From 1cf010b5a0ffe8b2b573375e1f092fcd891b4d77 Mon Sep 17 00:00:00 2001 From: kirbylife Date: Sat, 30 Jan 2021 21:11:54 -0600 Subject: [PATCH] add 'attr' method in the SelectorBase --- src/lib.rs | 36 ------------------------- src/selector.rs | 10 ++++++- tests/unit_tests.rs | 64 +++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 73 insertions(+), 37 deletions(-) create mode 100644 tests/unit_tests.rs diff --git a/src/lib.rs b/src/lib.rs index 51b9122..0e5b335 100644 --- a/src/lib.rs +++ b/src/lib.rs @@ -1,39 +1,3 @@ pub mod request; pub mod response; pub mod selector; - -#[cfg(test)] -mod tests { - use crate::request::{Request, RequestBase}; - use crate::selector::{Selector, SelectorBase}; - use http::StatusCode; - - #[test] - fn plain_text_selector() { - let html = " - - -

hello world

-

good bye

- simple text - - -" - .to_string(); - let sel = Selector::from_html(html); - assert_eq!(sel.css::("h1")[0].html(), "

hello world

"); - assert_eq!(sel.css::("#text")[0].content(), "good bye"); - assert_eq!( - sel.css_once::("body > a").unwrap().content(), - "simple text" - ); - } - - #[test] - fn simple_request() { - let req: Request = RequestBase::new("https://httpbin.org/").unwrap(); - let resp = req.launch(); - assert_eq!(resp.status_code, StatusCode::OK); - assert!(resp.css::("h2")[0].html().contains("httpbin.org")); - } -} diff --git a/src/selector.rs b/src/selector.rs index 7b40b02..989718f 100644 --- a/src/selector.rs +++ b/src/selector.rs @@ -20,12 +20,20 @@ pub trait SelectorBase { fn content(&self) -> String { let html = nipper::Document::from(self.html().as_str()); - html.select("*") + html.select("body > *") .iter() .map(|element| element.text().to_string()) .last() .unwrap() } + + fn attr(&self, attribute: &'static str) -> Option { + let html = nipper::Document::from(self.html().as_str()); + match html.select("body > *").attr(attribute) { + Some(text) => Some(format!("{}", text)), + None => None, + } + } } #[derive(Debug)] diff --git a/tests/unit_tests.rs b/tests/unit_tests.rs new file mode 100644 index 0000000..b96df90 --- /dev/null +++ b/tests/unit_tests.rs @@ -0,0 +1,64 @@ +use http::StatusCode; +use raspa::request::{Request, RequestBase}; +use raspa::selector::{Selector, SelectorBase}; + +#[test] +fn plain_text_selector() { + let html = " + + +

hello world

+

good bye

+ simple text + + +" + .to_string(); + let sel = Selector::from_html(html); + assert_eq!(sel.css::("h1")[0].html(), "

hello world

"); + assert_eq!(sel.css::("#text")[0].content(), "good bye"); + assert_eq!( + sel.css_once::("body > a").unwrap().content(), + "simple text" + ); +} + +#[test] +fn simple_request() { + let req: Request = RequestBase::new("https://httpbin.org/").unwrap(); + let resp = req.launch(); + assert_eq!(resp.status_code, StatusCode::OK); + assert!(resp.css::("h2")[0].html().contains("httpbin.org")); +} + +#[test] +fn complex_selectors() { + let html = " + + +

good bye

+ simple text +
    +
  • 1
    • +
  • 2
    • +
  • 3
    • +
+ + +" + .to_string(); + let sel = Selector::from_html(html); + assert_eq!( + sel.css_once::("p").unwrap().attr("id").unwrap(), + "text" + ); + assert_eq!( + sel.css::("a")[0].attr("href").unwrap(), + "http://google.com" + ); + for node in sel.css::("ul li").iter() { + let text = node.content(); + assert_eq!(node.attr("class").unwrap(), "item"); + assert!(node.attr("id").unwrap().contains(&text)); + } +}