safe_chains/cst/
mod.rs

1pub(crate) mod check;
2mod display;
3pub(crate) mod eval;
4mod explain;
5mod parse;
6#[cfg(test)]
7mod proptests;
8
9#[derive(Debug, Clone, PartialEq, Eq)]
10pub struct Script(pub Vec<Stmt>);
11
12#[derive(Debug, Clone, PartialEq, Eq)]
13pub struct Stmt {
14    pub pipeline: Pipeline,
15    pub op: Option<ListOp>,
16}
17
18#[derive(Debug, Clone, Copy, PartialEq, Eq)]
19pub enum ListOp {
20    And,
21    Or,
22    Semi,
23    Amp,
24}
25
26#[derive(Debug, Clone, PartialEq, Eq)]
27pub struct Pipeline {
28    pub bang: bool,
29    pub commands: Vec<Cmd>,
30}
31
32#[derive(Debug, Clone, PartialEq, Eq)]
33pub enum Cmd {
34    Simple(SimpleCmd),
35    Subshell {
36        body: Script,
37        redirs: Vec<Redir>,
38    },
39    BraceGroup {
40        body: Script,
41        redirs: Vec<Redir>,
42    },
43    For {
44        var: String,
45        items: Vec<Word>,
46        body: Script,
47        redirs: Vec<Redir>,
48    },
49    While {
50        cond: Script,
51        body: Script,
52        redirs: Vec<Redir>,
53    },
54    Until {
55        cond: Script,
56        body: Script,
57        redirs: Vec<Redir>,
58    },
59    If {
60        branches: Vec<Branch>,
61        else_body: Option<Script>,
62        redirs: Vec<Redir>,
63    },
64    DoubleBracket {
65        words: Vec<Word>,
66        redirs: Vec<Redir>,
67    },
68    /// `case WORD in PATTERN) BODY ;; … esac` (POSIX 2.9.4.3). Which arm runs depends on a value
69    /// resolved at runtime, so — like [`Cmd::If`] — every arm body is classified and the command
70    /// is only as safe as its worst arm.
71    Case {
72        subject: Word,
73        arms: Vec<CaseArm>,
74        redirs: Vec<Redir>,
75    },
76    /// `name() { body }` (or `function name { body }`). Defining a function has NO effect — it is
77    /// classified Inert. The body's safety matters only when the function is CALLED (resolved in
78    /// `check`), so it is stored, not flattened.
79    FunctionDef {
80        name: String,
81        body: Script,
82    },
83}
84
85#[derive(Debug, Clone, PartialEq, Eq)]
86pub struct Branch {
87    pub cond: Script,
88    pub body: Script,
89}
90
91/// One `PATTERN|PATTERN) BODY ;;` arm of a [`Cmd::Case`]. The patterns are glob words matched
92/// against the subject; they are never executed, so only `body` carries risk.
93#[derive(Debug, Clone, PartialEq, Eq)]
94pub struct CaseArm {
95    pub patterns: Vec<Word>,
96    pub body: Script,
97}
98
99#[derive(Debug, Clone, PartialEq, Eq)]
100pub struct SimpleCmd {
101    pub env: Vec<(String, Word)>,
102    pub words: Vec<Word>,
103    pub redirs: Vec<Redir>,
104}
105
106#[derive(Debug, Clone, PartialEq, Eq)]
107pub struct Word(pub Vec<WordPart>);
108
109#[derive(Debug, Clone, PartialEq, Eq)]
110pub enum WordPart {
111    Lit(String),
112    Escape(char),
113    SQuote(String),
114    DQuote(Word),
115    CmdSub(Script),
116    ProcSub(Script),
117    Backtick(String),
118    Arith(String),
119}
120
121/// How an output redirect opens its target. All three land the same bytes somewhere, so they
122/// classify identically — the distinction is kept so `--explain` can echo the command the user
123/// actually typed rather than a normalized one. Mutually exclusive by construction: `>>|` is not
124/// a redirect, and a bool pair would let a generator build one.
