Merge pull request #1852 from fischman/1727-ignore-contain

`--ignore-contain`: ignore directories containing a named entry.

CHANGELOG.md

@@ -1,7 +1,7 @@
 # Upcoming release
 
 ## Features
-
+- Add `--ignore-contain` option to ignore directories containing a named entry (e.g. to ignore [`CACHEDIR.TAG`](https://bford.info/cachedir/)); see #1727 (@fischman).
 
 ## Bugfixes
 

doc/fd.1

@@ -265,6 +265,10 @@ Examples:
   \-\-exclude '*.pyc'
   \-\-exclude node_modules
 .TP
+.BI "\-\-ignore-contain " name
+Exclude directories that (directly) contain the given name.
+This option can be specified multiple times.
+.TP
 .BI "\-\-ignore-file " path
 Add a custom ignore-file in '.gitignore' format.
 These files have a low precedence.

src/cli.rs

@@ -528,6 +528,10 @@ pub struct Opts {
     )]
     pub hyperlink: HyperlinkWhen,
 
+    /// Ignore directories containing the named entry.
+    #[arg(long, value_name = "name")]
+    pub ignore_contain: Vec<String>,
+
     /// Set number of threads to use for searching & executing (default: number
     /// of available CPU cores)
     #[arg(long, short = 'j', value_name = "num", hide_short_help = true, value_parser = str::parse::<NonZeroUsize>)]

src/config.rs

@@ -130,6 +130,9 @@ pub struct Config {
 
     /// Whether or not to use hyperlinks on paths
     pub hyperlink: bool,
+
+    /// Names that should stop traversal down their parent. (e.g. https://bford.info/cachedir/).
+    pub ignore_contain: Vec<String>,
 }
 
 impl Config {

src/main.rs

@@ -326,6 +326,7 @@ fn construct_config(mut opts: Opts, pattern_regexps: &[String]) -> Result<Config
         actual_path_separator,
         max_results: opts.max_results(),
         strip_cwd_prefix: opts.strip_cwd_prefix(|| !(opts.null_separator || has_command)),
+        ignore_contain: opts.ignore_contain,
     })
 }
 

src/walk.rs

@@ -461,11 +461,22 @@ impl WorkerState {
                     return WalkState::Quit;
                 }
 
-                let entry = match entry {
-                    Ok(ref e) if e.depth() == 0 => {
+                if let Ok(e) = &entry {
+                    let entry_path = e.path();
+                    if entry_path.is_dir()
+                        && config
+                            .ignore_contain
+                            .iter()
+                            .any(|ic| entry_path.join(ic).exists())
+                    {
+                        return WalkState::Skip;
+                    }
+                    if e.depth() == 0 {
                         // Skip the root directory entry.
                         return WalkState::Continue;
                     }
+                }
+                let entry = match entry {
                     Ok(e) => DirEntry::normal(e),
                     Err(ignore::Error::WithPath {
                         path,

tests/tests.rs

@@ -2707,3 +2707,54 @@ fn test_hyperlink() {
 
     te.assert_output(&["--hyperlink=always", "a.foo"], &expected);
 }
+
+#[test]
+fn test_ignore_contain() {
+    let te = TestEnv::new(
+        &["include", "exclude", "exclude/sub", "other"],
+        &[
+            "top",
+            "include/foo",
+            "exclude/CACHEDIR.TAG",
+            "exclude/sub/nope",
+            "other/ignoremyparent",
+        ],
+    );
+    let expected = "include/
+    include/foo
+    symlink
+    top";
+    te.assert_output(
+        &[
+            "--ignore-contain=CACHEDIR.TAG",
+            "--ignore-contain=ignoremyparent",
+            ".",
+        ],
+        expected,
+    );
+}
+
+#[test]
+fn test_ignore_contain_precedence_over_depth_check() {
+    let te = TestEnv::new(
+        &["include", "exclude", "exclude/sub"],
+        &[
+            "top",
+            "include/foo",
+            "exclude/CACHEDIR.TAG",
+            "exclude/sub/nope",
+        ],
+    );
+    let expected = "include/foo";
+    te.assert_output(
+        &["--ignore-contain=CACHEDIR.TAG", "--min-depth=2", "."],
+        expected,
+    );
+}
+
+#[test]
+fn test_ignore_contain_precedence_over_root_check() {
+    let te = TestEnv::new(&["include"], &["CACHEDIR.TAG", "top", "include/foo"]);
+    let expected = "";
+    te.assert_output(&["--ignore-contain=CACHEDIR.TAG", "."], expected);
+}