du: implement files0-from (#5721)

* du: implement files0-from

Should make tests/du/files0-from-dir pass

* du: prepare tests/du/files0-from.pl

* fix the build on Windows

* add testfile to the ignore list

* remove useless comment

Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com>

* mkdir is enough

Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com>

* address review comments

---------

Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com>

src/uu/du/src/du.rs

@@ -65,6 +65,7 @@ mod options {
     pub const INODES: &str = "inodes";
     pub const EXCLUDE: &str = "exclude";
     pub const EXCLUDE_FROM: &str = "exclude-from";
+    pub const FILES0_FROM: &str = "files0-from";
     pub const VERBOSE: &str = "verbose";
     pub const FILE: &str = "FILE";
 }
@@ -587,6 +588,49 @@ pub fn div_ceil(a: u64, b: u64) -> u64 {
     (a + b - 1) / b
 }
 
+// Read file paths from the specified file, separated by null characters
+fn read_files_from(file_name: &str) -> Result<Vec<PathBuf>, std::io::Error> {
+    let reader: Box<dyn BufRead> = if file_name == "-" {
+        // Read from standard input
+        Box::new(BufReader::new(std::io::stdin()))
+    } else {
+        // First, check if the file_name is a directory
+        let path = PathBuf::from(file_name);
+        if path.is_dir() {
+            return Err(std::io::Error::new(
+                std::io::ErrorKind::Other,
+                format!("{}: read error: Is a directory", file_name),
+            ));
+        }
+
+        // Attempt to open the file and handle the error if it does not exist
+        match File::open(file_name) {
+            Ok(file) => Box::new(BufReader::new(file)),
+            Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
+                return Err(std::io::Error::new(
+                    std::io::ErrorKind::Other,
+                    format!(
+                        "cannot open '{}' for reading: No such file or directory",
+                        file_name
+                    ),
+                ))
+            }
+            Err(e) => return Err(e),
+        }
+    };
+
+    let mut paths = Vec::new();
+
+    for line in reader.split(b'\0') {
+        let path = line?;
+        if !path.is_empty() {
+            paths.push(PathBuf::from(String::from_utf8_lossy(&path).to_string()));
+        }
+    }
+
+    Ok(paths)
+}
+
 #[uucore::main]
 #[allow(clippy::cognitive_complexity)]
 pub fn uumain(args: impl uucore::Args) -> UResult<()> {
@@ -601,13 +645,28 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
         summarize,
     )?;
 
-    let files = match matches.get_one::<String>(options::FILE) {
+    let files = if let Some(file_from) = matches.get_one::<String>(options::FILES0_FROM) {
-        Some(_) => matches
+        if file_from == "-" && matches.get_one::<String>(options::FILE).is_some() {
-            .get_many::<String>(options::FILE)
+            return Err(std::io::Error::new(
-            .unwrap()
+                std::io::ErrorKind::Other,
-            .map(PathBuf::from)
+                format!(
-            .collect(),
+                    "extra operand {}\nfile operands cannot be combined with --files0-from",
-        None => vec![PathBuf::from(".")],
+                    matches.get_one::<String>(options::FILE).unwrap().quote()
+                ),
+            )
+            .into());
+        }
+
+        read_files_from(file_from)?
+    } else {
+        match matches.get_one::<String>(options::FILE) {
+            Some(_) => matches
+                .get_many::<String>(options::FILE)
+                .unwrap()
+                .map(PathBuf::from)
+                .collect(),
+            None => vec![PathBuf::from(".")],
+        }
     };
 
     let time = matches.contains_id(options::TIME).then(|| {
@@ -954,6 +1013,14 @@ pub fn uu_app() -> Command {
                 .help("exclude files that match any pattern in FILE")
                 .action(ArgAction::Append)
         )
+        .arg(
+            Arg::new(options::FILES0_FROM)
+                .long("files0-from")
+                .value_name("FILE")
+                .value_hint(clap::ValueHint::FilePath)
+                .help("summarize device usage of the NUL-terminated file names specified in file F; if F is -, then read names from standard input")
+                .action(ArgAction::Append)
+        )
         .arg(
             Arg::new(options::TIME)
                 .long(options::TIME)

tests/by-util/test_du.rs

@@ -3,10 +3,9 @@
 // For the full copyright and license information, please view the LICENSE
 // file that was distributed with this source code.
 
