RGBCube/uutils-coreutils
1
Fork 0
mirror of https://github.com/RGBCube/uutils-coreutils synced 2025-07-29 03:57:44 +00:00
Code Issues Activity

Merge pull request #6575 from BenWiederhake/dev-cksum-nonutf8-filename

Browse source 
cksum: accept non-UTF-8 filenames
This commit is contained in:
Daniel Hofstetter 2024-07-30 14:31:46 +02:00 • committed by GitHub
commit 9fbfb2925f
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 105 additions and 63 deletions
Download patch file Download diff file Expand all files Collapse all files

133
src/uu/cksum/src/cksum.rs
View file

@ -4,8 +4,9 @@
// file that was distributed with this source code.
// spell-checker:ignore (ToDO) fname, algo
use clap::builder::ValueParser;
use clap::{crate_version, value_parser, Arg, ArgAction, Command};
use std::ffi::OsStr;
use std::ffi::{OsStr, OsString};
use std::fs::File;
use std::io::{self, stdin, stdout, BufReader, Read, Write};
use std::iter;
@ -18,7 +19,7 @@ use uucore::checksum::{
use uucore::{
encoding,
error::{FromIo, UResult, USimpleError},
format_usage, help_about, help_section, help_usage, show,
format_usage, help_about, help_section, help_usage, os_str_as_bytes, show,
sum::{div_ceil, Digest},
};
@ -116,52 +117,64 @@ where
};
// The BSD checksum output is 5 digit integer
let bsd_width = 5;
match (options.algo_name, not_file) {
(ALGORITHM_OPTIONS_SYSV, true) => println!(
"{} {}",
sum.parse::<u16>().unwrap(),
div_ceil(sz, options.output_bits)
let (before_filename, should_print_filename, after_filename) = match options.algo_name {
ALGORITHM_OPTIONS_SYSV => (
format!(
"{} {}{}",
sum.parse::<u16>().unwrap(),
div_ceil(sz, options.output_bits),
if not_file { "" } else { " " }
),
!not_file,
String::new(),
),
(ALGORITHM_OPTIONS_SYSV, false) => println!(
"{} {} {}",
sum.parse::<u16>().unwrap(),
div_ceil(sz, options.output_bits),
filename.display()
ALGORITHM_OPTIONS_BSD => (
format!(
"{:0bsd_width$} {:bsd_width$}{}",
sum.parse::<u16>().unwrap(),
div_ceil(sz, options.output_bits),
if not_file { "" } else { " " }
),
!not_file,
String::new(),
),
(ALGORITHM_OPTIONS_BSD, true) => println!(
"{:0bsd_width$} {:bsd_width$}",
sum.parse::<u16>().unwrap(),
div_ceil(sz, options.output_bits)
ALGORITHM_OPTIONS_CRC => (
format!("{sum} {sz}{}", if not_file { "" } else { " " }),
!not_file,
String::new(),
),
(ALGORITHM_OPTIONS_BSD, false) => println!(
"{:0bsd_width$} {:bsd_width$} {}",
sum.parse::<u16>().unwrap(),
div_ceil(sz, options.output_bits),
filename.display()
),
(ALGORITHM_OPTIONS_CRC, true) => println!("{sum} {sz}"),
(ALGORITHM_OPTIONS_CRC, false) => println!("{sum} {sz} {}", filename.display()),
(ALGORITHM_OPTIONS_BLAKE2B, _) if options.tag => {
if let Some(length) = options.length {
// Multiply by 8 here, as we want to print the length in bits.
println!("BLAKE2b-{} ({}) = {sum}", length * 8, filename.display());
} else {
println!("BLAKE2b ({}) = {sum}", filename.display());
}
ALGORITHM_OPTIONS_BLAKE2B if options.tag => {
(
if let Some(length) = options.length {
// Multiply by 8 here, as we want to print the length in bits.
format!("BLAKE2b-{} (", length * 8)
} else {
"BLAKE2b (".to_owned()
},
true,
format!(") = {sum}"),
)
}
_ => {
if options.tag {
println!(
"{} ({}) = {sum}",
options.algo_name.to_ascii_uppercase(),
filename.display()
);
(
format!("{} (", options.algo_name.to_ascii_uppercase()),
true,
format!(") = {sum}"),
)
} else {
let prefix = if options.asterisk { "*" } else { " " };
println!("{sum} {prefix}{}", filename.display());
(format!("{sum} {prefix}"), true, String::new())
}
}
};
print!("{}", before_filename);
if should_print_filename {
// The filename might not be valid UTF-8, and filename.display() would mangle the names.
// Therefore, emit the bytes directly to stdout, without any attempt at encoding them.
let _dropped_result = stdout().write_all(os_str_as_bytes(filename.as_os_str())?);
}
println!("{}", after_filename);
}
Ok(())
@ -209,7 +222,7 @@ fn prompt_asterisk(tag: bool, binary: bool, had_reset: bool) -> bool {
* Don't do it with clap because if it struggling with the --overrides_with
* marking the value as set even if not present
*/
fn had_reset(args: &[String]) -> bool {
fn had_reset(args: &[OsString]) -> bool {
// Indices where "--binary" or "-b", "--tag", and "--untagged" are found
let binary_index = args.iter().position(|x| x == "--binary" || x == "-b");
let tag_index = args.iter().position(|x| x == "--tag");
@ -234,7 +247,7 @@ fn handle_tag_text_binary_flags(matches: &clap::ArgMatches) -> UResult<(bool, bo
let binary_flag: bool = matches.get_flag(options::BINARY);
let args: Vec<String> = std::env::args().collect();
let args: Vec<OsString> = std::env::args_os().collect();
let had_reset = had_reset(&args);
let asterisk: bool = prompt_asterisk(tag, binary_flag, had_reset);
@ -298,7 +311,7 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
// Execute the checksum validation based on the presence of files or the use of stdin
let files = matches.get_many::<String>(options::FILE).map_or_else(
let files = matches.get_many::<OsString>(options::FILE).map_or_else(
|| iter::once(OsStr::new("-")).collect::<Vec<_>>(),
|files| files.map(OsStr::new).collect::<Vec<_>>(),
);
@ -337,7 +350,7 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
asterisk,
};
match matches.get_many::<String>(options::FILE) {
match matches.get_many::<OsString>(options::FILE) {
Some(files) => cksum(opts, files.map(OsStr::new))?,
None => cksum(opts, iter::once(OsStr::new("-")))?,
};
@ -356,6 +369,7 @@ pub fn uu_app() -> Command {
Arg::new(options::FILE)
.hide(true)
.action(clap::ArgAction::Append)
.value_parser(ValueParser::os_string())
.value_hint(clap::ValueHint::FilePath),
)
.arg(
@ -469,61 +483,62 @@ mod tests {
use super::had_reset;
use crate::calculate_blake2b_length;
use crate::prompt_asterisk;
use std::ffi::OsString;
#[test]
fn test_had_reset() {
let args = ["--binary", "--tag", "--untagged"]
.iter()
.map(|&s| s.to_string())
.collect::<Vec<String>>();
.map(|&s| s.into())
.collect::<Vec<OsString>>();
assert!(had_reset(&args));
let args = ["-b", "--tag", "--untagged"]
.iter()
.map(|&s| s.to_string())
.collect::<Vec<String>>();
.map(|&s| s.into())
.collect::<Vec<OsString>>();
assert!(had_reset(&args));
let args = ["-b", "--binary", "--tag", "--untagged"]
.iter()
.map(|&s| s.to_string())
.collect::<Vec<String>>();
.map(|&s| s.into())
.collect::<Vec<OsString>>();
assert!(had_reset(&args));
let args = ["--untagged", "--tag", "--binary"]
.iter()
.map(|&s| s.to_string())
.collect::<Vec<String>>();
.map(|&s| s.into())
.collect::<Vec<OsString>>();
assert!(!had_reset(&args));
let args = ["--untagged", "--tag", "-b"]
.iter()
.map(|&s| s.to_string())
.collect::<Vec<String>>();
.map(|&s| s.into())
.collect::<Vec<OsString>>();
assert!(!had_reset(&args));
let args = ["--binary", "--tag"]
.iter()
.map(|&s| s.to_string())
.collect::<Vec<String>>();
.map(|&s| s.into())
.collect::<Vec<OsString>>();
assert!(!had_reset(&args));
let args = ["--tag", "--untagged"]
.iter()
.map(|&s| s.to_string())
.collect::<Vec<String>>();
.map(|&s| s.into())
.collect::<Vec<OsString>>();
assert!(!had_reset(&args));
let args = ["--text", "--untagged"]
.iter()
.map(|&s| s.to_string())
.collect::<Vec<String>>();
.map(|&s| s.into())
.collect::<Vec<OsString>>();
assert!(!had_reset(&args));
let args = ["--binary", "--untagged"]
.iter()
.map(|&s| s.to_string())
.collect::<Vec<String>>();
.map(|&s| s.into())
.collect::<Vec<OsString>>();
assert!(!had_reset(&args));
}

