Implement XML deep equal function (#3)

* Implement XML deep equal function

* More tests

* More tests

tind_commons/tests/test_xmlutils.py

@@ -0,0 +1,85 @@
+import unittest
+import xml.etree.ElementTree as ET
+
+from tind_commons.xmlutils import deep_equal
+
+class DeepEqualTests(unittest.TestCase):
+    def test_obviously_equal_documents(self):
+        doc = "<a>1<b>2</b>d</a>"
+
+        a_element = ET.fromstring(doc)
+        b_element = ET.fromstring(doc)
+
+        self.assertTrue(deep_equal(a_element, b_element))
+
+    def test_different_contents(self):
+        a_doc = "<a>1<b>2</b></a>"
+        b_doc = "<a>2<b>2</b></a>"
+
+        a_element = ET.fromstring(a_doc)
+        b_element = ET.fromstring(b_doc)
+
+        self.assertFalse(deep_equal(a_element, b_element))
+
+    def test_different_tails(self):
+        a_doc = "<a>1<b>2</b></a>"
+        b_doc = "<a>1<b>2</b> </a>"
+
+        a_element = ET.fromstring(a_doc)
+        b_element = ET.fromstring(b_doc)
+
+        self.assertFalse(deep_equal(a_element, b_element))
+
+    def test_attribute_order_is_insignificant(self):
+        a_doc = """<a b="c" d="e"></a>"""
+        b_doc = """<a d="e" b="c"></a>"""
+
+        a_element = ET.fromstring(a_doc)
+        b_element = ET.fromstring(b_doc)
+
+        self.assertTrue(deep_equal(a_element, b_element))
+
+    def test_all_attributes(self):
+        a_doc = """<a b="c" d="e"></a>"""
+        b_doc = """<a b="c"></a>"""
+
+        a_element = ET.fromstring(a_doc)
+        b_element = ET.fromstring(b_doc)
+
+        self.assertFalse(deep_equal(a_element, b_element))
+
+    def test_attribute_contents(self):
+        a_doc = """<a b="c"></a>"""
+        b_doc = """<a b="d"></a>"""
+
+        a_element = ET.fromstring(a_doc)
+        b_element = ET.fromstring(b_doc)
+
+        self.assertFalse(deep_equal(a_element, b_element))
+
+    def test_namespace_aware(self):
+        a_doc = """<a xmlns="foo"></a>"""
+        b_doc = """<a xmlns="bar"></a>"""
+
+        a_element = ET.fromstring(a_doc)
+        b_element = ET.fromstring(b_doc)
+
+        self.assertFalse(deep_equal(a_element, b_element))
+
+    def test_namespace_prefix_aware(self):
+        a_doc = """<x:a xmlns:x="foo"></x:a>"""
+        b_doc = """<a xmlns="foo"></a>"""
+
+        a_element = ET.fromstring(a_doc)
+        b_element = ET.fromstring(b_doc)
+
+        self.assertTrue(deep_equal(a_element, b_element))
+
+    def test_comments_are_insignificant(self):
+        a_doc = """<a><!-- This is a comment --></a>"""
+        b_doc = """<a></a>"""
+
+        a_element = ET.fromstring(a_doc)
+        b_element = ET.fromstring(b_doc)
+
+        self.assertTrue(deep_equal(a_element, b_element))

tind_commons/xmlutils.py

@@ -0,0 +1,33 @@
+def deep_equal(a_element, b_element):
+    """Computes whether two different elements are deep equal.
+
+    Two elements are considered deep equal if:
+     1. Their tag names are equal
+     2. Their attributes are equal
+     3. Their text contents (including the tail contents) are equal
+     3. Their children are pairwise deep equal
+    """
+    if a_element.tag != b_element.tag:
+        return False
+
+    if len(a_element.items()) != len(b_element.items()):
+        return False
+
+    for key in a_element.keys():
+        if a_element.get(key) != b_element.get(key):
+            return False
+
+    if a_element.tail != b_element.tail:
+        return False
+
+    if a_element.text != b_element.text:
+        return False
+
+    if len(a_element) != len(b_element):
+        return False
+
+    for pair in zip(a_element, b_element):
+        if not deep_equal(*pair):
+            return False
+
+    return True