Merge pull request #7431 from RenjiSann/printf-fix-octal-escape

printf: fix octal escape parsing

Author: cakebaker

src/uu/echo/src/echo.rs

@@ -9,7 +9,7 @@ use std::env;
 use std::ffi::{OsStr, OsString};
 use std::io::{self, StdoutLock, Write};
 use uucore::error::{UResult, USimpleError};
-use uucore::format::{parse_escape_only, EscapedChar, FormatChar};
+use uucore::format::{parse_escape_only, EscapedChar, FormatChar, OctalParsing};
 use uucore::{format_usage, help_about, help_section, help_usage};
 
 const ABOUT: &str = help_about!("echo.md");
@@ -135,7 +135,7 @@ fn execute(
         }
 
         if escaped {
-            for item in parse_escape_only(bytes) {
+            for item in parse_escape_only(bytes, OctalParsing::ThreeDigits) {
                 match item {
                     EscapedChar::End => return Ok(()),
                     c => c.write(&mut *stdout_lock)?,

src/uucore/src/lib/features/format/escape.rs

@@ -17,24 +17,37 @@ pub enum EscapedChar {
     End,
 }
 
-#[repr(u8)]
+#[derive(Clone, Copy, Default)]
+pub enum OctalParsing {
+    #[default]
+    TwoDigits = 2,
+    ThreeDigits = 3,
+}
+
 #[derive(Clone, Copy)]
 enum Base {
-    Oct = 8,
-    Hex = 16,
+    Oct(OctalParsing),
+    Hex,
 }
 
 impl Base {
+    fn as_base(&self) -> u8 {
+        match self {
+            Base::Oct(_) => 8,
+            Base::Hex => 16,
+        }
+    }
+
     fn max_digits(&self) -> u8 {
         match self {
-            Self::Oct => 3,
+            Self::Oct(parsing) => *parsing as u8,
             Self::Hex => 2,
         }
     }
 
     fn convert_digit(&self, c: u8) -> Option<u8> {
         match self {
-            Self::Oct => {
+            Self::Oct(_) => {
                 if matches!(c, b'0'..=b'7') {
                     Some(c - b'0')
                 } else {
@@ -68,7 +81,7 @@ fn parse_code(input: &mut &[u8], base: Base) -> Option<u8> {
         let Some(n) = base.convert_digit(*c) else {
             break;
         };
-        ret = ret.wrapping_mul(base as u8).wrapping_add(n);
+        ret = ret.wrapping_mul(base.as_base()).wrapping_add(n);
         *input = rest;
     }
 
@@ -87,7 +100,9 @@ fn parse_unicode(input: &mut &[u8], digits: u8) -> Option<char> {
     for _ in 1..digits {
         let (c, rest) = input.split_first()?;
         let n = Base::Hex.convert_digit(*c)?;
-        ret = ret.wrapping_mul(Base::Hex as u32).wrapping_add(n as u32);
+        ret = ret
+            .wrapping_mul(Base::Hex.as_base() as u32)
+            .wrapping_add(n as u32);
         *input = rest;
     }
 
@@ -99,13 +114,16 @@ fn parse_unicode(input: &mut &[u8], digits: u8) -> Option<char> {
 pub struct EscapeError {}
 
 /// Parse an escape sequence, like `\n` or `\xff`, etc.
-pub fn parse_escape_code(rest: &mut &[u8]) -> Result<EscapedChar, EscapeError> {
+pub fn parse_escape_code(
+    rest: &mut &[u8],
+    zero_octal_parsing: OctalParsing,
+) -> Result<EscapedChar, EscapeError> {
     if let [c, new_rest @ ..] = rest {
         // This is for the \NNN syntax for octal sequences.
         // Note that '0' is intentionally omitted because that
         // would be the \0NNN syntax.
         if let b'1'..=b'7' = c {
-            if let Some(parsed) = parse_code(rest, Base::Oct) {
+            if let Some(parsed) = parse_code(rest, Base::Oct(OctalParsing::ThreeDigits)) {
                 return Ok(EscapedChar::Byte(parsed));
             }
         }
@@ -131,7 +149,7 @@ pub fn parse_escape_code(rest: &mut &[u8]) -> Result<EscapedChar, EscapeError> {
                 }
             }
             b'0' => Ok(EscapedChar::Byte(
-                parse_code(rest, Base::Oct).unwrap_or(b'\0'),
+                parse_code(rest, Base::Oct(zero_octal_parsing)).unwrap_or(b'\0'),
             )),
             b'u' => Ok(EscapedChar::Char(parse_unicode(rest, 4).unwrap_or('\0'))),
             b'U' => Ok(EscapedChar::Char(parse_unicode(rest, 8).unwrap_or('\0'))),

