2023-05-15 18:33:14 +00:00
|
|
|
/*
|
|
|
|
|
Program: Bibliofile
|
|
|
|
|
Purpose: This class is meant to process and return HTML formatted text as strings.
|
2023-06-28 17:16:04 +00:00
|
|
|
Last edited: 6/28/23
|
2023-05-15 18:33:14 +00:00
|
|
|
*/
|
2023-06-28 17:16:04 +00:00
|
|
|
|
|
|
|
|
use scraper::{Html, Selector};
|
2023-05-15 18:33:14 +00:00
|
|
|
|
2023-06-28 17:53:13 +00:00
|
|
|
pub fn main(content: String){
|
2023-06-28 17:16:04 +00:00
|
|
|
|
|
|
|
|
|
2023-06-28 17:53:13 +00:00
|
|
|
let str_content = Html::parse_document(&content);
|
|
|
|
|
|
2023-06-28 19:10:07 +00:00
|
|
|
//Selector is HTML tag. Can be <br> or <p> or anything else. To parse entire page, set selector to <html>
|
|
|
|
|
let selector = Selector::parse("html").unwrap();
|
2023-06-28 17:53:13 +00:00
|
|
|
let unwrapped_page = str_content.select(&selector).next().unwrap();
|
|
|
|
|
let page = unwrapped_page.text().collect::<Vec<_>>();
|
2023-06-28 19:10:07 +00:00
|
|
|
|
|
|
|
|
//every line in document is an entry into the vector. For loop iterates through every entry and displays it.
|
2023-06-28 17:53:13 +00:00
|
|
|
for i in 0..page.len() {
|
|
|
|
|
println!("{}", page[i]);
|
|
|
|
|
}
|
|
|
|
|
|
2023-06-28 17:16:04 +00:00
|
|
|
|
2023-05-15 18:33:14 +00:00
|
|
|
}
|
