kml: make kmlread more robust

this allows it to handle more complex kmz files from google earth

MAVProxy/modules/lib/kmlread.py

@@ -1,15 +1,18 @@
 #!/usr/bin/env python
 
+import lxml.etree as etree
+from io import BytesIO as SIO
 from zipfile import ZipFile
-from xml.dom.minidom import parseString
+
+namespaces = {'kml': 'http://www.opengis.net/kml/2.2'}
 
 def readkmz(filename):
     '''reads in a kmz file and returns xml nodes'''
     #Strip quotation marks if neccessary
     filename.strip('"')
     #Open the zip file (as applicable)    
     if filename[-4:] == '.kml':
-        fo = open(filename, "r")
+        fo = open(filename, "rb")
         fstring = fo.read()
         fo.close()
     elif filename[-4:] == '.kmz':
@@ -22,38 +25,56 @@ def readkmz(filename):
                 raise Exception("Could not find kml file in %s" % filename)
         else:
             raise Exception("Is not a valid kml or kmz file in %s" % filename)
-
-        #send into the xml parser
-        kmlstring = parseString(fstring)
-
-        #get all the placenames
-        nodes=kmlstring.getElementsByTagName('Placemark')
-
-        return nodes
-
+
+    parser = etree.XMLParser(encoding='utf-8', recover=True)
+    tree = etree.parse(SIO(fstring), parser)
+    xpath = xpath = ".//kml:Placemark"
+
+    return tree.findall(xpath, namespaces)
+
+def find_tag(node, tagname):
+    for c in node.getchildren():
+        if c.tag == "{" + namespaces['kml'] + "}" + tagname:
+            return c
+    return None
+
+def find_tag_recursive(node, tagname):
+    for c in node.getchildren():
+        if c.tag == "{" + namespaces['kml'] + "}" + tagname:
+            return c
+        if hasattr(c, 'getchildren'):
+            ret = find_tag_recursive(c, tagname)
+            if ret is not None:
+                return ret
+    return None
+
+
 def readObject(innode):
     '''reads in a node and returns as a tuple: (type, name, points[])'''
     #get name
-    names=innode.getElementsByTagName('name')[0].childNodes[0].data.strip()
-
-    #get type
-    pointType = 'Unknown'
-    if len(innode.getElementsByTagName('LineString')) == 0 and len(innode.getElementsByTagName('Point')) == 0:
-        pointType = 'Polygon'
-    elif len(innode.getElementsByTagName('Polygon')) == 0 and len(innode.getElementsByTagName('Point')) == 0:
-        pointType = 'Polygon'
-    elif len(innode.getElementsByTagName('LineString')) == 0 and len(innode.getElementsByTagName('Polygon')) == 0:
-        pointType = 'Point'
-
-    #get coords
-    coords = innode.getElementsByTagName('coordinates')[0].childNodes[0].data.strip()
-    coordsSplit = coords.split()
-    ret_s = []
-    for j in coordsSplit:
-        jcoord = j.split(',')
-        if len(jcoord) == 3 and jcoord[0] != '' and jcoord[1] != '':
-            #print("Got lon " + jcoord[0] + " and lat " + jcoord[1])
-            ret_s.append((float(jcoord[1]), float(jcoord[0])))
-
-    #return tuple
-    return (str(pointType), str(names), ret_s)
+    name = find_tag(innode, 'name')
+    if name is None:
+        return None
+    point = find_tag(innode, 'Point')
+    if point is not None:
+        coordinates = find_tag(point, 'coordinates')
+        if coordinates is None:
+            return None
+        s = coordinates.text.split(',')
+        return ("Point", name.text, [(float(s[1]), float(s[0]))])
+
+    coordinates = find_tag_recursive(innode, 'coordinates')
+    if coordinates is not None:
+        latlon = []
+        for c in coordinates.text.split():
+            s = c.split(',')
+            latlon.append((float(s[1]), float(s[0])))
+        return ("Polygon", name.text, latlon)
+
+    return ('Unknown', None, None)
+
+if __name__ == '__main__':
+    import sys
+    nodes = readkmz(sys.argv[1])
+    for n in nodes:
+        print(readObject(n))