-// spell-checker:ignore (paths) sublink subwords azerty azeaze xcwww azeaz amaz azea qzerty tazerty tsublink
+// spell-checker:ignore (paths) sublink subwords azerty azeaze xcwww azeaz amaz azea qzerty tazerty tsublink testfile1 testfile2 filelist testdir testfile
 #[cfg(not(windows))]
 use regex::Regex;
-#[cfg(not(windows))]
 use std::io::Write;
 
 #[cfg(any(target_os = "linux", target_os = "android"))]
@@ -991,3 +990,74 @@ fn test_du_symlink_multiple_fail() {
     assert_eq!(result.code(), 1);
     result.stdout_contains("4\tfile1\n");
 }
+
+#[test]
+// Disable on Windows because of different path separators and handling of null characters
+#[cfg(not(target_os = "windows"))]
+fn test_du_files0_from() {
+    let ts = TestScenario::new(util_name!());
+    let at = &ts.fixtures;
+
+    let mut file1 = at.make_file("testfile1");
+    file1.write_all(b"content1").unwrap();
+    let mut file2 = at.make_file("testfile2");
+    file2.write_all(b"content2").unwrap();
+
+    at.mkdir("testdir");
+    let mut file3 = at.make_file("testdir/testfile3");
+    file3.write_all(b"content3").unwrap();
+
+    let mut file_list = at.make_file("filelist");
+    write!(file_list, "testfile1\0testfile2\0testdir\0").unwrap();
+
+    ts.ucmd()
+        .arg("--files0-from=filelist")
+        .succeeds()
+        .stdout_contains("testfile1")
+        .stdout_contains("testfile2")
+        .stdout_contains("testdir");
+}
+
+#[test]
+fn test_du_files0_from_stdin() {
+    let ts = TestScenario::new(util_name!());
+    let at = &ts.fixtures;
+
+    let mut file1 = at.make_file("testfile1");
+    file1.write_all(b"content1").unwrap();
+    let mut file2 = at.make_file("testfile2");
+    file2.write_all(b"content2").unwrap();
+
+    let input = "testfile1\0testfile2\0";
+
+    ts.ucmd()
+        .arg("--files0-from=-")
+        .pipe_in(input)
+        .succeeds()
+        .stdout_contains("testfile1")
+        .stdout_contains("testfile2");
+}
+
+#[test]
+fn test_du_files0_from_dir() {
+    let ts = TestScenario::new(util_name!());
+    let at = &ts.fixtures;
+
+    at.mkdir("dir");
+
+    let result = ts.ucmd().arg("--files0-from=dir").fails();
+    assert_eq!(result.stderr_str(), "du: dir: read error: Is a directory\n");
+}
+
+#[test]
+fn test_du_files0_from_combined() {
+    let ts = TestScenario::new(util_name!());
+    let at = &ts.fixtures;
+
+    at.mkdir("dir");
+
+    let result = ts.ucmd().arg("--files0-from=-").arg("foo").fails();
+    let stderr = result.stderr_str();
+
+    assert!(stderr.contains("file operands cannot be combined with --files0-from"));
+}

util/build-gnu.sh

@@ -283,6 +283,10 @@ sed -i -E "s|^([^#]*2_31.*)$|#\1|g" tests/printf/printf-cov.pl
 
 sed -i -e "s/du: invalid -t argument/du: invalid --threshold argument/" -e "s/du: option requires an argument/error: a value is required for '--threshold <SIZE>' but none was supplied/" -e "/Try 'du --help' for more information./d" tests/du/threshold.sh
 
+# Remove the extra output check
+sed -i -e "s|Try '\$prog --help' for more information.\\\n||" tests/du/files0-from.pl
+sed -i -e "s|when reading file names from stdin, no file name of\"|-: No such file or directory\n\"|" -e "s| '-' allowed\\\n||" tests/du/files0-from.pl
+
 awk 'BEGIN {count=0} /compare exp out2/ && count < 6 {sub(/compare exp out2/, "grep -q \"cannot be used with\" out2"); count++} 1' tests/df/df-output.sh > tests/df/df-output.sh.tmp && mv tests/df/df-output.sh.tmp tests/df/df-output.sh
 
 # with ls --dired, in case of error, we have a slightly different error position