Merge pull request #2966 from allan-silva/wc-files0-from-opt

wc: implement files0-from option

src/uu/wc/src/wc.rs

@@ -20,12 +20,15 @@ use word_count::{TitledWordCount, WordCount};
 use clap::{crate_version, App, AppSettings, Arg, ArgMatches};
 
 use std::cmp::max;
+use std::error::Error;
+use std::ffi::OsStr;
+use std::fmt::Display;
 use std::fs::{self, File};
-use std::io::{self, Write};
+use std::io::{self, Read, Write};
 use std::path::{Path, PathBuf};
 
 use uucore::display::{Quotable, Quoted};
-use uucore::error::{UResult, USimpleError};
+use uucore::error::{UError, UResult, USimpleError};
 
 /// The minimum character width for formatting counts when reading from stdin.
 const MINIMUM_WIDTH: usize = 7;
@@ -83,12 +86,14 @@ more than one FILE is specified.";
 pub mod options {
     pub static BYTES: &str = "bytes";
     pub static CHAR: &str = "chars";
+    pub static FILES0_FROM: &str = "files0-from";
     pub static LINES: &str = "lines";
     pub static MAX_LINE_LENGTH: &str = "max-line-length";
     pub static WORDS: &str = "words";
 }
 
 static ARG_FILES: &str = "files";
+static STDIN_REPR: &str = "-";
 
 fn usage() -> String {
     format!(
@@ -115,12 +120,22 @@ enum Input {
     Stdin(StdinKind),
 }
 
+impl From<&OsStr> for Input {
+    fn from(input: &OsStr) -> Self {
+        if input == STDIN_REPR {
+            Self::Stdin(StdinKind::Explicit)
+        } else {
+            Self::Path(input.into())
+        }
+    }
+}
+
 impl Input {
     /// Converts input to title that appears in stats.
     fn to_title(&self) -> Option<&Path> {
         match self {
             Input::Path(path) => Some(path),
-            Input::Stdin(StdinKind::Explicit) => Some("-".as_ref()),
+            Input::Stdin(StdinKind::Explicit) => Some(STDIN_REPR.as_ref()),
             Input::Stdin(StdinKind::Implicit) => None,
         }
     }
@@ -133,29 +148,43 @@ impl Input {
     }
 }
 
+#[derive(Debug)]
+enum WcError {
+    FilesDisabled(String),
+    StdinReprNotAllowed(String),
+}
+
+impl UError for WcError {
+    fn code(&self) -> i32 {
+        match self {
+            WcError::FilesDisabled(_) | WcError::StdinReprNotAllowed(_) => 1,
+        }
+    }
+
+    fn usage(&self) -> bool {
+        matches!(self, WcError::FilesDisabled(_))
+    }
+}
+
+impl Error for WcError {}
+
+impl Display for WcError {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            WcError::FilesDisabled(message) | WcError::StdinReprNotAllowed(message) => {
+                write!(f, "{}", message)
+            }
+        }
+    }
+}
+
 #[uucore::main]
 pub fn uumain(args: impl uucore::Args) -> UResult<()> {
     let usage = usage();
 
     let matches = uu_app().override_usage(&usage[..]).get_matches_from(args);
 
-    let mut inputs: Vec<Input> = matches
-        .values_of_os(ARG_FILES)
-        .map(|v| {
-            v.map(|i| {
-                if i == "-" {
-                    Input::Stdin(StdinKind::Explicit)
-                } else {
-                    Input::Path(i.into())
-                }
-            })
-            .collect()
-        })
-        .unwrap_or_default();
-
-    if inputs.is_empty() {
-        inputs.push(Input::Stdin(StdinKind::Implicit));
-    }
+    let inputs = inputs(&matches)?;
 
     let settings = Settings::new(&matches);
 
@@ -179,6 +208,17 @@ pub fn uu_app<'a>() -> App<'a> {
                 .long(options::CHAR)
                 .help("print the character counts"),
         )
+        .arg(
+            Arg::new(options::FILES0_FROM)
+                .long(options::FILES0_FROM)
+                .takes_value(true)
+                .value_name("F")
+                .help(
+                    "read input from the files specified by
+    NUL-terminated names in file F;
+    If F is - then read names from standard input",
+                ),
+        )
         .arg(
             Arg::new(options::LINES)
                 .short('l')
@@ -205,6 +245,47 @@ pub fn uu_app<'a>() -> App<'a> {
         )
 }
 
