chore: Implement handling of relative URLs

Author: Stranger6667

Cargo.toml

@@ -17,6 +17,7 @@ categories = ["web-programming"]
 cssparser = "0"
 kuchiki = "0.8"
 attohttpc = "0"
+url = "2"
 
 [dev-dependencies]
 criterion = ">= 0.1"

python/CHANGELOG.md

@@ -10,6 +10,7 @@
 ### Changed
 
 - Skip selectors, that can't be parsed.
+- Validate `base_url` to be a valid URL.
 
 ### Fixed
 

python/Cargo.toml

@@ -15,6 +15,7 @@ built = { version = "0.4", features = ["chrono"] }
 
 [dependencies]
 css-inline = { path = "..", version = "= 0.1.0" }
+url = "2"
 rayon = "1"
 pyo3 = { version = ">= 0.10", features = ["extension-module"] }
 pyo3-built = "0.4"

python/src/lib.rs

@@ -15,6 +15,22 @@ fn to_pyerr(error: rust_inline::InlineError) -> PyErr {
     }
 }
 
+struct UrlError(url::ParseError);
+
+impl From<UrlError> for PyErr {
+    fn from(error: UrlError) -> Self {
+        exceptions::ValueError::py_err(format!("{}", error.0))
+    }
+}
+
+fn parse_url(url: Option<String>) -> PyResult<Option<url::Url>> {
+    Ok(if let Some(url) = url {
+        Some(url::Url::parse(url.as_str()).map_err(UrlError)?)
+    } else {
+        None
+    })
+}
+
 /// Customizable CSS inliner.
 #[pyclass]
 #[text_signature = "(remove_style_tags=False, base_url=None, load_remote_stylesheets=True)"]
@@ -29,15 +45,15 @@ impl CSSInliner {
         remove_style_tags: Option<bool>,
         base_url: Option<String>,
         load_remote_stylesheets: Option<bool>,
-    ) -> Self {
+    ) -> PyResult<Self> {
         let options = rust_inline::InlineOptions {
             remove_style_tags: remove_style_tags.unwrap_or(false),
-            base_url,
+            base_url: parse_url(base_url)?,
             load_remote_stylesheets: load_remote_stylesheets.unwrap_or(true),
         };
-        CSSInliner {
+        Ok(CSSInliner {
             inner: rust_inline::CSSInliner::new(options),
-        }
+        })
     }
 
     /// inline(html)
@@ -70,7 +86,7 @@ fn inline(
 ) -> PyResult<String> {
     let options = rust_inline::InlineOptions {
         remove_style_tags: remove_style_tags.unwrap_or(false),
-        base_url,
+        base_url: parse_url(base_url)?,
         load_remote_stylesheets: load_remote_stylesheets.unwrap_or(true),
     };
     let inliner = rust_inline::CSSInliner::new(options);
@@ -90,7 +106,7 @@ fn inline_many(
 ) -> PyResult<Vec<String>> {
     let options = rust_inline::InlineOptions {
         remove_style_tags: remove_style_tags.unwrap_or(false),
-        base_url,
+        base_url: parse_url(base_url)?,
         load_remote_stylesheets: load_remote_stylesheets.unwrap_or(true),
     };
     let inliner = rust_inline::CSSInliner::new(options);

python/tests-py/test_inlining.py

@@ -53,3 +53,8 @@ def test_no_existing_style(func, kwargs, expected):
 def test_inline_many_wrong_type():
     with pytest.raises(TypeError):
         css_inline.inline_many([1])
+
+
+def test_invalid_base_url():
+    with pytest.raises(ValueError):
+        css_inline.CSSInliner(base_url="foo")

src/lib.rs

@@ -133,9 +133,11 @@ pub mod error;
 mod parser;
 
 pub use error::InlineError;
+use std::borrow::Cow;
 use std::collections::HashMap;
 use std::fs::File;
 use std::io::{Read, Write};
+pub use url::{ParseError, Url};
 
 #[derive(Debug)]
 struct Rule<'i> {
@@ -161,7 +163,7 @@ pub struct InlineOptions {
     /// Remove "style" tags after inlining
     pub remove_style_tags: bool,
     /// Used for loading external stylesheets via relative URLs
-    pub base_url: Option<String>,
+    pub base_url: Option<Url>,
     /// Whether remote stylesheets should be loaded or not
     pub load_remote_stylesheets: bool,
 }
@@ -245,7 +247,7 @@ impl CSSInliner {
             {
                 if let Some(href) = &link_tag.attributes.borrow().get("href") {
                     let url = self.get_full_url(href);
-                    let css = self.load_external(url.as_str())?;
+                    let css = self.load_external(url.as_ref())?;
                     process_css(&document, css.as_str())?;
                 }
             }
@@ -254,20 +256,24 @@ impl CSSInliner {
         Ok(())
     }
 
