From f3293db7d91e53e161ca216a0bf67c2773cda4d2 Mon Sep 17 00:00:00 2001
From: Tommy Allen <tommy@esdf.io>
Date: Wed, 30 Dec 2015 01:42:13 -0500
Subject: [PATCH] Ignore <script> tag in content text

---
 microdata.py           | 2 ++
 test-data/example.html | 3 +++
 test.py                | 3 +++
 3 files changed, 8 insertions(+)

diff --git a/microdata.py b/microdata.py
index 906223c..6226ed7 100755
--- a/microdata.py
+++ b/microdata.py
@@ -209,6 +209,8 @@ def _text(e):
     chunks = []
     if e.nodeType == e.TEXT_NODE:
         chunks.append(e.data)
+    elif e.tagName == 'script':
+        return ''
     for child in e.childNodes:
         chunks.append(_text(child))
     return ''.join(chunks)
diff --git a/test-data/example.html b/test-data/example.html
index fbc1559..79eaf50 100644
--- a/test-data/example.html
+++ b/test-data/example.html
@@ -15,6 +15,9 @@
         <span itemprop="streetAddress">
           20341 Whitworth Institute
           405 N. Whitworth
+          <script>
+            // Unrelated text
+          </script>
         </span>
         <span itemprop="addressLocality">Seattle</span>,
         <span itemprop="addressRegion">WA</span>
diff --git a/test.py b/test.py
index 699b5d1..dc00a9c 100644
--- a/test.py
+++ b/test.py
@@ -41,6 +41,9 @@ def test_parse(self):
         self.assertEqual(item.address.itemtype, [URI("http://schema.org/PostalAddress")])
         self.assertTrue(item.address.addressLocality, "Seattle")
 
+        # <script> tag should be ignored in the content text
+        self.assertFalse("Unrelated text" in item.address.streetAddress)
+
         # json
         i = json.loads(item.json())
         self.assertEqual(i["properties"]["name"][0], "Jane Doe")