Add scaffolding for markdown

4 files changed, 54 insertions, 6 deletions

diff --git a/mammoth/__init__.py b/mammoth/__init__.py
index c7b305d..8ec2400 100644
--- a/mammoth/__init__.py
+++ b/mammoth/__init__.py
@@ -3,7 +3,15 @@ from . import docx, conversion, options, documents, images, underline
 __all__ = ["convert_to_html", "extract_raw_text", "images", "underline"]
 
 
-def convert_to_html(fileobj, transform_document=None, id_prefix=None, **kwargs):
+def convert_to_html(*args, **kwargs):
+    return convert(*args, output_format="html", **kwargs)
+
+
+def convert_to_markdown(*args, **kwargs):
+    return convert(*args, output_format="markdown", **kwargs)
+
+
+def convert(fileobj, transform_document=None, id_prefix=None, **kwargs):
     if transform_document is None:
         transform_document = lambda x: x
     return docx.read(fileobj).map(transform_document).bind(lambda document: 
@@ -13,7 +21,7 @@ def convert_to_html(fileobj, transform_document=None, id_prefix=None, **kwargs):
             **options.read_options(kwargs)
         )
     )
-
+    
 
 def extract_raw_text(fileobj):
     return docx.read(fileobj).map(_extract_raw_text_from_element)
diff --git a/mammoth/conversion.py b/mammoth/conversion.py
index 3c217e0..2932351 100644
--- a/mammoth/conversion.py
+++ b/mammoth/conversion.py
@@ -13,7 +13,8 @@ def convert_document_element_to_html(element,
         style_map=None,
         convert_image=None,
         convert_underline=None,
-        id_prefix=None):
+        id_prefix=None,
+        output_format=None):
             
     if style_map is None:
         style_map = []
@@ -21,7 +22,7 @@ def convert_document_element_to_html(element,
     if id_prefix is None:
         id_prefix = str(random.randint(0, 1000000000000000))
     
-    html_generator = HtmlGenerator(writers.writer)
+    html_generator = HtmlGenerator(lambda: writers.writer(output_format))
     converter = DocumentConverter(style_map,
         convert_image=convert_image,
         convert_underline=convert_underline,
diff --git a/mammoth/writers/__init__.py b/mammoth/writers/__init__.py
index 04a64d7..376b322 100644
--- a/mammoth/writers/__init__.py
+++ b/mammoth/writers/__init__.py
@@ -1,5 +1,19 @@
 from .html import HtmlWriter
+from .markdown import MarkdownWriter
 
 
-def writer():
-    return HtmlWriter()
+def writer(output_format=None):
+    if output_format is None:
+        output_format = "html"
+    
+    return _writers[output_format]()
+
+
+def formats():
+    return _writers.keys()
+
+
+_writers = {
+    "html": HtmlWriter,
+    "markdown": MarkdownWriter,
+}
diff --git a/mammoth/writers/markdown.py b/mammoth/writers/markdown.py
new file mode 100644
index 0000000..c4e1158
--- /dev/null
+++ b/mammoth/writers/markdown.py
@@ -0,0 +1,25 @@
+from __future__ import unicode_literals
+
+
+class MarkdownWriter(object):
+    def __init__(self):
+        self._fragments = []
+    
+    def text(self, text):
+        self._fragments.append(text)
+    
+    def start(self, name, attributes=None):
+        pass
+
+    def end(self, name):
+        if name == "p":
+            self._fragments.append("\n\n")
+    
+    def self_closing(self, name, attributes=None):
+        pass
+    
+    def append(self, other):
+        self._fragments.append(other)
+    
+    def as_string(self):
+        return "".join(self._fragments)