src/uucore/src/lib/features/format/mod.rs

@@ -51,7 +51,7 @@ use os_display::Quotable;
 use crate::error::UError;
 
 pub use self::{
-    escape::{parse_escape_code, EscapedChar},
+    escape::{parse_escape_code, EscapedChar, OctalParsing},
     num_format::Formatter,
 };
 
@@ -184,10 +184,12 @@ pub fn parse_spec_and_escape(
         }
         [b'\\', rest @ ..] => {
             current = rest;
-            Some(match parse_escape_code(&mut current) {
-                Ok(c) => Ok(FormatItem::Char(c)),
-                Err(_) => Err(FormatError::MissingHex),
-            })
+            Some(
+                match parse_escape_code(&mut current, OctalParsing::default()) {
+                    Ok(c) => Ok(FormatItem::Char(c)),
+                    Err(_) => Err(FormatError::MissingHex),
+                },
+            )
         }
         [c, rest @ ..] => {
             current = rest;
@@ -224,13 +226,19 @@ pub fn parse_spec_only(
 }
 
 /// Parse a format string containing escape sequences
-pub fn parse_escape_only(fmt: &[u8]) -> impl Iterator<Item = EscapedChar> + '_ {
+pub fn parse_escape_only(
+    fmt: &[u8],
+    zero_octal_parsing: OctalParsing,
+) -> impl Iterator<Item = EscapedChar> + '_ {
     let mut current = fmt;
     std::iter::from_fn(move || match current {
         [] => None,
         [b'\\', rest @ ..] => {
             current = rest;
-            Some(parse_escape_code(&mut current).unwrap_or(EscapedChar::Backslash(b'x')))
+            Some(
+                parse_escape_code(&mut current, zero_octal_parsing)
+                    .unwrap_or(EscapedChar::Backslash(b'x')),
+            )
         }
         [c, rest @ ..] => {
             current = rest;

src/uucore/src/lib/features/format/spec.rs

@@ -12,7 +12,7 @@ use super::{
         self, Case, FloatVariant, ForceDecimal, Formatter, NumberAlignment, PositiveSign, Prefix,
         UnsignedIntVariant,
     },
-    parse_escape_only, ArgumentIter, FormatChar, FormatError,
+    parse_escape_only, ArgumentIter, FormatChar, FormatError, OctalParsing,
 };
 use std::{io::Write, ops::ControlFlow};
 
@@ -348,7 +348,7 @@ impl Spec {
             Self::EscapedString => {
                 let s = args.get_str();
                 let mut parsed = Vec::new();
-                for c in parse_escape_only(s.as_bytes()) {
+                for c in parse_escape_only(s.as_bytes(), OctalParsing::default()) {
                     match c.write(&mut parsed)? {
                         ControlFlow::Continue(()) => {}
                         ControlFlow::Break(()) => {

tests/by-util/test_printf.rs

@@ -59,6 +59,14 @@ fn escaped_octal() {
     new_ucmd!().args(&["\\101"]).succeeds().stdout_only("A");
 }
 
+#[test]
+fn escaped_octal_and_newline() {
+    new_ucmd!()
+        .args(&["\\0377\\n"])
+        .succeeds()
+        .stdout_only("\x1F7\n");
+}
+
 #[test]
 fn escaped_unicode_four_digit() {
     new_ucmd!().args(&["\\u0125"]).succeeds().stdout_only("ĥ");