Experiment with jinja2 env instead of markdown replacement

[timvink]

Related to #14 and #20 , a different approach might work better: add macros to the jinja2 environment.

• https://www.mkdocs.org/dev-guide/plugins/#on_env to access the env
• To add a function: https://stackoverflow.com/questions/5034437/accessing-global-attributes-from-inside-a-macro-in-jinja2

[timvink]

The mkdocs-markdownextradata-plugin takes a different approach and creates its own jinja2 environment, which it applies during on_page_markdown (link).

That means it will give TemplatingErrors for any jinja tags or macros that are not defined. We would have the same problem..

[timvink]

A specific problem we have in #21 is that a tag like {{read_csv('path')}} is a valid jinja2 tag but not replaced by table_reader because of the missing space padding. Any other plugin that then uses jinja2 will complain about not knowing about the read_csv macro.

A more robust approach might be to make the regex replacement more robust and not enforce space padding:

https://github.com/timvink/mkdocs-table-reader-plugin/blob/1a35ceed87fdcdfe617cf27ebb47c6cd2841dc28/mkdocs_table_reader_plugin/plugin.py#L114-L119

[timvink]

Another alternative is using a completely different syntax, like ||| read_csv() ||| but that's not user friendly.

[timvink]

Addressed in https://github.com/timvink/mkdocs-table-reader-plugin/pull/21/commits/281de79acf6e13471abdbca7121ad23b8a37f1b7