Add transform_document argument to convert_to_html

4 files changed, 25 insertions, 10 deletions

diff --git a/mammoth/__init__.py b/mammoth/__init__.py
index 90ab764..59e349f 100644
--- a/mammoth/__init__.py
+++ b/mammoth/__init__.py
@@ -2,8 +2,10 @@ from .results import Result
 from . import docx, conversion, options
 
 
-def convert_to_html(fileobj, **kwargs):
-    return docx.read(fileobj).bind(lambda document: 
+def convert_to_html(fileobj, transform_document=None, **kwargs):
+    if transform_document is None:
+        transform_document = lambda x: x
+    return docx.read(fileobj).map(transform_document).bind(lambda document: 
         conversion.convert_document_element_to_html(
             document,
             **options.read_options(kwargs)
diff --git a/mammoth/documents.py b/mammoth/documents.py
index 0f2d767..65a9b84 100644
--- a/mammoth/documents.py
+++ b/mammoth/documents.py
@@ -1,10 +1,11 @@
-import collections
+import dodge
 
-Document = collections.namedtuple("Document", ["children"])
-Paragraph = collections.namedtuple("Paragraph", ["children", "style_name", "numbering"])
-Run = collections.namedtuple("Run", ["children", "style_name", "is_bold", "is_italic"])
-Text = collections.namedtuple("Text", ["value"])
-Hyperlink = collections.namedtuple("Hyperlink", ["href", "children"])
+
+Document = dodge.data_class("Document", ["children"])
+Paragraph = dodge.data_class("Paragraph", ["children", "style_name", "numbering"])
+Run = dodge.data_class("Run", ["children", "style_name", "is_bold", "is_italic"])
+Text = dodge.data_class("Text", ["value"])
+Hyperlink = dodge.data_class("Hyperlink", ["href", "children"])
 
 class Tab(object):
     pass
@@ -40,4 +41,4 @@ hyperlink = Hyperlink
 def numbering_level(level_index, is_ordered):
     return _NumberingLevel(str(level_index), bool(is_ordered))
 
-_NumberingLevel = collections.namedtuple("NumberingLevel", ["level_index", "is_ordered"])
+_NumberingLevel = dodge.data_class("NumberingLevel", ["level_index", "is_ordered"])
diff --git a/setup.py b/setup.py
index 16a0ab4..6c4f782 100644
--- a/setup.py
+++ b/setup.py
@@ -10,6 +10,7 @@ def read(fname):
 
 _install_requires = [
     "parsimonious>=0.5,<0.6",
+    "dodge>=0.1.5,<0.2",
 ]
 
 if sys.version_info[:2] <= (2, 6):
diff --git a/tests/mammoth_tests.py b/tests/mammoth_tests.py
index f2469a9..77e7ec2 100644
--- a/tests/mammoth_tests.py
+++ b/tests/mammoth_tests.py
@@ -27,4 +27,15 @@ def simple_list_is_converted_to_list_elements():
         result = mammoth.convert_to_html(fileobj=fileobj)
         assert_equal([], result.messages)
         assert_equal("<ul><li>Apple</li><li>Banana</li></ul>", result.value)
-        
+
+
+@istest
+def transform_document_is_applied_to_document_before_conversion():
+    def transform_document(document):
+        document.children[0].style_name = "Heading1"
+        return document
+    
+    with open(test_path("single-paragraph.docx"), "rb") as fileobj:
+        result = mammoth.convert_to_html(fileobj=fileobj, transform_document=transform_document)
+        assert_equal("<h1>Walking on imported air</h1>", result.value)
+        assert_equal([], result.messages)