oxml: add XmlString for xml __eq__ override

Steve Canny · Steve Canny · commit 259fcfd90000 · 2014-01-07T07:36:02.000-08:00
lxml on Python 3 is not so predictable as to how it orders namespace
declarations when serializing XML output. Because the unit tests
generally rely on a string compare to confirm to XML elements are
equal, this override on the OxmlBaseElement.xml.__eq__() is required to
avoid false positives when namespace and/or attribute order differs but
their contents do not.
diff --git a/docx/oxml/shared.py b/docx/oxml/shared.py
@@ -6,6 +6,8 @@
 
 from lxml import etree
 
+import re
+
 from .exceptions import ValidationError
 
 
@@ -145,13 +147,79 @@ def serialize_for_reading(element):
     Serialize *element* to human-readable XML suitable for tests. No XML
     declaration.
     """
-    return etree.tostring(element, encoding='unicode', pretty_print=True)
+    xml = etree.tostring(element, encoding='unicode', pretty_print=True)
+    return XmlString(xml)
 
 
 def _SubElement(parent, tag):
     return etree.SubElement(parent, qn(tag), nsmap=nsmap)
 
 
+class XmlString(str):
+    """
+    Provides string comparison override suitable for serialized XML that is
+    useful for tests.
+    """
+
+    # '    <w:xyz xmlns:a="http://ns/decl/a" attr_name="val">text</w:xyz>'
+    # |          |                                          ||           |
+    # +----------+------------------------------------------++-----------+
+    #  front      attrs                                     | text
+    #                                                     close
+
+    _xml_elm_line_patt = re.compile(
+        '( *</?[\w:]+)(.*?)(/?>)([^<]*</\w+>)?'
+    )
+
+    def __eq__(self, other):
+        lines = self.splitlines()
+        lines_other = other.splitlines()
+        if len(lines) != len(lines_other):
+            return False
+        for line, line_other in zip(lines, lines_other):
+            if not self._eq_elm_strs(line, line_other):
+                return False
+        return True
+
+    def __ne__(self, other):
+        return not self.__eq__(other)
+
+    def _attr_seq(self, attrs):
+        """
+        Return a sequence of attribute strings parsed from *attrs*. Each
+        attribute string is stripped of whitespace on both ends.
+        """
+        attrs = attrs.strip()
+        attr_lst = attrs.split()
+        return sorted(attr_lst)
+
+    def _eq_elm_strs(self, line, line_2):
+        """
+        Return True if the element in *line_2* is XML equivalent to the
+        element in *line*.
+        """
+        front, attrs, close, text = self._parse_line(line)
+        front_2, attrs_2, close_2, text_2 = self._parse_line(line_2)
+        if front != front_2:
+            return False
+        if self._attr_seq(attrs) != self._attr_seq(attrs_2):
+            return False
+        if close != close_2:
+            return False
+        if text != text_2:
+            return False
+        return True
+
+    def _parse_line(self, line):
+        """
+        Return front, attrs, close, text 4-tuple result of parsing XML element
+        string *line*.
+        """
+        match = self._xml_elm_line_patt.match(line)
+        front, attrs, close, text = [match.group(n) for n in range(1, 5)]
+        return front, attrs, close, text
+
+
 # ===========================================================================
 # shared custom element classes
 # ===========================================================================
diff --git a/tests/oxml/test_shared.py b/tests/oxml/test_shared.py
@@ -0,0 +1,64 @@
+# encoding: utf-8
+
+"""
+Test suite for docx.oxml.shared
+"""
+
+from __future__ import (
+    absolute_import, division, print_function, unicode_literals
+)
+
+import pytest
+
+from docx.oxml.shared import XmlString
+
+
+class DescribeXmlString(object):
+
+    def it_knows_if_two_xml_lines_are_equivalent(self, xml_line_case):
+        line, other, differs = xml_line_case
+        xml = XmlString(line)
+        assert xml == other
+        assert xml != differs
+
+    # fixtures ---------------------------------------------
+
+    @pytest.fixture(params=[
+        'simple_elm', 'nsp_tagname', 'indent', 'attrs', 'nsdecl_order',
+        'closing_elm',
+    ])
+    def xml_line_case(self, request):
+        cases = {
+            'simple_elm': (
+                '<name/>',
+                '<name/>',
+                '<name>',
+            ),
+            'nsp_tagname': (
+                '<xyz:name/>',
+                '<xyz:name/>',
+                '<abc:name/>',
+            ),
+            'indent': (
+                '  <xyz:name/>',
+                '  <xyz:name/>',
+                '<xyz:name/>',
+            ),
+            'attrs': (
+                '  <abc:Name foo="bar" bar="foo">',
+                '  <abc:Name bar="foo" foo="bar">',
+                '  <abc:Name far="boo" foo="bar">',
+            ),
+            'nsdecl_order': (
+                '    <name xmlns:a="http://ns/1" xmlns:b="http://ns/2"/>',
+                '    <name xmlns:b="http://ns/2" xmlns:a="http://ns/1"/>',
+                '    <name xmlns:b="http://ns/2" xmlns:a="http://ns/1">',
+            ),
+            'closing_elm': (
+                '</xyz:name>',
+                '</xyz:name>',
+                '<xyz:name>',
+            ),
+        }
+        line, other, differs = cases[request.param]
+        return line, other, differs
