Merge pull request #176 from snshn/img-srcset

IMG srcset

Cargo.lock

@@ -623,7 +623,7 @@ dependencies = [
 
 [[package]]
 name = "monolith"
-version = "2.2.5"
+version = "2.2.6"
 dependencies = [
  "assert_cmd 0.12.0 (registry+https://github.com/rust-lang/crates.io-index)",
  "base64 0.12.0 (registry+https://github.com/rust-lang/crates.io-index)",

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "monolith"
-version = "2.2.5"
+version = "2.2.6"
 edition = "2018"
 authors = [
     "Sunshine <sunshine@uberspace.net>",

src/html.rs

@@ -16,6 +16,11 @@ use sha2::{Digest, Sha256, Sha384, Sha512};
 use std::collections::HashMap;
 use std::default::Default;
 
+struct SrcSetItem<'a> {
+    path: &'a str,
+    descriptor: &'a str,
+}
+
 const ICON_VALUES: &[&str] = &[
     "icon",
     "shortcut icon",
@@ -58,6 +63,70 @@ pub fn has_proper_integrity(data: &[u8], integrity: &str) -> bool {
     }
 }
 
+pub fn embed_srcset(
+    cache: &mut HashMap<String, Vec<u8>>,
+    client: &Client,
+    parent_url: &str,
+    srcset: &str,
+    opt_no_images: bool,
+    opt_silent: bool,
+) -> String {
+    let mut array: Vec<SrcSetItem> = vec![];
+    let srcset_items: Vec<&str> = srcset.split(',').collect();
+    for srcset_item in srcset_items {
+        let parts: Vec<&str> = srcset_item.trim().split_whitespace().collect();
+        let path = parts[0].trim();
+        let descriptor = if parts.len() > 1 { parts[1].trim() } else { "" };
+        let srcset_real_item = SrcSetItem { path, descriptor };
+        array.push(srcset_real_item);
+    }
+
+    let mut result: String = str!();
+    let mut i: usize = array.len();
+    for part in array {
+        if opt_no_images {
+            result.push_str(empty_image!());
+        } else {
+            let image_full_url = resolve_url(&parent_url, part.path).unwrap_or_default();
+            let image_url_fragment = get_url_fragment(image_full_url.clone());
+            match retrieve_asset(cache, client, &parent_url, &image_full_url, opt_silent) {
+                Ok((image_data, image_final_url, image_media_type)) => {
+                    let image_data_url = data_to_data_url(
+                        &image_media_type,
+                        &image_data,
+                        &image_final_url,
+                        &image_url_fragment,
+                    );
+                    // Append retreved asset as a data URL
+                    result.push_str(image_data_url.as_ref());
+                }
+                Err(_) => {
+                    // Keep remote reference if unable to retrieve the asset
+                    if is_http_url(image_full_url.clone()) {
+                        result.push_str(image_full_url.as_ref());
+                    } else {
+                        // Avoid breaking the structure in case if not an HTTP(S) URL
+                        result.push_str(empty_image!());
+                    }
+                }
+            }
+        }
+
+        if !part.descriptor.is_empty() {
+            result.push_str(" ");
+            result.push_str(part.descriptor);
+        }
+
+        if i > 1 {
+            result.push_str(", ");
+        }
+
+        i -= 1;
+    }
+
+    result
+}
+
 pub fn walk_and_embed_assets(
     cache: &mut HashMap<String, Vec<u8>>,
     client: &Client,
@@ -352,15 +421,18 @@ pub fn walk_and_embed_assets(
                 }
                 "img" => {
                     // Find source attribute(s)
-                    let mut img_src: String = str!();
                     let mut img_data_src: String = str!();
+                    let mut img_src: String = str!();
+                    let mut img_srcset: String = str!();
                     let mut i = 0;
                     while i < attrs_mut.len() {
                         let attr_name: &str = &attrs_mut[i].name.local;
-                        if attr_name.eq_ignore_ascii_case("src") {
+                        if attr_name.eq_ignore_ascii_case("data-src") {
-                            img_src = str!(attrs_mut.remove(i).value.trim());
-                        } else if attr_name.eq_ignore_ascii_case("data-src") {
                             img_data_src = str!(attrs_mut.remove(i).value.trim());
+                        } else if attr_name.eq_ignore_ascii_case("src") {
+                            img_src = str!(attrs_mut.remove(i).value.trim());
+                        } else if attr_name.eq_ignore_ascii_case("srcset") {
+                            img_srcset = str!(attrs_mut.remove(i).value.trim());
                         } else {
                             i += 1;
                         }
@@ -416,6 +488,23 @@ pub fn walk_and_embed_assets(
                             }
                         }
                     }
+
+                    if !img_srcset.is_empty() {
+                        attrs_mut.push(Attribute {
+                            name: QualName::new(None, ns!(), local_name!("srcset")),
+                            value: Tendril::from_slice(
+                                embed_srcset(
+                                    cache,
+                                    client,
+                                    &url,
+                                    &img_srcset,
+                                    opt_no_images,
+                                    opt_silent,
+                                )
+                                .as_ref(),
+                            ),
+                        });
+                    }
                 }
                 "svg" => {
                     if opt_no_images {

