Add scaffolding for markdown

8 files changed, 76 insertions, 9 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)
diff --git a/scripts/mammoth b/scripts/mammoth
index b6fa056..f4865fb 100644
--- a/scripts/mammoth
+++ b/scripts/mammoth
@@ -7,6 +7,7 @@ import os
 import shutil
 
 import mammoth
+from mammoth import writers
 
 
 def main():
@@ -27,10 +28,11 @@ def main():
             output_filename = "{0}.html".format(os.path.basename(args.path).rpartition(".")[0])
             output_path = os.path.join(args.output_dir, output_filename)
         
-        result = mammoth.convert_to_html(
+        result = mammoth.convert(
             docx_fileobj,
             style_map=style_map,
             convert_image=convert_image,
+            output_format=args.output_format,
         )
         for message in result.messages:
             sys.stderr.write(message.message)
@@ -89,6 +91,7 @@ def _parse_args():
     output_group.add_argument("output", nargs="?")
     output_group.add_argument("--output-dir", nargs="?")
     
+    parser.add_argument("--output-format", required=False, choices=writers.formats())
     parser.add_argument("--style-map", required=False)
     return parser.parse_args()
 
diff --git a/setup.py b/setup.py
index 5456f2d..e194468 100644
--- a/setup.py
+++ b/setup.py
@@ -19,14 +19,14 @@ if sys.version_info[:2] <= (2, 6):
 setup(
     name='mammoth',
     version='0.3.9',
-    description='Convert Word documents to simple and clean HTML',
+    description='Convert Word documents from docx to simple and clean HTML and Markdown',
     long_description=read("README"),
     author='Michael Williamson',
     author_email='mike@zwobble.org',
     url='http://github.com/mwilliamson/python-mammoth',
     packages=['mammoth', 'mammoth.docx', 'mammoth.style_reader', 'mammoth.writers'],
     scripts=["scripts/mammoth"],
-    keywords="docx word office clean html",
+    keywords="docx word office clean html markdown md",
     install_requires=_install_requires,
     classifiers=[
         'Development Status :: 4 - Beta',
diff --git a/tests/cli_tests.py b/tests/cli_tests.py
index f76df45..4728c6b 100644
--- a/tests/cli_tests.py
+++ b/tests/cli_tests.py
@@ -66,3 +66,11 @@ def style_map_is_used_if_set():
         result = _local.run(["mammoth", docx_path, "--style-map", style_map_path])
         assert_equal(b"", result.stderr_output)
         assert_equal(b"<span>Walking on imported air</span>", result.output)
+
+
+@istest
+def output_format_markdown_option_generates_markdown_output():
+    docx_path = test_path("single-paragraph.docx")
+    result = _local.run(["mammoth", docx_path, "--output-format=markdown"])
+    assert_equal(b"", result.stderr_output)
+    assert_equal(b"Walking on imported air\n\n", result.output)
diff --git a/tests/mammoth_tests.py b/tests/mammoth_tests.py
index c5f2557..ce76221 100644
--- a/tests/mammoth_tests.py
+++ b/tests/mammoth_tests.py
@@ -99,6 +99,14 @@ def transform_document_is_applied_to_document_before_conversion():
 
 
 @istest
+def docx_containing_one_paragraph_can_be_converted_to_markdown():
+    with open(test_path("single-paragraph.docx"), "rb") as fileobj:
+        result = mammoth.convert_to_markdown(fileobj=fileobj)
+        assert_equal("Walking on imported air\n\n", result.value)
+        assert_equal([], result.messages)
+
+
+@istest
 def can_extract_raw_text():
     with open(test_path("simple-list.docx"), "rb") as fileobj:
         result = mammoth.extract_raw_text(fileobj=fileobj)