Merge pull request #105 from mailgun/sergey/fromstring

html with comment that has no parent crashes html_tree_to_text

setup.py

@@ -29,7 +29,7 @@ class InstallCommand(install):
 
 
 setup(name='talon',
-      version='1.2.15',
+      version='1.2.16',
       description=("Mailgun library "
                    "to extract message quotations and signatures."),
       long_description=open("README.rst").read(),

talon/utils.py

@@ -117,7 +117,13 @@ def html_tree_to_text(tree):
         style.getparent().remove(style)
 
     for c in tree.xpath('//comment()'):
-        c.getparent().remove(c)
+        parent = c.getparent()
+
+        # comment with no parent does not impact produced text
+        if not parent:
+            continue
+
+        parent.remove(c)
 
     text   = ""
     for el in tree.iter():

tests/utils_test.py

@@ -6,6 +6,7 @@ from . import *
 from talon import utils as u
 import cchardet
 import six
+from lxml import html
 
 
 def test_get_delimiter():
@@ -107,3 +108,9 @@ font: 13px 'Lucida Grande', Arial, sans-serif;
 <p>TEXT 2 <!-- COMMENT 2 --></p>
 </div>"""
     eq_(b'TEXT 1 \nTEXT 2', u.html_to_text(html))
+
+
+def test_comment_no_parent():
+    s = "<!-- COMMENT 1 --> no comment"
+    d = html.document_fromstring(s)
+    eq_("no comment", u.html_tree_to_text(d))