Updated code that actually works for links

pelican/plugins/obsidian/obsidian.py

@@ -1,25 +1,103 @@
+from pathlib import Path
+
+from itertools import chain
+import os
+import re
 from pelican import signals
+from pelican.readers import MarkdownReader
+from pelican.utils import pelican_open
 
+from markdown import Markdown
 
-def pre_taxonomy(article_generator):
-    """
-    Modify the tags of the article
-    """
-    pass
+ARTICLES = {}
+FILES = {}
+
+link = r'\[\[\s*(?P<filename>[\w+\s.]+)(\|\s*(?P<linkname>[\w\s]+))?\]\]'
+file_re = re.compile(r'!' + link)
+link_re = re.compile(link)
+
+
+"""
+# Test cases
+[[my link]]
+[[ my work ]]
+[[ my work | is finished ]]
 
+![[ a file.jpg ]]
+![[file.jpg]]
+"""
 
-def modify_article_content(article_generator, content):
+
+class ObsidianMarkdownReader(MarkdownReader):
     """
     Change the format of various links to the accepted case of pelican.
     """
-    pass
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+
+    def replace_obsidian_links(self, text):
+        def replacement(match):
+            nonlocal self
+            group = match.groupdict()
+            filename = group['filename'].strip()
+            linkname = group['linkname'] if group['linkname'] else filename
+            linkname = linkname.strip()
+            path = ARTICLES[filename]
+            link_structure = '[{linkname}]({{filename}}/{path}/{filename}.md)'.format(
+                linkname=linkname, path=path, filename=filename
+            )
+            return link_structure
+
+        text = link_re.sub(replacement, text)
+        return text
+
+    def read(self, source_path):
+        """Parse content and metadata of markdown files
+
+        It also changes the links to the acceptable format for pelican
+        """
+
+        self._source_path = source_path
+        self._md = Markdown(**self.settings['MARKDOWN'])
+
+        with pelican_open(source_path) as text:
+            text = self.replace_obsidian_links(text)
+            content = self._md.convert(text)
+
+        if hasattr(self._md, 'Meta'):
+            metadata = self._parse_metadata(self._md.Meta)
+        else:
+            metadata = {}
+        return content, metadata
+
+
+def populate_files_and_articles(article_generator):
+    global ARTICLES
+    global FILES
+
+    base_path = Path(article_generator.path)
+    if not ARTICLES:
+        articles = chain(base_path.glob('**/*.md'), base_path.glob('**/*.md'))
+        for article in articles:
+            full_path, filename_w_ext = os.path.split(article)
+            filename, ext = os.path.splitext(filename_w_ext)
+            full_path = str(full_path).replace(str(base_path) + '/', '')
+            ARTICLES[filename] = full_path
+
+    if not FILES:
+        pass
+
+
+def modify_reader(article_generator):
+    populate_files_and_articles(article_generator)
+    article_generator.readers.readers['md'] = ObsidianMarkdownReader(article_generator.settings)
 
 
 def modify_metadata(article_generator, metadata):
     """
     Modify the tags so we can define the tags as we are used to in obsidian.
     """
-
     for tag in metadata['tags']:
         if '#' in tag.name:
             tag.name = tag.name.replace('#', '')
@@ -27,5 +105,4 @@ def modify_metadata(article_generator, metadata):
 
 def register():
     signals.article_generator_context.connect(modify_metadata)
-    signals.article_generator_pretaxonomy.connect(pre_taxonomy)
-    signals.article_generator_write_article.connect(modify_article_content)
+    signals.article_generator_init.connect(modify_reader)