forcefully set document's charset to UTF-8

Makefile

@@ -10,7 +10,7 @@ build:
 test: build
 	@cargo test --locked
 	@cargo fmt --all -- --check
-.PHONY: test_code_formatting
+.PHONY: test
 
 lint:
 	@cargo fmt --all --

src/css.rs

@@ -26,7 +26,7 @@ const CSS_PROPS_WITH_IMAGE_URLS: &[&str] = &[
     "suffix",
     "symbols",
 ];
-const CSS_SPECIAL_CHARS: &str = "~!@$%^&*()+=,./'\";:?><[]{}|`#";
+const CSS_SPECIAL_CHARS: &'static str = "~!@$%^&*()+=,./'\";:?><[]{}|`#";
 
 pub fn is_image_url_prop(prop_name: &str) -> bool {
     CSS_PROPS_WITH_IMAGE_URLS

src/html.rs

@@ -28,7 +28,7 @@ struct SrcSetItem<'a> {
     descriptor: &'a str,
 }
 
-const ICON_VALUES: &[&str] = &["icon", "shortcut icon"];
+const ICON_VALUES: &'static [&str] = &["icon", "shortcut icon"];
 
 pub fn add_favicon(document: &Handle, favicon_data_url: String) -> RcDom {
     let mut buf: Vec<u8> = Vec::new();
@@ -105,7 +105,7 @@ pub fn compose_csp(options: &Options) -> String {
     }
 
     if options.no_images {
-        // Note: data: is needed for transparent pixels
+        // Note: "data:" is required for transparent pixel images to work
         string_list.push("img-src data:;");
     }
 
@@ -127,23 +127,18 @@ pub fn create_metadata_tag(url: &str) -> String {
                 clean_url.set_password(None).unwrap();
             }
 
-            if is_http_url(url) {
             format!(
                 "<!-- Saved from {} at {} using {} v{} -->",
-                    &clean_url,
-                    timestamp,
-                    env!("CARGO_PKG_NAME"),
-                    env!("CARGO_PKG_VERSION"),
-                )
+                if is_http_url(url) {
+                    &clean_url.as_str()
                 } else {
-                format!(
-                    "<!-- Saved from local source at {} using {} v{} -->",
+                    "local source"
+                },
                 timestamp,
                 env!("CARGO_PKG_NAME"),
                 env!("CARGO_PKG_VERSION"),
             )
         }
-        }
         Err(_) => str!(),
     }
 }
@@ -498,12 +493,12 @@ pub fn walk_and_embed_assets(
         } => {
             match name.local.as_ref() {
                 "meta" => {
-                    // Remove http-equiv attributes from META nodes if they're able to control the page
                     if let Some(meta_attr_http_equiv_value) = get_node_attr(node, "http-equiv") {
                         let meta_attr_http_equiv_value: &str = &meta_attr_http_equiv_value;
                         if meta_attr_http_equiv_value.eq_ignore_ascii_case("refresh")
                             || meta_attr_http_equiv_value.eq_ignore_ascii_case("location")
                         {
+                            // Remove http-equiv attributes from META nodes if they're able to control the page
                             set_node_attr(
                                 &node,
                                 "http-equiv",
@@ -512,8 +507,21 @@ pub fn walk_and_embed_assets(
                                     meta_attr_http_equiv_value
                                 )),
                             );
+                        } else if meta_attr_http_equiv_value.eq_ignore_ascii_case("Content-Type") {
+                            // Enforce charset to be set to UTF-8
+                            if let Some(_attr_value) = get_node_attr(node, "content") {
+                                set_node_attr(
+                                    &node,
+                                    "content",
+                                    Some(str!("text/html; charset=utf-8")),
+                                );
                             }
                         }
+                    } else if let Some(_meta_attr_http_equiv_value) = get_node_attr(node, "charset")
+                    {
+                        // Enforce charset to be set to UTF-8
+                        set_node_attr(&node, "charset", Some(str!("utf-8")));
+                    }
                 }
                 "link" => {
                     // Read and remember integrity attribute value of this LINK node

src/js.rs

@@ -1,4 +1,4 @@
-const JS_DOM_EVENT_ATTRS: &[&str] = &[
+const JS_DOM_EVENT_ATTRS: &'static [&str] = &[
     // From WHATWG HTML spec 8.1.5.2 "Event handlers on elements, Document objects, and Window objects":
     //   https://html.spec.whatwg.org/#event-handlers-on-elements,-document-objects,-and-window-objects
     //   https://html.spec.whatwg.org/#attributes-3 (table "List of event handler content attributes")

src/main.rs

@@ -212,6 +212,9 @@ fn main() {
         }
     }
 
+    // Remove charset meta-tag
+    // set_charset_meta_to_utf8(&dom.document);
+
     // Serialize DOM tree
     let mut result: String = stringify_document(&dom.document, &options);
 

src/opts.rs

@@ -21,7 +21,7 @@ pub struct Options {
     pub target: String,
 }
 
-const ASCII: &str = " \
+const ASCII: &'static str = " \
  _____     ______________    __________      ___________________    ___
 |     \\   /              \\  |          |    |                   |  |   |
 |      \\_/       __       \\_|    __    |    |    ___     ___    |__|   |
@@ -31,7 +31,7 @@ const ASCII: &str = " \
 |___|       |__________|  \\_____________________|   |___|   |___|  |___|
 ";
 const DEFAULT_NETWORK_TIMEOUT: u64 = 120;
-const DEFAULT_USER_AGENT: &str =
+const DEFAULT_USER_AGENT: &'static str =
     "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:73.0) Gecko/20100101 Firefox/73.0";
 
