From 6c2d3df3ff956481cacfee43276fd23a5fb872aa Mon Sep 17 00:00:00 2001
From: Holger Schurig <holgerschurig@gmx.de>
Date: Fri, 26 Jun 2009 10:20:10 +0200
Subject: [PATCH] new plugin: google_sitemap.py, which auto-generates a
 sitemap.xml

---
 in/plugins/google_sitemap.md | 38 +++++++++++++++++++
 in/plugins/index.md          |  3 +-
 in/webber.conf               |  4 ++
 plugins/google_sitemap.py    | 71 ++++++++++++++++++++++++++++++++++++
 4 files changed, 115 insertions(+), 1 deletion(-)
 create mode 100644 in/plugins/google_sitemap.md
 create mode 100644 plugins/google_sitemap.py

diff --git a/in/plugins/google_sitemap.md b/in/plugins/google_sitemap.md
new file mode 100644
index 0000000..913ec2e
--- /dev/null
+++ b/in/plugins/google_sitemap.md
@@ -0,0 +1,38 @@
+title: Google Sitemap generation
+linktitle: google_sitemap.py
+parent: Plugins
+keywords: Google, XML, Sitemap generator
+ctime: 2009-06-26
+
+This plugins write an XML "`sitemap.xml`" file into the out-directory. The
+format is documented at [[http://www.sitemaps.org|http://www.sitemaps.org]].
+
+
+= Configuration =
+
+The sitemap generator needs three [[Configuration]] items:
+
+	main_url: "www.holgerschurig.de"
+	sitemap_changefreq: "monthly"
+	sitemap_priority: 0.5
+
+
+== main_url ==
+
+This is the main URL of your website, without the "`http://`" stuff.
+
+
+== sitemap_changefreq ==
+
+You can define an estimated change frequency on each page by specifying
+this keyword at the header of each [[page|Page format]]. However, the
+"`sitemap_changefreq`" from the configuration file will be used as a
+default.
+
+
+== sitemap_priority ==
+
+You can define an relative page importance on each page by specifying
+this keyword at the header of each [[page|Page format]]. However, the
+"`sitemap_priority`" from the configuration file will be used as a
+default.
diff --git a/in/plugins/index.md b/in/plugins/index.md
index 588e653..176cde1 100644
--- a/in/plugins/index.md
+++ b/in/plugins/index.md
@@ -12,8 +12,9 @@ to plugins. Those plugins do:
 * Update internal state or modify HTML snippets
   ([[hierarchy.py|hierarchy]], [[link.py|link]])
 * Create HTML pages ([[template_mako.py|template_mako]])
+* Create additional output files ([[google_sitemap.py||Google Sitemap generation]])
 
-There's another plugin there ([[skeleton.py|skeleton]], which is
+There's another plugin there ([[skeleton.py|skeleton]]), which is
 is just a demo for plugin-programmers.
 
 Plugins simply reside in the "`plugins/`" directory. However, webber
diff --git a/in/webber.conf b/in/webber.conf
index 225a2a3..2a7866b 100644
--- a/in/webber.conf
+++ b/in/webber.conf
@@ -1,5 +1,6 @@
 template: "history"
 subtitle: "Webber"
+main_url: "www.holgerschurig.de/projects/webber"
 date_format: "%Y-%m.%d"
 input_encoding: "iso-8859-1"
 output_encoding: "iso-8859-1"
@@ -12,6 +13,7 @@ plugins: [
 	"read_copyonly",
 	"read_markdown",
 	"template_mako",
+	"google_sitemap",
 	]
 exclude_dir: [
 	]
@@ -27,3 +29,5 @@ copy_files: [
 	"*.css",
 	"robots.txt",
 ]
+sitemap_changefreq: "monthly"
+sitemap_priority: 0.5
diff --git a/plugins/google_sitemap.py b/plugins/google_sitemap.py
new file mode 100644
index 0000000..0db3df5
--- /dev/null
+++ b/plugins/google_sitemap.py
@@ -0,0 +1,71 @@
+# -*- coding: iso-8859-1 -*-
+from webber import *
+import os, sys, time
+
+f = None
+
+@set_hook("checkconfig")
+def checkconfig(params):
+	if not cfg.has_key("main_url"):
+		error('no "main_url:" configured:')
+		error('  this should be something like "www.yourpage.org"')
+		sys.exit(1)
+	if not cfg.has_key("sitemap_changefreq"):
+		warning('no default "sitemap_changefreq:" configured, using default "monthly"')
+		cfg.sitemap_changefreq = "monthly"
+	if not cfg.has_key("sitemap_priority"):
+		warning('no default "sitemap_priority:" configured, using default "0.5"')
+		cfg.sitemap_priority = "0.5"
+
+
+def write_initial(params):
+	global f
+	try:
+		os.makedirs(params.file.out_dir)
+	except:
+		pass
+	f = open(os.path.join(params.file.out_dir, "sitemap.xml"), "w")
+	print >>f, '<?xml version="1.0" encoding="UTF-8"?>'
+	print >>f, '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"'
+	print >>f, '        xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"'
+	print >>f, '        xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9'
+	print >>f, '                            http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd">'
+
+
+@set_hook("scan")
+def sitemap_scan(params):
+	global f
+	file = params.file
+	if not file.has_key("linktitle"):
+		return
+	if f is None:
+		write_initial(params)
+
+	# Sanity checks
+	ok = True
+	try:
+		prio = float(file.sitemap_priority)
+	except:
+		ok = False
+	if not ok or (prio < 0.0 or prio > 1.0):
+		error("%s: sitemap_priority '%s' is invalid" % (file.rel_path, file.sitemap_priority))
+		return
+	if file.sitemap_changefreq not in ("always", "hourly", "daily", "weekly", "monthly", "yearly", "never"):
+		error("%s: sitemap_changefreq '%s' is invalid" % (file.rel_path, file.sitemap_changefreq))
+		return
+
+	#print file.sitemap_priority, file.sitemap_changefreq, file.rel_path
+	f.write('<url>\n')
+	f.write('   <loc>http://%s/%s</loc>\n' % (file.main_url, file.rel_path))
+	f.write('   <lastmod>%s</lastmod>\n' % time.strftime( "%Y-%m-%d", time.localtime(file.mtime)) )
+	f.write('   <changefreq>%s</changefreq>\n' % file.sitemap_changefreq)
+	f.write('   <priority>%s</changefreq>\n' % file.sitemap_priority)
+	f.write('</url>\n')
+
+
+@set_hook("finish")
+def finish(params):
+	global f
+	if f:
+		print >>f, '</urlset>'
+		f.close()
-- 
2.39.2