tr: Add ambiguous octal escape warning, issue #6821

Author: OshinoShinobu-Chan

src/uu/tr/src/operation.rs

@@ -16,13 +16,15 @@ use nom::{
     IResult,
 };
 use std::{
+    char,
     collections::{HashMap, HashSet},
     error::Error,
     fmt::{Debug, Display},
     io::{BufRead, Write},
     ops::Not,
 };
 use uucore::error::UError;
+use uucore::show_warning;
 
 #[derive(Debug, Clone)]
 pub enum BadSequence {
@@ -293,7 +295,9 @@ impl Sequence {
             Self::parse_class,
             Self::parse_char_equal,
             // NOTE: This must be the last one
-            map(Self::parse_backslash_or_char, |s| Ok(Self::Char(s))),
+            map(Self::parse_backslash_or_char_with_warning, |s| {
+                Ok(Self::Char(s))
+            }),
         )))(input)
         .map(|(_, r)| r)
         .unwrap()
@@ -302,10 +306,16 @@ impl Sequence {
     }
 
     fn parse_octal(input: &[u8]) -> IResult<&[u8], u8> {
+        // For `parse_char_range`, `parse_char_star`, `parse_char_repeat`, `parse_char_equal`.
+        // Bcause in these patterns, there's no ambiguous cases.
+        preceded(tag("\\"), Self::parse_octal_up_to_three_digits)(input)
+    }
+
+    fn parse_octal_with_warning(input: &[u8]) -> IResult<&[u8], u8> {
         preceded(
             tag("\\"),
             alt((
-                Self::parse_octal_up_to_three_digits,
+                Self::parse_octal_up_to_three_digits_with_warning,
                 // Fallback for if the three digit octal escape is greater than \377 (0xFF), and therefore can't be
                 // parsed as as a byte
                 // See test `test_multibyte_octal_sequence`
@@ -319,13 +329,31 @@ impl Sequence {
             recognize(many_m_n(1, 3, one_of("01234567"))),
             |out: &[u8]| {
                 let str_to_parse = std::str::from_utf8(out).unwrap();
+                match u8::from_str_radix(str_to_parse, 8) {
+                    Ok(ue) => Some(ue),
+                    Err(_pa) => None,
+                }
+            },
+        )(input)
+    }
 
+    fn parse_octal_up_to_three_digits_with_warning(input: &[u8]) -> IResult<&[u8], u8> {
+        map_opt(
+            recognize(many_m_n(1, 3, one_of("01234567"))),
+            |out: &[u8]| {
+                let str_to_parse = std::str::from_utf8(out).unwrap();
                 match u8::from_str_radix(str_to_parse, 8) {
                     Ok(ue) => Some(ue),
                     Err(_pa) => {
-                        // TODO
-                        // A warning needs to be printed here
-                        // See https://github.com/uutils/coreutils/issues/6821
+                        let origin_octal: &str = std::str::from_utf8(input).unwrap();
+                        let actual_octal_tail: &str = std::str::from_utf8(&input[0..2]).unwrap();
+                        let outstand_char: char = char::from_u32(input[2] as u32).unwrap();
+                        show_warning!(
+                            "the ambiguous octal escape \\{} is being\n        interpreted as the 2-byte sequence \\0{}, {}",
+                            origin_octal,
+                            actual_octal_tail,
+                            outstand_char
+                        );
                         None
                     }
                 }
@@ -360,6 +388,14 @@ impl Sequence {
         alt((Self::parse_octal, Self::parse_backslash, Self::single_char))(input)
     }
 
+    fn parse_backslash_or_char_with_warning(input: &[u8]) -> IResult<&[u8], u8> {
+        alt((
+            Self::parse_octal_with_warning,
+            Self::parse_backslash,
+            Self::single_char,
+        ))(input)
+    }
+
     fn single_char(input: &[u8]) -> IResult<&[u8], u8> {
         take(1usize)(input).map(|(l, a)| (l, a[0]))
     }

tests/by-util/test_tr.rs

@@ -1494,9 +1494,7 @@ fn test_multibyte_octal_sequence() {
         .args(&["-d", r"\501"])
         .pipe_in("(1Ł)")
         .succeeds()
-        // TODO
-        // A warning needs to be printed here
-        // See https://github.com/uutils/coreutils/issues/6821
+        .stderr_is("tr: warning: the ambiguous octal escape \\501 is being\n        interpreted as the 2-byte sequence \\050, 1\n")
         .stdout_is("Ł)");
 }