Skip to main content

openmoat_core/
policy.rs

1//! Policy file model, loading and linting (docs/POLICY.md).
2
3use std::collections::{BTreeMap, BTreeSet};
4
5/// Fallback verdicts when no rule matches: one verdict for everything, or a map
6/// from action kind (`shell`, `fs.read`, `fs.write`, `net`, `fetch`,
7/// `web.search`, `tool.unknown`, `env.read`, `env.set`, `mcp`, or `"*"`) to a verdict; a `fetch`
8/// without its own entry takes the `net` one. This is how "outbound network is
9/// deny-by-default" is expressed, because deny rules are absolute and cannot
10/// be punched through by allow rules (ADR-002).
11#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
12#[serde(untagged)]
13pub enum Defaults {
14    /// The same verdict for every kind.
15    All(Verdict),
16    /// A verdict per kind name (`net`, `fs.read`, …), with `*` as the fallback.
17    PerKind(BTreeMap<String, Verdict>),
18}
19
20impl Defaults {
21    /// Verdict and synthetic rule id (`default` or `default.<kind>`) for a kind.
22    #[must_use]
23    pub fn for_kind(&self, kind: Kind) -> (Verdict, String) {
24        match self {
25            Self::All(v) => (*v, "default".to_owned()),
26            Self::PerKind(map) => kind
27                .rule_kinds()
28                .iter()
29                .find_map(|k| map.get(k.as_str()).map(|v| (*v, format!("default.{k}"))))
30                .unwrap_or_else(|| {
31                    (
32                        map.get("*").copied().unwrap_or(Verdict::Review),
33                        "default".to_owned(),
34                    )
35                }),
36        }
37    }
38
39    /// The default for an atom no rule matched: [`Self::for_kind`], except that a
40    /// word that only looks like a host is no network evidence and gets
41    /// [`Self::unclassified`] (ADR-025).
42    #[must_use]
43    pub fn for_atom(&self, atom: &crate::AtomicAction) -> (Verdict, String) {
44        match atom {
45            crate::AtomicAction::NetWord { .. } => (self.unclassified(), "default".to_owned()),
46            _ => self.for_kind(atom.kind()),
47        }
48    }
49
50    /// What OpenMoat gives a call it cannot classify, or a word that only looks
51    /// like a host: `defer` under a catch-all (`"*"`) of `allow` or `defer`, else
52    /// `review` (ADR-025). Never `allow`, and never a deny nothing named.
53    #[must_use]
54    pub fn unclassified(&self) -> Verdict {
55        let catch_all = match self {
56            Self::All(v) => *v,
57            Self::PerKind(map) => map.get("*").copied().unwrap_or(Verdict::Review),
58        };
59        catch_all.clamp(Verdict::Defer, Verdict::Review)
60    }
61}
62
63use serde::{Deserialize, Serialize};
64use thiserror::Error;
65
66use crate::kind::Kind;
67use crate::pattern::{GlobPattern, ShellPattern};
68use crate::secret::{self, Secret};
69use crate::staging::{self, StagingSettings};
70use crate::taint::{self, TaintSettings};
71use crate::verdict::Verdict;
72
73mod env_name;
74mod hook_only;
75mod old_names;
76mod require;
77mod text;
78
79pub use env_name::env_name_problem;
80pub use require::{Require, RequiredBy, RequiredLevel, Requirement, RunBoundary};
81pub use text::{TITLE_MAX_CHARS, WHY_MAX_CHARS};
82
83pub use old_names::OldNames;
84
85/// The policy schema version (`version:`) this build reads; any other is an error.
86pub const SUPPORTED_VERSION: u32 = 1;
87
88/// Why a policy file cannot be used. Any of these keeps `guard` from deciding.
89///
90/// Messages include their cause, so no variant also exposes it as `source`:
91/// a chained report (`{:#}`) would print it twice.
92#[derive(Debug, Error)]
93#[non_exhaustive]
94pub enum PolicyError {
95    /// The file is not valid YAML or does not fit the schema.
96    #[error("policy parse error: {0}")]
97    Parse(crate::yaml::Error),
98    /// The `version` key names a schema this build does not know.
99    #[error("unsupported policy version {found}; this build supports {supported}")]
100    Version {
101        /// The version in the file.