+fn inputs(matches: &ArgMatches) -> UResult<Vec<Input>> {
+    match matches.values_of_os(ARG_FILES) {
+        Some(os_values) => {
+            if matches.is_present(options::FILES0_FROM) {
+                return Err(WcError::FilesDisabled(
+                    "file operands cannot be combined with --files0-from".into(),
+                )
+                .into());
+            }
+
+            Ok(os_values.map(Input::from).collect())
+        }
+        None => match matches.value_of(options::FILES0_FROM) {
+            Some(files_0_from) => create_paths_from_files0(files_0_from),
+            None => Ok(vec![Input::Stdin(StdinKind::Implicit)]),
+        },
+    }
+}
+
+fn create_paths_from_files0(files_0_from: &str) -> UResult<Vec<Input>> {
+    let mut paths = String::new();
+    let read_from_stdin = files_0_from == STDIN_REPR;
+
+    if read_from_stdin {
+        io::stdin().lock().read_to_string(&mut paths)?;
+    } else {
+        File::open(files_0_from)?.read_to_string(&mut paths)?;
+    }
+
+    let paths: Vec<&str> = paths.split_terminator('\0').collect();
+
+    if read_from_stdin && paths.contains(&STDIN_REPR) {
+        return Err(WcError::StdinReprNotAllowed(
+            "when reading file names from stdin, no file name of '-' allowed".into(),
+        )
+        .into());
+    }
+
+    Ok(paths.iter().map(OsStr::new).map(Input::from).collect())
+}
+
 fn word_count_from_reader<T: WordCountable>(
     mut reader: T,
     settings: &Settings,

tests/by-util/test_wc.rs

@@ -245,3 +245,57 @@ fn test_files_from_pseudo_filesystem() {
     let result = new_ucmd!().arg("-c").arg("/proc/version").succeeds();
     assert_ne!(result.stdout_str(), "0 /proc/version\n");
 }
+
+#[test]
+fn test_files0_disabled_files_argument() {
+    const MSG: &str = "file operands cannot be combined with --files0-from";
+    new_ucmd!()
+        .args(&["--files0-from=files0_list.txt"])
+        .arg("lorem_ipsum.txt")
+        .fails()
+        .stderr_contains(MSG)
+        .stdout_is("");
+}
+
+#[test]
+fn test_files0_from() {
+    new_ucmd!()
+        .args(&["--files0-from=files0_list.txt"])
+        .run()
+        .stdout_is(
+            "  13  109  772 lorem_ipsum.txt\n  18  204 1115 moby_dick.txt\n   5   57  302 \
+             alice_in_wonderland.txt\n  36  370 2189 total\n",
+        );
+}
+
+#[test]
+fn test_files0_from_with_stdin() {
+    new_ucmd!()
+        .args(&["--files0-from=-"])
+        .pipe_in("lorem_ipsum.txt")
+        .run()
+        .stdout_is(" 13 109 772 lorem_ipsum.txt\n");
+}
+
+#[test]
+fn test_files0_from_with_stdin_in_file() {
+    new_ucmd!()
+        .args(&["--files0-from=files0_list_with_stdin.txt"])
+        .pipe_in_fixture("alice_in_wonderland.txt")
+        .run()
+        .stdout_is(
+            "     13     109     772 lorem_ipsum.txt\n     18     204    1115 moby_dick.txt\n      5      57     302 \
+             -\n     36     370    2189 total\n",
+        );
+}
+
+#[test]
+fn test_files0_from_with_stdin_try_read_from_stdin() {
+    const MSG: &str = "when reading file names from stdin, no file name of '-' allowed";
+    new_ucmd!()
+        .args(&["--files0-from=-"])
+        .pipe_in("-")
+        .fails()
+        .stderr_contains(MSG)
+        .stdout_is("");
+}

tests/fixtures/wc/files0_list.txt

@@ -0,0 +1 @@
+lorem_ipsum.txtmoby_dick.txtalice_in_wonderland.txt

tests/fixtures/wc/files0_list_with_stdin.txt

@@ -0,0 +1 @@
+lorem_ipsum.txtmoby_dick.txt-