Merge pull request #3455 from mike-kfed/ptx_breakfile

ptx: implement breakfile option

src/uu/ptx/src/ptx.rs

@@ -5,7 +5,7 @@
 //  * For the full copyright and license information, please view the LICENSE
 //  * file that was distributed with this source code.
 
-// spell-checker:ignore (ToDOs) corasick memchr Roff trunc oset iset
+// spell-checker:ignore (ToDOs) corasick memchr Roff trunc oset iset CHARCLASS
 
 use clap::{crate_version, Arg, Command};
 use regex::Regex;
@@ -31,6 +31,8 @@ const ABOUT: &str = "\
     Mandatory arguments to long options are mandatory for short options too.\n\
     With no FILE, or when FILE is -, read standard input. Default is '-F /'.";
 
+const REGEX_CHARCLASS: &str = "^-]\\";
+
 #[derive(Debug)]
 enum OutFormat {
     Dumb,
@@ -88,6 +90,18 @@ fn read_word_filter_file(
     Ok(words)
 }
 
+/// reads contents of file as unique set of characters to be used with the break-file option
+fn read_char_filter_file(
+    matches: &clap::ArgMatches,
+    option: &str,
+) -> std::io::Result<HashSet<char>> {
+    let filename = matches.value_of(option).expect("parsing options failed!");
+    let mut reader = File::open(filename)?;
+    let mut buffer = String::new();
+    reader.read_to_string(&mut buffer)?;
+    Ok(buffer.chars().collect())
+}
+
 #[derive(Debug)]
 struct WordFilter {
     only_specified: bool,
@@ -113,9 +127,23 @@ impl WordFilter {
         } else {
             (false, HashSet::new())
         };
-        if matches.is_present(options::BREAK_FILE) {
-            return Err(PtxError::NotImplemented("-b").into());
-        }
+        let break_set: Option<HashSet<char>> = if matches.is_present(options::BREAK_FILE)
+            && !matches.is_present(options::WORD_REGEXP)
+        {
+            let chars =
+                read_char_filter_file(matches, options::BREAK_FILE).map_err_context(String::new)?;
+            let mut hs: HashSet<char> = if config.gnu_ext {
+                HashSet::new() // really only chars found in file
+            } else {
+                // GNU off means at least these are considered
+                [' ', '\t', '\n'].iter().cloned().collect()
+            };
+            hs.extend(chars);
+            Some(hs)
+        } else {
+            // if -W takes precedence or default
+            None
+        };
         // Ignore empty string regex from cmd-line-args
         let arg_reg: Option<String> = if matches.is_present(options::WORD_REGEXP) {
             match matches.value_of(options::WORD_REGEXP) {
@@ -134,7 +162,21 @@ impl WordFilter {
         let reg = match arg_reg {
             Some(arg_reg) => arg_reg,
             None => {
-                if config.gnu_ext {
+                if break_set.is_some() {
+                    format!(
+                        "[^{}]+",
+                        break_set
+                            .unwrap()
+                            .into_iter()
+                            .map(|c| if REGEX_CHARCLASS.contains(c) {
+                                format!("\\{}", c)
+                            } else {
+                                c.to_string()
+                            })
+                            .collect::<Vec<String>>()
+                            .join("")
+                    )
+                } else if config.gnu_ext {
                     "\\w+".to_owned()
                 } else {
                     "[^ \t\n]+".to_owned()

tests/by-util/test_ptx.rs

@@ -71,3 +71,35 @@ fn gnu_ext_disabled_ignore_and_only_file() {
         .succeeds()
         .stdout_only_fixture("gnu_ext_disabled_ignore_and_only_file.expected");
 }
+
+#[test]
+fn gnu_ext_disabled_output_width_50() {
+    new_ucmd!()
+        .args(&["-G", "-w", "50", "input"])
+        .succeeds()
+        .stdout_only_fixture("gnu_ext_disabled_output_width_50.expected");
+}
+
+#[test]
+fn gnu_ext_disabled_output_width_70() {
+    new_ucmd!()
+        .args(&["-G", "-w", "70", "input"])
+        .succeeds()
+        .stdout_only_fixture("gnu_ext_disabled_output_width_70.expected");
+}
+
+#[test]
+fn gnu_ext_disabled_break_file() {
+    new_ucmd!()
+        .args(&["-G", "-b", "break_file", "input"])
+        .succeeds()
+        .stdout_only_fixture("gnu_ext_disabled_break_file.expected");
+}
+
+#[test]
+fn gnu_ext_disabled_empty_word_regexp_ignores_break_file() {
+    new_ucmd!()
+        .args(&["-G", "-b", "break_file", "-R", "-W", "", "input"])
+        .succeeds()
+        .stdout_only_fixture("gnu_ext_disabled_rightward_no_ref.expected");
+}