102        found: u32,
103        /// The version this build supports.
104        supported: u32,
105    },
106    /// Two rule groups share an id.
107    #[error("duplicate rule id `{0}`")]
108    DuplicateId(String),
109    /// A rule group lists no patterns.
110    #[error("rule `{0}` has no patterns")]
111    EmptyRule(String),
112    /// A pattern is empty (or only `!`/`$`).
113    #[error("empty pattern")]
114    EmptyPattern,
115    /// A glob does not compile.
116    #[error("bad glob `{pattern}`: {error}")]
117    BadGlob {
118        /// The pattern as written.
119        pattern: String,
120        /// The glob compiler's explanation.
121        error: globset::Error,
122    },
123    /// A shell pattern does not tokenise (unbalanced quotes, a here-document).
124    #[error("bad shell pattern `{pattern}` (unbalanced quotes?)")]
125    BadShellPattern {
126        /// The pattern as written.
127        pattern: String,
128    },
129    /// The file has both `ask:` and `review:` (ADR-053): `ask` is the old name
130    /// of `review`, and which list was meant cannot be told.
131    #[error(
132        "the policy has both `{old}:` and `{new}:`; `{old}` is the old name of `{new}`: \
133         move the `{old}:` rules into `{new}:` (`openmoat policy upgrade` renames a policy \
134         that has only `{old}:`)"
135    )]
136    OldAndNewName {
137        /// The old key.
138        old: &'static str,
139        /// The current key.
140        new: &'static str,
141    },
142    /// A rule group or `executables` entry is invalid; `problem` says why.
143    #[error("rule `{rule}`: {problem}")]
144    Rule {
145        /// Rule id, or `executables.<name>`.
146        rule: String,
147        /// What is wrong.
148        problem: String,
149    },
150}
151
152/// One named group of patterns. A group matches when **any** of its patterns
153/// matches the atomic action of the corresponding kind.
154#[derive(Debug, Clone, Default, PartialEq, Eq, Serialize, Deserialize)]
155#[serde(deny_unknown_fields)]
156pub struct RuleGroup {
157    /// Stable identifier shown in decisions and the audit log.
158    pub id: String,
159    /// Text prepended to the reason when this group decides.
160    #[serde(default, skip_serializing_if = "Option::is_none")]
161    pub reason: Option<String>,
162    /// What the rule stops or holds, in the person's words ("reading your SSH
163    /// private key"): the first line of the message an agent and its person
164    /// see when this group denies or holds a call (ADR-059). One line.
165    #[serde(default, skip_serializing_if = "Option::is_none")]
166    pub title: Option<String>,
167    /// Why the rule exists, in one line, shown under [`Self::title`].
168    #[serde(default, skip_serializing_if = "Option::is_none")]
169    pub why: Option<String>,
170    /// Shell patterns (token prefixes, `*`, `!`, trailing `$`).
171    #[serde(default, skip_serializing_if = "Vec::is_empty")]
172    pub shell: Vec<String>,
173    /// Path globs for file reads.
174    #[serde(default, rename = "fs.read", skip_serializing_if = "Vec::is_empty")]
175    pub fs_read: Vec<String>,
176    /// Path globs for file writes.
177    #[serde(default, rename = "fs.write", skip_serializing_if = "Vec::is_empty")]
178    pub fs_write: Vec<String>,
179    /// Host globs for any network access, fetches included.
180    #[serde(default, skip_serializing_if = "Vec::is_empty")]
181    pub net: Vec<String>,
182    /// Host globs for read-only fetches by a host's fetch tool only.
183    #[serde(default, skip_serializing_if = "Vec::is_empty")]
184    pub fetch: Vec<String>,
185    /// Variable-name globs for reads.
186    #[serde(default, rename = "env.read", skip_serializing_if = "Vec::is_empty")]
187    pub env_read: Vec<String>,
188    /// Variable-name globs for sets.
189    #[serde(default, rename = "env.set", skip_serializing_if = "Vec::is_empty")]
190    pub env_set: Vec<String>,
191    /// MCP tool-name globs.
192    #[serde(default, skip_serializing_if = "Vec::is_empty")]
193    pub mcp: Vec<String>,
194    /// Applied by the hook only (ADR-034): OS layers leave this group's
195    /// `fs.read` and `fs.write` paths to the other rules, so a build tool a
196    /// sandboxed command runs can still read them. `deny` groups only, never
197    /// with `net` or `fetch`.
