Read vertical alignment from document.xml

3 files changed, 32 insertions, 4 deletions

diff --git a/mammoth/documents.py b/mammoth/documents.py
index 49b2396..e7a46c4 100644
--- a/mammoth/documents.py
+++ b/mammoth/documents.py
@@ -3,7 +3,10 @@ import dodge
 
 Document = dodge.data_class("Document", ["children", "footnotes"])
 Paragraph = dodge.data_class("Paragraph", ["children", "style_id", "style_name", "numbering"])
-Run = dodge.data_class("Run", ["children", "style_id", "style_name", "is_bold", "is_italic", "is_underline"])
+Run = dodge.data_class("Run", [
+    "children", "style_id", "style_name", "is_bold", "is_italic",
+    "is_underline", "vertical_alignment",
+])
 Text = dodge.data_class("Text", ["value"])
 Hyperlink = dodge.data_class("Hyperlink", ["href", "children"])
 Table = dodge.data_class("Table", ["children"])
@@ -30,8 +33,18 @@ def document(children, footnotes=None):
 def paragraph(children, style_id=None, style_name=None, numbering=None):
     return Paragraph(children, style_id, style_name, numbering)
 
-def run(children, style_id=None, style_name=None, is_bold=None, is_italic=None, is_underline=None):
-    return Run(children, style_id, style_name, bool(is_bold), bool(is_italic), bool(is_underline))
+def run(children, style_id=None, style_name=None, is_bold=None, is_italic=None, is_underline=None, vertical_alignment=None):
+    if vertical_alignment is None:
+        vertical_alignment = VerticalAlignment.baseline
+    return Run(
+        children, style_id, style_name, bool(is_bold), bool(is_italic),
+        bool(is_underline), vertical_alignment,
+    )
+
+class VerticalAlignment(object):
+    baseline = "baseline"
+    superscript = "superscript"
+    subscript = "subscript"
 
 text = Text
 
diff --git a/mammoth/docx/document_xml.py b/mammoth/docx/document_xml.py
index e909a86..cdc870a 100644
--- a/mammoth/docx/document_xml.py
+++ b/mammoth/docx/document_xml.py
@@ -70,7 +70,11 @@ def _create_reader(numbering, content_types, relationships, styles, footnote_ele
             style_name = None
         else:
             style_name = styles.find_character_style_by_id(style_id).name
-            
+        
+        vertical_alignment = properties \
+            .find_child_or_null("w:vertAlign") \
+            .attributes.get("w:val")
+        
         is_bold = properties.find_child("w:b")
         is_italic = properties.find_child("w:i")
         is_underline = properties.find_child("w:u")
@@ -83,6 +87,7 @@ def _create_reader(numbering, content_types, relationships, styles, footnote_ele
                 is_bold=is_bold,
                 is_italic=is_italic,
                 is_underline=is_underline,
+                vertical_alignment=vertical_alignment,
             ))
 
 
diff --git a/tests/docx/document_xml_tests.py b/tests/docx/document_xml_tests.py
index f95419b..4c1f330 100644
--- a/tests/docx/document_xml_tests.py
+++ b/tests/docx/document_xml_tests.py
@@ -143,6 +143,16 @@ class ReadXmlElementTests(object):
         run = self._read_run_with_properties([xml_element("w:u")])
         assert_equal(True, run.is_underline)
     
+    @istest
+    def run_has_baseline_vertical_alignment_if_vertical_alignment_element_is_not_present(self):
+        run = self._read_run_with_properties([])
+        assert_equal(documents.VerticalAlignment.baseline, run.vertical_alignment)
+        
+    @istest
+    def run_has_vertical_alignment_read_from_vertical_alignment_element(self):
+        run = self._read_run_with_properties([xml_element("w:vertAlign", {"w:val": "superscript"})])
+        assert_equal(documents.VerticalAlignment.superscript, run.vertical_alignment)
+    
     def _read_run_with_properties(self, properties, styles=None):
         properties_xml = xml_element("w:rPr", {}, properties)
         run_xml = xml_element("w:r", {}, [properties_xml])