Always check files explicitly if specified by user or matched by user glob (#1816)

See: lycheeverse/lychee-action#305

Author: mre

fixtures/example_dir/example.html

@@ -0,0 +1 @@
+<a href="https://example.com/example_dir/html">This is an HTML file to test glob patterns in lychee.</a>

fixtures/example_dir/example.md

@@ -0,0 +1,3 @@
+# Example Markdown file to test glob patterns in lychee.
+
+It contains a link: https://example.com/example_dir/md

fixtures/example_dir/example.ts

@@ -0,0 +1,2 @@
+// This is a TypeScript file to test glob patterns in lychee.
+// It contains a link: https://example.com/example_dir/ts

fixtures/example_dir/example.tsx

@@ -0,0 +1,14 @@
+import React from 'react';
+
+const Example: React.FC = () => {
+    return (
+        <div>
+            <h1>Example Page</h1>
+            <a href="https://example.com/example_dir/tsx" target="_blank" rel="noopener noreferrer">
+                Visit Example.com
+            </a>
+        </div>
+    );
+};
+
+export default Example;

fixtures/glob_dir/example.html

@@ -0,0 +1 @@
+<a href="https://example.com/glob_dir/html">This is an HTML file to test glob patterns in lychee.</a>

fixtures/glob_dir/example.md

@@ -0,0 +1,3 @@
+# Example Markdown file to test glob patterns in lychee.
+
+It contains a link: https://example.com/glob_dir/md

fixtures/glob_dir/example.ts

@@ -0,0 +1,2 @@
+// This is a TypeScript file to test glob patterns in lychee.
+// It contains a link: https://example.com/glob_dir/ts

fixtures/glob_dir/example.tsx

@@ -0,0 +1,14 @@
+import React from 'react';
+
+const Example: React.FC = () => {
+    return (
+        <div>
+            <h1>Example Page</h1>
+            <a href="https://example.com/glob_dir/tsx" target="_blank" rel="noopener noreferrer">
+                Visit Example.com
+            </a>
+        </div>
+    );
+};
+
+export default Example;

lychee-bin/src/commands/dump_inputs.rs

@@ -16,7 +16,7 @@ pub(crate) async fn dump_inputs(
     inputs: HashSet<Input>,
     output: Option<&PathBuf>,
     excluded_paths: &[String],
-    valid_extensions: &FileExtensions,
+    file_extensions: &FileExtensions,
     skip_hidden: bool,
     skip_gitignored: bool,
 ) -> Result<ExitCode> {
@@ -34,7 +34,7 @@ pub(crate) async fn dump_inputs(
 
     for input in inputs {
         let sources_stream = input.get_sources(
-            valid_extensions.clone(),
+            file_extensions.clone(),
             skip_hidden,
             skip_gitignored,
             &excluded_path_filter,

lychee-bin/tests/cli.rs

@@ -2622,4 +2622,104 @@ mod cli {
             .stdout(contains("fixtures/invalid_utf8/index.html"))
             .stdout(contains("fixtures/invalid_utf8/invalid_utf8.txt"));
     }
+
+    /// Check that files specified via glob patterns are always checked
+    /// no matter their extension. I.e. extensions are ignored for files
+    /// explicitly specified by the user.
+    ///
+    /// See https://github.com/lycheeverse/lychee-action/issues/305
+    #[test]
+    fn test_globbed_files_are_always_checked() {
+        let input = fixtures_path().join("glob_dir/**/*.tsx");
+
+        // The directory contains:
+        // - example.ts
+        // - example.tsx
+        // - example.md
+        // - example.html
+        // But the user only specified the .tsx file via the glob pattern.
+        main_command()
+            .arg("--verbose")
+            // Only check ts, js, and html files by default.
+            // However, all files explicitly specified by the user
+            // should always be checked so this should be ignored.
+            .arg("--extensions=ts,js,html")
+            .arg(input)
+            .assert()
+            .failure()
+            .stdout(contains("1 Total"))
+            .stderr(contains("https://example.com/glob_dir/tsx"));
+    }
+
+    #[test]
+    fn test_extensions_work_on_glob_files_directory() {
+        let input = fixtures_path().join("glob_dir");
+
+        // Make sure all files matching the given extensions are checked
+        // if we specify a directory (and not a glob pattern).
+        main_command()
+            .arg("--verbose")
+            .arg("--extensions=ts,html")
+            .arg(input)
+            .assert()
+            .failure()
+            .stdout(contains("2 Total"))
+            // Note: The space is intentional to avoid matching tsx.
+            .stderr(contains("https://example.com/glob_dir/ts "))
+            // TSX files are ignored because we did not specify
+            // that extension. So `https://example.com/tsx"` should be missing from the output.
+            .stderr(contains("https://example.com/glob_dir/tsx").not())
+            // Markdown is also ignored because we did not specify that extension.
+            .stderr(contains("https://example.com/glob_dir/md").not())
+            .stderr(contains("https://example.com/glob_dir/html"));
+    }
+
+    /// We define two inputs, one being a glob pattern, the other a directory path.
+    /// The extensions should only apply to the directory path, not the glob pattern.
+    #[test]
+    fn test_extensions_apply_to_files_not_globs() {
+        let glob_input = fixtures_path().join("glob_dir/**/*.tsx");
+        let dir_input = fixtures_path().join("example_dir");
+
+        main_command()
+            .arg("--verbose")
+            .arg("--extensions=html,md")
+            .arg(glob_input)
+            .arg(dir_input)
+            .assert()
+            .failure()
+            .stdout(contains("3 Total"))
+            // Only TSX files are matched by the glob pattern.
+            .stderr(contains("https://example.com/glob_dir/tsx"))
+            .stderr(contains("https://example.com/glob_dir/ts ").not())
+            .stderr(contains("https://example.com/glob_dir/md").not())
+            .stderr(contains("https://example.com/glob_dir/html").not())
+            // For the example_dir, the extensions should apply.
+            .stderr(contains("https://example.com/example_dir/html"))
+            .stderr(contains("https://example.com/example_dir/md"))
+            // TS files in example_dir are ignored because we did not specify that extension.
+            .stderr(contains("https://example.com/example_dir/ts ").not())
+            // TSX files in examle_dir are ignored because we did not specify that extension.
+            .stderr(contains("https://example.com/example_dir/tsx").not());
+    }
+
+    /// Individual files should always be checked, even if their
+    /// extension does not match the given extensions.
+    #[test]
+    fn test_file_inputs_always_get_checked_no_matter_their_extension() {
+        let ts_input_file = fixtures_path().join("glob_dir/example.ts");
+        let md_input_file = fixtures_path().join("glob_dir/example.md");
+
+        main_command()
+            .arg("--verbose")
+            .arg("--dump")
+            .arg("--extensions=html,md")
+            .arg(ts_input_file)
+            .arg(md_input_file)
+            .assert()
+            .success()
+            .stderr("") // Ensure stderr is empty
+            .stdout(contains("https://example.com/glob_dir/ts"))
+            .stdout(contains("https://example.com/glob_dir/md"));
+    }
 }