198    #[serde(default, skip_serializing_if = "std::ops::Not::not")]
199    pub hook_only: bool,
200}
201
202impl RuleGroup {
203    /// The patterns this group lists for `kind`.
204    #[must_use]
205    pub fn patterns(&self, kind: Kind) -> &[String] {
206        match kind {
207            Kind::Shell => &self.shell,
208            Kind::FsRead => &self.fs_read,
209            Kind::FsWrite => &self.fs_write,
210            Kind::Net => &self.net,
211            Kind::Fetch => &self.fetch,
212            // A query or an unknown tool names no host or path to match; only
213            // `defaults` decide them.
214            Kind::WebSearch | Kind::UnknownTool => &[],
215            Kind::EnvRead => &self.env_read,
216            Kind::EnvSet => &self.env_set,
217            Kind::Mcp => &self.mcp,
218        }
219    }
220
221    fn is_empty(&self) -> bool {
222        Kind::ALL.iter().all(|k| self.patterns(*k).is_empty())
223    }
224
225    /// Check that every pattern of this group parses, naming the group in the
226    /// error: what [`Policy::lint`] does for each group.
227    ///
228    /// # Errors
229    /// [`PolicyError::Rule`] for the first pattern that does not parse.
230    pub fn check_patterns(&self) -> Result<(), PolicyError> {
231        for kind in Kind::ALL {
232            for p in self.patterns(kind) {
233                let compiled = if kind == Kind::Shell {
234                    ShellPattern::compile(p).map(drop)
235                } else {
236                    GlobPattern::compile(p, false).map(drop)
237                };
238                compiled.map_err(|e| rule_err(&self.id, &e))?;
239            }
240        }
241        Ok(())
242    }
243}
244
245#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
246#[serde(deny_unknown_fields)]
247pub struct Scope {
248    #[serde(default = "default_roots")]
249    pub project_roots: Vec<String>,
250}
251
252fn default_roots() -> Vec<String> {
253    vec![".".to_owned()]
254}
255
256impl Default for Scope {
257    fn default() -> Self {
258        Self {
259            project_roots: default_roots(),
260        }
261    }
262}
263
264#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
265#[serde(rename_all = "lowercase")]
266pub enum ApprovalChannel {
267    Terminal,
268    Telegram,
269}
270
271#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
272#[serde(rename_all = "lowercase")]
273pub enum RememberScope {
274    Once,
275    Session,
276    Permanent,
277}
278
279#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
280#[serde(deny_unknown_fields)]
281pub struct Approval {
282    #[serde(default = "default_channel")]
283    pub channel: ApprovalChannel,
284    #[serde(default = "default_remember")]
285    pub remember: RememberScope,
286    #[serde(default = "default_timeout")]
287    pub timeout_s: u32,
288}
289
290fn default_channel() -> ApprovalChannel {
291    ApprovalChannel::Terminal
292}
293fn default_remember() -> RememberScope {
294    RememberScope::Session
295}
296fn default_timeout() -> u32 {
297    300
298}
299
300impl Default for Approval {
301    fn default() -> Self {
302        Self {
303            channel: default_channel(),
304            remember: default_remember(),
305            timeout_s: default_timeout(),
306        }
307    }
308}
309
310/// The policy document (`policy.yaml`), as written by users.
311#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
312#[serde(deny_unknown_fields)]
313pub struct Policy {
314    /// Schema version; only 1 is supported.
315    pub version: u32,
316    /// Verdict when no rule matches.
317    #[serde(default = "default_verdict")]
318    pub defaults: Defaults,
319    /// Reserved for multi-root projects; parsed so existing files stay valid,
320    /// not yet consulted (`policy lint` warns).
321    #[serde(default, skip_serializing_if = "Option::is_none")]
322    pub scope: Option<Scope>,
323    /// Evaluated first; a match is final.
324    #[serde(default)]
325    pub deny: Vec<RuleGroup>,
326    /// Evaluated after `deny`.
327    #[serde(default)]
328    pub allow: Vec<RuleGroup>,
329    /// Evaluated after `allow`: calls a person reviews. Read from `ask:` too,
330    /// its name before ADR-053 ([`OldNames`]).
