safe_chains/cst/
parse.rs

1use super::*;
2use winnow::ModalResult;
3use winnow::combinator::{alt, delimited, not, opt, preceded, repeat, separated, terminated};
4use winnow::error::{ContextError, ErrMode};
5use winnow::prelude::*;
6use winnow::token::{any, take_while};
7
8pub fn parse(input: &str) -> Option<Script> {
9    script.parse(input).ok()
10}
11
12fn backtrack<T>() -> ModalResult<T> {
13    Err(ErrMode::Backtrack(ContextError::new()))
14}
15
16fn comment(input: &mut &str) -> ModalResult<()> {
17    if input.starts_with('#') {
18        if let Some(pos) = input.find('\n') {
19            *input = &input[pos + 1..];
20        } else {
21            *input = "";
22        }
23    }
24    Ok(())
25}
26
27fn ws(input: &mut &str) -> ModalResult<()> {
28    loop {
29        take_while(0.., [' ', '\t']).void().parse_next(input)?;
30        if input.starts_with('#') {
31            comment(input)?;
32        } else {
33            break;
34        }
35    }
36    Ok(())
37}
38
39fn sep(input: &mut &str) -> ModalResult<()> {
40    loop {
41        take_while(0.., [' ', '\t', ';', '\n']).void().parse_next(input)?;
42        if input.starts_with('#') {
43            comment(input)?;
44        } else {
45            break;
46        }
47    }
48    Ok(())
49}
50
51fn eat_keyword(input: &mut &str, kw: &str) -> ModalResult<()> {
52    if !input.starts_with(kw) {
53        return backtrack();
54    }
55    if input
56        .as_bytes()
57        .get(kw.len())
58        .is_some_and(|&b| b.is_ascii_alphanumeric() || b == b'_')
59    {
60        return backtrack();
61    }
62    *input = &input[kw.len()..];
63    Ok(())
64}
65
66const SCRIPT_STOPS: &[&str] = &["do", "done", "elif", "else", "fi", "then"];
67
68fn at_script_stop(input: &str) -> bool {
69    input.starts_with(')')
70        || SCRIPT_STOPS.iter().any(|kw| {
71            input.starts_with(kw)
72                && !input
73                    .as_bytes()
74                    .get(kw.len())
75                    .is_some_and(|&b| b.is_ascii_alphanumeric() || b == b'_')
76        })
77}
78
79fn is_word_boundary(c: char) -> bool {
80    matches!(c, ' ' | '\t' | '\n' | ';' | '|' | '&' | ')' | '>' | '<')
81}
82
83fn is_word_literal(c: char) -> bool {
84    !is_word_boundary(c) && !matches!(c, '\'' | '"' | '`' | '\\' | '(' | '$')
85}
86
87fn is_dq_literal(c: char) -> bool {
88    !matches!(c, '"' | '\\' | '`' | '$')
89}
90
91// === Script ===
92
93fn script(input: &mut &str) -> ModalResult<Script> {
94    sep.parse_next(input)?;
95    let mut stmts = Vec::new();
96    while let Some(pl) = opt(pipeline).parse_next(input)? {
97        ws.parse_next(input)?;
98        let op = opt(list_op).parse_next(input)?;
99        stmts.push(Stmt { pipeline: pl, op });
100        if op.is_none() {
101            break;
102        }
103    }
104    Ok(Script(stmts))
105}
106
107fn list_op(input: &mut &str) -> ModalResult<ListOp> {
108    ws.parse_next(input)?;
109    alt((
110        "&&".value(ListOp::And),
111        "||".value(ListOp::Or),
112        '\n'.value(ListOp::Semi),
113        ';'.value(ListOp::Semi),
114        ('&', not('>')).value(ListOp::Amp),
115    ))
116    .parse_next(input)
117}
118
119fn pipe_sep(input: &mut &str) -> ModalResult<()> {
120    (ws, '|', not('|'), ws).void().parse_next(input)
121}
