refactor: refactor update_imgs_base64

chore: add doc comment on ResourceType alias

fix: add error when image MIME type is invalid on an image

src/extractor.rs

@@ -4,6 +4,7 @@ use kuchiki::{traits::*, NodeRef};
 use crate::errors::PaperoniError;
 use crate::moz_readability::{MetaData, Readability};
 
+/// A tuple of the url and an Option of the resource's MIME type
 pub type ResourceInfo = (String, Option<String>);
 
 pub struct Article {

src/html.rs

@@ -91,14 +91,9 @@ pub fn generate_html_exports(
                     *id_attr = format!("readability-page-{}", idx);
                 }
 
-                for (img_url, mime_type_opt) in &article.img_urls {
                 if app_config.is_inlining_images {
                     info!("Inlining images for {}", title);
-                        let result = update_imgs_base64(
+                    let result = update_imgs_base64(article);
-                            article,
-                            img_url,
-                            mime_type_opt.as_deref().unwrap_or("image/*"),
-                        );
 
                     if let Err(e) = result {
                         let mut err: PaperoniError = e.into();
@@ -122,7 +117,7 @@ pub fn generate_html_exports(
                         info!("Successfully copied images to imgs dir for {}", title);
                     }
                 }
-                }
+
                 bar.inc(1);
                 successful_articles_table.add_row(vec![title]);
                 body_elem.as_node().append(article_elem.as_node().clone());
@@ -200,13 +195,7 @@ pub fn generate_html_exports(
                     let mut out_file = File::create(&file_name)?;
 
                     if app_config.is_inlining_images {
-                        for (img_url, mime_type_opt) in &article.img_urls {
+                        update_imgs_base64(article)?;
-                            update_imgs_base64(
-                                article,
-                                img_url,
-                                mime_type_opt.as_deref().unwrap_or("image/*"),
-                            )?
-                        }
                     } else {
                         let base_path =
                             Path::new(app_config.output_directory.as_deref().unwrap_or("."));
@@ -270,15 +259,16 @@ fn create_qualname(name: &str) -> QualName {
 }
 
 /// Updates the src attribute of `<img>` elements with a base64 encoded string of the image data
-fn update_imgs_base64(
+fn update_imgs_base64(article: &Article) -> Result<(), std::io::Error> {
-    article: &Article,
-    img_url: &str,
-    mime_type: &str,
-) -> Result<(), std::io::Error> {
     let temp_dir = std::env::temp_dir();
+    for (img_url, mime_type) in &article.img_urls {
         let img_path = temp_dir.join(img_url);
         let img_bytes = std::fs::read(img_path)?;
-    let img_base64_str = format!("data:image:{};base64,{}", mime_type, encode(img_bytes));
+        let img_base64_str = format!(
+            "data:image:{};base64,{}",
+            mime_type.as_deref().unwrap_or("image/*"),
+            encode(img_bytes)
+        );
 
         let img_elems = article
             .node_ref()
@@ -290,6 +280,7 @@ fn update_imgs_base64(
                 *src_attr = img_base64_str.clone();
             }
         }
+    }
     Ok(())
 }
 

src/http.rs

@@ -150,6 +150,15 @@ async fn process_img_response<'a>(
     let img_mime = img_response
         .content_type()
         .map(|mime| mime.essence().to_string());
+    if let Some(mime_str) = &img_mime {
+        if !mime_str.starts_with("image/") {
+            return Err(ErrorKind::HTTPError(format!(
+                "Invalid image MIME type: {} for {}",
+                mime_str, url
+            ))
+            .into());
+        }
+    }
     let img_ext = match img_response
         .content_type()
         .map(|mime| map_mime_subtype_to_ext(mime.subtype()).to_string())