331    #[serde(default, alias = "ask")]
332    pub review: Vec<RuleGroup>,
333    /// Evaluated after `review`: calls the agent's own permission flow decides
334    /// (ADR-025). Absent in policies written before it existed.
335    #[serde(default, skip_serializing_if = "Vec::is_empty")]
336    pub defer: Vec<RuleGroup>,
337    /// Reserved for the kernel's own approval prompt; parsed so existing files
338    /// stay valid, not yet consulted (`policy lint` warns).
339    #[serde(default, skip_serializing_if = "Option::is_none")]
340    pub approval: Option<Approval>,
341    /// Basename → allowed absolute paths (defeats PATH poisoning).
342    #[serde(default)]
343    pub executables: BTreeMap<String, Vec<String>>,
344    /// What the operating-system layers OpenMoat configures need beyond the rules
345    /// (ADR-018). Absent in policies written before it existed.
346    #[serde(default, skip_serializing_if = "Option::is_none")]
347    pub sandbox: Option<SandboxSettings>,
348    /// Secrets `moat proxy` keeps from the agent and injects for one host each.
349    #[serde(default, skip_serializing_if = "Vec::is_empty")]
350    pub secrets: Vec<Secret>,
351    /// What session taint protects beyond the built-in list. Absent in older policies.
352    #[serde(default, skip_serializing_if = "Option::is_none")]
353    pub taint: Option<TaintSettings>,
354    /// When a session counts as having collected files outside the project
355    /// (ADR-061, proposed): the `@staging-egress` atom's thresholds. Absent in
356    /// older policies, which means the defaults.
357    #[serde(default, skip_serializing_if = "Option::is_none")]
358    pub staging: Option<StagingSettings>,
359    /// Trust a repository policy only when its detached signature
360    /// (`policy.yaml.sig`) verifies against a key the person trusted (ADR-031):
361    /// a signed one needs no `moat trust`, and an unsigned one is not trusted even
362    /// after `moat trust`. Off by default: then only `moat trust` lets a
363    /// repository policy allow.
364    #[serde(default, skip_serializing_if = "std::ops::Not::not")]
365    pub require_signed_repo_policy: bool,
366    /// How `moat mcp proxy` treats tool definitions (ADR-032). Absent in older policies.
367    #[serde(default, skip_serializing_if = "Option::is_none")]
368    pub mcp: Option<McpSettings>,
369    /// The levels `openmoat policy check-diff` reports its content checks at
370    /// (ADR-055). Absent in older policies: each check's own level.
371    #[serde(default, skip_serializing_if = "Option::is_none")]
372    pub check_diff: Option<crate::diffscan::Settings>,
373    /// The least protection level each agent must be at for `openmoat guard`
374    /// to decide its calls; below it, every call is denied
375    /// (`protection-required`, ADR-064). Absent in older policies and the
376    /// default: nothing is required.
377    #[serde(default, skip_serializing_if = "Option::is_none")]
378    pub require: Option<Require>,
379    /// The `require:` sections of the repository policy and the packs merged
380    /// in ([`Policy::add_requirement`]), read with [`Self::require`] by
381    /// [`Policy::required_for`], which takes the strictest: they raise a
382    /// level, never lower one. Never read from or written to a file.
383    #[serde(skip)]
384    pub require_added: Vec<(RequiredBy, Require)>,
385    /// Review rules merged in from a repository policy ([`crate::RepoPolicy`],
386    /// ADR-022) or a policy pack (ADR-031). Tried after `deny` and before
387    /// `allow`, so a repository can make an action the user allows need review.
388    /// Never read from or written to a file, and not lowered to OS layers, which
389    /// are generated from the user policy alone.
390    #[serde(skip)]
391    pub repo_review: Vec<RuleGroup>,
392    /// Where the file used `ask`, the old name of `review` (ADR-053), which
393    /// `policy lint` warns about. Set by [`Policy::parse`]; never written.
394    #[serde(skip)]
395    pub old_names: OldNames,
396}
397
398/// The `sandbox:` section (ADR-018, ADR-019).
