Add options to remove descriptive elements (#102)

* --remove-titles (removes <title>)
* --remove-descriptions (removes <desc>)
* --remove-descriptive-elements (removes <title>, <desc> and <metadata>)

scour/scour.py

@@ -910,11 +910,25 @@ def removeNamespacedElements(node, namespaces):
          num += removeNamespacedElements(child, namespaces)
    return num
 
-def removeMetadataElements(doc):
+def removeDescriptiveElements(doc, options):
+   elementTypes = []
+   if options.remove_descriptive_elements:
+      elementTypes.extend(("title", "desc", "metadata"))
+   else:
+      if options.remove_titles:
+         elementTypes.append("title")
+      if options.remove_descriptions:
+         elementTypes.append("desc")
+      if options.remove_metadata:
+         elementTypes.append("metadata")
+   if not elementTypes:
+      return
+
    global numElemsRemoved
    num = 0
-   # clone the list, as the tag list is live from the DOM
+   elementsToRemove = []
-   elementsToRemove = [element for element in doc.documentElement.getElementsByTagName('metadata')]
+   for elementType in elementTypes:
+      elementsToRemove.extend(doc.documentElement.getElementsByTagName(elementType))
 
    for element in elementsToRemove:
       element.parentNode.removeChild(element)
@@ -3059,9 +3073,8 @@ def scourString(in_string, options=None):
       else:
          print("WARNING: {}".format(errmsg), file = options.ensure_value("stdout", sys.stdout))
 
-   # remove <metadata> if the user wants to
+   # remove descriptive elements
-   if options.remove_metadata:
+   removeDescriptiveElements(doc, options)
-      removeMetadataElements(doc)
 
    # for whatever reason this does not always remove all inkscape/sodipodi attributes/elements
    # on the first pass, so we do it multiple times
@@ -3142,7 +3155,7 @@ def scourString(in_string, options=None):
 
    # remove empty defs, metadata, g
    # NOTE: these elements will be removed if they just have whitespace-only text nodes
-   for tag in ['defs', 'metadata', 'g'] :
+   for tag in ['defs', 'title', 'desc', 'metadata', 'g'] :
       for elem in doc.documentElement.getElementsByTagName(tag) :
          removeElem = not elem.hasChildNodes()
          if removeElem == False :
@@ -3369,9 +3382,18 @@ _option_group_document = optparse.OptionGroup(_options_parser, "SVG document")
 _option_group_document.add_option("--strip-xml-prolog",
    action="store_true", dest="strip_xml_prolog", default=False,
    help="won't output the XML prolog (<?xml ?>)")
+_option_group_document.add_option("--remove-titles",
+   action="store_true", dest="remove_titles", default=False,
+   help="remove <title> elements")
+_option_group_document.add_option("--remove-descriptions",
+   action="store_true", dest="remove_descriptions", default=False,
+   help="remove <desc> elements")
 _option_group_document.add_option("--remove-metadata",
    action="store_true", dest="remove_metadata", default=False,
    help="remove <metadata> elements (which may contain license/author information etc.)")
+_option_group_document.add_option("--remove-descriptive-elements",
+   action="store_true", dest="remove_descriptive_elements", default=False,
+   help="remove <title>, <desc> and <metadata> elements")
 _option_group_document.add_option("--enable-comment-stripping",
    action="store_true", dest="strip_comments", default=False,
    help="remove all comments (<!-- -->)")

testscour.py

@@ -141,15 +141,59 @@ class NoAdobeXPathElements(unittest.TestCase):
 			lambda e: e.namespaceURI != 'http://ns.adobe.com/XPath/1.0/'), False,
 			'Found Adobe XPath elements' )
 
+class DoNotRemoveTitleWithOnlyText(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/descriptive-elements-with-text.svg')
+		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'title')), 1,
+			'Removed title element with only text child' )
+
+class RemoveEmptyTitleElement(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/empty-descriptive-elements.svg')
+		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'title')), 0,
+			'Did not remove empty title element' )
+
+class DoNotRemoveDescriptionWithOnlyText(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/descriptive-elements-with-text.svg')
+		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'desc')), 1,
+			'Removed description element with only text child' )
+
+class RemoveEmptyDescriptionElement(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/empty-descriptive-elements.svg')
+		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'desc')), 0,
+			'Did not remove empty description element' )
+
 class DoNotRemoveMetadataWithOnlyText(unittest.TestCase):
 	def runTest(self):
-		doc = scour.scourXmlFile('unittests/metadata-with-text.svg')
+		doc = scour.scourXmlFile('unittests/descriptive-elements-with-text.svg')
 		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'metadata')), 1,
 			'Removed metadata element with only text child' )
 
