[bug] HTML encoding issue for data-url

Describe the bug

Html content containing special characters are not properly encoded when creating a new webview window using data url.

Reproduction

let html = r#"
    <html>
      <body>
       special character incoming ###
       thai characters incoming สระ
      </body>
    </html>
  "#;
  let data = format!("data:text/html,{}", html);
  let _docs_window = tauri::WindowBuilder::new(
    &handle,
    "myWindow",
    tauri::WindowUrl::External(data.parse().unwrap())
  )

  .build()
  .unwrap();

does not work, while reading from an external file

  let html = r#"
    <html>
      <body>
       special character incoming ###
       thai characters incoming สระ
      </body>
    </html>
  "#;
  std::fs::write(path.clone(), html).expect("Unable to write file");
  let data = format!("data:text/html,{}", html);
  let _docs_window = tauri::WindowBuilder::new(
    &handle,
    "myWindow",
    tauri::WindowUrl::App("external.html".parse().unwrap())
  )
  .build()
  .unwrap();

works

Expected behavior

No response

Full `tauri info` output

[✔] Environment
    - OS: Windows 10.0.19045 X64
    ✔ WebView2: 120.0.2210.144
    ✔ MSVC: Visual Studio Build Tools 2022
    ✔ rustc: 1.74.1 (a28077b28 2023-12-04)
    ✔ cargo: 1.74.1 (ecb9851af 2023-10-18)
    ✔ rustup: 1.26.0 (5af9b9484 2023-04-05)
    ✔ Rust toolchain: stable-x86_64-pc-windows-msvc (default)
    - node: 16.16.0
    - pnpm: 6.11.0
    - yarn: 1.22.15
    - npm: 8.11.0

[-] Packages
    - tauri [RUST]: 1.5.3
    - tauri-build [RUST]: 1.5.0
    - wry [RUST]: 0.24.6
    - tao [RUST]: 0.16.5
    - @tauri-apps/api [NPM]: 1.5.1 (outdated, latest: 1.5.3)
    - @tauri-apps/cli [NPM]: 1.5.6 (outdated, latest: 1.5.9)

[-] App
    - build-type: bundle
    - CSP: unset
    - distDir: Set automatically by Vue CLI plugin
    - devPath: Set automatically by Vue CLI plugin
    - framework: Vue.js (Nuxt)
    - bundler: Webpack

Stack trace

No response

Additional context

After some initial investigation, and help from @i-c-b on Discord, I have identified some problems

data-url is not parsing html correctly at https://github.com/tauri-apps/tauri/blob/dev/core/tauri/src/manager/webview.rs#L441.

    let html = r#"
        <html>
        <body>
        special character incoming ###
        thai characters incoming สระ
        </body>
        </html>
    "#;
    let data = format!("data:text/html,{}", html);
    let result = data_url::DataUrl::process(&data).unwrap();
    let (body, _) = result.decode_to_vec().unwrap();
    let res = String::from_utf8(body).unwrap();

Will produce <html> <body> special character incoming

As a temporary workaround, we can percent encode the HTML on tauri level, before running the process function - that seems to work

    let html = r#"
        <html>
        <body>
        special character incoming ###
        thai characters incoming สระ
        </body>
        </html>
    "#;
    let html = urlencoding::encode(&html);
    let data = format!("data:text/html,{}", html);
    let result = data_url::DataUrl::process(&data).unwrap();
    let (body, _) = result.decode_to_vec().unwrap();
    let res = String::from_utf8(body).unwrap();

Will produce back the same html `

special character incoming ### thai characters incoming สระ

    </html>`

url.to_string() / to_path() converts the html into percent encoded ASCII string, which in turn is passed to the unified webview of WRY, that does encoding again on webview layer depending on the OS. As a result, the html is encoded twice on OS like windows, https://github.com/tauri-apps/wry/blob/dev/src/webview2/mod.rs#L842
On the tauri-runtime-wry layer, its using .with_url(&url) even for data url with html content. There is another method .with_html(&html) in wry we can make use of, for data url containing html content.

tauri-apps / tauri