Start to fetch page content

src/utils/mod.rs

@@ -10,4 +10,5 @@ pub mod string_utils;
 pub mod pdf_utils;
 pub mod mp3_utils;
 pub mod mp4_utils;
-pub mod zip_utils;
+pub mod zip_utils;
+pub mod webpage_utils;

src/utils/webpage_utils.rs

@@ -0,0 +1,32 @@
+//! # Webpage utilities
+//!
+//! @author Pierre Hubert
+
+use std::time::Duration;
+
+use webpage::{Webpage, WebpageOptions};
+
+use crate::data::error::{ExecError, Res};
+use crate::data::post::PostWebLink;
+
+/// Attempt to get information about a URL given by a user
+pub fn get_post_web_link(url: &str) -> Res<PostWebLink> {
+    let mut options = WebpageOptions::default();
+    options.max_redirections = 2;
+    options.follow_location = true;
+    options.useragent = "OpenGraph Fetcher v1.0".to_string();
+    options.timeout = Duration::from_secs(5);
+
+    let page = Webpage::from_url(url, options)?;
+
+    /*if !page.http.content_type.to_lowercase().starts_with("text/html") {
+        return Err(ExecError::boxed_string(format!("The page at {} as an invalid content type: {}!", url, page.http.content_type)));
+    }*/
+
+    Ok(PostWebLink {
+        url: url.to_string(),
+        title: page.html.opengraph.properties.get("title").map(String::to_string),
+        description: page.html.opengraph.properties.get("description").map(String::to_string),
+        image: page.html.opengraph.images.first().map(|m| m.url.to_string()),
+    })
+}