1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
//! Quote strings for use with `/bin/sh`.
//!
//! # Notes
//!
//! The following escapes seem to be "okay":
//!
//! ```text
//! \a     alert (bell)
//! \b     backspace
//! \f     form feed
//! \n     new line
//! \r     carriage return
//! \t     horizontal tab
//! \v     vertical tab
//! \\     backslash
//! \nnn   the eight-bit character whose value is the octal value nnn
//! ```
//!
//! I wasn't able to find any definitive statement of exactly how Bourne Shell
//! strings should be escaped, mainly because "Bourne Shell" or `/bin/sh` can
//! refer to many different pieces of software: Bash has a Bourne Shell mode,
//! `/bin/sh` on Ubuntu is actually Dash, and on macOS 12.3 (and later, and
//! possibly earlier) all bets are off:
//!
//! > `sh` is a POSIX-compliant command interpreter (shell). It is implemented
//! > by re-execing as either `bash(1)`, `dash(1)`, or `zsh(1)` as determined by
//! > the symbolic link located at `/private/var/select/sh`. If
//! > `/private/var/select/sh` does not exist or does not point to a valid
//! > shell, `sh` will use one of the supported shells.
//!
//! ⚠️ In practice, however, bytes between 0x80 and 0xff inclusive **cannot** be
//! escaped with `\nnn` notation. The shell simply ignores these escapes and
//! treats `\nnn` as a literal string of 4 characters. Hence, in this module,
//! these bytes are reproduced as-is within the quoted string output, with no
//! special escaping.
//!
//! The code in this module sticks to escape sequences that I consider
//! "standard" by a heuristic known only to me. It operates byte by byte, making
//! no special allowances for multi-byte character sets. In other words, it's up
//! to the caller to figure out encoding for non-ASCII characters. A significant
//! use case for this code is to escape filenames into scripts, and on *nix
//! variants I understand that filenames are essentially arrays of bytes, even
//! if the OS adds some normalisation and case-insensitivity on top.
//!
//! If you have some expertise in this area I would love to hear from you.
//!

use std::ffi::OsString;
use std::os::unix::ffi::OsStringExt;

use crate::ascii::Char;

/// Escape a string of *bytes* into a new `Vec<u8>`.
///
/// This will return one of the following:
/// - The string as-is, if no escaping is necessary.
/// - A quoted string containing ANSI-C-like escapes, like `'foo\nbar'`.
///
/// See [`escape_into`] for a variant that extends an existing `Vec` instead of
/// allocating a new one.
///
/// The input argument is `Into<OsString>`, so you can pass in regular Rust
/// strings, `PathBuf`, and so on. For a regular Rust string it will be quoted
/// byte for byte.
///
/// # Examples
///
/// ```
/// # use shell_quote::sh;
/// assert_eq!(sh::escape("foobar"), b"foobar");
/// assert_eq!(sh::escape("foo bar"), b"'foo bar'");
/// ```
///
pub fn escape<T: Into<OsString>>(s: T) -> Vec<u8> {
    let sin = s.into().into_vec();
    if let Some(esc) = escape_prepare(&sin) {
        // Maybe pointless optimisation, but here we calculate the memory we need to
        // avoid reallocations as we construct the output string. Since we now know
        // we're going to use single quotes, we also add 2 bytes.
        let size: usize = esc.iter().map(escape_size).sum();
        let mut sout = Vec::with_capacity(size + 2);
        escape_chars(esc, &mut sout); // Do the work.
        sout
    } else {
        sin
    }
}

/// Escape a string of *bytes* into a new `OsString`.
///
/// Same as [`escape`], but returns an `OsString`.
///
pub fn quote<T: Into<OsString>>(s: T) -> OsString {
    OsString::from_vec(escape(s))
}

/// Escape a string of *bytes* into an existing `Vec<u8>`.
///
/// See [`escape`] for more details.
///
/// # Examples
///
/// ```
/// # use shell_quote::sh;
/// let mut buf = Vec::with_capacity(128);
/// sh::escape_into("foobar", &mut buf);
/// buf.push(b' ');  // Add a space.
/// sh::escape_into("foo bar", &mut buf);
/// assert_eq!(buf, b"foobar 'foo bar'");
/// ```
///
pub fn escape_into<T: Into<OsString>>(s: T, sout: &mut Vec<u8>) {
    let sin = s.into().into_vec();
    if let Some(esc) = escape_prepare(&sin) {
        // Maybe pointless optimisation, but here we calculate the memory we
        // need to avoid reallocations as we construct the output string. Since
        // we now know we're going to use single quotes, we also add 2 bytes.
        let size: usize = esc.iter().map(escape_size).sum();
        sout.reserve(size + 2);
        escape_chars(esc, sout); // Do the work.
    } else {
        sout.extend(sin);
    }
}