399#[derive(Debug, Clone, Default, PartialEq, Eq, Serialize, Deserialize)]
400#[serde(deny_unknown_fields)]
401pub struct SandboxSettings {
402    /// Paths (directories or files) the OS layers let commands read although no
403    /// allow rule covers them: system and toolchain directories every build
404    /// reads. The hook decision ignores them, and deny rules win over them in
405    /// every layer. Absolute or `~/…`, never a glob.
406    #[serde(default)]
407    pub read_roots: Vec<String>,
408    /// The loopback port `moat proxy` listens on and the host sandboxes send
409    /// their commands' traffic to; absent, the CLI's default.
410    #[serde(default, skip_serializing_if = "Option::is_none")]
411    pub proxy_port: Option<std::num::NonZeroU16>,
412    /// Let `moat proxy` forward a TLS `ClientHello` that carries an encrypted
413    /// server name (ECH, ESNI). Off, such a tunnel is refused: the server acts
414    /// on the encrypted name, which may be another site than the allowed host.
415    #[serde(default, skip_serializing_if = "std::ops::Not::not")]
416    pub proxy_allow_ech: bool,
417    /// Let `moat proxy` serve a request that carries no session credential,
418    /// with no session (ADR-044). Off, such a request gets 407: a session's
419    /// taint and connection share cannot be escaped by sending no credential,
420    /// and other local processes cannot use the proxy. `moat run` ignores it.
421    #[serde(default, skip_serializing_if = "std::ops::Not::not")]
422    pub proxy_serve_without_credential: bool,
423    /// Days a package version must have been published before the package
424    /// managers `moat run` starts install it (ADR-030); `0` turns this off,
425    /// absent is 7 days ([`Policy::package_min_release_age_days`]).
426    #[serde(default, skip_serializing_if = "Option::is_none")]
427    pub package_min_release_age_days: Option<u16>,
428    /// Names of inherited environment variables `openmoat run` passes on
429    /// under `require.run: full-boundary`, beyond its own list and the
430    /// agent's (ADR-077). Names only, never a pattern. Read from the person's
431    /// own policy alone: a repository policy or a pack has no `sandbox:`.
432    #[serde(default, skip_serializing_if = "Vec::is_empty")]
433    pub env_allow: Vec<String>,
434    /// What `openmoat run`'s Lightweight tier on Linux without a network
435    /// namespace grants at the top of the project (ADR-090): `protected`
436    /// grants it entry by entry, so nothing can be created there and `.env`,
437    /// `.git` and `.moat` stay covered; `writable` grants it whole, so new
438    /// entries and save-by-rename work there and those paths are open to the
439    /// agent's commands. Never read by the Isolated tier or the network
440    /// namespace. Read from the person's own policy alone: a repository policy
441    /// or a pack has no `sandbox:`.
442    #[serde(default, skip_serializing_if = "ProjectTop::is_protected")]
443    pub project_top: ProjectTop,
444}
445
446/// `sandbox.project_top` (ADR-090).
447#[derive(Debug, Clone, Copy, Default, PartialEq, Eq, Serialize, Deserialize)]
448#[serde(rename_all = "lowercase")]
449pub enum ProjectTop {
450    /// Granted entry by entry: the policy's denies at the top hold.
451    #[default]
452    Protected,
453    /// Granted whole: the top is writable, and `.env` and `.git/hooks` there
454    /// are open to the agent's commands.
455    Writable,
456}
457
458impl ProjectTop {
459    /// Whether this is the default, [`ProjectTop::Protected`].
460    #[must_use]
461    pub fn is_protected(&self) -> bool {
462        *self == Self::Protected
463    }
464}
465
466/// The `mcp:` section (ADR-032).
467#[derive(Debug, Clone, Default, PartialEq, Eq, Serialize, Deserialize)]
468#[serde(deny_unknown_fields)]
469pub struct McpSettings {
470    /// Pin every tool `moat mcp proxy` sees from a server it has no pins for,
471    /// as it is, instead of holding them until `moat mcp accept`. Off by
472    /// default: a server that is malicious on first contact would get its
473    /// tool descriptions pinned as good.
474    #[serde(default, skip_serializing_if = "std::ops::Not::not")]
475    pub trust_first_sight: bool,
476    /// How an MCP call that only opens or reads one web page is decided
477    /// (ADR-047): `fetch` (absent) as a `fetch` of its host, like `WebFetch`;
478    /// `net` as a `net` connection, which the default policy denies for
479    /// unlisted hosts.
