From 03f7d77da6ce807effa5494f161aab07fdd2498c Mon Sep 17 00:00:00 2001
From: Michael Williamson <mike@zwobble.org>
Date: Sat, 26 Dec 2015 11:01:22 +0000
Subject: Always use mc:Fallback when reading mc:AlternateContent elements

---
 NEWS                           |  4 ++++
 mammoth/docx/office_xml.py     | 15 +++++++++++++--
 mammoth/lists.py               |  8 ++++++++
 tests/docx/office_xml_tests.py | 26 ++++++++++++++++++++++++++
 4 files changed, 51 insertions(+), 2 deletions(-)
 create mode 100644 tests/docx/office_xml_tests.py

diff --git a/NEWS b/NEWS
index 1765b3a..f5fabcf 100644
--- a/NEWS
+++ b/NEWS
@@ -1,3 +1,7 @@
+# 0.3.26
+
+* Always use mc:Fallback when reading mc:AlternateContent elements.
+
 # 0.3.25
 
 * Remove duplicate messages from results.
diff --git a/mammoth/docx/office_xml.py b/mammoth/docx/office_xml.py
index 4dffb44..0a6a8a8 100644
--- a/mammoth/docx/office_xml.py
+++ b/mammoth/docx/office_xml.py
@@ -1,4 +1,5 @@
-from .xmlparser import parse_xml
+from ..lists import flat_map
+from .xmlparser import parse_xml, XmlElement
 
 
 _namespaces = [
@@ -16,5 +17,15 @@ _namespaces = [
 
 
 def read(fileobj):
-    return parse_xml(fileobj, _namespaces)
+    return _collapse_alternate_content(parse_xml(fileobj, _namespaces))[0]
     
+
+def _collapse_alternate_content(node):
+    if isinstance(node, XmlElement):
+        if node.name == "mc:AlternateContent":
+            return node.find_child("mc:Fallback").children
+        else:
+            node.children = flat_map(_collapse_alternate_content, node.children)
+            return [node]
+    else:
+        return [node]
diff --git a/mammoth/lists.py b/mammoth/lists.py
index 6e39ae0..e8477cb 100644
--- a/mammoth/lists.py
+++ b/mammoth/lists.py
@@ -24,6 +24,14 @@ def unique(values):
     return output
 
 
+def flat_map(func, values):
+    return [
+        element
+        for value in values
+        for element in func(value)
+    ]
+
+
 if sys.version_info[0] == 2:
     map = map
     filter = filter
diff --git a/tests/docx/office_xml_tests.py b/tests/docx/office_xml_tests.py
new file mode 100644
index 0000000..5b8b1e1
--- /dev/null
+++ b/tests/docx/office_xml_tests.py
@@ -0,0 +1,26 @@
+from __future__ import unicode_literals
+
+import io
+
+from nose.tools import istest, assert_equal
+
+from mammoth.docx import xmlparser as xml, office_xml
+
+
+@istest
+def alternate_content_is_replaced_by_contents_of_fallback():
+    xml_string = (
+        '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>' +
+        '<numbering xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006">' +
+        '<mc:AlternateContent>' +
+        '<mc:Choice Requires="w14">' +
+        '<choice/>' +
+        '</mc:Choice>' +
+        '<mc:Fallback>' +
+        '<fallback/>' +
+        '</mc:Fallback>' +
+        '</mc:AlternateContent>' +
+        '</numbering>')
+    
+    result = office_xml.read(io.StringIO(xml_string))
+    assert_equal([xml.element("fallback")], result.children)
-- 
cgit v1.2.3