Skip to content
506 linesCodeBlameRaw
1//! Dependencies: reading a repository's lockfiles, asking OSV about every
2//! package in them, and recording what is vulnerable. The security updates
3//! that fix them are `security_updates`'s.
4
5use std::collections::{BTreeSet, HashMap};
6
7use g1t_contracts::repos::RepoPath;
8use g1t_contracts::repos::{BlobArgs, BlobView};
9use g1t_contracts::security::{AlertState, FindLockfilesArgs, Lockfiles, VersionUpdatesState, VulnStatus, Vulnerability};
10use g1t_contracts::updates::DEPENDABOT_PATHS;
11use g1t_contracts::time::rfc3339;
12use g1t_contracts::work::{AddCommentArgs, Issue, IssueDetail, ViewArgs};
13use g1t_contracts::{Outcome, User};
14use g1t_kit::now_ms;
15use g1t_scan::lockfiles::{Lockfile, Package, still_locked_check, test_command};
16use g1t_scan::osv::{self, Advisory};
17use serde_json::{Value, json};
18use worker::{Fetch, Headers, Method, Request, RequestInit, Result};
19
20use crate::Security;
21use crate::store::{RepoRow, VulnRow};
22use crate::updates;
23use crate::version_updates::Loaded;
24
25/// OSV's records are fetched again after this long.
26const ADVISORY_MAX_AGE_MS: u64 = 7 * 24 * 60 * 60 * 1000;
27/// A record that names no fixed version is fetched again after a day, so
28/// a fix is noticed the day it is published.
29const UNFIXED_ADVISORY_MAX_AGE_MS: u64 = 24 * 60 * 60 * 1000;
30/// Records fetched per scan, at most; the rest wait for the next one.
31const MAX_ADVISORY_FETCHES: usize = 150;
32/// CPU one call to OSV takes, sending it and reading its answer, in
33/// milliseconds (an estimate, rounded up). OSV itself is free, and a
34/// Worker's outgoing requests are not charged.
35const CPU_MS_PER_OSV_CALL: f64 = 2.0;
36
37/// What a dependency check cost g1t, in millionths of a dollar, rounded
38/// up, at the prices in `history`: the CPU of its OSV calls, and the rows
39/// it writes (an advisory kept for each call at most, each vulnerability
40/// found, where the check stands and the month's usage).
41pub fn dependency_check_cost(calls: u32, found: usize) -> i64 {
42 use crate::history::{MICROS_PER_CPU_MS, MICROS_PER_ROW_WRITTEN};
43 let cpu = f64::from(calls) * CPU_MS_PER_OSV_CALL * MICROS_PER_CPU_MS;
44 let rows = (calls as usize + found + 2) as f64 * MICROS_PER_ROW_WRITTEN;
45 (cpu + rows).ceil() as i64
46}
47
48async fn osv_call(method: Method, url: &str, body: Option<&Value>) -> Result<Option<Value>> {
49 let headers = Headers::new();
50 headers.set("user-agent", "g1t (+https://g1t.sh)")?;
51 headers.set("accept", "application/json")?;
52 if body.is_some() {
53 headers.set("content-type", "application/json")?;
54 }
55 let mut init = RequestInit::new();
56 init.with_method(method).with_headers(headers);
57 if let Some(body) = body {
58 init.with_body(Some(body.to_string().into()));
59 }
60 let mut response = Fetch::Request(Request::new_with_init(url, &init)?).send().await?;
61 if response.status_code() == 404 {
62 return Ok(None);
63 }
64 if !(200..300).contains(&response.status_code()) {
65 return Err(worker::Error::RustError(format!("OSV answered {}", response.status_code())));
66 }
67 Ok(Some(response.json().await?))
68}
69
70/// Whether an OSV record names a version that fixes it, for any package.
71fn names_a_fix(record: &Value) -> bool {
72 record["affected"].as_array().into_iter().flatten().any(|affected| {
73 affected["ranges"]
74 .as_array()
75 .into_iter()
76 .flatten()
77 .any(|range| range["events"].as_array().into_iter().flatten().any(|event| event.get("fixed").is_some()))
78 })
79}
80
81/// One package in one lockfile.
