Skip to main content

spec_drift/analyzers/
ci.rs

1use super::DriftAnalyzer;
2use crate::context::ProjectContext;
3use crate::domain::{Divergence, Location, RuleId, Severity};
4use regex::Regex;
5use std::path::Path;
6use std::process::Command;
7use std::sync::OnceLock;
8
9/// CiAnalyzer — enforces the `ghost_command` rule (deterministic).
10///
11/// Strategy: scan Makefile / justfile / GitHub workflow YAMLs for `cargo`
12/// invocations. For each `--package <name>` / `-p <name>` / `--bin <name>`
13/// argument, confirm the target exists in `cargo metadata`. Anything else
14/// (unknown packages, unknown bins) is a ghost command.
15#[derive(Default)]
16pub struct CiAnalyzer {
17    metadata_override: Option<CargoMetadata>,
18}
19
20impl CiAnalyzer {
21    /// Construct with pre-loaded metadata (used by tests so they don't shell out).
22    pub fn with_metadata(md: CargoMetadata) -> Self {
23        Self {
24            metadata_override: Some(md),
25        }
26    }
27}
28
29#[derive(Debug, Clone, Default)]
30pub struct CargoMetadata {
31    pub packages: Vec<String>,
32    pub bins: Vec<String>,
33}
34
35impl CargoMetadata {
36    fn knows_package(&self, name: &str) -> bool {
37        self.packages.iter().any(|p| p == name)
38    }
39    fn knows_bin(&self, name: &str) -> bool {
40        self.bins.iter().any(|b| b == name)
41    }
42
43    fn load(manifest_dir: &Path) -> Self {
44        #[derive(serde::Deserialize)]
45        struct RawMetadata {
46            packages: Vec<RawPackage>,
47        }
48        #[derive(serde::Deserialize)]
49        struct RawPackage {
50            name: String,
51            targets: Vec<RawTarget>,
52        }
53        #[derive(serde::Deserialize)]
54        struct RawTarget {
55            name: String,
56            kind: Vec<String>,
57        }
58
59        let out = Command::new("cargo")
60            .current_dir(manifest_dir)
61            .args(["metadata", "--format-version=1", "--no-deps"])
62            .output();
63        let Ok(out) = out else {
64            return Self::default();
65        };
66        if !out.status.success() {
67            return Self::default();
68        }
69        let Ok(md) = serde_json::from_slice::<RawMetadata>(&out.stdout) else {
70            return Self::default();
71        };
72
73        let mut packages = Vec::new();
74        let mut bins = Vec::new();
75        for pkg in &md.packages {
76            packages.push(pkg.name.clone());
77            for t in &pkg.targets {
78                if t.kind.iter().any(|k| k == "bin") {
79                    bins.push(t.name.clone());
80                }
81            }
82        }
83        Self { packages, bins }
84    }
85}
86
87impl DriftAnalyzer for CiAnalyzer {
88    fn analyze(&self, ctx: &ProjectContext) -> Vec<Divergence> {
89        let metadata = self
90            .metadata_override
91            .clone()
92            .unwrap_or_else(|| CargoMetadata::load(&ctx.root));
93
94        let mut out = Vec::new();
95
96        // Makefiles and justfiles — line-oriented.
97        for mk in &ctx.makefile_files {
98            let src = match std::fs::read_to_string(mk) {
99                Ok(s) => s,
100                Err(e) => {
101                    eprintln!("spec-drift: skipping {} (ci): {e}", mk.display());
102                    continue;
103                }
104            };
105            for (idx, line) in src.lines().enumerate() {
106                let line_no = (idx + 1) as u32;
107                let trimmed = line.trim_start_matches(['\t', ' ']);
108                // Strip `@` echo prefix common in make recipes.
109                let trimmed = trimmed.strip_prefix('@').unwrap_or(trimmed);
110                inspect_cargo_line(trimmed, mk, line_no, &metadata, &mut out);
111            }
112        }
113
114        // YAML workflows — treat every line containing a cargo invocation.
115        for yaml in &ctx.yaml_files {
116            let src = match std::fs::read_to_string(yaml) {
117                Ok(s) => s,
118                Err(e) => {
119                    eprintln!("spec-drift: skipping {} (ci): {e}", yaml.display());
120                    continue;
121                }
122            };
123            if !is_workflow_path(yaml) {
124                continue;
125            }
126            for (idx, line) in src.lines().enumerate() {
127                let line_no = (idx + 1) as u32;
128                inspect_cargo_line(line, yaml, line_no, &metadata, &mut out);
129            }
130        }
131
132        out
133    }
134}
135
136fn is_workflow_path(path: &Path) -> bool {
137    // Only scan files inside a `.github/workflows/` directory to avoid noisy
138    // matches in unrelated YAML (k8s manifests, docker-compose, etc.).
139    path.components().any(|c| c.as_os_str() == ".github")
140        && path.components().any(|c| c.as_os_str() == "workflows")
141}
142
143fn cargo_command_re() -> &'static Regex {
144    static RE: OnceLock<Regex> = OnceLock::new();
145    RE.get_or_init(|| Regex::new(r"(?:^|\s)cargo\s+([a-zA-Z][a-zA-Z0-9\-]*)\b").unwrap())
146}
147
148fn package_arg_re() -> &'static Regex {
149    static RE: OnceLock<Regex> = OnceLock::new();
150    RE.get_or_init(|| Regex::new(r"(?:--package|-p)[ =]([A-Za-z_][A-Za-z0-9_\-]*)").unwrap())
151}
152
153fn bin_arg_re() -> &'static Regex {
154    static RE: OnceLock<Regex> = OnceLock::new();
155    RE.get_or_init(|| Regex::new(r"--bin[ =]([A-Za-z_][A-Za-z0-9_\-]*)").unwrap())
156}
157
158fn inspect_cargo_line(
159    line: &str,
160    path: &Path,
161    line_no: u32,
162    md: &CargoMetadata,
163    out: &mut Vec<Divergence>,
164) {
165    if cargo_command_re().find(line).is_none() {
166        return;
167    }
168    // Skip cargo-generated or unknown metadata — only flag when we have facts.
