I have some improvements...

README.md

@@ -7,7 +7,6 @@ A pandoc [markdown] reader plugin for [pelican]
 Requirements
 ------------
 
-  - [pypandoc]
   - [pandoc] in $PATH
 
 
@@ -30,6 +29,13 @@ Additional command line parameters can be passed to pandoc via the PANDOC_ARGS p
       '--number-sections',
     ]
 
+Pandoc's markdown extensions can be enabled or disabled via the
+PANDOC_EXTENSIONS parameter.
+
+    PANDOC_EXTENSIONS = [
+      '+hard_line_breaks',
+      '-citations'
+    ]
 
 Contributing
 ------------
@@ -44,4 +50,3 @@ Contributing
 [markdown]: http://daringfireball.net/projects/markdown/
 [pandoc]: http://johnmacfarlane.net/pandoc/
 [pelican]: http://getpelican.com
-[pypandoc]: https://github.com/bebraw/pypandoc

pandoc_reader.py

@@ -1,43 +1,48 @@
-import os
+import subprocess
+
 from pelican import signals
 from pelican.readers import BaseReader
 from pelican.utils import pelican_open
-import pypandoc
-
 
 class PandocReader(BaseReader):
     enabled = True
     file_extensions = ['md', 'markdown', 'mkd', 'mdown']
 
     def read(self, filename):
-        with pelican_open(filename) as text:
-            metadata_items = []
-            in_content = False
-            MD = ''
-            for line in text.splitlines():
-                splitted = line.split(':', 1)
-                if len(splitted) == 2 and not in_content:
-                    metadata_items.append(splitted)
-                else:
-                    in_content = True
-                    MD += line + '\n'
-
-            metadata = {}
-            for name, value in metadata_items:
-                name = name.lower()
-                value = value.strip()
+        with pelican_open(filename) as fp:
+            text = list(fp.splitlines())
+
+        metadata = {}
+        for i, line in enumerate(text):
+            kv = line.split(':', 1)
+            if len(kv) == 2:
+                name, value = kv[0].lower(), kv[1].strip()
                 metadata[name] = self.process_metadata(name, value)
+            else:
+                content = "\n".join(text[i:])
+                break
 
-        os.chdir(self.settings['PATH']) # change the cwd to the content dir
-        if not 'PANDOC_ARGS' in self.settings: self.settings['PANDOC_ARGS'] = []
-        output = pypandoc.convert(MD, 'html5', format='md', extra_args=self.settings['PANDOC_ARGS'])
+        extra_args = self.settings.get('PANDOC_ARGS', [])
+        extensions = self.settings.get('PANDOC_EXTENSIONS', '')
+        if isinstance(extensions, list):
+            extensions = ''.join(extensions)
 
-        return output, metadata
+        pandoc_cmd = ["pandoc", "--from=markdown" + extensions, "--to=html5"]
+        pandoc_cmd.extend(extra_args)
+
+        proc = subprocess.Popen(pandoc_cmd,
+                                stdin = subprocess.PIPE,
+                                stdout = subprocess.PIPE)
 
+        output = proc.communicate(content.encode('utf-8'))[0].decode('utf-8')
+        status = proc.wait()
+        if status:
+            raise subprocess.CalledProcessError(status, pandoc_cmd)
+
+        return output, metadata
 
 def add_reader(readers):
     readers.reader_classes['md'] = PandocReader
 
-
 def register():
     signals.readers_init.connect(add_reader)