480    #[serde(default, skip_serializing_if = "Option::is_none")]
481    pub browser_open: Option<BrowserOpen>,
482    /// MCP servers, by the name `openmoat mcp proxy --server` gives them,
483    /// whose data flows are trusted (ADR-058): text such a server returned may
484    /// reach another server's tool that writes or sends, and a call to such a
485    /// server may carry another server's text, without a `mcp-cross-server`
486    /// review.
487    #[serde(default, skip_serializing_if = "Vec::is_empty")]
488    pub flow_trusted: Vec<String>,
489}
490
491/// `mcp.browser_open`: the kind a read-only page open of an MCP tool is decided as.
492#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
493#[serde(rename_all = "lowercase")]
494pub enum BrowserOpen {
495    /// A `fetch` of the page's host (the default).
496    Fetch,
497    /// A `net` connection to the page's host, as any other MCP URL argument.
498    Net,
499}
500
501/// The release age `moat run` requires of new package versions when the policy
502/// does not say (ADR-030).
503pub const DEFAULT_PACKAGE_MIN_RELEASE_AGE_DAYS: u16 = 7;
504
505impl Policy {
506    /// The days a package version must have been published before `moat run`
507    /// lets npm and uv install it; `0` is off.
508    #[must_use]
509    pub fn package_min_release_age_days(&self) -> u16 {
510        self.sandbox
511            .as_ref()
512            .and_then(|s| s.package_min_release_age_days)
513            .unwrap_or(DEFAULT_PACKAGE_MIN_RELEASE_AGE_DAYS)
514    }
515
516    /// What `openmoat run` without a network namespace grants at the project's
517    /// top (`sandbox.project_top`, ADR-090); [`ProjectTop::Protected`] unless
518    /// the person's own policy says otherwise.
519    #[must_use]
520    pub fn project_top(&self) -> ProjectTop {
521        self.sandbox
522            .as_ref()
523            .map(|s| s.project_top)
524            .unwrap_or_default()
525    }
526
527    /// Whether `moat mcp proxy` pins a new server's tools without a person
528    /// accepting them (`mcp.trust_first_sight`, ADR-032).
529    #[must_use]
530    pub fn trust_mcp_first_sight(&self) -> bool {
531        self.mcp.as_ref().is_some_and(|m| m.trust_first_sight)
532    }
533
534    /// Whether an MCP call that only opens a web page is decided as a `fetch`
535    /// (`mcp.browser_open`, ADR-047): yes unless the policy says `net`.
536    #[must_use]
537    pub fn browser_open_is_fetch(&self) -> bool {
538        self.mcp.as_ref().and_then(|m| m.browser_open) != Some(BrowserOpen::Net)
539    }
540
541    /// The MCP servers whose data flows need no `mcp-cross-server` review
542    /// (`mcp.flow_trusted`, ADR-058).
543    #[must_use]
544    pub fn mcp_flow_trusted(&self) -> &[String] {
545        self.mcp.as_ref().map_or(&[], |m| m.flow_trusted.as_slice())
546    }
547}
548
549fn default_verdict() -> Defaults {
550    Defaults::All(Verdict::Review)
551}
552
553impl Policy {
554    /// Parse and lint a policy document. Pure: takes the YAML text.
555    pub fn parse(yaml: &str) -> Result<Self, PolicyError> {
556        let old_names = old_names::find(yaml)?;
557        let mut policy: Self = crate::yaml::from_str(yaml).map_err(PolicyError::Parse)?;
558        policy.old_names = old_names;
559        policy.lint()?;
560        Ok(policy)
561    }
562
563    /// Structural validation independent of any environment: [`Self::lint_structure`]
564    /// and every pattern's syntax ([`RuleGroup::check_patterns`]).
565    pub fn lint(&self) -> Result<(), PolicyError> {
566        self.lint_structure()?;
567        self.groups().try_for_each(RuleGroup::check_patterns)
568    }
569
570    /// [`Self::lint`] without parsing each pattern again, for a policy that
571    /// gained rule groups after a full lint, such as a merge of rules each
572    /// checked when they were read (a repository policy, a pack, `openmoat allow`
573    /// rules): the new groups' patterns are checked with
574    /// [`RuleGroup::check_patterns`], and nothing else needs reparsing. Parsing
575    /// every pattern costs more than the rest of the lint together, and
576    /// [`crate::CompiledPolicy::compile`] reports a bad one in any case.
