From 710170c46b7761f5f09cda91c83ad397ff3e5396 Mon Sep 17 00:00:00 2001 From: Nulo Date: Tue, 6 Feb 2024 18:54:40 -0300 Subject: [PATCH] permitir scrapear solo n productos para testear --- scraper-rs/src/main.rs | 26 ++++++++++++++++++++------ 1 file changed, 20 insertions(+), 6 deletions(-) diff --git a/scraper-rs/src/main.rs b/scraper-rs/src/main.rs index 3e61f9c..43e03c6 100644 --- a/scraper-rs/src/main.rs +++ b/scraper-rs/src/main.rs @@ -47,7 +47,10 @@ struct ScrapUrlArgs { url: String, } #[derive(clap::Args)] -struct AutoArgs {} +struct AutoArgs { + #[arg(long)] + n_products: Option, +} #[tokio::main] async fn main() -> anyhow::Result<()> { @@ -59,7 +62,7 @@ async fn main() -> anyhow::Result<()> { Args::GetUrlList(a) => get_url_list_cli(a.supermercado).await, Args::ScrapUrl(a) => scrap_url_cli(a.url).await, Args::ScrapBestSelling => scrap_best_selling_cli().await, - Args::Auto(_) => auto_cli().await, + Args::Auto(a) => auto_cli(a).await, Args::Cron(_) => cron_cli().await, } } @@ -287,6 +290,7 @@ struct AutoTelegram { struct Auto { db: Db, telegram: Option, + limit_n_products: Option, } impl Auto { async fn download_supermercado(self, supermercado: Supermercado) -> anyhow::Result<()> { @@ -300,7 +304,13 @@ impl Auto { )) .await; } - let links: Vec = self.db.get_urls_by_domain(supermercado.host()).await?; + let links: Vec = { + let mut links = self.db.get_urls_by_domain(supermercado.host()).await?; + if let Some(n) = self.limit_n_products { + links.truncate(n); + } + links + }; // { // let debug_path = PathBuf::from("debug/"); // tokio::fs::create_dir_all(&debug_path).await.unwrap(); @@ -355,7 +365,7 @@ impl Auto { } } -async fn auto_cli() -> anyhow::Result<()> { +async fn auto_cli(args: AutoArgs) -> anyhow::Result<()> { let auto = { let db = Db::connect().await?; let telegram = { @@ -370,7 +380,11 @@ async fn auto_cli() -> anyhow::Result<()> { } } }; - Auto { db, telegram } + Auto { + db, + telegram, + limit_n_products: args.n_products, + } }; auto.inform("[auto] Empezando scrap").await; let handles: Vec<_> = Supermercado::value_variants() @@ -405,7 +419,7 @@ async fn cron_cli() -> anyhow::Result<()> { .unwrap(); println!("Waiting for {:?}", t); tokio::time::sleep(t).await; - auto_cli().await.unwrap(); + auto_cli(AutoArgs { n_products: None }).await.unwrap(); } }