gen_sitemap

#!/usr/bin/python

import sys, os
import datetime
import xml.dom.minidom

online_url = "http://www.cosmozhang.com"

doc = xml.dom.minidom.Document()
urlsetEle = doc.createElement('urlset')
urlsetEle.setAttribute("xmlns", "http://www.sitemaps.org/schemas/sitemap/0.9")
urlsetEle.setAttribute("xmlns:mobile", "http://www.baidu.com/schemas/sitemap-mobile/1/")
doc.appendChild(urlsetEle)

def traverseDir(path):
    if not os.path.exists(path):
        return []
    elif os.path.isdir(path):
        files = []
        for subpath in os.listdir(path):
            subpath = os.path.join(path, subpath)
            files += traverseDir(subpath)
        return files
    elif os.path.isfile(path):
        return [path]
    return []

cwd = os.getcwd()
site_path = os.path.join(cwd, '_site')
if (site_path[-1] == '/'):
    site_path = site_path[0:-1]

all_files = map(lambda x: os.path.join(site_path, x), [
    "index.html", "newbies.html", "about.html", "blog/index.html"
]);

priorities = [1.0, 0.9, 0.9, 0.9]

if os.path.exists(site_path):
    for dirpath in os.listdir(site_path):
        dirfullpath = os.path.join(site_path, dirpath)
        if os.path.isdir(dirfullpath) and not (dirpath in ["blog", "public", "stylesheets"]):
            all_files += traverseDir(dirfullpath)

def filepath_to_xml(path, priority = 0.8):
    url = online_url + path.replace(site_path, '')
    time = datetime.datetime.fromtimestamp(os.path.getmtime(path)).strftime("%Y-%m-%d")
    node = doc.createElement("url")
    snode = doc.createElement("loc")
    snode.appendChild(doc.createTextNode(url))
    node.appendChild(snode)
    snode = doc.createElement("mobile:mobile")
    snode.setAttribute("type", "pc,mobile")
    node.appendChild(snode)
    snode = doc.createElement("lastmod")
    snode.appendChild(doc.createTextNode(time))
    node.appendChild(snode)
    snode = doc.createElement("changefreq")
    snode.appendChild(doc.createTextNode("daily"))
    node.appendChild(snode)
    snode = doc.createElement("priority")
    snode.appendChild(doc.createTextNode("%1.1f" % priority))
    node.appendChild(snode)
    return node

nodes = map(filepath_to_xml, all_files)

for i in range(0,len(priorities)):
    nodes[i] = filepath_to_xml(all_files[i], priorities[i])

for n in nodes:
    urlsetEle.appendChild(n)

f = open('./sitemap.xml', 'w')
doc.writexml(f, addindent='  ', newl='\n', encoding='utf-8')