From 7151ac23fd2f2e04c69d1fb642081201c8a16ce6 Mon Sep 17 00:00:00 2001
From: ryanss
Date: Mon, 12 Jan 2015 23:59:00 -0500
Subject: Add readability parsing for reading linked pages

---
 plugin/readability/url_helpers.py | 52 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 52 insertions(+)
 create mode 100755 plugin/readability/url_helpers.py

(limited to 'plugin/readability/url_helpers.py')

diff --git a/plugin/readability/url_helpers.py b/plugin/readability/url_helpers.py
new file mode 100755
index 0000000..8234c8d
--- /dev/null
+++ b/plugin/readability/url_helpers.py
@@ -0,0 +1,52 @@
+import logging
+from urlparse import urlparse
+
+def host_for_url(url):
+	"""
+	>>> host_for_url('http://base/whatever/fdsh')
+	'base'
+	>>> host_for_url('invalid')
+	"""
+	host = urlparse(url)[1]
+	if not host:
+		logging.error("could not extract host from URL: %r" % (url,))
+		return None
+	return host
+
+def absolute_url(url, base_href):
+	"""
+	>>> absolute_url('foo', 'http://base/whatever/ooo/fdsh')
+	'http://base/whatever/ooo/foo'
+
+	>>> absolute_url('foo/bar/', 'http://base')
+	'http://base/foo/bar/'
+
+	>>> absolute_url('/foo/bar', 'http://base/whatever/fdskf')
+	'http://base/foo/bar'
+
+	>>> absolute_url('\\n/foo/bar', 'http://base/whatever/fdskf')
+	'http://base/foo/bar'
+
+	>>> absolute_url('http://localhost/foo', 'http://base/whatever/fdskf')
+	'http://localhost/foo'
+	"""
+	url = url.strip()
+	proto = urlparse(url)[0]
+	if proto:
+		return url
+
+	base_url_parts = urlparse(base_href)
+	base_server = '://'.join(base_url_parts[:2])
+	if url.startswith('/'):
+		return base_server + url
+	else:
+		path = base_url_parts[2]
+		if '/' in path:
+			path = path.rsplit('/', 1)[0] + '/'
+		else:
+			path = '/'
+		return base_server + path + url
+
+if __name__ == '__main__':
+	import doctest
+	doctest.testmod()
\ No newline at end of file
-- 
cgit v1.2.3