82pub(crate) struct Located {
83 pub(crate) package: Package,
84 pub(crate) lockfile: Lockfile,
85 pub(crate) path: String,
86}
87
88/// Every package the lockfiles resolve. A directory with a `go.mod` is read
89/// from it rather than from its `go.sum`, which lists versions not built.
90fn packages(files: &Lockfiles) -> Vec<Located> {
91 let go_mods: BTreeSet<&str> = files
92 .files
93 .iter()
94 .filter(|file| file.path.ends_with("go.mod"))
95 .map(|file| file.path.trim_end_matches("go.mod"))
96 .collect();
97 let mut located = Vec::new();
98 for file in &files.files {
99 let Some(lockfile) = Lockfile::for_path(&file.path) else { continue };
100 if lockfile == Lockfile::GoSum && go_mods.contains(file.path.trim_end_matches("go.sum")) {
101 continue;
102 }
103 for package in lockfile.parse(&file.text) {
104 located.push(Located { package, lockfile, path: file.path.clone() });
105 }
106 }
107 located
108}
109
110fn directory(path: &str) -> &str {
111 path.rsplit_once('/').map_or("", |(directory, _)| directory)
112}
113
114impl Security {
115 /// The ids of the vulnerabilities affecting each package, from OSV.
116 pub(crate) async fn query_osv(&self, packages: &[Package]) -> Result<(Vec<Vec<String>>, u32)> {
117 let mut ids = Vec::with_capacity(packages.len());
118 let mut calls = 0;
119 for (body, chunk) in osv::batch_bodies(packages).iter().zip(packages.chunks(osv::MAX_BATCH)) {
120 calls += 1;
121 let answer = osv_call(Method::Post, osv::QUERY_BATCH_URL, Some(body)).await?.unwrap_or(Value::Null);
122 let (mut found, more) = osv::read_batch(&answer, chunk.len());
123 // A package with many advisories is paged; fetch the rest.
124 for (index, token) in more.into_iter().take(20) {
125 let package = &chunk[index];
126 let query = json!({
127 "package": {"name": package.name, "ecosystem": package.ecosystem.osv()},
128 "version": package.version,
129 "page_token": token,
130 });
131 calls += 1;
132 if let Some(page) = osv_call(Method::Post, "https://api.osv.dev/v1/query", Some(&query)).await? {
133 found[index].extend(
134 page["vulns"].as_array().into_iter().flatten().filter_map(|v| v["id"].as_str().map(str::to_owned)),
135 );
136 }
137 }
138 ids.extend(found);
139 }
140 Ok((ids, calls))
141 }
142
143 /// OSV's record of each id, from the cache when it is fresh.
144 pub(crate) async fn advisories(&self, ids: &BTreeSet<String>) -> Result<(HashMap<String, Value>, u32)> {
145 let fresh_after = rfc3339(now_ms().saturating_sub(ADVISORY_MAX_AGE_MS));
146 let unfixed_fresh_after = rfc3339(now_ms().saturating_sub(UNFIXED_ADVISORY_MAX_AGE_MS));
147 let mut records = HashMap::new();
148 let mut fetched = 0u32;
149 for id in ids {
150 if let Some(record) = self.store.advisory(id, &fresh_after).await? {
151 // One without a fix is asked about again daily; until then,
152 // or if asking fails below, the kept record stands.
153 let fixed = names_a_fix(&record);
154 let recent = !fixed && self.store.advisory(id, &unfixed_fresh_after).await?.is_some();
155 if fixed || recent || fetched as usize >= MAX_ADVISORY_FETCHES {
156 records.insert(id.clone(), record);
157 continue;
158 }
159 fetched += 1;
160 let refreshed = osv_call(Method::Get, &osv::vuln_url(id), None).await.ok().flatten();
161 if let Some(refreshed) = &refreshed {
162 self.store.keep_advisory(id, refreshed).await?;
163 }
164 records.insert(id.clone(), refreshed.unwrap_or(record));
165 continue;
166 }
167 if fetched as usize >= MAX_ADVISORY_FETCHES {
168 continue;
169 }
170 fetched += 1;
171 if let Some(record) = osv_call(Method::Get, &osv::vuln_url(id), None).await? {
172 self.store.keep_advisory(id, &record).await?;
173 records.insert(id.clone(), record);
174 }
175 }
176 Ok((records, fetched))
177 }
178
179 /// Reads a repository's dependencies, records which are vulnerable,
180 /// and starts security updates for those with a fix. Also reads
181 /// `.g1t/dependencies.yml`. Returns what went wrong, for the Security
182 /// page, if anything did.
