Update parsing of extensions, to handle filenames with dots

src/asset_util/src/lib.rs

@@ -422,19 +422,18 @@ pub fn collect_assets(dir: &Dir, html_transformer: Option<fn(&str) -> String>) -
 /// NOTE: behavior is undefined with symlinks (and esp. symlink loops)!
 fn collect_assets_rec(dir: &Dir, assets: &mut Vec<Asset>) {
     for asset in dir.files() {
-        let file_bytes = asset.contents().to_vec();
-        let (content, encoding, content_type) = match file_extension(asset) {
-            "css" => (file_bytes, ContentEncoding::Identity, ContentType::CSS),
-            "html" => (file_bytes, ContentEncoding::Identity, ContentType::HTML),
-            "ico" => (file_bytes, ContentEncoding::Identity, ContentType::ICO),
-            "json" => (file_bytes, ContentEncoding::Identity, ContentType::JSON),
-            "js" => (file_bytes, ContentEncoding::Identity, ContentType::JS),
-            "js.gz" => (file_bytes, ContentEncoding::GZip, ContentType::JS),
-            "png" => (file_bytes, ContentEncoding::Identity, ContentType::PNG),
-            "svg" => (file_bytes, ContentEncoding::Identity, ContentType::SVG),
-            "webp" => (file_bytes, ContentEncoding::Identity, ContentType::WEBP),
-            "woff2" => (file_bytes, ContentEncoding::Identity, ContentType::WOFF2),
-            "woff2.gz" => (file_bytes, ContentEncoding::GZip, ContentType::WOFF2),
+        let content = asset.contents().to_vec();
+        let (extension, encoding) = file_extension(asset);
+        let content_type = match extension {
+            "css" => ContentType::CSS,
+            "html" => ContentType::HTML,
+            "ico" => ContentType::ICO,
+            "json" => ContentType::JSON,
+            "js" => ContentType::JS,
+            "png" => ContentType::PNG,
+            "svg" => ContentType::SVG,
+            "webp" => ContentType::WEBP,
+            "woff2" => ContentType::WOFF2,
             ext => panic!(
                 "Unknown asset type '{}' for asset '{}'",
                 ext,
@@ -465,21 +464,29 @@ fn collect_assets_rec(dir: &Dir, assets: &mut Vec<Asset>) {
     }
 }
 
-/// Returns the portion of the filename after the first dot.
-/// This corresponds to the file extension for the assets handled by this canister.
-///
-/// The builtin `extension` method on `Path` does not work for file extensions with multiple dots
-/// such as `.js.gz`.
-fn file_extension<'a>(asset: &'a File) -> &'a str {
-    asset
-        .path()
-        .file_name()
-        .unwrap()
-        .to_str()
-        .unwrap()
-        .split_once('.')
-        .unwrap()
-        .1
+/// Returns the extension of the given file, and the encoding type.
+/// If the text after the last dot is "gz" (i.e. this is a gzipped file), then the extension
+/// is the text after the second to last dot and the last dot in the file name
+/// (e.g. "js" for "some.gzipped.file.js.gz"), and the encoding is `ContentEncoding::GZip`.
+/// Otherwise the extension is the text after the last dot in the file name,
+/// and the encoding is `ContentEncoding::Identity`.
+/// This corresponds to the file extension for the assets handled by this crate.
+fn file_extension<'a>(asset: &'a File) -> (&'a str, ContentEncoding) {
+    let extension = asset.path().extension().unwrap().to_str().unwrap();
+    if extension == "gz" {
+        let type_extension = asset
+            .path()
+            .file_name()
+            .unwrap()
+            .to_str()
+            .unwrap()
+            .split('.')
+            .nth_back(1)
+            .unwrap();
+        return (type_extension, ContentEncoding::GZip);
+    } else {
+        return (extension, ContentEncoding::Identity);
+    }
 }
 
 /// Returns the URL paths for a given asset filepath. For instance:
@@ -605,3 +612,30 @@ fn test_filepath_urlpaths() {
         ],
     );
 }
+
+#[test]
+fn should_return_correct_extension() {
+    let path_extension_encoding = [
+        ("path1/some_css_file.css", "css", ContentEncoding::Identity),
+        ("path2/an_html_file.html", "html", ContentEncoding::Identity),
+        ("path3/an_ico_file.ico", "ico", ContentEncoding::Identity),
+        ("path4/json_file.json", "json", ContentEncoding::Identity),
+        ("path5/js_file.js", "js", ContentEncoding::Identity),
+        ("path6/gzipped_js_file.js.gz", "js", ContentEncoding::GZip),
+        ("path7/a_png_file.png", "png", ContentEncoding::Identity),
+        ("path8/svg_file.svg", "svg", ContentEncoding::Identity),
+        ("path9/webp_file.webp", "webp", ContentEncoding::Identity),
+        ("path10/a_file.woff2", "woff2", ContentEncoding::Identity),
+        ("path11/gz_woff2.woff2.gz", "woff2", ContentEncoding::GZip),
+        ("path.dots/ico_file.ico", "ico", ContentEncoding::Identity),
+        ("path13/file.with.dots.js", "js", ContentEncoding::Identity),
+        ("path14.dot/gz_js_file.js.gz", "js", ContentEncoding::GZip),
+        ("path15.dots/.dots.woff2.gz", "woff2", ContentEncoding::GZip),
+    ];
+    for (path, expected_extension, expected_encoding) in path_extension_encoding {
+        assert_eq!(
+            file_extension(&File::new(path, &[42])),
+            (expected_extension, expected_encoding)
+        );
+    }
+}