122
123// === Pipeline ===
124
125fn pipeline(input: &mut &str) -> ModalResult<Pipeline> {
126    ws.parse_next(input)?;
127    if at_script_stop(input) {
128        return backtrack();
129    }
130    let bang = opt(terminated('!', ws)).parse_next(input)?.is_some();
131    let commands: Vec<Cmd> = separated(1.., command, pipe_sep).parse_next(input)?;
132    Ok(Pipeline { bang, commands })
133}
134
135// === Command ===
136
137fn command(input: &mut &str) -> ModalResult<Cmd> {
138    ws.parse_next(input)?;
139    if at_script_stop(input) {
140        return backtrack();
141    }
142    alt((
143        subshell,
144        for_cmd,
145        while_cmd,
146        until_cmd,
147        if_cmd,
148        simple_cmd.map(Cmd::Simple),
149    ))
150    .parse_next(input)
151}
152
153fn subshell(input: &mut &str) -> ModalResult<Cmd> {
154    delimited(('(', ws), script, (ws, ')'))
155        .map(Cmd::Subshell)
156        .parse_next(input)
157}
158
159// === Simple Command ===
160
161fn simple_cmd(input: &mut &str) -> ModalResult<SimpleCmd> {
162    let env: Vec<(String, Word)> =
163        repeat(0.., terminated(assignment, ws)).parse_next(input)?;
164    let mut words = Vec::new();
165    let mut redirs = Vec::new();
166
167    loop {
168        ws.parse_next(input)?;
169        if at_cmd_end(input) {
170            break;
171        }
172        if let Some(r) = opt(redirect).parse_next(input)? {
173            redirs.push(r);
174        } else if let Some(w) = opt(word).parse_next(input)? {
175            words.push(w);
176        } else {
177            break;
178        }
179    }
180
181    if env.is_empty() && words.is_empty() && redirs.is_empty() {
182        return backtrack();
183    }
184    Ok(SimpleCmd { env, words, redirs })
185}
186
187fn at_cmd_end(input: &str) -> bool {
188    input.is_empty()
189        || matches!(
190            input.as_bytes().first(),
191            Some(b'\n' | b';' | b'|' | b'&' | b')')
192        )
193}
194
195fn assignment(input: &mut &str) -> ModalResult<(String, Word)> {
196    let n: &str = take_while(1.., |c: char| c.is_ascii_alphanumeric() || c == '_')
197        .parse_next(input)?;
198    '='.parse_next(input)?;
199    let value = opt(word)
200        .parse_next(input)?
201        .unwrap_or(Word(vec![WordPart::Lit(String::new())]));
202    Ok((n.to_string(), value))
203}
204
205// === Redirect ===
206
207fn redirect(input: &mut &str) -> ModalResult<Redir> {
208    let fd = opt(fd_prefix).parse_next(input)?;
209    alt((
210        preceded("<<<", (ws, word)).map(|(_, target)| Redir::HereStr(target)),
211        heredoc,
212        preceded(">>", (ws, word)).map(move |(_, target)| Redir::Write {
213            fd: fd.unwrap_or(1),
214            target,
215            append: true,
216        }),
217        preceded(">&", fd_target).map(move |dst| Redir::DupFd {
218            src: fd.unwrap_or(1),
219            dst,
220        }),
221        preceded('>', (ws, word)).map(move |(_, target)| Redir::Write {
222            fd: fd.unwrap_or(1),
223            target,
224            append: false,
225        }),
226        preceded('<', (ws, word)).map(move |(_, target)| Redir::Read {
227            fd: fd.unwrap_or(0),
228            target,
229        }),
230    ))
231    .parse_next(input)
232}
233