183 pub async fn scan_dependencies(&self, repo: &RepoRow) -> Result<Option<String>> {
184 let files: Lockfiles = g1t_kit::call(&self.repos, "find_lockfiles", &FindLockfilesArgs { repo_id: repo.repo_id.clone(), git_ref: None }).await?;
185 let paths: Vec<String> = files.files.iter().map(|file| file.path.clone()).collect();
186 let located = packages(&files);
187 let unique: Vec<Package> = located.iter().map(|l| l.package.clone()).collect::<BTreeSet<_>>().into_iter().collect();
188 let outcome = async {
189 let (ids, calls) = self.query_osv(&unique).await?;
190 let by_package: HashMap<&Package, &Vec<String>> = unique.iter().zip(ids.iter()).collect();
191 let wanted: BTreeSet<String> = ids.iter().flatten().cloned().collect();
192 let (records, fetched) = self.advisories(&wanted).await?;
193 let mut found = Vec::new();
194 for item in &located {
195 for id in by_package.get(&item.package).into_iter().flat_map(|ids| ids.iter()) {
196 let Some(advisory) = records.get(id).and_then(|record| osv::read_vuln(record, &item.package)) else {
197 continue;
198 };
199 found.push(vulnerability(&repo.repo_id, item, &advisory));
200 }
201 }
202 Ok::<_, worker::Error>((found, calls + fetched))
203 }
204 .await;
205 let (found, calls) = match outcome {
206 Ok(result) => result,
207 Err(error) => {
208 let problem = format!("The dependencies could not be checked: {error}");
209 self.store
210 .set_dependencies_scanned(&repo.repo_id, files.commit.as_deref(), &paths, Some(&problem))
211 .await?;
212 return Ok(Some(problem));
213 }
214 };
215 let before: BTreeSet<String> = self.store.open_vulnerabilities(&repo.repo_id).await?.into_iter().map(|row| row.id).collect();
216 self.store.replace_vulnerabilities(&repo.repo_id, &found).await?;
217 self.store.set_dependencies_scanned(&repo.repo_id, files.commit.as_deref(), &paths, None).await?;
218 // g1t's update pull requests whose alerts are now fixed or dismissed,
219 // or whose versions the lockfiles already have, close first, so that
220 // nothing later in the scan can keep them open.
221 if let Err(error) = self.resolve_updates(repo, Some((&located, &paths))).await {
222 worker::console_error!("security: updates of {} not resolved: {error}", repo.repo_id);
223 }
224 if let Err(error) = self.record_graph(repo, &files).await {
225 worker::console_error!("security: dependency graph of {} not kept: {error}", repo.repo_id);
226 }
227 self.vulnerabilities_changed(repo, &before).await?;
228 self.meter(&repo.namespace, 0, 0, calls, dependency_check_cost(calls, found.len())).await?;
229 // The dependency update file says how security updates are made too.
230 let rules = match files.commit.as_deref() {
231 Some(commit) => self.read_version_updates(repo, commit).await?,
232 None => None,
233 };
234 // No security updates on an archived (read-only) or deleted
235 // repository; ones in flight for packages no longer vulnerable are
236 // closed either way.
237 let active = self.active(&repo.repo_id).await?;
238 self.security_updates(repo, repo.upkeep != 0 && active, rules.as_ref()).await?;
239 Ok(None)
240 }
241
242 /// Reads the dependency update file at `commit` and keeps what it
243 /// says; see `updates`. Its entries' schedules start or move. Returns
244 /// it, when it has no problems.