fn escape_prepare(sin: &[u8]) -> Option<Vec<Char>> {
    let esc: Vec<_> = sin.iter().map(Char::from).collect();
    // An optimisation: if the string only contains "safe" characters we can
    // avoid further work.
    if esc.iter().all(Char::is_inert) {
        None
    } else {
        Some(esc)
    }
}

fn escape_chars(esc: Vec<Char>, sout: &mut Vec<u8>) {
    // Push a Bourne-style '...' escaped string into `sout`.
    sout.extend(b"'");
    for mode in esc {
        use Char::*;
        match mode {
            Bell => sout.extend(b"\\a"),
            Backspace => sout.extend(b"\\b"),
            Escape => sout.extend(b"\\033"),
            FormFeed => sout.extend(b"\\f"),
            NewLine => sout.extend(b"\\n"),
            CarriageReturn => sout.extend(b"\\r"),
            HorizontalTab => sout.extend(b"\\t"),
            VerticalTab => sout.extend(b"\\v"),
            Control(ch) => sout.extend(format!("\\{:03o}", ch).bytes()),
            Backslash => sout.extend(b"\\\\"),
            SingleQuote => sout.extend(b"\\047"),
            DoubleQuote => sout.extend(b"\""),
            Delete => sout.push(0x7F),
            PrintableInert(ch) => sout.push(ch),
            Printable(ch) => sout.push(ch),
            Extended(ch) => sout.push(ch),
        }
    }
    sout.push(b'\'');
}

fn escape_size(char: &Char) -> usize {
    use Char::*;
    match char {
        Bell => 2,
        Backspace => 2,
        Escape => 4,
        FormFeed => 2,
        NewLine => 2,
        CarriageReturn => 2,
        HorizontalTab => 2,
        VerticalTab => 2,
        Control(_) => 4,
        Backslash => 2,
        SingleQuote => 4,
        DoubleQuote => 1,
        Delete => 4,
        PrintableInert(_) => 1,
        Printable(_) => 1,
        Extended(_) => 4,
    }
}

#[cfg(test)]
mod tests {
    use std::ffi::OsString;
    use std::os::unix::prelude::OsStringExt;

    use crate::find_bins;

    use super::{escape, escape_into, quote};

    #[test]
    fn test_lowercase_ascii() {
        assert_eq!(
            escape("abcdefghijklmnopqrstuvwxyz"),
            b"abcdefghijklmnopqrstuvwxyz"
        );
    }

    #[test]
    fn test_uppercase_ascii() {
        assert_eq!(
            escape("ABCDEFGHIJKLMNOPQRSTUVWXYZ"),
            b"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
        );
    }

    #[test]
    fn test_numbers() {
        assert_eq!(escape("0123456789"), b"0123456789");
    }

    #[test]
    fn test_punctuation() {
        assert_eq!(escape("-_=/,.+"), b"'-_=/,.+'");
    }

    #[test]
    fn test_basic_escapes() {
        assert_eq!(escape(r#"woo'wah""#), br#"'woo\047wah"'"#);
    }

    #[test]
    #[allow(non_snake_case)]
    fn test_control_characters() {
        assert_eq!(escape(&"\x07"), b"'\\a'");
        assert_eq!(escape(&"\x00"), b"'\\000'");
        assert_eq!(escape(&"\x06"), b"'\\006'");
        assert_eq!(escape(&"\x7F"), b"'\x7F'");
        assert_eq!(escape(&"\x1B"), b"'\\033'");
    }

    #[test]
    fn test_escape_into_plain() {
        let mut buffer = Vec::new();
        escape_into("hello", &mut buffer);
        assert_eq!(buffer, b"hello");
    }

    #[test]
    fn test_escape_into_with_escapes() {
        let mut buffer = Vec::new();
        escape_into("-_=/,.+", &mut buffer);
        assert_eq!(buffer, b"'-_=/,.+'");
    }

    #[test]
    fn test_roundtrip() {
        use std::process::Command;
        // In Bash it doesn't seem possible to roundtrip NUL, but in the Bourne
        // shell, or whatever is masquerading as `sh`, it seems to be fine.
        let string: OsString = OsString::from_vec((u8::MIN..=u8::MAX).collect());
        let mut script = b"echo ".to_vec();
        escape_into(&string, &mut script);
        let script = OsString::from_vec(script);
        for bin in find_bins("sh") {
            let output = Command::new(bin).arg("-c").arg(&script).output().unwrap();
            let mut result = output.stdout;
            result.resize(result.len() - 1, 0); // Remove trailing newline.
            let result = OsString::from_vec(result);
            assert_eq!(result, string);
        }
    }

    #[test]
    fn test_quote() {
        assert_eq!(
            quote("abcdefghijklmnopqrstuvwxyz"),
            OsString::from("abcdefghijklmnopqrstuvwxyz"),
        );
    }
}