tr: reject unknown character classes during sequence parsing

This commit is contained in:
can1357
2026-03-19 21:57:14 +01:00
committed by Sylvestre Ledru
parent 593f5b191e
commit 7369f4d4fb
4 changed files with 46 additions and 26 deletions
+1
View File
@@ -27,6 +27,7 @@ tr-warning-invalid-utf8 = invalid utf8 sequence
# Sequence parsing error messages
tr-error-missing-char-class-name = missing character class name '[::]'
tr-error-invalid-char-class = invalid character class { $class }
tr-error-missing-equivalence-class-char = missing equivalence class character '[==]'
tr-error-multiple-char-repeat-in-set2 = only one [c*] repeat construct may appear in string2
tr-error-char-repeat-in-set1 = the [c*] repeat construct may not appear in string1
+1
View File
@@ -28,6 +28,7 @@ tr-warning-invalid-utf8 = séquence UTF-8 non valide
# Messages d'erreur d'analyse de séquence
tr-error-missing-char-class-name = nom de classe de caractères manquant '[::]'
tr-error-invalid-char-class = classe de caractères non valide { $class }
tr-error-missing-equivalence-class-char = caractère de classe d'équivalence manquant '[==]'
tr-error-multiple-char-repeat-in-set2 = seule une construction de répétition [c*] peut apparaître dans string2
tr-error-char-repeat-in-set1 = la construction de répétition [c*] ne peut pas apparaître dans string1
+34 -25
View File
@@ -34,6 +34,7 @@ pub trait ChunkProcessor {
#[derive(Debug, Clone)]
pub enum BadSequence {
MissingCharClassName,
InvalidCharClass(String),
MissingEquivalentClassChar,
MultipleCharRepeatInSet2,
CharRepeatInSet1,
@@ -53,6 +54,13 @@ impl Display for BadSequence {
Self::MissingCharClassName => {
write!(f, "{}", translate!("tr-error-missing-char-class-name"))
}
Self::InvalidCharClass(class) => {
write!(
f,
"{}",
translate!("tr-error-invalid-char-class", "class" => format!("'{}'", class))
)
}
Self::MissingEquivalentClassChar => {
write!(
f,
@@ -499,31 +507,32 @@ impl Sequence {
}
fn parse_class(input: &[u8]) -> IResult<&[u8], Result<Self, BadSequence>> {
delimited(
tag("[:"),
alt((
map(
alt((
value(Self::Class(Class::Alnum), tag("alnum")),
value(Self::Class(Class::Alpha), tag("alpha")),
value(Self::Class(Class::Blank), tag("blank")),
value(Self::Class(Class::Control), tag("cntrl")),
value(Self::Class(Class::Digit), tag("digit")),
value(Self::Class(Class::Graph), tag("graph")),
value(Self::Class(Class::Lower), tag("lower")),
value(Self::Class(Class::Print), tag("print")),
value(Self::Class(Class::Punct), tag("punct")),
value(Self::Class(Class::Space), tag("space")),
value(Self::Class(Class::Upper), tag("upper")),
value(Self::Class(Class::Xdigit), tag("xdigit")),
)),
Ok,
),
value(Err(BadSequence::MissingCharClassName), tag("")),
)),
tag(":]"),
)
.parse(input)
preceded(tag("[:"), terminated(take_until(":]"), tag(":]")))
.parse(input)
.map(|(l, class_name)| {
(
l,
match class_name {
b"" => Err(BadSequence::MissingCharClassName),
b"alnum" => Ok(Self::Class(Class::Alnum)),
b"alpha" => Ok(Self::Class(Class::Alpha)),
b"blank" => Ok(Self::Class(Class::Blank)),
b"cntrl" => Ok(Self::Class(Class::Control)),
b"digit" => Ok(Self::Class(Class::Digit)),
b"graph" => Ok(Self::Class(Class::Graph)),
b"lower" => Ok(Self::Class(Class::Lower)),
b"print" => Ok(Self::Class(Class::Print)),
b"punct" => Ok(Self::Class(Class::Punct)),
b"space" => Ok(Self::Class(Class::Space)),
b"upper" => Ok(Self::Class(Class::Upper)),
b"xdigit" => Ok(Self::Class(Class::Xdigit)),
_ => Err(BadSequence::InvalidCharClass(format!(
"[:{}:]",
String::from_utf8_lossy(class_name)
))),
},
)
})
}
fn parse_char_equal(input: &[u8]) -> IResult<&[u8], Result<Self, BadSequence>> {
+10 -1
View File
@@ -2,7 +2,7 @@
//
// For the full copyright and license information, please view the LICENSE
// file that was distributed with this source code.
// spell-checker:ignore aabbaa aabbcc aabc abbb abbbcddd abcc abcdefabcdef abcdefghijk abcdefghijklmn abcdefghijklmnop ABCDEFGHIJKLMNOPQRS abcdefghijklmnopqrstuvwxyz ABCDEFGHIJKLMNOPQRSTUVWXYZ ABCDEFZZ abcxyz ABCXYZ abcxyzabcxyz ABCXYZABCXYZ acbdef alnum amzamz AMZXAMZ bbbd cclass cefgm cntrl compl dabcdef dncase Gzabcdefg PQRST upcase wxyzz xdigit XXXYYY xycde xyyye xyyz xyzzzzxyzzzz ZABCDEF Zamz Cdefghijkl Cdefghijklmn asdfqqwweerr qwerr asdfqwer qwer aassddffqwer asdfqwer
// spell-checker:ignore aabbaa aabbcc aabc abbb abbbcddd abcc abcdefabcdef abcdefghijk abcdefghijklmn abcdefghijklmnop ABCDEFGHIJKLMNOPQRS abcdefghijklmnopqrstuvwxyz ABCDEFGHIJKLMNOPQRSTUVWXYZ ABCDEFZZ abcxyz ABCXYZ abcxyzabcxyz ABCXYZABCXYZ acbdef alnum amzamz AMZXAMZ bbbd cclass cefgm cntrl compl dabcdef dncase fooclass Gzabcdefg PQRST upcase wxyzz xdigit XXXYYY xycde xyyye xyyz xyzzzzxyzzzz ZABCDEF Zamz Cdefghijkl Cdefghijklmn asdfqqwweerr qwerr asdfqwer qwer aassddffqwer asdfqwer
use uutests::at_and_ucmd;
use uutests::new_ucmd;
@@ -1185,6 +1185,15 @@ fn check_against_gnu_tr_tests_empty_cc() {
.stderr_is("tr: missing character class name '[::]'\n");
}
#[test]
fn check_against_gnu_tr_tests_invalid_cc() {
new_ucmd!()
.args(&["[:fooclass:]", "x"])
.pipe_in("")
.fails()
.stderr_is("tr: invalid character class '[:fooclass:]'\n");
}
#[test]
fn check_against_gnu_tr_tests_repeat_set1() {
new_ucmd!()