245 pub(crate) async fn read_version_updates(&self, repo: &RepoRow, commit: &str) -> Result<Option<Loaded>> {
246 let path = RepoPath { namespace: repo.namespace.clone(), name: repo.name.clone() };
247 let asks: Vec<BlobArgs> = DEPENDABOT_PATHS
248 .iter()
249 .map(|file| BlobArgs {
250 path: path.clone(),
251 viewer: Some(User::system(&repo.namespace)),
252 git_ref: commit.to_owned(),
253 file_path: (*file).to_owned(),
254 })
255 .collect();
256 let reads = asks.iter().map(|ask| g1t_kit::call::<_, Outcome<BlobView>>(&self.repos, "blob", ask));
257 let mut found = Vec::new();
258 let mut default_branch = None;
259 for (file, read) in DEPENDABOT_PATHS.iter().zip(futures_util::future::join_all(reads).await) {
260 if let Outcome::Ok(blob) = read? {
261 default_branch = Some(blob.repo.default_branch.clone());
262 found.push(((*file).to_owned(), blob.text));
263 }
264 }
265 let (mut state, config) = match updates::choose(found) {
266 Some(((file, text), ignored)) => {
267 let (state, config) = updates::state(&file, text.as_deref(), ignored, &repo.repo_id, default_branch.as_deref());
268 (state, config.map(|config| (config, file)))
269 }
270 None => (VersionUpdatesState::default(), None),
271 };
272 state.read_at = Some(crate::store::now());
273 state.commit = Some(commit.to_owned());
274 self.store.set_version_updates(&repo.repo_id, &state).await?;
275 self.schedule_entries(repo, config.as_ref().map(|(config, _)| config), default_branch.as_deref()).await?;
276 Ok(config.map(|(config, file)| Loaded { config, file, default_branch: default_branch.unwrap_or_default() }))
277 }
278
279 pub(crate) async fn issue(&self, actor: &User, repo: &RepoPath, number: u32) -> Result<Option<Issue>> {
280 let found: Outcome<IssueDetail> = g1t_kit::call(
281 &self.work,
282 "get_issue",
283 &ViewArgs { repo: repo.clone(), number, viewer: Some(actor.clone()), after_seq: 0 },
284 )
285 .await?;
286 Ok(found.into_result().ok().map(|detail| detail.issue))
287 }
288
289 pub(crate) async fn comment(&self, actor: &User, repo: &RepoPath, number: u32, body: String) -> Result<()> {
290 let _: Outcome<Value> = g1t_kit::call(
291 &self.work,
292 "add_comment",
293 &AddCommentArgs {
294 actor: actor.clone(),
295 repo: repo.clone(),
296 number,
297 body,
298 path: None,
299 line: None,
300 verdict: None,
301 },
302 )
303 .await?;
304 Ok(())
305 }
306}
307
308/// The event that tells of a vulnerability alert.
309pub(crate) fn vulnerability_event(repo: &RepoRow, vuln: &Vulnerability) -> g1t_contracts::security_suite::SecurityEvent {
310 g1t_contracts::security_suite::SecurityEvent {
311 repo_id: repo.repo_id.clone(),
312 alert_id: vuln.id.clone(),
313 alert_type: g1t_contracts::security_suite::AlertType::Vulnerability.as_str().to_owned(),
314 severity: vuln.severity.clone(),
315 title: format!("{} {} in {}: {}", vuln.package, vuln.version, vuln.manifest, vuln.advisory),
316 link: crate::suite::link(repo, &format!("vulnerabilities?finding={}", vuln.id)),
317 path: Some(vuln.manifest.clone()),
318 state: vuln.state.as_str().to_owned(),
319 ..Default::default()
320 }
321}
322
323impl Security {
324 /// Tells of vulnerability alerts a dependency read opened or fixed,
325 /// given the ids that were open before it.
326 async fn vulnerabilities_changed(&self, repo: &RepoRow, before: &BTreeSet<String>) -> Result<()> {
327 use g1t_contracts::security_suite::AlertType;
328 let after: Vec<VulnRow> = self.store.open_vulnerabilities(&repo.repo_id).await?;
329 let now: BTreeSet<&str> = after.iter().map(|row| row.id.as_str()).collect();
330 let opened: Vec<&str> = now.iter().copied().filter(|id| !before.contains(*id)).take(20).collect();
331 let fixed: Vec<&String> = before.iter().filter(|id| !now.contains(id.as_str())).take(20).collect();
332 for id in opened {
333 if let Some(vuln) = self.store.vulnerability(&repo.repo_id, id).await? {
334 self.alert_event(AlertType::Vulnerability, "created", repo, vulnerability_event(repo, &vuln), None).await;
335 }
336 }
337 for id in fixed {
338 if let Some(vuln) = self.store.vulnerability(&repo.repo_id, id).await?
