5 mins
I'm making my own blog system, which allows me to write my posts in Obsidian and then it transforms them into html documents. I am writing it is Rust, the best language of all time. Currently its barely functional and missing a lot of basic features like serving the posts, here are some features im planning on adding:
Urgent
- fix huge performance regression with a lot of codeblocks
Shiny
-
replace handlebars with tera, and improve the templating
- add helper functions
-
improve error handling
- add warning for invalid posts, and languages without tree-sitter parsers
- [?] A theme to my website, making it look better than just html
- Adding a sql database, pushing new posts and only parsing new ones
- Little "mini avatars", which can show different expressions
- tags list, where you can see what posts have what tags
- A server where it uploads to something like cloudflare pages so people can view my site.
- A tagging system, so people can find posts based on tags.
- A searching system, using wasm
- Live reloading the websites html, whenever changes happen
- Add a way to generate a new post easily, and admin options when inside the website.
- Use a better font, like Iosevka
- theme mini codeblocks with a nice border
- add a config file, so other people can use easily.
- improve how i add modifications as the current implementation is quite bad
- add a system to make images smaller and use them
This list will grow over time, but i currently just implemented syntax highlighting using the crate syntastica which allows you to easily parse and render code using tree sitter. to do this i needed to decrypt the inner working of pulldown_cmark, which converts markdown to html, and does the bulk of the work.
Pulldown works using (to my understanding at least) on an event system, where each part of the markdown file is broken up and turned into an event, which then you can iterate over using iterators, like the following
rust
let parser = parser.map(move |event| {
match event {
Event::Start(Tag::CodeBlock) => { /* Gets the start of a codeblock */ }
// Just pass through every other event
_ => event,
}
});
Thats what they thought... mwahahahaha
Yeah? Well it took me 2 hours to decyphyr the system to be able to insert the syntax highlighting.
This basic system expands with Event::End, so i ended up having to keep track of where the code block started, trap all the text in-between until i reached the end event, and add syntax highlighting there.
rust
let mut in_code_block = false;
// Stores all the text inside the codeblock
let mut code_buffer = String::new();
// Have to use 'move' for some reason
// its out of my understanding as im still learning
let parser = parser.map(move |event| {
match event {
Event::Start(Tag::CodeBlock) => {
// Start tracking text
in_code_block = true;
// Clean the code block buffer,
// since there might have been a code block before
code_buffer.clear();
}
Event::Text(text) if in_code_block => {
// Collect the text
code_buffer.push_str(&text);
}
Event::End(TagEnd::CodeBlock) if in_code_block => {
// Leaving code block, so stop tracking text
in_code_block = false;
// formatting code...
// (cut...)
}
_ => event,
}
});
The formatting code itself? that is an abomination, and i know theres probably a much better way of achieving it. BUT THATS FOR FUTURE ME so ill deal with that later™
The next feature im planning on adding is a tag list showing what posts have what tags
Adding tags was done by finding every mention of a tag in the metadata for markdown files and counting how often each tag was found and in what files, then making a new page for each tag
Thinking about adding a database
Adding a database would mean i could query posts and image inside the html without having to manually find them in code, i could also speed up building the pages as i could cache compiled codeblocks and other files so that we wouldnt have to modify them every time.
Getting inspiration from fasterthanli.me i could also query the database in tenplates to show latest posts and other data. I can also using lol_html as it seems like it can do cool stuff.
Some problems are that i dont know how i can keep track of a specific versions changes and make sure that each version gets the files they need and reduce the amount of querying needed i also still want the source of truth being my markdown files, not a database im not skilled enough in rust and it might be a shiny object causing me to become demotivated with the project.
It would be awome thpugh, i can query posts and videos and tags inside the templates without having to do weird parsing and passing variables throughout the program I could access the htmk transformed files and the non-modified ones with great ease
Revision An input file is split up into the individual elements Text, italics, codeblock, blockquote etc
Clock SVG by Dazzle UI