improve parsing of data URLs

src/css.rs

@@ -2,7 +2,7 @@ use cssparser::{ParseError, Parser, ParserInput, SourcePosition, Token};
 use reqwest::blocking::Client;
 use std::collections::HashMap;
 
-use crate::utils::{data_to_data_url, decode_url, get_url_fragment, resolve_url, retrieve_asset};
+use crate::utils::{data_to_data_url, get_url_fragment, resolve_url, retrieve_asset};
 
 const CSS_PROPS_WITH_IMAGE_URLS: &[&str] = &[
     "background",
@@ -142,12 +142,11 @@ pub fn process_css<'a>(
 
                     let full_url = resolve_url(&parent_url, value).unwrap_or_default();
                     let url_fragment = get_url_fragment(full_url.clone());
-                    let full_url_decoded = decode_url(full_url);
                     let (css, final_url) = retrieve_asset(
                         cache,
                         client,
                         &parent_url,
-                        &full_url_decoded,
+                        &full_url,
                         false,
                         "",
                         opt_silent,
@@ -261,12 +260,11 @@ pub fn process_css<'a>(
                 if is_import {
                     let full_url = resolve_url(&parent_url, value).unwrap_or_default();
                     let url_fragment = get_url_fragment(full_url.clone());
-                    let full_url_decoded = decode_url(full_url);
                     let (css, final_url) = retrieve_asset(
                         cache,
                         client,
                         &parent_url,
-                        &full_url_decoded,
+                        &full_url,
                         false,
                         "",
                         opt_silent,

src/tests/css/embed_css.rs

@@ -163,9 +163,9 @@ fn passing_import_string() {
         "\
 @charset 'UTF-8';\n\
 \n\
-@import 'data:text/css;base64,ZGF0YTp0ZXh0L2NzcyxodG1se2JhY2tncm91bmQtY29sb3I6IzAwMH0=';\n\
+@import 'data:text/css;base64,aHRtbHtiYWNrZ3JvdW5kLWNvbG9yOiMwMDB9';\n\
 \n\
-@import url('data:text/css;base64,ZGF0YTp0ZXh0L2NzcyxodG1se2NvbG9yOiNmZmZ9')\n\
+@import url('data:text/css;base64,aHRtbHtjb2xvcjojZmZmfQ==')\n\
 "
     );
 }

src/tests/utils/data_url_to_text.rs

@@ -45,6 +45,14 @@ fn passing_parse_text_html_charset_utf_8_between_two_whitespaces() {
     );
 }
 
+#[test]
+fn passing_parse_text_css_url_encoded() {
+    assert_eq!(
+        utils::data_url_to_text("data:text/css,div{background-color:%23000}"),
+        "div{background-color:#000}"
+    );
+}
+
 //  ███████╗ █████╗ ██╗██╗     ██╗███╗   ██╗ ██████╗
 //  ██╔════╝██╔══██╗██║██║     ██║████╗  ██║██╔════╝
 //  █████╗  ███████║██║██║     ██║██╔██╗ ██║██║  ███╗

src/tests/utils/decode_url.rs

@@ -24,3 +24,13 @@ fn passing_decode_file_url() {
         "file:///tmp/space here/test#1.html"
     );
 }
+
+#[test]
+fn passing_plus_sign() {
+    assert_eq!(
+        utils::decode_url(str!(
+            "fonts.somewhere.com/css?family=Open+Sans:300,400,400italic,600,600italic"
+        )),
+        "fonts.somewhere.com/css?family=Open+Sans:300,400,400italic,600,600italic"
+    );
+}

src/tests/utils/file_url_to_fs_path.rs

@@ -21,3 +21,18 @@ fn passing_remove_protocl_and_fragment() {
         );
     }
 }
+
+#[test]
+fn passing_decodes_urls() {
+    if cfg!(windows) {
+        assert_eq!(
+            utils::file_url_to_fs_path("file:///C:/Documents%20and%20Settings/some-file.html"),
+            "C:\\Documents and Settings\\some-file.html"
+        );
+    } else {
+        assert_eq!(
+            utils::file_url_to_fs_path("file:///home/user/My%20Documents"),
+            "/home/user/My Documents"
+        );
+    }
+}

src/utils.rs

@@ -30,6 +30,14 @@ const MAGIC: [[&[u8]; 2]; 18] = [
     [b"\x1A\x45\xDF\xA3", b"video/webm"],
 ];
 
+const PLAINTEXT_MEDIA_TYPES: &[&str] = &[
+    "image/svg+xml",
+    "text/css",
+    "text/html",
+    "text/javascript",
+    "text/plain",
+];
+
 pub fn data_to_data_url(media_type: &str, data: &[u8], url: &str, fragment: &str) -> String {
     let media_type: String = if media_type.is_empty() {
         detect_media_type(data, &url)
@@ -88,6 +96,10 @@ pub fn is_http_url<T: AsRef<str>>(url: T) -> bool {
         .unwrap_or(false)
 }
 
+pub fn is_plaintext_media_type(media_type: &str) -> bool {
+    PLAINTEXT_MEDIA_TYPES.contains(&media_type.to_lowercase().as_str())
+}
+
 pub fn resolve_url<T: AsRef<str>, U: AsRef<str>>(from: T, to: U) -> Result<String, ParseError> {
     let result = if is_http_url(to.as_ref()) {
         to.as_ref().to_string()
@@ -139,10 +151,11 @@ pub fn data_url_to_text<T: AsRef<str>>(url: T) -> String {
     let mut media_type: &str = "";
     let mut encoding: &str = "";
 
+    // Detect media type and encoding
     let mut i: i8 = 0;
     for item in &meta_data_items {
         if i == 0 {
-            if item.eq_ignore_ascii_case("text/html") {
+            if is_plaintext_media_type(item) {
                 media_type = item;
                 continue;
             }
@@ -155,7 +168,7 @@ pub fn data_url_to_text<T: AsRef<str>>(url: T) -> String {
         i = i + 1;
     }
 
-    if media_type.eq_ignore_ascii_case("text/html") {
+    if is_plaintext_media_type(media_type) {
         if encoding.eq_ignore_ascii_case("base64") {
             String::from_utf8(base64::decode(&data).unwrap_or(vec![])).unwrap_or(str!())
         } else {
@@ -167,6 +180,8 @@ pub fn data_url_to_text<T: AsRef<str>>(url: T) -> String {
 }
 
 pub fn decode_url(input: String) -> String {
+    let input: String = input.replace("+", "%2B");
+
     form_urlencoded::parse(input.as_bytes())
         .map(|(key, val)| {
             [
@@ -200,7 +215,8 @@ pub fn file_url_to_fs_path(url: &str) -> String {
         fs_file_path = fs_file_path.replace("/", "\\");
     }
 
-    fs_file_path
+    // File paths should not be %-encoded
+    decode_url(fs_file_path)
 }
 
 pub fn retrieve_asset(
@@ -219,7 +235,11 @@ pub fn retrieve_asset(
     let cache_key = clean_url(&url);
 
     if is_data_url(&url) {
-        Ok((url.to_string(), url.to_string()))
+        if as_data_url {
+            Ok((url.to_string(), url.to_string()))
+        } else {
+            Ok((data_url_to_text(url), url.to_string()))
+        }
     } else if is_file_url(&url) {
         // Check if parent_url is also file:///
         // (if not, then we don't embed the asset)