add index, limit options, order arguments asc

README.md

@@ -28,6 +28,8 @@ pulsarss --source https://path/to/feed.rss
 * `source`, `s` - RSS feed source (required)
 * `target`, `t` - Destination directory (`public` by default)
 * `update`, `u` - Update timeout in seconds (`60` by default)
+* `index`, `i` - Generate `index.gmi` file (`true` by default)
+* `limit`, `l` - Limit channel items (unlimited by default)
 * `output`, `o` - Print output (`dw` by default):
   * `d` - debug
   * `w` - warning

src/argument.rs

@@ -3,6 +3,18 @@ use clap::Parser;
 #[derive(Parser, Debug)]
 #[command(version, about, long_about = None)]
 pub struct Argument {
+    /// Generate `index.gmi` file (`true` by default)
+    #[arg(short, long, default_value_t = true)]
+    pub index: bool,
+
+    /// Limit channel items (unlimited by default)
+    #[arg(short, long, default_value_t = 0)]
+    pub limit: usize,
+
+    /// Show output (`dw` by default)
+    #[arg(short, long, default_value_t = String::from("dw"))]
+    pub output: String,
+
     /// RSS feed source (required)
     #[arg(short, long)]
     pub source: String,
@@ -14,8 +26,4 @@ pub struct Argument {
     /// Update timeout in seconds (`60` by default)
     #[arg(short, long, default_value_t = 60)]
     pub update: u64,
-
-    /// Show output (`dw` by default)
-    #[arg(short, long, default_value_t = String::from("dw"))]
-    pub output: String,
 }

src/main.rs

@@ -2,11 +2,11 @@ mod argument;
 mod destination;
 mod output;
 
+use argument::Argument;
 use output::Output;
 use std::error::Error;
 
 fn main() -> Result<(), Box<dyn Error>> {
-    use argument::Argument;
     use clap::Parser;
     use std::{thread::sleep, time::Duration};
 
@@ -16,12 +16,12 @@ fn main() -> Result<(), Box<dyn Error>> {
     output.debug("crawler started");
 
     loop {
-        crawl(&argument.source, &argument.target, &output)?;
+        crawl(&argument, &output)?;
         sleep(Duration::from_secs(argument.update));
     }
 }
 
-fn crawl(source: &str, target: &str, output: &Output) -> Result<(), Box<dyn Error>> {
+fn crawl(argument: &Argument, output: &Output) -> Result<(), Box<dyn Error>> {
     use destination::Destination;
     use reqwest::blocking::get;
     use rss::Channel;
@@ -39,17 +39,21 @@ fn crawl(source: &str, target: &str, output: &Output) -> Result<(), Box<dyn Erro
     let mut index = HashSet::new();
 
     // collect feed items
-    for item in Channel::read_from(&get(source)?.bytes()?[..])?
+    for item in Channel::read_from(&get(&argument.source)?.bytes()?[..])?
         .items()
         .iter()
     {
+        if argument.limit > 0 && total >= argument.limit {
+            break;
+        }
+
         total += 1;
 
         let mut data = Vec::new();
 
         let destination = match item.pub_date() {
             Some(pub_date) => {
-                let destination = Destination::build(target, pub_date, true)?;
+                let destination = Destination::build(&argument.target, pub_date, true)?;
                 if metadata(destination.item()).is_ok() {
                     exist += 1;
                     continue;
@@ -90,7 +94,9 @@ fn crawl(source: &str, target: &str, output: &Output) -> Result<(), Box<dyn Erro
 
         File::create(destination.item())?.write_all(data.join("\n\n").as_bytes())?;
 
-        index.insert(destination.path); // request `index.gmi` update
+        if argument.index {
+            index.insert(destination.path); // request `index.gmi` update
+        }
     }
 
     // renew pending `index.gmi` files on items crawl completed