download the posts

Cargo.toml

@@ -4,3 +4,7 @@ version = "0.1.0"
 edition = "2021"
 
 [dependencies]
+clap = { version = "3.2", features = ["derive"] }
+reqwest = { version = "0.11" }
+tokio = { version = "1", features = ["full"] }
+scraper = "0.13"

src/download.rs

@@ -0,0 +1,23 @@
+/// Download all the posts from the raw endpoint
+pub async fn download_posts(posts: (String, Vec<String>), dir: String) {
+    // Create folder, silently ignore if already exists
+    std::fs::create_dir(&dir).unwrap_or_default();
+
+    // Endpoint name
+    let endpoint = "raw";
+
+    for post in posts.1 {
+        let mut file = std::fs::File::create(format!("{}/{}.md", dir, post)).unwrap();
+        std::io::Write::write_all(
+            &mut file,
+            reqwest::get(format!("{}/{}/{}", posts.0, endpoint, post))
+                .await
+                .unwrap()
+                .text()
+                .await
+                .unwrap()
+                .as_bytes(),
+        )
+        .unwrap();
+    }
+}

src/main.rs

@@ -1,3 +1,51 @@
-fn main() {
+use clap::Parser;
-    println!("Hello, prose!");
+
+mod download;
+mod parse;
+
+#[derive(Parser)]
+#[clap(version, about, long_about = None)]
+struct Cli {
+    /// Your username
+    #[clap(value_parser)]
+    username: String,
+
+    /// Directory output [default: the username]
+    #[clap(short, long, value_parser, value_name = "DIRECTORY")]
+    directory: Option<String>,
+
+    /// Domain name
+    #[clap(
+        long,
+        value_parser,
+        value_name = "DOMAIN NAME",
+        default_value = "prose.sh"
+    )]
+    domain: String,
+
+    /// Scheme: HTTP/HTTPS
+    #[clap(long, value_parser, default_value = "https")]
+    scheme: String,
+}
+
+#[tokio::main]
+async fn main() {
+    let cli = Cli::parse();
+
+    // Retrieve user's posts
+    let posts = parse::get_posts(
+        cli.scheme.to_lowercase(),
+        cli.username.to_lowercase(),
+        cli.domain.to_lowercase(),
+    )
+    .await;
+
+    // Defines the output folder name
+    let directory = match cli.directory {
+        Some(loc) => loc,
+        None => cli.username,
+    };
+
+    // Download the posts
+    download::download_posts(posts, directory).await;
 }

src/parse.rs

@@ -0,0 +1,25 @@
+use scraper::{Html, Selector};
+
+/// Return vector of all the posts of the user
+pub async fn get_posts(scheme: String, username: String, domain: String) -> (String, Vec<String>) {
+    // Defines the address
+    let url = format!("{}://{}.{}", scheme, username, domain);
+
+    // Parse index page: sheme://username.domain
+    let document = Html::parse_document(&reqwest::get(&url).await.unwrap().text().await.unwrap());
+
+    // Look at the posts
+    let raw_posts = document
+        .select(&Selector::parse("section.posts").unwrap())
+        .next()
+        .unwrap();
+
+    // Get the name of them and push them into the vector
+    let mut posts = Vec::new();
+    for link in raw_posts.select(&Selector::parse("a").unwrap()) {
+        posts.push(link.inner_html());
+    }
+
+    // Return the vector
+    (url, posts)
+}