Merge pull request #180 from ebfe/tr

tr: speed up and add '--complement'

tr/test.rs

@@ -1,7 +1,7 @@
 use std::io::process::Command;
 
-fn run(input: &str, set1: &str, set2: &str) -> Vec<u8> {
-    let mut process = Command::new("build/tr").arg(set1).arg(set2).spawn().unwrap();
+fn run(input: &str, args: &[&'static str]) -> Vec<u8> {
+    let mut process = Command::new("build/tr").args(args).spawn().unwrap();
 
     process.stdin.take_unwrap().write_str(input).unwrap();
 
@@ -14,20 +14,32 @@ fn run(input: &str, set1: &str, set2: &str) -> Vec<u8> {
 
 #[test]
 fn test_toupper() {
-    let out = run("!abcd!", "a-z", "A-Z");
+    let out = run("!abcd!", ["a-z", "A-Z"]);
     assert_eq!(out.as_slice(), bytes!("!ABCD!"));
 }
 
 #[test]
 fn test_small_set2() {
-    let out = run("@0123456789", "0-9", "X");
+    let out = run("@0123456789", ["0-9", "X"]);
     assert_eq!(out.as_slice(), bytes!("@XXXXXXXXXX"));
 }
 
 #[test]
 fn test_unicode() {
-    let out = run("(,°□°）, ┬─┬", ", ┬─┬", "╯︵┻━┻");
+    let out = run("(,°□°）, ┬─┬", [", ┬─┬", "╯︵┻━┻"]);
     assert_eq!(out.as_slice(), bytes!("(╯°□°）╯︵┻━┻"));
 }
 
+#[test]
+fn test_delete() {
+    let out = run("aBcD", ["-d", "a-z"]);
+    assert_eq!(out.as_slice(), bytes!("BD"));
+}
+
+#[test]
+fn test_delete_complement() {
+    let out = run("aBcD", ["-d", "-c", "a-z"]);
+    assert_eq!(out.as_slice(), bytes!("ac"));
+}
+
 

tr/tr.rs

@@ -1,4 +1,5 @@
 #![crate_id(name="tr", vers="1.0.0", author="Michael Gehring")]
+#![feature(macro_rules)]
 
 /*
  * This file is part of the uutils coreutils package.
@@ -12,7 +13,8 @@
 extern crate collections;
 extern crate getopts;
 
-use collections::hashmap::{HashMap, HashSet};
+use collections::bitv::BitvSet;
+use collections::smallintmap::SmallIntMap;
 use getopts::OptGroup;
 use std::char::from_u32;
 use std::io::print;
@@ -21,6 +23,9 @@ use std::iter::FromIterator;
 use std::os;
 use std::vec::Vec;
 
+#[path="../common/util.rs"]
+mod util;
+
 static NAME : &'static str = "tr";
 static VERSION : &'static str = "1.0.0";
 
@@ -81,17 +86,23 @@ fn expand_set(s: &str) -> Vec<char> {
     set
 }
 
-fn delete(set: Vec<char>) {
-    let mut hset = HashSet::new();
+fn delete(set: Vec<char>, complement: bool) {
+    let mut bset = BitvSet::new();
     let mut out = stdout();
 
     for &c in set.iter() {
-        hset.insert(c);
+        bset.insert(c as uint);
     }
 
+    let is_allowed = if complement {
+        |c: char| bset.contains(&(c as uint))
+    } else {
+        |c: char| !bset.contains(&(c as uint))
+    };
+
     for c in stdin().chars() {
         match c {
-            Ok(c) if !hset.contains(&c) => out.write_char(c).unwrap(),
+            Ok(c) if is_allowed(c) => out.write_char(c).unwrap(),
             Ok(_) => (),
             Err(err) => fail!("{}", err),
         };
@@ -99,21 +110,21 @@ fn delete(set: Vec<char>) {
 }
 
 fn tr(set1: &[char], set2: &[char]) {
-    let mut map = HashMap::new();
+    let mut map = SmallIntMap::<char>::new();
     let mut out = stdout();
 
     for i in range(0, set1.len()) {
         if i >= set2.len() {
-            map.insert(set1[i], set2[set2.len()-1]);
+            map.insert(set1[i] as uint, set2[set2.len()-1]);
         } else {
-            map.insert(set1[i], set2[i]);
+            map.insert(set1[i] as uint, set2[i]);
         }
     }
 
     for c in stdin().chars() {
         match c {
             Ok(inc) => {
-                let trc = match map.find(&inc) {
+                let trc = match map.find(&(inc as uint)) {
                     Some(t) => *t,
                     None => inc,
                 };
@@ -138,6 +149,8 @@ fn usage(opts: &[OptGroup]) {
 pub fn main() {
     let args: Vec<StrBuf> = os::args().iter().map(|x| x.to_strbuf()).collect();
     let opts = [
+        getopts::optflag("c", "complement", "use the complement of SET1"),
+        getopts::optflag("C", "", "same as -c"),
         getopts::optflag("d", "delete", "delete characters in SET1"),
         getopts::optflag("h", "help", "display this help and exit"),
         getopts::optflag("V", "version", "output version information and exit"),
@@ -145,12 +158,15 @@ pub fn main() {
 
     let matches = match getopts::getopts(args.tail(), opts) {
         Ok(m) => m,
-        Err(err) => fail!("{}", err.to_err_msg()),
+        Err(err) => {
+            show_error!(1, "{}", err.to_err_msg());
+            return;
+        }
     };
 
     if matches.opt_present("help") {
         usage(opts);
-        return
+        return;
     }
 
     if matches.opt_present("version") {
@@ -161,15 +177,21 @@ pub fn main() {
     if matches.free.len() == 0 {
         usage(opts);
         os::set_exit_status(1);
-        return
+        return;
     }
 
     let dflag = matches.opt_present("d");
+    let cflag = matches.opts_present(["c".to_strbuf(), "C".to_strbuf()]);
     let sets = matches.free;
 
+    if cflag && !dflag {
+        show_error!(1, "-c is only supported with -d");
+        return;
+    }
+
     if dflag {
         let set1 = expand_set(sets.get(0).as_slice());
-        delete(set1);
+        delete(set1, cflag);
     } else {
         let set1 = expand_set(sets.get(0).as_slice());
         let set2 = expand_set(sets.get(1).as_slice());