Merge pull request #37 from wri/master

Find best suited element, in case multiple elements with the same path exist

arcpy_metadata/metadata_constructors.py

@@ -165,35 +165,55 @@ def _require_tree_elements(self):
         e_tree = self.path.split('/')
         root = self.parent.elements.getroot()
 
-        done = False
-        while not done:
-            d = {}
-            i = 0
-            d[i] = root
-            for e in e_tree:
-                # remove index for multi items, looks goofy. Is there a better way?
-                if "[" in e:
-                    p = e.index('[')
-                    e_name = e[:p]
-                    e_attrib = e[p:][1:-1].split('][')
-                else:
-                    e_name = e
-                    e_attrib = []
-                i += 1
-                d[i] = d[i-1].find(e)
-                if d[i] is None:
-                    child = ET.Element(e_name)
-                    for attrib in e_attrib:
-                        if attrib[0] == "@":
-                            kv = attrib.split('=')
-                            key = kv[0][1:]
-                            value = kv[1][1:-1]
-                            child.set(key, value)
-                    d[i-1].append(child)
+        elements = root.findall(self.path)
+
+        # if there is already exactly one element, take this one
+        if len(elements) == 1:
+            self.element = elements[0]
+            return
+
+        # if there is more than one, take the first one with a value, attribut or children.
+        # if all are empty, just take the last one
+        elif len(elements) > 1:
+            for element in elements:
+                if element.text.strip() != '' or element.attrib is not None or len(element.getchildren()) > 0:
                     break
-                elif i == len(e_tree):
-                    self.element = d[i]
-                    done = True
+            self.element = element
+            return
+
+        # otherwise build the tree
+        else:
+
+            done = False
+            while not done:
+                d = {}
+                i = 0
+                d[i] = root
+
+                for e in e_tree:
+                    # remove index for multi items, looks goofy. Is there a better way?
+                    if "[" in e:
+                        p = e.index('[')
+                        e_name = e[:p]
+                        e_attrib = e[p:][1:-1].split('][')
+                    else:
+                        e_name = e
+                        e_attrib = []
+                    i += 1
+                    d[i] = d[i-1].find(e)
+                    if d[i] is None:
+                        child = ET.Element(e_name)
+                        for attrib in e_attrib:
+                            if attrib[0] == "@":
+                                kv = attrib.split('=')
+                                key = kv[0][1:]
+                                value = kv[1][1:-1]
+                                child.set(key, value)
+                        d[i-1].append(child)
+                        break
+                    elif i == len(e_tree):
+                        self.element = d[i]
+                        done = True
 
 
 class MetadataValueListConstructor(MetadataItemConstructor):

tests/test_metadata_editor.py

@@ -3,7 +3,6 @@
 import unittest
 import os
 import sys
-import shutil
 import distutils
 from distutils import dir_util
 import tempfile
@@ -21,8 +20,6 @@
 from test_elements import test_elements
 
 
-
-
 class TestMetadataWriteRead(unittest.TestCase):
     """
     To start with, let's just get a simple test in that tests the code from the perspective of how someone might use it.
@@ -204,6 +201,10 @@ def test_raster_file(self):
         self._read_metadata(os.path.join(self.temp_data_folder, r"simple_raster.tif"))
 
 
+
+
+
+
 if __name__ == '__main__':
     suite = unittest.TestLoader().loadTestsFromTestCase(TestMetadataWriteRead)
     unittest.TextTestRunner(verbosity=2).run(suite)