src/tests/cli.rs

@@ -227,9 +227,9 @@ fn passing_local_file_target_input() -> Result<(), Box<dyn std::error::Error>> {
     let out = cmd
         .arg("-M")
         .arg(if cfg!(windows) {
-            "src\\tests\\data\\local-file.html"
+            "src\\tests\\data\\basic\\local-file.html"
         } else {
-            "src/tests/data/local-file.html"
+            "src/tests/data/basic/local-file.html"
         })
         .output()
         .unwrap();
@@ -257,9 +257,9 @@ fn passing_local_file_target_input() -> Result<(), Box<dyn std::error::Error>> {
         std::str::from_utf8(&out.stderr).unwrap(),
         format!(
             "\
-{file}{cwd}/src/tests/data/local-file.html\n\
+{file}{cwd}/src/tests/data/basic/local-file.html\n\
-{file}{cwd}/src/tests/data/local-style.css\n\
+{file}{cwd}/src/tests/data/basic/local-style.css\n\
-{file}{cwd}/src/tests/data/local-script.js\n\
+{file}{cwd}/src/tests/data/basic/local-script.js\n\
 ",
             file = file_url_protocol,
             cwd = cwd_normalized
@@ -284,12 +284,12 @@ fn passing_local_file_target_input_absolute_target_path() -> Result<(), Box<dyn
         .arg("-jciI")
         .arg(if cfg!(windows) {
             format!(
-                "{cwd}\\src\\tests\\data\\local-file.html",
+                "{cwd}\\src\\tests\\data\\basic\\local-file.html",
                 cwd = cwd.to_str().unwrap()
             )
         } else {
             format!(
-                "{cwd}/src/tests/data/local-file.html",
+                "{cwd}/src/tests/data/basic/local-file.html",
                 cwd = cwd.to_str().unwrap()
             )
         })
@@ -322,7 +322,7 @@ fn passing_local_file_target_input_absolute_target_path() -> Result<(), Box<dyn
     assert_eq!(
         std::str::from_utf8(&out.stderr).unwrap(),
         format!(
-            "{file}{cwd}/src/tests/data/local-file.html\n",
+            "{file}{cwd}/src/tests/data/basic/local-file.html\n",
             file = file_url_protocol,
             cwd = cwd_normalized,
         )
@@ -345,13 +345,13 @@ fn passing_local_file_url_target_input() -> Result<(), Box<dyn std::error::Error
         .arg("-cji")
         .arg(if cfg!(windows) {
             format!(
-                "{file}{cwd}/src/tests/data/local-file.html",
+                "{file}{cwd}/src/tests/data/basic/local-file.html",
                 file = file_url_protocol,
                 cwd = cwd_normalized,
             )
         } else {
             format!(
-                "{file}{cwd}/src/tests/data/local-file.html",
+                "{file}{cwd}/src/tests/data/basic/local-file.html",
                 file = file_url_protocol,
                 cwd = cwd_normalized,
             )
@@ -385,13 +385,13 @@ fn passing_local_file_url_target_input() -> Result<(), Box<dyn std::error::Error
         std::str::from_utf8(&out.stderr).unwrap(),
         if cfg!(windows) {
             format!(
-                "{file}{cwd}/src/tests/data/local-file.html\n",
+                "{file}{cwd}/src/tests/data/basic/local-file.html\n",
                 file = file_url_protocol,
                 cwd = cwd_normalized,
             )
         } else {
             format!(
-                "{file}{cwd}/src/tests/data/local-file.html\n",
+                "{file}{cwd}/src/tests/data/basic/local-file.html\n",
                 file = file_url_protocol,
                 cwd = cwd_normalized,
             )
@@ -410,7 +410,7 @@ fn passing_security_disallow_local_assets_within_data_url_targets(
     let mut cmd = Command::cargo_bin(env!("CARGO_PKG_NAME"))?;
     let out = cmd
         .arg("-M")
-        .arg("data:text/html,%3Cscript%20src=\"src/tests/data/local-script.js\"%3E%3C/script%3E")
+        .arg("data:text/html,%3Cscript%20src=\"src/tests/data/basic/local-script.js\"%3E%3C/script%3E")
         .output()
         .unwrap();
 

