Merge pull request #3501 from ickc/issue-3492

dispatch PANDOC_OPTIONS base on input extensions

AUTHORS.txt

@@ -62,6 +62,7 @@
 * `James C. McPherson <https://github.com/jmcp>`_
 * `Kade For <https://github.com/kadefor>`_
 * `Kay Hayen <https://github.com/kayhayen>`_
+* `Kolen Cheung <https://github.com/ickc>`_
 * `lbiaggi <https://github.com/lbiaggi>`_
 * `Leandro Poblet <https://github.com/DoctorMalboro>`_
 * `Luis Miguel Morillas <https://github.com/lmorillas>`_

CHANGES.txt

@@ -7,6 +7,7 @@ Features
 * Provide the full ``GLOBAL_CONTEXT`` to the post list shortcode plugin
   (Issue #3481)
 * Add ``BasePlugin.register_auto_watched_folder()``
+* dispatch PANDOC_OPTIONS base on input extensions (Issue #3492)
 
 New in v8.1.2
 =============

nikola/conf.py.in

@@ -1046,8 +1046,16 @@ MARKDOWN_EXTENSIONS = ['markdown.extensions.fenced_code', 'markdown.extensions.c
 
 
 # Extra options to pass to the pandoc command.
-# by default, it's empty, is a list of strings, for example
+# by default, it's an empty list
+# it can be is a list of strings, for example
 # ['-F', 'pandoc-citeproc', '--bibliography=/Users/foo/references.bib']
+# it can be a dict, where the keys are the extensions in COMPILERS['pandoc'], for example
+# COMPILERS['pandoc'] = ['.rst', '.md', '.txt']
+# PANDOC_OPTIONS = {
+#     '.rst': ['-t', 'rst'],
+#     '.md': ['-t', 'markdown'],
+#     '.txt': ['-t', 'markdown-raw_html'],
+# }
 # Pandoc does not demote headers by default.  To enable this, you can use, for example
 # ['--base-header-level=2']
 # PANDOC_OPTIONS = []

nikola/plugins/compile/pandoc.py

@@ -33,6 +33,8 @@
 import io
 import os
 import subprocess
+from typing import List
+from pathlib import Path
 
 from nikola.plugin_categories import PageCompiler
 from nikola.utils import req_missing, makedirs, write_metadata
@@ -49,11 +51,30 @@ def set_site(self, site):
         self.config_dependencies = [str(site.config['PANDOC_OPTIONS'])]
         super().set_site(site)
 
+    def _get_pandoc_options(self, source: str) -> List[str]:
+        """Obtain pandoc args from config depending on type and file extensions."""
+        # Union[List[str], Dict[str, List[str]]]
+        config_options = self.site.config['PANDOC_OPTIONS']
+        type_ = type(config_options)
+        if type_ is list:
+            pandoc_options = config_options
+        elif type_ is dict:
+            ext = Path(source).suffix
+            try:
+                pandoc_options = config_options[ext]
+            except KeyError:
+                self.logger.warn('Setting PANDOC_OPTIONS to [], because extension {} is not defined in PANDOC_OPTIONS: {}.'.format(ext, config_options))
+                pandoc_options = []
+        else:
+            self.logger.warn('Setting PANDOC_OPTIONS to [], because PANDOC_OPTIONS is expected to be of type Union[List[str], Dict[str, List[str]]] but this is not: {}'.format(config_options))
+            pandoc_options = []
+        return pandoc_options
+
     def compile(self, source, dest, is_two_file=True, post=None, lang=None):
         """Compile the source file into HTML and save as dest."""
         makedirs(os.path.dirname(dest))
         try:
-            subprocess.check_call(['pandoc', '-o', dest, source] + self.site.config['PANDOC_OPTIONS'])
+            subprocess.check_call(['pandoc', '-o', dest, source] + self._get_pandoc_options(source))
             with open(dest, 'r', encoding='utf-8-sig') as inf:
                 output, shortcode_deps = self.site.apply_shortcodes(inf.read())
             with open(dest, 'w', encoding='utf-8') as outf:

tests/test_plugins.py

@@ -14,3 +14,7 @@ def test_command_version():
 
 def test_importing_plugin_task_galleries():
     import nikola.plugins.task.galleries  # NOQA
+
+
+def test_importing_plugin_compile_pandoc():
+    import nikola.plugins.compile.pandoc  # NOQA