Today I worked on adding a new post script to my blogging setup. I was realizing yesterday that I was procrastinating on writing my blog post. Usually when this happens my first reaction is to make it as easy as possible to get the thing I don't want to do done. This sparks some interest in doing the task again, and makes it easier next time (assuming I complete the optimization).

So today I did just that, I worked on writing a simple script which finds the next day number for the blog post, makes a new post directory with a passed in name, and makes an index.md file with the standard header in it. I ended up using a lightweight script preprocessor called rust-script so that I could write the logic in Rust using some helper crates for the casing changes.

Rust-Script

This is my first time using rust-script for this sort of thing, but I'm very impressed. The runner parses the rust code looking for doc comments with cargo code blocks and uses the contents for the Cargo.toml file.

//! ```cargo
//! [dependencies]
//! chrono = "0.4.24"
//! convert_case = "0.6.0"
//! ```

use std::io::Write;

use convert_case::{Case, Casing};

fn main() {
    // Go to ./content/trio/maple/ and find all the files and directories that
    // start with "day##-"
    let mut days: Vec<String> = Vec::new();
    for entry in std::fs::read_dir("./content/trio/maple/").unwrap() {

It then compiles the code using a generated project on the fly in a hidden directory so that it doesn't have to be recompiled.

Convert Case

The code itself is pretty straight forward. Only cool thing is the use of a crate I found called ConvertCase which provides easy automatic conversion between different case styles. The upside is that I can write the title however I like and it will get converted to the kebab (words with dashes) and title (Capitalized text with spaces) cases automatically.

Copilot

As a fun extra challenge for this task, I decided to rely on copilot for the majority of the programming. To do this, I split the task into high level steps I expected to need, and wrote comments for them as I would normally for documenting a piece of code, but with some extra detail.

fn main()
    // Go to ./content/trio/maple/ and find all the files and directories that
    // start with "day##-"

    // Parse the ## from each day and find the max

    // Create the new day name using the passed in commandline argument
    // as the suffix

    // Create a new directory for the new day and create an index.md file
    // in it with a header formatted like so:
    //
    //   +++
    //   title = "Day<number> - <suffix>"
    //   description = ""
    //   date = <date>
    //   +++
    //
    //  With the date being the current date in the format of YYYY-MM-DD
}

Then I navigated to each of the blank lines after a comment and accepted the first copilot suggestion. For example, copilot completed the following for the first one.

let mut files = std::fs::read_dir("./content/trio/maple/")
    .unwrap()
    .filter_map(|entry| {
        let entry = entry.unwrap();
        let path = entry.path();
        let file_name = path.file_name().unwrap().to_str().unwrap();
        if file_name.starts_with("day") {
            Some((path, file_name.to_string()))
        } else {
            None
        }
    })
    .collect::<Vec<_>>();

Which is close except for an error in the lifetime of the path file_name. But with a quick fix, the step is written flawlessly. I did the entire script like that and I estimate that I had to fix 5 separate bugs in the final code. Imho this is pretty good and I'm pretty impressed.

Okeydokey script

With the script written I then used Okeydokey (my directory script runner that I talk about here). I already had an okeydokey script for the blog source, so it was as simple as adding a new command with a whole for the title.

new: e "$(rust-script .\scripts\new.rs "{0}")"

Tricky bit here is that I wrote the rust code to output the index.md path when done setting everything up so that I could open it using my e function in powershell. With this done I just have to type ok new "Title for a new blog post" and it will create a properly named file in the right place with the header properly filled out and edit the file automatically. Pretty clean!

Till tomorrow,
Kay