 class RemoveEmptyMetadataElement(unittest.TestCase):
 	def runTest(self):
-		doc = scour.scourXmlFile('unittests/empty-metadata.svg')
+		doc = scour.scourXmlFile('unittests/empty-descriptive-elements.svg')
+		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'metadata')), 0,
+			'Did not remove empty metadata element' )
+
+class DoNotRemoveDescriptiveElementsWithOnlyText(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/descriptive-elements-with-text.svg')
+		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'title')), 1,
+			'Removed title element with only text child' )
+		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'desc')), 1,
+			'Removed description element with only text child')
+		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'metadata')), 1,
+			'Removed metadata element with only text child' )
+
+class RemoveEmptyDescriptiveElements(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/empty-descriptive-elements.svg')
+		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'title')), 0,
+			'Did not remove empty title element' )
+		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'desc')), 0,
+			'Did not remove empty description element' )
 		self.assertEqual(len(doc.getElementsByTagNameNS(SVGNS, 'metadata')), 0,
 			'Did not remove empty metadata element' )
 
@@ -1152,13 +1196,34 @@ class PathImplicitLineWithMoveCommands(unittest.TestCase):
 		self.assertEqual( path.getAttribute('d'), "m100 100v100m200-100h-200m200 100v-100",
 			"Implicit line segments after move not preserved")
 
+class RemoveTitlesOption(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/full-descriptive-elements.svg',
+			scour.parse_args(['--remove-titles']))
+		self.assertEqual(doc.childNodes.length, 1,
+			'Did not remove <title> tag with --remove-titles')
+
+class RemoveDescriptionsOption(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/full-descriptive-elements.svg',
+			scour.parse_args(['--remove-descriptions']))
+		self.assertEqual(doc.childNodes.length, 1,
+			'Did not remove <desc> tag with --remove-descriptions')
+
 class RemoveMetadataOption(unittest.TestCase):
 	def runTest(self):
-		doc = scour.scourXmlFile('unittests/full-metadata.svg',
+		doc = scour.scourXmlFile('unittests/full-descriptive-elements.svg',
 			scour.parse_args(['--remove-metadata']))
 		self.assertEqual(doc.childNodes.length, 1,
 			'Did not remove <metadata> tag with --remove-metadata')
 
+class RemoveDescriptiveElementsOption(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/full-descriptive-elements.svg',
+			scour.parse_args(['--remove-descriptive-elements']))
+		self.assertEqual(doc.childNodes.length, 1,
+			'Did not remove <title>, <desc> and <metadata> tags with --remove-descriptive-elements')
+
 class EnableCommentStrippingOption(unittest.TestCase):
 	def runTest(self):
 		with open('unittests/comment-beside-xml-decl.svg') as f:

unittests/descriptive-elements-with-text.svg

@@ -1,4 +1,6 @@
 <?xml version="1.0" encoding="UTF-8" standalone="no"?>
 <svg xmlns="http://www.w3.org/2000/svg">
+  <title>This is a title element with only text node children</title>
+  <desc>This is a desc element with only text node children</desc>
   <metadata>This is a metadata element with only text node children</metadata>
 </svg>

unittests/empty-descriptive-elements.svg

@@ -1,3 +1,5 @@
 <svg xmlns="http://www.w3.org/2000/svg">
+  <title></title>
+  <desc></desc>
   <metadata></metadata>
 </svg>

unittests/full-descriptive-elements.svg

@@ -1,4 +1,13 @@
 <svg xmlns="http://www.w3.org/2000/svg">
+    <title xmlns:mytitle="http://example.org/mytitle">
+      <mytitle:title>This is an example SVG file</mytitle:title>
+      <mytitle:desc>Unit test for Scour's --remove-titles option</mytitle:desc>
+    </title>
+    <desc xmlns:mydesc="http://example.org/mydesc">
+      <mydesc:title>This is an example SVG file</mydesc:title>
+      <mydesc:para>Unit test for Scour's
+      <mydesc:emph>--remove-descriptions</mydesc:emph> option</mydesc:para>
+    </desc>
     <metadata>
       <rdf:RDF
            xmlns:rdf = "http://www.w3.org/1999/02/22-rdf-syntax-ns#"