125#[derive(Debug, Clone, Copy, PartialEq, Eq)]
126pub enum WriteMode {
127    /// `>` — truncate.
128    Truncate,
129    /// `>>` — append.
130    Append,
131    /// `&>` (and the equivalent `>&FILE`) — stdout AND stderr to the file, truncating. Both
132    /// streams land on ONE target, so the locus gate has exactly one path to judge, the same as
133    /// `>`; the variant exists so `--explain` echoes the operator that was typed.
134    TruncateBoth,
135    /// `&>>` — stdout AND stderr to the file, appending.
136    AppendBoth,
137    /// `>|` — truncate, overriding `noclobber` (POSIX 2.7.2).
138    Clobber,
139}
140
141#[derive(Debug, Clone, PartialEq, Eq)]
142pub enum Redir {
143    Write {
144        fd: u32,
145        target: Word,
146        mode: WriteMode,
147    },
148    Read {
149        fd: u32,
150        target: Word,
151    },
152    /// `<>` — the target is opened for reading AND writing (POSIX 2.7.5), so it is gated on both
153    /// faces. Neither alone is sufficient: the write gate would miss the disclosure of reading a
154    /// secret, and the read gate would miss the overwrite.
155    ReadWrite {
156        fd: u32,
157        target: Word,
158    },
159    HereStr(Word),
160    HereDoc {
161        delimiter: String,
162        strip_tabs: bool,
163        /// The body's parsed EXPANSIONS. A heredoc body is data only when the delimiter is quoted
164        /// (`<<'EOF'`, `<<"EOF"`, `<<\EOF`, `<<E"O"F`); with a bare `<<EOF` the shell expands the
165        /// body exactly as it would a double-quoted string, so `$(…)` and backticks in it RUN.
166        /// Empty when the delimiter is quoted, so a quoted body stays pure data.
167        body: Word,
168    },
169    DupFd {
170        src: u32,
171        dst: String,
172    },
173}
174
175pub use check::{command_verdict, is_safe_command, is_safe_pipeline};
176pub use explain::{Explanation, SegmentReport, explain, explain_with_coverage};
177pub use parse::parse;
178
179impl Word {
180    pub fn eval(&self) -> String {
181        eval::eval_word(self)
182    }
183
184    /// The set of literal words this word produces under UNQUOTED brace expansion (`{a,b}` → two
185    /// words). Every produced word must be classified, so a braced alternative can't hide a system
186    /// path from the gate (`cat {/etc/shadow,x}`). Non-braced words expand to `[self.eval()]`.
187    pub fn expand(&self) -> Vec<String> {
188        eval::expand_word(self)
189    }
190
191    pub fn literal(s: &str) -> Self {
192        Word(vec![WordPart::Lit(s.to_string())])
193    }
194
195    pub fn normalize(&self) -> Self {
196        let mut parts = Vec::new();
197        for part in &self.0 {
198            let part = match part {
199                WordPart::DQuote(inner) => WordPart::DQuote(inner.normalize()),
200                WordPart::CmdSub(s) => WordPart::CmdSub(s.normalize()),
201                WordPart::ProcSub(s) => WordPart::ProcSub(s.normalize()),
202                other => other.clone(),
203            };
204            if let WordPart::Lit(s) = &part
205                && let Some(WordPart::Lit(prev)) = parts.last_mut()
206            {
207                prev.push_str(s);
208                continue;
209            }
210            parts.push(part);
211        }
212        Word(parts)
213    }
214}
215
216impl Script {
217    pub fn is_empty(&self) -> bool {
218        self.0.is_empty()
219    }
220
221    pub fn normalize(&self) -> Self {
222        Script(
223            self.0
224                .iter()
225                .map(|stmt| Stmt {
226                    pipeline: stmt.pipeline.normalize(),
227                    op: stmt.op,
228                })
229                .collect(),
230        )
231    }
232
233    pub fn normalize_as_body(&self) -> Self {
234        let mut s = self.normalize();
235        if let Some(last) = s.0.last_mut()
236            && last.op.is_none()
237        {
238            last.op = Some(ListOp::Semi);
239        }
240        s
241    }
242}
243
244impl Pipeline {
245    fn normalize(&self) -> Self {
246        Pipeline {