169    if md.packages.is_empty() && md.bins.is_empty() {
170        return;
171    }
172
173    if let Some(caps) = package_arg_re().captures(line) {
174        let name = caps.get(1).unwrap().as_str();
175        if !md.knows_package(name) {
176            out.push(Divergence {
177                rule: RuleId::GhostCommand,
178                severity: Severity::Warning,
179                location: Location::new(path.to_path_buf(), line_no),
180                stated: format!("CI runs `cargo` against package `{name}`"),
181                reality: format!("`{name}` is not a member of the workspace"),
182                risk: "CI exercises a target that no longer exists; the step is a no-op at best."
183                    .to_string(),
184                attribution: None,
185            });
186        }
187    }
188
189    if let Some(caps) = bin_arg_re().captures(line) {
190        let name = caps.get(1).unwrap().as_str();
191        if !md.knows_bin(name) {
192            out.push(Divergence {
193                rule: RuleId::GhostCommand,
194                severity: Severity::Warning,
195                location: Location::new(path.to_path_buf(), line_no),
196                stated: format!("CI builds/runs bin `{name}`"),
197                reality: format!("no bin target named `{name}` in the workspace"),
198                risk: "CI step refers to a bin that doesn't exist.".to_string(),
199                attribution: None,
200            });
201        }
202    }
203}
204
205#[cfg(test)]
206mod tests_inner {
207    use super::*;
208    use std::path::PathBuf;
209
210    fn ctx_with(path: PathBuf, mk: bool, yml: bool) -> ProjectContext {
211        let tmp = path.parent().unwrap().to_path_buf();
212        let mut ctx = ProjectContext::new(tmp);
213        if mk {
214            ctx.makefile_files.push(path);
215        } else if yml {
216            ctx.yaml_files.push(path);
217        }
218        ctx
219    }
220
221    fn md(packages: &[&str], bins: &[&str]) -> CargoMetadata {
222        CargoMetadata {
223            packages: packages.iter().map(|s| s.to_string()).collect(),
224            bins: bins.iter().map(|s| s.to_string()).collect(),
225        }
226    }
227
228    #[test]
229    fn flags_unknown_package_in_makefile() {
230        let tmp = tempfile::tempdir().unwrap();
231        let mk = tmp.path().join("Makefile");
232        std::fs::write(&mk, "test:\n\tcargo test --package legacy_crate\n").unwrap();
233
234        let ctx = ctx_with(mk, true, false);
235        let analyzer = CiAnalyzer::with_metadata(md(&["spec-drift"], &["spec-drift"]));
236        let divs = analyzer.analyze(&ctx);
237
238        assert_eq!(divs.len(), 1);
239        assert_eq!(divs[0].rule, RuleId::GhostCommand);
240        assert_eq!(divs[0].severity, Severity::Warning);
241    }
242
243    #[test]
244    fn accepts_known_package() {
245        let tmp = tempfile::tempdir().unwrap();
246        let mk = tmp.path().join("Makefile");
247        std::fs::write(&mk, "test:\n\tcargo test --package spec-drift\n").unwrap();
248
249        let ctx = ctx_with(mk, true, false);
250        let analyzer = CiAnalyzer::with_metadata(md(&["spec-drift"], &[]));
251        assert!(analyzer.analyze(&ctx).is_empty());
252    }
253
254    #[test]
255    fn flags_unknown_bin_in_workflow() {
256        let tmp = tempfile::tempdir().unwrap();
257        let workflows = tmp.path().join(".github").join("workflows");
258        std::fs::create_dir_all(&workflows).unwrap();
259        let yml = workflows.join("ci.yml");
260        std::fs::write(
261            &yml,
262            "jobs:\n  t:\n    steps:\n      - run: cargo run --bin legacy\n",
263        )
264        .unwrap();
265
266        let mut ctx = ProjectContext::new(tmp.path());
267        ctx.yaml_files.push(yml);
268        let analyzer = CiAnalyzer::with_metadata(md(&["spec-drift"], &["spec-drift"]));
269        let divs = analyzer.analyze(&ctx);
270
271        assert_eq!(divs.len(), 1);
272        assert_eq!(divs[0].rule, RuleId::GhostCommand);
273    }
274
275    #[test]
276    fn ignores_yaml_outside_github_workflows() {
277        let tmp = tempfile::tempdir().unwrap();
278        let yml = tmp.path().join("deploy.yml");
279        std::fs::write(&yml, "run: cargo run --bin legacy\n").unwrap();
280
281        let mut ctx = ProjectContext::new(tmp.path());
282        ctx.yaml_files.push(yml);
283        let analyzer = CiAnalyzer::with_metadata(md(&["spec-drift"], &[]));
284        assert!(analyzer.analyze(&ctx).is_empty());
285    }
286
287    #[test]
288    fn noop_when_metadata_unavailable() {
289        let tmp = tempfile::tempdir().unwrap();
290        let mk = tmp.path().join("Makefile");
291        std::fs::write(&mk, "cargo test --package ghost\n").unwrap();
292        let ctx = ctx_with(mk, true, false);
293        // Empty metadata — the analyzer must fail-open, never spurious-positive.
294        let analyzer = CiAnalyzer::with_metadata(CargoMetadata::default());
295        assert!(analyzer.analyze(&ctx).is_empty());
296    }
297}