Allow elements to be found via Document.getElementById() in the minidom document returned by scourXmlFile()

scour/scour.py

@@ -3126,7 +3126,20 @@ def scourXmlFile(filename, options=None):
    with open(filename, "rb") as f:
       in_string = f.read()
    out_string = scourString(in_string, options)
-   return xml.dom.minidom.parseString(out_string.encode('utf-8'))
+
+   doc = xml.dom.minidom.parseString(out_string.encode('utf-8'))
+
+   # since minidom does not seem to parse DTDs properly
+   # manually declare all attributes with name "id" to be of type ID
+   # (otherwise things like doc.getElementById() won't work)
+   all_nodes = doc.getElementsByTagName("*")
+   for node in all_nodes:
+      try:
+         node.setIdAttribute('id')
+      except:
+         pass
+
+   return doc
 
 
 

testscour.py

@@ -71,18 +71,25 @@ class InvalidOptions(unittest.TestCase):
 			fail = True
 		self.assertEqual(fail, False, 'Exception when calling Scour with invalid options')
 
+class GetElementById(unittest.TestCase):
+	def runTest(self):
+		doc = scour.scourXmlFile('unittests/ids.svg')
+		self.assertIsNotNone(doc.getElementById('svg1'), 'Root SVG element not found by ID')
+		self.assertIsNotNone(doc.getElementById('linearGradient1'), 'linearGradient not found by ID')
+		self.assertIsNotNone(doc.getElementById('layer1'), 'g not found by ID')
+		self.assertIsNotNone(doc.getElementById('rect1'), 'rect not found by ID')
+		self.assertIsNone(doc.getElementById('rect2'), 'Non-existing element found by ID')
+
 class NoInkscapeElements(unittest.TestCase):
 	def runTest(self):
 		self.assertNotEqual(walkTree(scour.scourXmlFile('unittests/sodipodi.svg').documentElement,
 			lambda e: e.namespaceURI != 'http://www.inkscape.org/namespaces/inkscape'), False,
 			'Found Inkscape elements' )
-
 class NoSodipodiElements(unittest.TestCase):
 	def runTest(self):
 		self.assertNotEqual(walkTree(scour.scourXmlFile('unittests/sodipodi.svg').documentElement,
 			lambda e: e.namespaceURI != 'http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd'), False,
 			'Found Sodipodi elements' )
-
 class NoAdobeIllustratorElements(unittest.TestCase):
 	def runTest(self):
 		self.assertNotEqual(walkTree(scour.scourXmlFile('unittests/adobe.svg').documentElement,

unittests/ids.svg

@@ -0,0 +1,12 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<svg id="svg1" xmlns="http://www.w3.org/2000/svg" height="100" width="100" viewBox="0 0 100 100">
+  <defs>
+    <linearGradient id="linearGradient1">
+      <stop offset="0"/>
+      <stop offset="1" stop-color="blue"/>
+    </linearGradient>
+  </defs>
+  <g id="layer1">
+    <rect id="rect1" height="50" width="50" x="10" y="10" fill="url(#linearGradient1)"/>
+  </g>
+</svg>