paperoni/src/main.rs

78 lines
2.5 KiB
Rust
Raw Normal View History

#[macro_use]
extern crate lazy_static;
use async_std::stream;
use async_std::task;
use futures::stream::StreamExt;
2021-04-17 15:27:38 +01:00
use indicatif::{ProgressBar, ProgressStyle};
2020-05-02 16:33:45 +01:00
use url::Url;
2020-05-16 08:09:44 +01:00
mod cli;
2021-02-06 09:59:03 +00:00
mod epub;
mod errors;
mod extractor;
2021-02-06 09:59:03 +00:00
/// This module is responsible for async HTTP calls for downloading
/// the HTML content and images
mod http;
mod moz_readability;
use cli::AppConfig;
use epub::generate_epubs;
use extractor::Extractor;
use http::{download_images, fetch_html};
2021-02-06 09:59:03 +00:00
fn main() {
2021-02-06 09:59:03 +00:00
let app_config = cli::cli_init();
2021-02-06 09:59:03 +00:00
if !app_config.urls().is_empty() {
download(app_config);
2020-05-16 08:09:44 +01:00
}
}
fn download(app_config: AppConfig) {
2021-04-17 15:27:38 +01:00
let bar = ProgressBar::new(app_config.urls().len() as u64);
let style = ProgressStyle::default_bar().template(
"{spinner:.cyan} [{elapsed_precise}] {bar:40.white} {:>8} link {pos}/{len:7} {msg:.yellow/white}",
);
bar.set_style(style);
bar.enable_steady_tick(500);
let articles = task::block_on(async {
let urls_iter = app_config.urls().iter().map(|url| fetch_html(url));
let mut responses = stream::from_iter(urls_iter).buffered(app_config.max_conn());
let mut articles = Vec::new();
while let Some(fetch_result) = responses.next().await {
match fetch_result {
Ok((url, html)) => {
2021-04-17 15:27:38 +01:00
// println!("Extracting");
let mut extractor = Extractor::from_html(&html);
2021-04-17 15:27:38 +01:00
bar.set_message("Extracting...");
extractor.extract_content(&url);
2021-02-06 09:59:03 +00:00
if extractor.article().is_some() {
extractor.extract_img_urls();
if let Err(img_errors) =
2021-04-17 15:27:38 +01:00
download_images(&mut extractor, &Url::parse(&url).unwrap(), &bar).await
{
eprintln!(
"{} image{} failed to download for {}",
img_errors.len(),
if img_errors.len() > 1 { "s" } else { "" },
url
);
}
articles.push(extractor);
}
}
Err(e) => eprintln!("{}", e),
}
2021-04-17 15:27:38 +01:00
bar.inc(1);
}
articles
});
2021-04-17 15:27:38 +01:00
bar.finish_with_message("Downloaded articles");
match generate_epubs(articles, app_config.merged()) {
Ok(_) => (),
Err(e) => eprintln!("{}", e),
};
}