6
src/uucore/src/lib/lib.rs
View file

@ -222,9 +222,9 @@ pub fn read_yes() -> bool {
}
}
// Helper function for processing delimiter values (which could be non UTF-8)
// It converts OsString to &[u8] for unix targets only
// On non-unix (i.e. Windows) it will just return an error if delimiter value is not UTF-8
/// Helper function for processing delimiter values (which could be non UTF-8)
/// It converts OsString to &[u8] for unix targets only
/// On non-unix (i.e. Windows) it will just return an error if delimiter value is not UTF-8
pub fn os_str_as_bytes(os_string: &OsStr) -> mods::error::UResult<&[u8]> {
#[cfg(unix)]
let bytes = os_string.as_bytes();

29
tests/by-util/test_cksum.rs
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 (words) asdf algo algos mgmt
// spell-checker:ignore (words) asdf algo algos asha mgmt xffname
use crate::common::util::TestScenario;
@ -1250,3 +1250,30 @@ fn test_several_files_error_mgmt() {
.stderr_contains("empty: no properly ")
.stderr_contains("incorrect: no properly ");
}
#[cfg(target_os = "linux")]
#[test]
fn test_non_utf8_filename() {
use std::ffi::OsString;
use std::os::unix::ffi::OsStringExt;
let scene = TestScenario::new(util_name!());
let at = &scene.fixtures;
let filename: OsString = OsStringExt::from_vec(b"funky\xffname".to_vec());
at.touch(&filename);
scene
.ucmd()
.arg(&filename)
.succeeds()
.stdout_is_bytes(b"4294967295 0 funky\xffname\n")
.no_stderr();
scene
.ucmd()
.arg("-asha256")
.arg(filename)
.succeeds()
.stdout_is_bytes(b"SHA256 (funky\xffname) = e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855\n")
.no_stderr();
}