"
	s += '\t</url>\n'
	args.write(s)

class SiteMapper(object) :
	def __init__(self, wwwPrefix, startDir, includeFiles) :
		self.wwwPrefix, self.startDir, self.includeFiles = wwwPrefix, startDir, includeFiles
		self.sitemapHttp = os.path.join(self.wwwPrefix, "sitemap.xml")
		self.sitemapPath = os.path.join(self.startDir, "sitemap.xml")
		excludeDirs = [ ]
		self.walker = ReWalker(wwwPrefix, startDir, includeFiles, generateUrl, excludeDirs)
	def generateMap(self) :
		f = open (self.sitemapPath, "w")
		f.write('<?xml version="1.0" encoding="UTF-8"?>\n')
		f.write('<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n')
		self.walker.execute(f)
		f.write('</urlset>\n')
		f.close()

class SiteMapIndex(object) :
	def __init__(self, root, maps) :
		self.root, self.maps = root, maps
		self.siteMapIndex = os.path.join(self.root, "sitemapindex.xml")
	def generateIndex(self) :
		f = open(self.siteMapIndex, "w")
		s = '<?xml version="1.0" encoding="UTF-8"?>\n'
		s += '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n'

		for map in maps :
			sm = SiteMapper(map["wwwPrefix"], map["startDir"], map["includeFiles"])
			sm.generateMap()
			s += "\t<sitemap>\n"
			s += "\t\t<loc>%s</loc>\n" % sm.sitemapHttp
			#s += "\t\t<lastmod>%s</lastmod>" % lastModTime(sm.sitemapPath)
			s += "\t</sitemap>\n"

		s += '</sitemapindex>\n'
		f.write(s)
		f.close()

if __name__ == "__main__":
	root = "/vservers/hughdbrown/htdocs/"
	iwebthereforeiam = dict(zip(("wwwPrefix", "startDir", "includeFiles"), ("http://www.iwebthereforeiam.com/iwebthereforeiam/", '/vservers/hughdbrown/htdocs/iwebthereforeiam/', r".*\.html$")))
	maps = [ iwebthereforeiam ]
	smi = SiteMapIndex(root, maps)
	smi.generateIndex()