From dc0493f99bdcbdb8f8e356d0a777ee0ea9b326de Mon Sep 17 00:00:00 2001
From: Richard Harding <rharding@mitechie.com>
Date: Fri, 29 Nov 2013 12:08:26 -0500
Subject: [PATCH] Update to catch back up to craig's image helper

---
 readability/scoring.py                               | 10 +++++++++-
 readability/scripts/client.py                        |  5 ++++-
 tests/test_articles/test_businessinsider-com/test.py |  2 +-
 tests/test_articles/test_sweetshark/test.py          |  2 +-
 4 files changed, 15 insertions(+), 4 deletions(-)

diff --git a/readability/scoring.py b/readability/scoring.py
index 65344c3..3b90db8 100644
--- a/readability/scoring.py
+++ b/readability/scoring.py
@@ -90,7 +90,15 @@ def get_link_density(node, node_text=None):
     if text_length == 0:
         return 0.0
 
-    links_length = sum(map(_get_normalized_text_length, node.findall(".//a")))
+
+    link_length = sum([len(a.text_content()) or 0
+        for a in node.findall(".//a")])
+
+    # For each img, give 50 bonus chars worth of length.
+    # Tweaking this 50 down a notch should help if we hit false positives.
+    links_length = max(link_length -
+                      sum([50 for img in node.findall(".//img")]), 0)
+
     return links_length / text_length
 
 
diff --git a/readability/scripts/client.py b/readability/scripts/client.py
index acb1783..8025b98 100644
--- a/readability/scripts/client.py
+++ b/readability/scripts/client.py
@@ -37,7 +37,10 @@ from ..readable import Article
 
 
 HEADERS = {
-    "User-Agent": "Readability (Readable content parser) Version/%s" % __version__,
+    "User-Agent": 'breadability/{version} ({url})'.format(
+        url="https://github.com/bookieio/breadability",
+        version=__version__
+    )
 }
 
 
diff --git a/tests/test_articles/test_businessinsider-com/test.py b/tests/test_articles/test_businessinsider-com/test.py
index 54d5570..0d850be 100644
--- a/tests/test_articles/test_businessinsider-com/test.py
+++ b/tests/test_articles/test_businessinsider-com/test.py
@@ -5,7 +5,7 @@ try:
 except ImportError:
     import unittest
 
-from breadability.readable import Article
+from readability.readable import Article
 
 
 class TestBusinessInsiderArticle(unittest.TestCase):
diff --git a/tests/test_articles/test_sweetshark/test.py b/tests/test_articles/test_sweetshark/test.py
index e4e498c..7580e05 100644
--- a/tests/test_articles/test_sweetshark/test.py
+++ b/tests/test_articles/test_sweetshark/test.py
@@ -5,7 +5,7 @@ try:
 except ImportError:
     import unittest
 
-from breadability.readable import Article
+from readability.readable import Article
 
 
 class TestSweetsharkBlog(unittest.TestCase):