-    fn get_full_url(&self, href: &str) -> String {
-        if href.starts_with("//") {
-            if let Some(base_url) = &self.options.base_url {
-                if base_url.starts_with("https://") {
-                    format!("https:{}", href)
-                } else {
-                    format!("http:{}", href)
-                }
+    fn get_full_url<'u>(&self, href: &'u str) -> Cow<'u, str> {
+        // Valid absolute URL
+        if Url::parse(href).is_ok() {
+            return Cow::Borrowed(href);
+        };
+        if let Some(base_url) = &self.options.base_url {
+            // Use the same scheme as the base URL
+            if href.starts_with("//") {
+                return Cow::Owned(format!("{}:{}", base_url.scheme(), href));
             } else {
-                format!("http:{}", href)
+                // Not a URL, then it is a relative URL
+                if let Ok(new_url) = base_url.join(href) {
+                    return Cow::Owned(new_url.to_string());
+                }
             }
-        } else {
-            href.to_string()
-        }
+        };
+        // If it is not a valid URL and there is no base URL specified, we assume a local path
+        Cow::Borrowed(href)
     }
 
     fn load_external(&self, url: &str) -> Result<String, InlineError> {

tests/server.py

@@ -2,13 +2,10 @@
 
 app = Flask(__name__)
 
-with open("tests/external.css") as fd:
-    STYLESHEET = fd.read()
-
 
 @app.route("/external.css")
 def stylesheet():
-    return STYLESHEET
+    return "h1 { color: blue; }"
 
 
 if __name__ == "__main__":

tests/test_inlining.rs

@@ -1,4 +1,4 @@
-use css_inline::{inline, CSSInliner, InlineOptions};
+use css_inline::{inline, CSSInliner, InlineOptions, Url};
 
 macro_rules! html {
     ($style: expr, $body: expr) => {
@@ -151,11 +151,6 @@ h2 { color: red; }
 <h2>Smaller Text</h2>
 </body>
 </html>"#;
-    let options = InlineOptions {
-        remove_style_tags: false,
-        base_url: None,
-        load_remote_stylesheets: false,
-    };
     let result = inline(&html).unwrap();
     assert!(result.ends_with(
         r#"<body>
@@ -192,6 +187,68 @@ h2 { color: red; }
     ))
 }
 
+#[test]
+fn remote_network_stylesheet_same_scheme() {
+    let html = r#"
+<html>
+<head>
+<link href="//127.0.0.1:5000/external.css" rel="stylesheet" type="text/css">
+<link rel="alternate" type="application/rss+xml" title="RSS" href="/rss.xml">
+<style type="text/css">
+h2 { color: red; }
+</style>
+</head>
+<body>
+<h1>Big Text</h1>
+<h2>Smaller Text</h2>
+</body>
+</html>"#;
+    let options = InlineOptions {
+        base_url: Some(Url::parse("http://127.0.0.1:5000").unwrap()),
+        ..Default::default()
+    };
+    let inliner = CSSInliner::new(options);
+    let result = inliner.inline(&html).unwrap();
+    assert!(result.ends_with(
+        r#"<body>
+<h1 style="color: blue;">Big Text</h1>
+<h2 style="color: red;">Smaller Text</h2>
+
+</body></html>"#
+    ))
+}
+
+#[test]
+fn remote_network_relative_stylesheet() {
+    let html = r#"
+<html>
+<head>
+<link href="external.css" rel="stylesheet" type="text/css">
+<link rel="alternate" type="application/rss+xml" title="RSS" href="/rss.xml">
+<style type="text/css">
+h2 { color: red; }
+</style>
+</head>
+<body>
+<h1>Big Text</h1>
+<h2>Smaller Text</h2>
+</body>
+</html>"#;
+    let options = InlineOptions {
+        base_url: Some(Url::parse("http://127.0.0.1:5000").unwrap()),
+        ..Default::default()
+    };
+    let inliner = CSSInliner::new(options);
+    let result = inliner.inline(&html).unwrap();
+    assert!(result.ends_with(
+        r#"<body>
+<h1 style="color: blue;">Big Text</h1>
+<h2 style="color: red;">Smaller Text</h2>
+
+</body></html>"#
+    ))
+}
+
 #[test]
 fn customize_inliner() {
     let options = InlineOptions {