Add ImgError struct for representing errors in downloading article images

src/errors.rs

@@ -14,6 +14,53 @@ pub enum ErrorKind {
     ReadabilityError(String),
 }
 
+#[derive(Error, Debug)]
+#[error("{kind}")]
+/// Used to represent errors from downloading images. Errors from here are used solely for debugging
+/// as they are considered recoverable.
+pub struct ImgError {
+    kind: ErrorKind,
+    url: Option<String>,
+}
+
+impl ImgError {
+    pub fn with_kind(kind: ErrorKind) -> Self {
+        ImgError { url: None, kind }
+    }
+
+    pub fn set_url(&mut self, url: &str) {
+        self.url = Some(url.to_string());
+    }
+
+    pub fn url(&self) -> &Option<String> {
+        &self.url
+    }
+}
+
+impl From<ErrorKind> for ImgError {
+    fn from(kind: ErrorKind) -> Self {
+        ImgError::with_kind(kind)
+    }
+}
+
+impl From<surf::Error> for ImgError {
+    fn from(err: surf::Error) -> Self {
+        ImgError::with_kind(ErrorKind::HTTPError(err.to_string()))
+    }
+}
+
+impl From<url::ParseError> for ImgError {
+    fn from(err: url::ParseError) -> Self {
+        ImgError::with_kind(ErrorKind::HTTPError(err.to_string()))
+    }
+}
+
+impl From<std::io::Error> for ImgError {
+    fn from(err: std::io::Error) -> Self {
+        ImgError::with_kind(ErrorKind::IOError(err.to_string()))
+    }
+}
+
 #[derive(Error, Debug)]
 #[error("{kind}")]
 pub struct PaperoniError {

src/http.rs

@@ -5,8 +5,8 @@ use indicatif::ProgressBar;
 use log::{debug, info};
 use url::Url;
 
-use crate::{errors::ErrorKind, errors::PaperoniError, extractor::Extractor};
-
+use crate::errors::{ErrorKind, ImgError, PaperoniError};
+use crate::extractor::Extractor;
 type HTMLResource = (String, String);
 
 pub async fn fetch_html(url: &str) -> Result<HTMLResource, PaperoniError> {
@@ -76,7 +76,7 @@ pub async fn download_images(
     extractor: &mut Extractor,
     article_origin: &Url,
     bar: &ProgressBar,
-) -> Result<(), Vec<PaperoniError>> {
+) -> Result<(), Vec<ImgError>> {
     if extractor.img_urls.len() > 0 {
         debug!(
             "Downloading {} images for {}",
@@ -102,7 +102,7 @@ pub async fn download_images(
             bar.set_message(format!("Downloading images [{}/{}]", img_idx + 1, img_count).as_str());
             match req.await {
                 Ok(mut img_response) => {
-                    // let mut img_response = req.await.expect("Unable to retrieve image");
+                    let process_response = async {
                         let img_content: Vec<u8> = match img_response.body_bytes().await {
                             Ok(bytes) => bytes,
                             Err(e) => return Err(e.into()),
@@ -147,8 +147,17 @@ pub async fn download_images(
                                 .unwrap(),
                             img_mime,
                         ))
+                    };
+                    process_response.await.map_err(|mut e: ImgError| {
+                        e.set_url(url);
+                        e
+                    })
+                }
+                Err(e) => {
+                    let mut img_err: ImgError = e.into();
+                    img_err.set_url(url);
+                    Err(img_err)
                 }
-                Err(e) => Err(e.into()),
             }
         });
 
@@ -170,7 +179,7 @@ pub async fn download_images(
 
     let imgs_req_iter = stream::from_iter(imgs_req_iter)
         .buffered(10)
-        .collect::<Vec<Result<_, PaperoniError>>>()
+        .collect::<Vec<Result<_, ImgError>>>()
         .await;
     let mut errors = Vec::new();
     let mut replaced_imgs = Vec::new();

src/main.rs

@@ -70,6 +70,13 @@ fn download(app_config: AppConfig) {
                                     if img_errors.len() > 1 { "s" } else { "" },
                                     url
                                 );
+                                for img_error in img_errors {
+                                    warn!(
+                                        "{}\n\t\tReason {}",
+                                        img_error.url().as_ref().unwrap(),
+                                        img_error
+                                    );
+                                }
                             }
                             articles.push(extractor);
                         }