247            bang: self.bang,
248            commands: self.commands.iter().map(|c| c.normalize()).collect(),
249        }
250    }
251}
252
253impl Cmd {
254    fn normalize(&self) -> Self {
255        match self {
256            Cmd::Simple(s) => Cmd::Simple(s.normalize()),
257            Cmd::Subshell { body, redirs } => Cmd::Subshell {
258                body: body.normalize(),
259                redirs: normalize_redirs(redirs),
260            },
261            Cmd::BraceGroup { body, redirs } => Cmd::BraceGroup {
262                body: body.normalize_as_body(),
263                redirs: normalize_redirs(redirs),
264            },
265            Cmd::For { var, items, body, redirs } => Cmd::For {
266                var: var.clone(),
267                items: items.iter().map(|w| w.normalize()).collect(),
268                body: body.normalize_as_body(),
269                redirs: normalize_redirs(redirs),
270            },
271            Cmd::While { cond, body, redirs } => Cmd::While {
272                cond: cond.normalize_as_body(),
273                body: body.normalize_as_body(),
274                redirs: normalize_redirs(redirs),
275            },
276            Cmd::Until { cond, body, redirs } => Cmd::Until {
277                cond: cond.normalize_as_body(),
278                body: body.normalize_as_body(),
279                redirs: normalize_redirs(redirs),
280            },
281            Cmd::If { branches, else_body, redirs } => Cmd::If {
282                branches: branches
283                    .iter()
284                    .map(|b| Branch {
285                        cond: b.cond.normalize_as_body(),
286                        body: b.body.normalize_as_body(),
287                    })
288                    .collect(),
289                else_body: else_body.as_ref().map(|e| e.normalize_as_body()),
290                redirs: normalize_redirs(redirs),
291            },
292            Cmd::DoubleBracket { words, redirs } => Cmd::DoubleBracket {
293                words: words.iter().map(|w| w.normalize()).collect(),
294                redirs: normalize_redirs(redirs),
295            },
296            Cmd::Case { subject, arms, redirs } => Cmd::Case {
297                subject: subject.normalize(),
298                arms: arms
299                    .iter()
300                    .map(|a| CaseArm {
301                        patterns: a.patterns.iter().map(|w| w.normalize()).collect(),
302                        body: a.body.normalize_as_body(),
303                    })
304                    .collect(),
305                redirs: normalize_redirs(redirs),
306            },
307            Cmd::FunctionDef { name, body } => Cmd::FunctionDef {
308                name: name.clone(),
309                body: body.normalize_as_body(),
310            },
311        }
312    }
313}
314
315impl SimpleCmd {
316    fn normalize(&self) -> Self {
317        SimpleCmd {
318            env: self
319                .env
320                .iter()
321                .map(|(k, v)| (k.clone(), v.normalize()))
322                .collect(),
323            words: self.words.iter().map(|w| w.normalize()).collect(),
324            redirs: normalize_redirs(&self.redirs),
325        }
326    }
327}
328
329fn normalize_redirs(redirs: &[Redir]) -> Vec<Redir> {
330    redirs
331        .iter()
332        .map(|r| match r {
333            Redir::Write { fd, target, mode } => Redir::Write {
334                fd: *fd,
335                target: target.normalize(),
336                mode: *mode,
337            },
338            Redir::Read { fd, target } => Redir::Read {
339                fd: *fd,
340                target: target.normalize(),
341            },
342            Redir::ReadWrite { fd, target } => Redir::ReadWrite {
343                fd: *fd,
344                target: target.normalize(),
345            },
346            Redir::HereStr(w) => Redir::HereStr(w.normalize()),
347            Redir::HereDoc { .. } | Redir::DupFd { .. } => r.clone(),
348        })
349        .collect()
350}
safe_chains/cst/mod.rs

safe_chains/cst/
mod.rs