234fn heredoc(input: &mut &str) -> ModalResult<Redir> {
235    "<<".parse_next(input)?;
236    let strip_tabs = opt('-').parse_next(input)?.is_some();
237    ws.parse_next(input)?;
238    let delimiter = heredoc_delimiter.parse_next(input)?;
239    let needle = format!("\n{delimiter}");
240    if let Some(pos) = input.find(&needle) {
241        let after = pos + needle.len();
242        *input = input[after..].trim_start_matches([' ', '\t', '\n']);
243    }
244    Ok(Redir::HereDoc { delimiter, strip_tabs })
245}
246
247fn heredoc_delimiter(input: &mut &str) -> ModalResult<String> {
248    alt((
249        delimited('\'', take_while(0.., |c| c != '\''), '\'').map(|s: &str| s.to_string()),
250        delimited('"', take_while(0.., |c| c != '"'), '"').map(|s: &str| s.to_string()),
251        take_while(1.., |c: char| c.is_ascii_alphanumeric() || c == '_').map(|s: &str| s.to_string()),
252    ))
253    .parse_next(input)
254}
255
256fn fd_prefix(input: &mut &str) -> ModalResult<u32> {
257    let b = input.as_bytes();
258    if b.len() >= 2 && b[0].is_ascii_digit() && matches!(b[1], b'>' | b'<') {
259        let d = (b[0] - b'0') as u32;
260        *input = &input[1..];
261        Ok(d)
262    } else {
263        backtrack()
264    }
265}
266
267fn fd_target(input: &mut &str) -> ModalResult<String> {
268    alt((
269        '-'.value("-".to_string()),
270        take_while(1.., |c: char| c.is_ascii_digit()).map(|s: &str| s.to_string()),
271    ))
272    .parse_next(input)
273}
274
275// === Word ===
276
277fn word(input: &mut &str) -> ModalResult<Word> {
278    repeat(1.., word_part)
279        .map(Word)
280        .parse_next(input)
281}
282
283fn word_part(input: &mut &str) -> ModalResult<WordPart> {
284    if input.is_empty() {
285        return backtrack();
286    }
287    if input.starts_with("<(") || input.starts_with(">(") {
288        return proc_sub(input);
289    }
290    if is_word_boundary(input.as_bytes()[0] as char) {
291        return backtrack();
292    }
293    alt((single_quoted, double_quoted, arith_sub, cmd_sub, backtick_part, escaped, dollar_lit(is_word_literal), lit(is_word_literal)))
294        .parse_next(input)
295}
296
297fn single_quoted(input: &mut &str) -> ModalResult<WordPart> {
298    delimited('\'', take_while(0.., |c| c != '\''), '\'')
299        .map(|s: &str| WordPart::SQuote(s.to_string()))
300        .parse_next(input)
301}
302
303fn double_quoted(input: &mut &str) -> ModalResult<WordPart> {
304    delimited('"', repeat(0.., dq_part).map(Word), '"')
305        .map(WordPart::DQuote)
306        .parse_next(input)
307}
308
309fn cmd_sub(input: &mut &str) -> ModalResult<WordPart> {
310    delimited(("$(", ws), script, (ws, ')'))
311        .map(WordPart::CmdSub)
312        .parse_next(input)
313}
314
315fn proc_sub(input: &mut &str) -> ModalResult<WordPart> {
316    if !(input.starts_with("<(") || input.starts_with(">(")) {
317        return backtrack();
318    }
319    *input = &input[1..];
320    delimited(('(', ws), script, (ws, ')'))
321        .map(WordPart::ProcSub)
322        .parse_next(input)
323}
324
325fn arith_sub(input: &mut &str) -> ModalResult<WordPart> {
326    if !input.starts_with("$((") {
327        return backtrack();
328    }
329    let body_start = 3;
330    let bytes = input.as_bytes();
331    let mut depth: i32 = 1;