tests/fixtures/ptx/break_file

@@ -0,0 +1 @@
+abc_e^-]\

tests/fixtures/ptx/gnu_ext_disabled_break_file.expected

@@ -0,0 +1,42 @@
+.xx "" "" """quotes"", for roff" ""
+.xx "" "and some other like %a, b" "#, c$c" ""
+.xx "" "and some other like %a, b#, c" "$c" ""
+.xx "" "and some other like" "%a, b#, c$c" ""
+.xx "" "and some other like %a" ", b#, c$c" ""
+.xx "" """quotes""," "for roff" ""
+.xx "" "{brackets}" "for tex" ""
+.xx "" "" "hello world!" ""
+.xx "" "let's c" "heck special characters:" ""
+.xx "" "let's check special c" "haracters:" ""
+.xx "" "let's check spec" "ial characters:" ""
+.xx "" "let's chec" "k special characters:" ""
+.xx "" "{brac" "kets} for tex" ""
+.xx "" "oh, and bac" "k\slash" ""
+.xx "" "" "let's check special characters:" ""
+.xx "" "let's check specia" "l characters:" ""
+.xx "" "and some other" "like %a, b#, c$c" ""
+.xx "" "he" "llo world!" ""
+.xx "" "maybe a" "lso~or^" ""
+.xx "" "" "maybe also~or^" ""
+.xx "" "a" "nd some other like %a, b#, c$c" ""
+.xx "" "oh, a" "nd back\slash" ""
+.xx "" "" "oh, and back\slash" ""
+.xx "" "and some" "other like %a, b#, c$c" ""
+.xx "" "let's check special cha" "racters:" ""
+.xx "" "{b" "rackets} for tex" ""
+.xx "" "and some othe" "r like %a, b#, c$c" ""
+.xx "" """quotes"", for" "roff" ""
+.xx "" "let's check special characte" "rs:" ""
+.xx "" """quote" "s"", for roff" ""
+.xx "" "oh, and back\sla" "sh" ""
+.xx "" "oh, and back\" "slash" ""
+.xx "" "and" "some other like %a, b#, c$c" ""
+.xx "" "let's check" "special characters:" ""
+.xx "" "let's check special charac" "ters:" ""
+.xx "" "{brackets} for" "tex" ""
+.xx "" "le" "t's check special characters:" ""
+.xx "" "{bracke" "ts} for tex" ""
+.xx "" "hello" "world!" ""
+.xx "" "{brackets} for te" "x" ""
+.xx "" "ma" "ybe also~or^" ""
+.xx "" "" "{brackets} for tex" ""

tests/fixtures/ptx/gnu_ext_disabled_output_width_50.expected

@@ -0,0 +1,24 @@
+.xx "" "" """quotes"", for roff" ""
+.xx "" "and some other like" "%a, b#, c$c" ""
+.xx "" "maybe" "also~or^" ""
+.xx "%a, b#, c$c" "" "and some other like" ""
+.xx "" "oh," "and back\slash" ""
+.xx "" "some other like %a," "b#, c$c" "and"
+.xx "" "oh, and" "back\slash" ""
+.xx "" "other like %a, b#," "c$c" "and some"
+.xx "" "let's check special" "characters:" ""
+.xx "characters:" "let's" "check special" ""
+.xx "" """quotes""," "for roff" ""
+.xx "" "{brackets}" "for tex" ""
+.xx "" "" "hello world!" ""
+.xx "characters:" "" "let's check special" ""
+.xx "" "and some other" "like %a, b#, c$c" ""
+.xx "" "" "maybe also~or^" ""
+.xx "" "" "oh, and back\slash" ""
+.xx "" "and some" "other like %a, b#, c$c" ""
+.xx "" """quotes"", for" "roff" ""
+.xx "b#, c$c" "and" "some other like %a," ""
+.xx "" "let's check" "special characters:" ""
+.xx "" "{brackets} for" "tex" ""
+.xx "" "hello" "world!" ""
+.xx "" "" "{brackets} for tex" ""

tests/fixtures/ptx/gnu_ext_disabled_output_width_70.expected

@@ -0,0 +1,24 @@
+.xx "" "" """quotes"", for roff" ""
+.xx "" "and some other like" "%a, b#, c$c" ""
+.xx "" "maybe" "also~or^" ""
+.xx "" "" "and some other like %a, b#, c$c" ""
+.xx "" "oh," "and back\slash" ""
+.xx "" "and some other like %a," "b#, c$c" ""
+.xx "" "oh, and" "back\slash" ""
+.xx "" "and some other like %a, b#," "c$c" ""
+.xx "" "let's check special" "characters:" ""
+.xx "" "let's" "check special characters:" ""
+.xx "" """quotes""," "for roff" ""
+.xx "" "{brackets}" "for tex" ""
+.xx "" "" "hello world!" ""
+.xx "" "" "let's check special characters:" ""
+.xx "" "and some other" "like %a, b#, c$c" ""
+.xx "" "" "maybe also~or^" ""
+.xx "" "" "oh, and back\slash" ""
+.xx "" "and some" "other like %a, b#, c$c" ""
+.xx "" """quotes"", for" "roff" ""
+.xx "" "and" "some other like %a, b#, c$c" ""
+.xx "" "let's check" "special characters:" ""
+.xx "" "{brackets} for" "tex" ""
+.xx "" "hello" "world!" ""
+.xx "" "" "{brackets} for tex" ""