Skip to content

todo.md

Latest commit

 

History

History
103 lines (73 loc) · 4.43 KB

todo.md

File metadata and controls

103 lines (73 loc) · 4.43 KB

TODO

Wishlist for stuff to add in the future.

Paste full correctly rendered article summaries in RSS description field

This should be accomplished without re-rendering every indexed article twice.

  1. This would require spinning off macro expansion into it's own function
  2. Another possibility is processing index pages last and concatting from finished pages without processing twice.
  3. Or better yet, index pages get their own process_dir() function where threads are backgrounded until all pre-requsite pages have been built.
  4. Or a correctly rendered temporary file without header/footer can be dropped for a hypothetical process_index_dir() to slurp up later.

Prevent automatic rewrite features from affecting <pre> blocks

  • Currently, # comments and *.md references within <pre> blocks are being modified and introducing page issues.
    • Prevent md_toc_anchor() from affecting # comments within code blocks.

# comments

This can be avoided by using markdown single-tab delimited <pre> blocks (not ```) at the expense of losing syntax highlighting hints.

*.md => *.htm rewrites

This can be suppressed by replacing period with HTML entity code &period; to link to literal .md documents.

implement {MACRO} expansion in CSS and JS files (probably stupid)

Would this be re-implementing a bloated CSS preprocessor too? idk

friendly URL scheme (non-feature)

  • rewrite output filename into friendly URL based on md_title
    • strip non-URL characters at the same time

Having files with and without creation dates will complicate things, requiring edge case logic in md_title and process_dir.

  • wordpress-style URLs eg. "site.net/2020/6/20/my-blogpost.htm"

These can be implemented as symlinks, but DOC_ROOT becomes a non-optional macro.

IMPLEMENTED

Old implementation notes for reference. Stricken notes represent outdated assumptions.

automatic RSS feed generation

I was overthinking this, RSS feed is now built automatically if optional macro SITE_HOSTNAME with a domain name prefix exists.

  • This would require the non-optional use of SITE_AUTHOR, SITE_TITLE, SITE_DESCRIPTION and SITE_HOSTNAME user provided macros to form a valid RSS document during the md_index() process, as this is when most article metadata like title, page contents and creation date are all in scope.

just SITE_TITLE and SITE_DESCRIPTION and they're still optional

  • Relative filenames are not ideal, many RSS readers choke on use of xml:base for relative URL basenames and it's officially deprecated from the XML standard. Additional local {MACROS} including full canonical URL and a description would have to be implemented.
  • Currently, md_index() has no awareness of how deep within the .src/ directory it's operating in, for f in *.md only returns immediate filenames.

full canonical URLs are derived by subtracting $source_dir from $PWD when needed

  • At this point, it seems easier to re-implement a simplistic version of what pandoc does, appending metadata at the top of the markdown document instead of hiding it in HTML inline comments as currently done.

demanding more information from the user wasn't neccessary

  • Generating an RSS file from multiple invocations of md_index() would also violate DRY principles, as it would be tacked on before and after execution of the main process_dir() loop

add support for generating table of contents with inline anchor links to every heading

cmark doesn't support generating table of contents or even inline anchor links for headings

Forcibly add support for toc's even if it's not supported in cmark.

Append <span id="0001"><span> to the end of # headings via IFS= while read -r or similar

# a <span id="0001"></span>
## b
### c
#### d
##### e
###### f

<div class="toc">

* [a](#0001)
	* [b](#0002)
		* [c](#0003)
			* [d](#0004)
				* [e](#0005)
					* [f](#0006)

</div>

Maintain markdown list of these in {TOC}, which can be optionally added in document.

fix issue where newlines and stray grave characters can cause issues with sed

grave characters can appear in markdown toc but it must be converted to HTML before expansion of {TOC}

rewrite metadata functions as stateless filter functions

Performance gain from reading input file only once seem negligible. This would be a style choice more than anything else.

Performance is marginally worse, but this was neccessary to write md_toc* series functions