332    let mut i = body_start;
333    while i < bytes.len() {
334        match bytes[i] {
335            b'(' => depth += 1,
336            b')' => {
337                if depth == 1 && i + 1 < bytes.len() && bytes[i + 1] == b')' {
338                    let body = input[body_start..i].to_string();
339                    if body.contains("$(") || body.contains('`') {
340                        return backtrack();
341                    }
342                    *input = &input[i + 2..];
343                    return Ok(WordPart::Arith(body));
344                }
345                depth -= 1;
346                if depth < 0 {
347                    return backtrack();
348                }
349            }
350            _ => {}
351        }
352        i += 1;
353    }
354    backtrack()
355}
356
357fn backtick_part(input: &mut &str) -> ModalResult<WordPart> {
358    delimited('`', backtick_inner, '`')
359        .map(WordPart::Backtick)
360        .parse_next(input)
361}
362
363fn escaped(input: &mut &str) -> ModalResult<WordPart> {
364    preceded('\\', any).map(WordPart::Escape).parse_next(input)
365}
366
367fn lit(pred: fn(char) -> bool) -> impl FnMut(&mut &str) -> ModalResult<WordPart> {
368    move |input: &mut &str| {
369        take_while(1.., pred)
370            .map(|s: &str| WordPart::Lit(s.to_string()))
371            .parse_next(input)
372    }
373}
374
375fn dollar_lit(pred: fn(char) -> bool) -> impl FnMut(&mut &str) -> ModalResult<WordPart> {
376    move |input: &mut &str| {
377        ('$', not('(')).void().parse_next(input)?;
378        let rest: &str = take_while(0.., pred).parse_next(input)?;
379        Ok(WordPart::Lit(format!("${rest}")))
380    }
381}
382
383// === Double-quoted parts ===
384
385fn dq_part(input: &mut &str) -> ModalResult<WordPart> {
386    if input.is_empty() || input.starts_with('"') {
387        return backtrack();
388    }
389    alt((dq_escape, arith_sub, cmd_sub, backtick_part, dollar_lit(is_dq_literal), lit(is_dq_literal)))
390        .parse_next(input)
391}
392
393fn dq_escape(input: &mut &str) -> ModalResult<WordPart> {
394    preceded('\\', any)
395        .map(|c: char| match c {
396            '"' | '\\' | '$' | '`' => WordPart::Escape(c),
397            _ => WordPart::Lit(format!("\\{c}")),
398        })
399        .parse_next(input)
400}
401
402// === Backtick inner content ===
403
404fn backtick_inner(input: &mut &str) -> ModalResult<String> {
405    repeat(0.., alt((bt_escape, bt_literal)))
406        .fold(String::new, |mut acc, chunk: &str| {
407            acc.push_str(chunk);
408            acc
409        })
410        .parse_next(input)
411}
412
413fn bt_escape<'a>(input: &mut &'a str) -> ModalResult<&'a str> {
414    ('\\', any).take().parse_next(input)
415}
416
417fn bt_literal<'a>(input: &mut &'a str) -> ModalResult<&'a str> {
418    take_while(1.., |c: char| c != '`' && c != '\\').parse_next(input)
419}
420
421// === Compound Commands ===
422
423fn for_cmd(input: &mut &str) -> ModalResult<Cmd> {
424    eat_keyword(input, "for")?;
425    ws.parse_next(input)?;
426    let var = name.parse_next(input)?;
427    ws.parse_next(input)?;
428
429    let items = if eat_keyword(input, "in").is_ok() {
430        ws.parse_next(input)?;
431        repeat(0.., terminated(word, ws)).parse_next(input)?