339 && vuln.status == VulnStatus::Fixed
340 {
341 self.alert_event(AlertType::Vulnerability, "fixed", repo, vulnerability_event(repo, &vuln), None).await;
342 }
343 }
344 Ok(())
345 }
346}
347
348fn vulnerability(repo_id: &str, item: &Located, advisory: &Advisory) -> Vulnerability {
349 Vulnerability {
350 id: String::new(),
351 repo_id: repo_id.to_owned(),
352 ecosystem: item.package.ecosystem.osv().to_owned(),
353 package: item.package.name.clone(),
354 version: item.package.version.clone(),
355 manifest: item.path.clone(),
356 advisory: advisory.display_id.clone(),
357 osv_id: advisory.id.clone(),
358 summary: advisory.summary.clone(),
359 severity: advisory.severity.as_str().to_owned(),
360 fixed_version: advisory.fixed.clone(),
361 status: VulnStatus::Open,
362 issue: None,
363 found_at: String::new(),
364 fixed_at: None,
365 state: AlertState::Open,
366 dismissed_by: None,
367 dismissed_reason: None,
368 dismissed_comment: None,
369 dismissed_at: None,
370 update: None,
371 }
372}
373
374/// The issue's body, for when raising the version is not enough, written
375/// for the agent that takes it as much as for a person, ending with what
376/// done means: commands that show no lockfile still resolves a vulnerable
377/// version, and the project's tests. The pull request merges on the
378/// repository's required checks, like any other.
379pub(crate) fn issue_text(ecosystem: &str, package: &str, target: &str, vulns: &[&VulnRow], located: &[Located]) -> String {
380 let mut body = format!(
381 "`{package}` ({ecosystem}) has known vulnerabilities with a fix in **{target}**, and raising its version alone \
382 does not pass this project's checks. Upgrade it to {target} or later everywhere it is locked, and change the \
383 code that depends on it, keeping other changes to what the upgrade needs.\n\n"
384 );
385 body.push_str(&advisory_table(vulns));
386 let mut checks = Vec::new();
387 let mut manifests = BTreeSet::new();
388 let mut tests = BTreeSet::new();
389 for vuln in vulns {
390 let Some(item) = located
391 .iter()
392 .find(|item| item.path == vuln.manifest && item.package.name == vuln.package && item.package.version == vuln.version)
393 else {
394 continue;
395 };
396 if manifests.insert((item.path.clone(), item.package.version.clone())) {
397 checks.push(still_locked_check(item.lockfile, &item.path, &item.package.name, &item.package.version));
398 }
399 if let Some(test) = test_command(item.lockfile, directory(&item.path)) {
400 tests.insert(test);
401 }
402 }
403 let locked: Vec<String> = manifests.iter().map(|(path, version)| format!("`{path}` ({version})")).collect();
404 body.push_str(&format!("\nLocked in: {}.\n", locked.join(", ")));
405 body.push_str(
406 "\nIf the fix needs a major upgrade that breaks the build, change the code that depends on it in the same pull request.",
407 );
408 checks.extend(tests);
409 let mut done = vec!["No lockfile resolves a vulnerable version, and the tests still pass.".to_owned()];
410 done.extend(g1t_contracts::work::commands_pass(&checks));
411 let mut body = g1t_contracts::work::with_definition_of_done(&body, &done);
412 body.push_str("\n\n---\n_Opened by g1t's security updates. Turn them off for this project on its Security page._");
413 body
414}
415
416/// The advisories a package's vulnerabilities name, as a table.