577    pub fn lint_structure(&self) -> Result<(), PolicyError> {
578        if self.version != SUPPORTED_VERSION {
579            return Err(PolicyError::Version {
580                found: self.version,
581                supported: SUPPORTED_VERSION,
582            });
583        }
584        let mut ids = BTreeSet::new();
585        for group in self.groups() {
586            if group.id.trim().is_empty() {
587                return Err(PolicyError::Rule {
588                    rule: "<unnamed>".into(),
589                    problem: "missing id".into(),
590                });
591            }
592            if !ids.insert(group.id.clone()) {
593                return Err(PolicyError::DuplicateId(group.id.clone()));
594            }
595            if group.is_empty() {
596                return Err(PolicyError::EmptyRule(group.id.clone()));
597            }
598            text::check(group)?;
599        }
600        hook_only::check(self)?;
601        for (name, paths) in &self.executables {
602            if paths.is_empty() {
603                return Err(PolicyError::Rule {
604                    rule: format!("executables.{name}"),
605                    problem: "no paths listed".into(),
606                });
607            }
608            if let Some(bad) = paths.iter().find(|p| !crate::paths::is_absolute(p)) {
609                return Err(PolicyError::Rule {
610                    rule: format!("executables.{name}"),
611                    problem: format!("`{bad}` is not an absolute path"),
612                });
613            }
614        }
615        for root in self.sandbox.iter().flat_map(|s| &s.read_roots) {
616            lint_read_root(root)?;
617        }
618        for name in self.sandbox.iter().flat_map(|s| &s.env_allow) {
619            if let Some(problem) = env_name_problem(name) {
620                return Err(PolicyError::Rule {
621                    rule: "sandbox.env_allow".to_owned(),
622                    problem: format!("`{name}` {problem}"),
623                });
624            }
625        }
626        taint::check(self.taint.as_ref())?;
627        staging::check(self.staging.as_ref())?;
628        crate::diffscan::check(self.check_diff.as_ref())?;
629        require::check(self.require.as_ref())?;
630        for (_, added) in &self.require_added {
631            require::check(Some(added))?;
632        }
633        secret::check(&self.secrets)
634    }
635
636    /// Every rule group, in the order the engine tries the lists.
637    fn groups(&self) -> impl Iterator<Item = &RuleGroup> {
638        self.deny
639            .iter()
640            .chain(&self.repo_review)
641            .chain(&self.allow)
642            .chain(&self.review)
643            .chain(&self.defer)
644    }
645
646    /// Number of rule groups in `deny`, `allow` and `review`.
647    #[must_use]
648    pub fn rule_count(&self) -> (usize, usize, usize) {
649        (self.deny.len(), self.allow.len(), self.review.len())
650    }
651}
652
653/// An OS layer takes a read root literally, and a root that is the disk or the
654/// home directory would undo every default deny.
655fn lint_read_root(root: &str) -> Result<(), PolicyError> {
656    let trimmed = root.trim_end_matches(['/', '\\']);
657    let problem = if !(crate::paths::is_absolute(root) || root.starts_with("~/")) {
658        "must be absolute or start with `~/`"
659    } else if root.contains(['*', '?', '[', ']', '{', '}', '!', '$']) {
660        "must be a plain path, not a glob or a variable"
661    } else if root.split(['/', '\\']).any(|s| s == "." || s == "..") {
662        "must not contain `.` or `..` segments"
663    } else if trimmed == "~" {
664        "must not be the home directory"
665    } else if trimmed.is_empty() || (trimmed.len() == 2 && trimmed.ends_with(':')) {
666        "must not be a filesystem root"
667    } else {
668        return Ok(());
669    };
670    Err(PolicyError::Rule {
671        rule: "sandbox.read_roots".to_owned(),
672        problem: format!("`{root}` {problem}"),
673    })
674}
675
676fn rule_err(rule: &str, e: &PolicyError) -> PolicyError {
677    PolicyError::Rule {
678        rule: rule.to_owned(),
679        problem: e.to_string(),
680    }
681}
682
683#[cfg(test)]
684mod tests;