432    } else {
433        vec![]
434    };
435
436    let body = do_done_body.parse_next(input)?;
437    Ok(Cmd::For { var, items, body })
438}
439
440fn while_cmd(input: &mut &str) -> ModalResult<Cmd> {
441    eat_keyword(input, "while")?;
442    ws.parse_next(input)?;
443    let cond = script.parse_next(input)?;
444    let body = do_done_body.parse_next(input)?;
445    Ok(Cmd::While { cond, body })
446}
447
448fn until_cmd(input: &mut &str) -> ModalResult<Cmd> {
449    eat_keyword(input, "until")?;
450    ws.parse_next(input)?;
451    let cond = script.parse_next(input)?;
452    let body = do_done_body.parse_next(input)?;
453    Ok(Cmd::Until { cond, body })
454}
455
456fn do_done_body(input: &mut &str) -> ModalResult<Script> {
457    sep.parse_next(input)?;
458    eat_keyword(input, "do")?;
459    sep.parse_next(input)?;
460    let body = script.parse_next(input)?;
461    sep.parse_next(input)?;
462    eat_keyword(input, "done")?;
463    Ok(body)
464}
465
466fn if_cmd(input: &mut &str) -> ModalResult<Cmd> {
467    eat_keyword(input, "if")?;
468    ws.parse_next(input)?;
469    let mut branches = vec![cond_then_body.parse_next(input)?];
470    let mut else_body = None;
471
472    loop {
473        sep.parse_next(input)?;
474        if eat_keyword(input, "elif").is_ok() {
475            ws.parse_next(input)?;
476            branches.push(cond_then_body.parse_next(input)?);
477        } else if eat_keyword(input, "else").is_ok() {
478            sep.parse_next(input)?;
479            else_body = Some(script.parse_next(input)?);
480            break;
481        } else {
482            break;
483        }
484    }
485
486    sep.parse_next(input)?;
487    eat_keyword(input, "fi")?;
488    Ok(Cmd::If { branches, else_body })
489}
490
491fn cond_then_body(input: &mut &str) -> ModalResult<Branch> {
492    let cond = script.parse_next(input)?;
493    sep.parse_next(input)?;
494    eat_keyword(input, "then")?;
495    sep.parse_next(input)?;
496    let body = script.parse_next(input)?;
497    Ok(Branch { cond, body })
498}
499
500fn name(input: &mut &str) -> ModalResult<String> {
501    take_while(1.., |c: char| c.is_ascii_alphanumeric() || c == '_')
502        .map(|s: &str| s.to_string())
503        .parse_next(input)
504}
505
506#[cfg(test)]
507mod tests {
508    use super::*;
509
510    fn p(input: &str) -> Script {
511        parse(input).unwrap_or_else(|| panic!("failed to parse: {input}"))
512    }
513
514    fn words(script: &Script) -> Vec<String> {
515        match &script.0[0].pipeline.commands[0] {
516            Cmd::Simple(s) => s.words.iter().map(|w| w.eval()).collect(),
517            _ => panic!("expected simple command"),
518        }
519    }
520
521    fn simple(script: &Script) -> &SimpleCmd {
522        match &script.0[0].pipeline.commands[0] {
523            Cmd::Simple(s) => s,
524            _ => panic!("expected simple command"),
525        }
526    }
527
528    #[test]
529    fn simple_command() { assert_eq!(words(&p("echo hello")), ["echo", "hello"]); }
530    #[test]
531    fn flags() { assert_eq!(words(&p("ls -la")), ["ls", "-la"]); }
532    #[test]
533    fn single_quoted() { assert_eq!(words(&p("echo 'hello world'")), ["echo", "hello world"]); }
534    #[test]
535    fn double_quoted() { assert_eq!(words(&p("echo \"hello world\"")), ["echo", "hello world"]); }
536    #[test]
537    fn mixed_quotes() { assert_eq!(words(&p("jq '.key' file.json")), ["jq", ".key", "file.json"]); }
538