417pub(crate) fn advisory_table(vulns: &[&VulnRow]) -> String {
418 let mut table = "| Advisory | Severity | Affected | Fixed in | Summary |\n| --- | --- | --- | --- | --- |\n".to_owned();
419 let mut seen = BTreeSet::new();
420 for vuln in vulns {
421 if !seen.insert((vuln.advisory.clone(), vuln.version.clone())) {
422 continue;
423 }
424 table.push_str(&format!(
425 "| [{}]({}) | {} | {} | {} | {} |\n",
426 vuln.advisory,
427 osv::page_url(&vuln.osv_id),
428 vuln.severity,
429 vuln.version,
430 vuln.fixed_version.as_deref().unwrap_or("none yet"),
431 vuln.summary.replace('|', "\\|").replace('\n', " "),
432 ));
433 }
434 table
435}
436
437#[cfg(test)]
438mod tests {
439 use super::*;
440 use g1t_contracts::security::LockfileText;
441
442 #[test]
443 fn scans_cost_their_cpu_and_the_rows_they_write() {
444 // 10 OSV calls and 3 vulnerabilities: 0.4 of CPU, 15 rows.
445 assert_eq!(dependency_check_cost(10, 3), 16);
446 assert_eq!(dependency_check_cost(0, 0), 2);
447 // A page of 25 commits that read 100 objects and found nothing:
448 // 10 of CPU and 2 rows. The old placeholder charged 100.
449 assert_eq!(crate::history::history_page_cost(100, 0), 12);
450 assert_eq!(crate::history::history_page_cost(0, 1), 3);
451 }
452
453 #[test]
454 fn go_sum_is_skipped_beside_go_mod() {
455 let files = Lockfiles {
456 commit: None,
457 files: vec![
458 LockfileText { path: "go.mod".into(), text: "require golang.org/x/net v0.7.0\n".into() },
459 LockfileText { path: "go.sum".into(), text: "golang.org/x/net v0.1.0 h1:x=\n".into() },
460 LockfileText { path: "tools/go.sum".into(), text: "golang.org/x/text v0.3.0 h1:x=\n".into() },
461 ],
462 };
463 let found: Vec<String> = packages(&files).iter().map(|l| format!("{}:{}", l.path, l.package.version)).collect();
464 assert_eq!(found, ["go.mod:v0.7.0", "tools/go.sum:v0.3.0"]);
465 }
466
467 #[test]
468 fn the_issue_names_the_advisories_and_checks_the_lockfile() {
469 let row = VulnRow {
470 id: "vul_1".into(),
471 repo_id: "rep_1".into(),
472 ecosystem: "npm".into(),
473 package: "lodash".into(),
474 version: "4.17.20".into(),
475 manifest: "web/package-lock.json".into(),
476 osv_id: "GHSA-35jh-r3h4-6jhm".into(),
477 advisory: "GHSA-35jh-r3h4-6jhm".into(),
478 summary: "Command Injection in lodash".into(),
479 severity: "high".into(),
480 fixed_version: Some("4.17.21".into()),
481 status: "open".into(),
482 found_at: "2026-10-04T00:00:00Z".into(),
483 fixed_at: None,
484 number: None,
485 dismiss_reason: None,
486 dismiss_comment: None,
487 dismissed_by: None,
488 dismissed_at: None,
489 };
490 let located = vec![Located {
491 package: Package { ecosystem: g1t_scan::lockfiles::Ecosystem::Npm, name: "lodash".into(), version: "4.17.20".into() },
492 lockfile: Lockfile::PackageLock,
493 path: "web/package-lock.json".into(),
494 }];
495 let body = issue_text("npm", "lodash", "4.17.21", &[&row], &located);
496 assert!(body.contains("[GHSA-35jh-r3h4-6jhm](https://osv.dev/vulnerability/GHSA-35jh-r3h4-6jhm) | high | 4.17.20 | 4.17.21"));
497 assert!(body.contains("`web/package-lock.json` (4.17.20)"));
498 let (_, done) = body.split_once("## Definition of done\n\n").unwrap();
499 let items: Vec<&str> = done.lines().take_while(|line| line.starts_with("- ")).collect();
500 assert_eq!(items.len(), 3);
501 assert!(items[1].contains("node_modules/lodash") && items[1].contains("'web/package-lock.json'"));
502 assert_eq!(items[2], "- `cd 'web' && npm ci && npm test --if-present` passes.");
503 assert!(body.ends_with("on its Security page._"));
504 assert!(advisory_table(&[&row]).contains("| [GHSA-35jh-r3h4-6jhm](https://osv.dev/vulnerability/GHSA-35jh-r3h4-6jhm) | high |"));
505 }
506}