Merge pull request #586 from jerome-poisson/SD-5475_2

[NITF] [FORMATTER] fixed bad handling of root element
superdesk · Sep 26, 2016 · a7f9258 · a7f9258
2 parents cc00473 + fd6bd41
commit a7f9258
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 17 deletions.
diff --git a/superdesk/publish/formatters/nitf_formatter.py b/superdesk/publish/formatters/nitf_formatter.py
@@ -147,21 +147,25 @@ def html2nitf(self, html_elem, root_elem=True):
         """convert HTML elements to NITF compatible elements
 
         :param ET.Element: HTML to clean/transform
-        :param bool: True if its the main element (i.e. must no be deleted)
-        :return ET.Element: NITF compliant element
+        :param bool: True if its the main element (must be a <div>)
+        :return ET.Element: <div> element with NITF compliant children
         """
-        if html_elem.tag not in self.HTML2NITF:
-            if root_elem:
-                html_elem.tag = 'p'
-            else:
-                raise ValueError("Unhandled HTML element")
-        else:
+        if root_elem:
+            assert html_elem.tag == 'div'
+            # we change children of root element in place
+            for c in html_elem:
+                self.html2nitf(c, root_elem=False)
+            return html_elem
+
+        try:
             nitf_map = self.HTML2NITF[html_elem.tag]
-            nitf_elem = nitf_map.get('nitf')
-            if nitf_elem is not None:
-                if nitf_elem == '':
-                    raise ValueError("Element need to be removed")
-                html_elem.tag = nitf_elem
+        except KeyError:
+            raise ValueError("Unhandled HTML element")
+        nitf_elem = nitf_map.get('nitf')
+        if nitf_elem is not None:
+            if nitf_elem == '':
+                raise ValueError("Element need to be removed")
+            html_elem.tag = nitf_elem
 
         html_elem.attrib.update(nitf_map.get('attrib', {}))
 

diff --git a/tests/publish/nitf_formatter_tests.py b/tests/publish/nitf_formatter_tests.py
@@ -61,15 +61,15 @@ def test_formatter(self):
 
     def test_html2nitf(self):
         html = etree.fromstring(
-            '<p><strong>this text should be <i>modified</i></strong> so '
+            '<div><p><strong>this text should be <i>modified</i></strong> so '
             '<span>[this should not be removed]</span> unkown <em unknown_attribute="toto">'
             'elements</em> and <a bad_attribute="to_remove">attributes</a> are <h6>'
-            'removed</h6></p>')
+            'removed</h6></p></div>')
 
         nitf = self.formatter.html2nitf(html)
-        expected = ('<p><em class="bold">this text should be <em class="italic">modified</em>'
+        expected = ('<div><p><em class="bold">this text should be <em class="italic">modified</em>'
                     '</em> so [this should not be removed] unkown <em class="italic">elements</em>'
-                    ' and <a>attributes</a> are <hl2>removed</hl2></p>')
+                    ' and <a>attributes</a> are <hl2>removed</hl2></p></div>')
         self.assertEqual(etree.tostring(nitf, 'unicode'), expected)
 
     def test_company_codes(self):