539    #[test]
540    fn pipeline_test() { assert_eq!(p("grep foo | head -5").0[0].pipeline.commands.len(), 2); }
541    #[test]
542    fn sequence_and() { assert_eq!(p("ls && echo done").0[0].op, Some(ListOp::And)); }
543    #[test]
544    fn sequence_semi() { assert_eq!(p("ls; echo done").0.len(), 2); }
545    #[test]
546    fn newline_separator() { assert_eq!(p("echo foo\necho bar").0.len(), 2); }
547    #[test]
548    fn background() { assert_eq!(p("ls & echo done").0[0].op, Some(ListOp::Amp)); }
549
550    #[test]
551    fn redirect_dev_null() {
552        let s = p("echo hello > /dev/null");
553        let cmd = simple(&s);
554        assert_eq!(cmd.words.len(), 2);
555        assert!(matches!(&cmd.redirs[0], Redir::Write { fd: 1, append: false, .. }));
556    }
557    #[test]
558    fn redirect_stderr() {
559        assert!(matches!(&simple(&p("echo hello 2>&1")).redirs[0], Redir::DupFd { src: 2, dst } if dst == "1"));
560    }
561    #[test]
562    fn here_string() {
563        assert!(matches!(&simple(&p("grep -c , <<< 'hello,world,test'")).redirs[0], Redir::HereStr(_)));
564    }
565    #[test]
566    fn heredoc_bare() {
567        assert!(matches!(&simple(&p("cat <<EOF")).redirs[0], Redir::HereDoc { delimiter, strip_tabs: false } if delimiter == "EOF"));
568    }
569    #[test]
570    fn heredoc_with_content() {
571        let s = p("cat <<EOF\nhello world\nEOF");
572        assert!(matches!(&simple(&s).redirs[0], Redir::HereDoc { delimiter, .. } if delimiter == "EOF"));
573    }
574    #[test]
575    fn heredoc_quoted_delimiter() {
576        assert!(matches!(&simple(&p("cat <<'EOF'")).redirs[0], Redir::HereDoc { delimiter, .. } if delimiter == "EOF"));
577    }
578    #[test]
579    fn heredoc_strip_tabs() {
580        assert!(matches!(&simple(&p("cat <<-EOF")).redirs[0], Redir::HereDoc { strip_tabs: true, .. }));
581    }
582    #[test]
583    fn heredoc_then_pipe() {
584        let s = p("cat <<EOF\nhello\nEOF | grep hello");
585        assert_eq!(s.0[0].pipeline.commands.len(), 2);
586    }
587    #[test]
588    fn heredoc_then_pipe_next_line() {
589        let s = p("cat <<EOF\nhello\nEOF\n| grep hello");
590        assert_eq!(s.0[0].pipeline.commands.len(), 2);
591    }
592
593    #[test]
594    fn env_prefix() {
595        let s = p("FOO='bar baz' ls -la");
596        let cmd = simple(&s);
597        assert_eq!(cmd.env[0].0, "FOO");
598        assert_eq!(cmd.env[0].1.eval(), "bar baz");
599    }
600    #[test]
601    fn cmd_substitution() { assert!(matches!(&simple(&p("echo $(ls)")).words[1].0[0], WordPart::CmdSub(_))); }
602    #[test]
603    fn backtick_substitution() { assert_eq!(simple(&p("ls `pwd`")).words[1].eval(), "__SAFE_CHAINS_SUB__"); }
604    #[test]
605    fn nested_substitution() {
606        if let WordPart::CmdSub(inner) = &simple(&p("echo $(echo $(ls))")).words[1].0[0] {
607            assert!(matches!(&simple(inner).words[1].0[0], WordPart::CmdSub(_)));
608        } else { panic!("expected CmdSub"); }
609    }
610
611    #[test]
612    fn subshell_test() { assert!(matches!(&p("(echo hello)").0[0].pipeline.commands[0], Cmd::Subshell(_))); }
613    #[test]
614    fn negation() { assert!(p("! echo hello").0[0].pipeline.bang); }
615
616    #[test]
617    fn for_loop() { assert!(matches!(&p("for x in 1 2 3; do echo $x; done").0[0].pipeline.commands[0], Cmd::For { var, .. } if var == "x")); }
618    #[test]
619    fn while_loop() { assert!(matches!(&p("while test -f /tmp/foo; do sleep 1; done").0[0].pipeline.commands[0], Cmd::While { .. })); }