 impl Options {

src/tests/cli/local_files.rs

@@ -69,8 +69,7 @@ mod passing {
     #[test]
     fn local_file_target_input_absolute_target_path() -> Result<(), Box<dyn std::error::Error>> {
         let cwd = env::current_dir().unwrap();
-        let cwd_normalized: String =
-            str!(env::current_dir().unwrap().to_str().unwrap()).replace("\\", "/");
+        let cwd_normalized: String = str!(cwd.to_str().unwrap()).replace("\\", "/");
         let mut cmd = Command::cargo_bin(env!("CARGO_PKG_NAME"))?;
         let out = cmd
             .arg("-M")

src/tests/cli/mod.rs

@@ -2,3 +2,4 @@ mod base_url;
 mod basic;
 mod data_url;
 mod local_files;
+mod unusual_encodings;

src/tests/cli/unusual_encodings.rs

@@ -0,0 +1,51 @@
+//  ██████╗  █████╗ ███████╗███████╗██╗███╗   ██╗ ██████╗
+//  ██╔══██╗██╔══██╗██╔════╝██╔════╝██║████╗  ██║██╔════╝
+//  ██████╔╝███████║███████╗███████╗██║██╔██╗ ██║██║  ███╗
+//  ██╔═══╝ ██╔══██║╚════██║╚════██║██║██║╚██╗██║██║   ██║
+//  ██║     ██║  ██║███████║███████║██║██║ ╚████║╚██████╔╝
+//  ╚═╝     ╚═╝  ╚═╝╚══════╝╚══════╝╚═╝╚═╝  ╚═══╝ ╚═════╝
+
+#[cfg(test)]
+mod passing {
+    use assert_cmd::prelude::*;
+    use std::env;
+    use std::process::Command;
+
+    #[test]
+    fn change_encoding_to_utf_8() -> Result<(), Box<dyn std::error::Error>> {
+        let cwd = env::current_dir().unwrap();
+        let cwd_normalized: String = str!(cwd.to_str().unwrap()).replace("\\", "/");
+        let mut cmd = Command::cargo_bin(env!("CARGO_PKG_NAME"))?;
+        let out = cmd
+            .arg("-M")
+            .arg(if cfg!(windows) {
+                "src\\tests\\data\\unusual_encodings\\iso-8859-1.html"
+            } else {
+                "src/tests/data/unusual_encodings/iso-8859-1.html"
+            })
+            .output()
+            .unwrap();
+        let file_url_protocol: &str = if cfg!(windows) { "file:///" } else { "file://" };
+
+        // STDOUT should contain newly added base URL
+        assert_eq!(
+            std::str::from_utf8(&out.stdout).unwrap(),
+            "<html><head>\n        <meta http-equiv=\"Content-Type\" content=\"text/html; charset=utf-8\">\n    </head>\n    <body>\n        © Some Company\n    \n\n</body></html>\n"
+        );
+
+        // STDERR should contain only the target file
+        assert_eq!(
+            std::str::from_utf8(&out.stderr).unwrap(),
+            format!(
+                "{file}{cwd}/src/tests/data/unusual_encodings/iso-8859-1.html\n",
+                file = file_url_protocol,
+                cwd = cwd_normalized,
+            )
+        );
+
+        // The exit code should be 0
+        out.assert().code(0);
+
+        Ok(())
+    }
+}

src/tests/data/unusual_encodings/iso-8859-1.html

@@ -0,0 +1,8 @@
+<html>
+    <head>
+        <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
+    </head>
+    <body>
+        &copy; Some Company
+    </body>
+</html>

src/utils.rs

@@ -7,7 +7,7 @@ use std::path::Path;
 use crate::opts::Options;
 use crate::url::{clean_url, file_url_to_fs_path, is_data_url, is_file_url, parse_data_url};
 
-const INDENT: &str = " ";
+const INDENT: &'static str = " ";
 
 const MAGIC: [[&[u8]; 2]; 18] = [
     // Image
@@ -34,11 +34,13 @@ const MAGIC: [[&[u8]; 2]; 18] = [
 ];
 
 const PLAINTEXT_MEDIA_TYPES: &[&str] = &[
+    "application/javascript",
     "image/svg+xml",
-    "text/css",
-    "text/html",
-    "text/javascript",
-    "text/plain",
+    // "text/css",
+    // "text/csv",
+    // "text/html",
+    // "text/javascript",
+    // "text/plain",
 ];
 
 pub fn detect_media_type(data: &[u8], url: &str) -> String {
@@ -56,7 +58,8 @@ pub fn detect_media_type(data: &[u8], url: &str) -> String {
 }
 
 pub fn is_plaintext_media_type(media_type: &str) -> bool {
-    PLAINTEXT_MEDIA_TYPES.contains(&media_type.to_lowercase().as_str())
+    media_type.to_lowercase().as_str().starts_with("text/")
+        || PLAINTEXT_MEDIA_TYPES.contains(&media_type.to_lowercase().as_str())
 }
 
 pub fn indent(level: u32) -> String {
@@ -125,7 +128,7 @@ pub fn retrieve_asset(
                 Ok(mut response) => {
                     if !options.ignore_errors && response.status() != 200 {
                         if !options.silent {
-                            eprintln!("Unable to retrieve {} ({})", &url, response.status());
+                            eprintln!("Unable to retrieve {} (error: {})", &url, response.status());
                         }
                         // Provoke error
                         return Err(client.get("").send().unwrap_err());