src/tests/html/embed_srcset.rs

@@ -0,0 +1,26 @@
+//  ██████╗  █████╗ ███████╗███████╗██╗███╗   ██╗ ██████╗
+//  ██╔══██╗██╔══██╗██╔════╝██╔════╝██║████╗  ██║██╔════╝
+//  ██████╔╝███████║███████╗███████╗██║██╔██╗ ██║██║  ███╗
+//  ██╔═══╝ ██╔══██║╚════██║╚════██║██║██║╚██╗██║██║   ██║
+//  ██║     ██║  ██║███████║███████║██║██║ ╚████║╚██████╔╝
+//  ╚═╝     ╚═╝  ╚═╝╚══════╝╚══════╝╚═╝╚═╝  ╚═══╝ ╚═════╝
+
+#[cfg(test)]
+mod passing {
+    use crate::html;
+    use reqwest::blocking::Client;
+    use std::collections::HashMap;
+
+    #[test]
+    fn replace_with_empty_images() {
+        let cache = &mut HashMap::new();
+        let client = Client::new();
+        let srcset_value = "small.png 1x, large.png 2x";
+        let embedded_css = html::embed_srcset(cache, &client, "", &srcset_value, true, true);
+
+        assert_eq!(
+            format!("{} 1x, {} 2x", empty_image!(), empty_image!()),
+            embedded_css
+        );
+    }
+}

src/tests/html/mod.rs

@@ -1,3 +1,4 @@
+mod embed_srcset;
 mod get_node_name;
 mod has_proper_integrity;
 mod is_icon;

src/tests/utils/retrieve_asset.rs

@@ -49,12 +49,12 @@ fn passing_read_local_file_with_file_url_parent() {
         cache,
         &client,
         &format!(
-            "{file}{cwd}/src/tests/data/local-file.html",
+            "{file}{cwd}/src/tests/data/basic/local-file.html",
             file = file_url_protocol,
             cwd = cwd.to_str().unwrap()
         ),
         &format!(
-            "{file}{cwd}/src/tests/data/local-script.js",
+            "{file}{cwd}/src/tests/data/basic/local-script.js",
             file = file_url_protocol,
             cwd = cwd.to_str().unwrap()
         ),
@@ -65,7 +65,7 @@ fn passing_read_local_file_with_file_url_parent() {
     assert_eq!(
         &final_url,
         &format!(
-            "{file}{cwd}/src/tests/data/local-script.js",
+            "{file}{cwd}/src/tests/data/basic/local-script.js",
             file = file_url_protocol,
             cwd = cwd.to_str().unwrap()
         )