620    #[test]
621    fn if_then_fi() {
622        if let Cmd::If { branches, else_body } = &p("if test -f foo; then echo exists; fi").0[0].pipeline.commands[0] {
623            assert_eq!(branches.len(), 1);
624            assert!(else_body.is_none());
625        } else { panic!("expected If"); }
626    }
627    #[test]
628    fn if_elif_else() {
629        if let Cmd::If { branches, else_body } = &p("if test -f a; then echo a; elif test -f b; then echo b; else echo c; fi").0[0].pipeline.commands[0] {
630            assert_eq!(branches.len(), 2);
631            assert!(else_body.is_some());
632        } else { panic!("expected If"); }
633    }
634
635    #[test]
636    fn escaped_outside_quotes() { assert_eq!(words(&p("echo hello\\ world")), ["echo", "hello world"]); }
637    #[test]
638    fn double_quoted_escape() { assert_eq!(words(&p("echo \"hello\\\"world\"")), ["echo", "hello\"world"]); }
639    #[test]
640    fn assign_subst() { assert_eq!(simple(&p("out=$(ls)")).env[0].0, "out"); }
641
642    #[test]
643    fn unmatched_single_quote_fails() { assert!(parse("echo 'hello").is_none()); }
644    #[test]
645    fn unmatched_double_quote_fails() { assert!(parse("echo \"hello").is_none()); }
646    #[test]
647    fn unclosed_subshell_fails() { assert!(parse("(echo hello").is_none()); }
648    #[test]
649    fn unclosed_cmd_sub_fails() { assert!(parse("echo $(ls").is_none()); }
650    #[test]
651    fn for_missing_do_fails() { assert!(parse("for x in 1 2 3; echo $x; done").is_none()); }
652    #[test]
653    fn if_missing_fi_fails() { assert!(parse("if true; then echo hello").is_none()); }
654
655    #[test]
656    fn subshell_for() {
657        if let Cmd::Subshell(inner) = &p("(for x in 1 2; do echo $x; done)").0[0].pipeline.commands[0] {
658            assert!(matches!(&inner.0[0].pipeline.commands[0], Cmd::For { .. }));
659        } else { panic!("expected Subshell"); }
660    }
661    #[test]
662    fn proc_sub_input() {
663        let s = p("diff <(sort a.txt) <(sort b.txt)");
664        let cmd = simple(&s);
665        assert_eq!(cmd.words.len(), 3);
666        assert!(matches!(&cmd.words[1].0[0], WordPart::ProcSub(_)));
667        assert!(matches!(&cmd.words[2].0[0], WordPart::ProcSub(_)));
668    }
669    #[test]
670    fn proc_sub_output() {
671        let s = p("tee >(grep error > /dev/null)");
672        let cmd = simple(&s);
673        assert_eq!(cmd.words.len(), 2);
674        assert!(matches!(&cmd.words[1].0[0], WordPart::ProcSub(_)));
675    }
676    #[test]
677    fn comment_only() {
678        let s = p("# just a comment");
679        assert!(s.0.is_empty());
680    }
681    #[test]
682    fn comment_before_command() {
683        let s = p("# comment\necho hello");
684        assert_eq!(words(&s), ["echo", "hello"]);
685    }
686    #[test]
687    fn inline_comment() {
688        let s = p("echo hello # this is a comment");
689        assert_eq!(words(&s), ["echo", "hello"]);
690    }
691    #[test]
692    fn comment_between_commands() {
693        let s = p("echo hello\n# middle comment\necho world");
694        assert_eq!(s.0.len(), 2);
695    }
696    #[test]
697    fn comment_after_semicolon() {
698        let s = p("echo hello; # comment\necho world");
699        assert_eq!(s.0.len(), 2);
700    }
701    #[test]
702    fn comment_in_for_loop() {
703        assert!(parse("for x in 1 2; do\n# loop body\necho $x\ndone").is_some());
704    }
705    #[test]
706    fn quoted_redirect_in_echo() {
707        let s = p("echo 'greater > than' test");
708        let cmd = simple(&s);
709        assert_eq!(cmd.words.len(), 3);
710        assert_eq!(cmd.redirs.len(), 0);
711    }
712
713    #[test]
714    fn parses_all_safe_commands() {
715        let cmds = [
716            "grep foo file.txt", "cat /etc/hosts", "jq '.key' file.json", "base64 -d",
717            "ls -la", "wc -l file.txt", "ps aux", "echo hello", "cat file.txt",
718            "echo $(ls)", "ls `pwd`", "echo $(echo $(ls))", "echo \"$(ls)\"",
719            "out=$(ls)", "out=$(git status)", "a=$(ls) b=$(pwd)",
720            "(echo hello)", "(ls)", "(ls && echo done)", "(echo hello; echo world)",
721            "(ls | grep foo)", "(echo hello) | grep hello", "(ls) && echo done",
722            "((echo hello))", "(for x in 1 2; do echo $x; done)",
723            "echo 'greater > than' test", "echo '$(safe)' arg",
724            "FOO='bar baz' ls -la", "FOO=\"bar baz\" ls -la",
725            "RACK_ENV=test bundle exec rspec spec/foo_spec.rb",
726            "grep foo file.txt | head -5", "cat file | sort | uniq",
727            "ls && echo done", "ls; echo done", "ls & echo done",
728            "grep -c , <<< 'hello,world,test'",
729            "cat <<EOF\nhello world\nEOF",
730            "cat <<'MARKER'\nsome text\nMARKER",
731            "cat <<-EOF\n\thello\nEOF",
732            "echo foo\necho bar", "ls\ncat file.txt",
733            "git log --oneline -20 | head -5",
734            "echo hello > /dev/null", "echo hello 2> /dev/null",
735            "echo hello >> /dev/null", "git log > /dev/null 2>&1",
736            "ls 2>&1", "cargo clippy 2>&1", "git log < /dev/null",
737            "for x in 1 2 3; do echo $x; done",
738            "for f in *.txt; do cat $f | grep pattern; done",
739            "for x in 1 2 3; do; done",
740            "for x in 1 2; do echo $x; done; for y in a b; do echo $y; done",
741            "for x in 1 2; do for y in a b; do echo $x $y; done; done",
742            "for x in 1 2; do echo $x; done && echo finished",
743            "for x in $(seq 1 5); do echo $x; done",
744            "while test -f /tmp/foo; do sleep 1; done",
745            "while ! test -f /tmp/done; do sleep 1; done",
746            "until test -f /tmp/ready; do sleep 1; done",
747            "if test -f foo; then echo exists; fi",
748            "if test -f foo; then echo yes; else echo no; fi",
749            "if test -f a; then echo a; elif test -f b; then echo b; else echo c; fi",
750            "for x in 1 2; do if test $x = 1; then echo one; fi; done",
751            "if true; then for x in 1 2; do echo $x; done; fi",
752            "diff <(sort a.txt) <(sort b.txt)",
753            "comm -23 file.txt <(sort other.txt)",
754            "cat <(echo hello)",
755            "# comment only",
756            "# comment\necho hello",
757            "echo hello # inline comment",
758            "echo one\n# between\necho two",
759            "! echo hello", "! test -f foo",
760            "echo for; echo done; echo if; echo fi",
761        ];
762        let mut failures = Vec::new();
763        for cmd in &cmds {
764            if parse(cmd).is_none() { failures.push(*cmd); }
765        }
766        assert!(failures.is_empty(), "failed on {} commands:\n{}", failures.len(), failures.join("\n"));
767    }
768}
safe_chains/cst/parse.rs

safe_chains/cst/
parse.rs
