| 1 | //! Artifacts and the cache: `actions/upload-artifact`, its `merge`, |
| 2 | //! `actions/download-artifact` and `actions/cache`, done natively against |
| 3 | //! g1t. Artifacts belong to the run; cache entries to the repository, found |
| 4 | //! by exact key or by the newest under a `restore-keys` prefix. |
| 5 | //! |
| 6 | //! An artifact is a ZIP file, as GitHub's v4 actions make it (zip.rs), of |
| 7 | //! the files its `path` finds (glob.rs): uploaded in parts with its SHA-256, |
| 8 | //! and downloaded straight to a file before it is unpacked. |
| 9 | //! |
| 10 | //! A cache entry is a tar archive, compressed with zstd where the machine |
| 11 | //! has it (gzip otherwise), of up to 2 GB: uploaded in parts, and |
| 12 | //! downloaded straight to a file. Its `path` takes globs (`**` included) |
| 13 | //! and `!` patterns that leave paths out, as `actions/cache` does. Each is |
| 14 | //! saved and found under a version, the hash of its `path` and compression, |
| 15 | //! and g1t keeps it under the ref whose run saved it. |
| 16 | |
| 17 | use std::collections::BTreeMap; |
| 18 | use std::io::{Read, Write}; |
| 19 | use std::path::Path; |
| 20 | use std::process::Command; |
| 21 | use std::time::{Duration, Instant}; |
| 22 | |
| 23 | use sha2::{Digest, Sha256}; |
| 24 | |
| 25 | use super::process::{self, Commands, Ended}; |
| 26 | use super::{Job, glob, zip}; |
| 27 | |
| 28 | |
| 29 | /// How often a long transfer says how far it has got. |
| 30 | const PROGRESS_EVERY: Duration = Duration::from_secs(5); |
| 31 | |
| 32 | /// A transfer's progress as GitHub's cache action words it: |
| 33 | /// `Received 46137344 of 523190823 (8.8%), 43.9 MBs/sec`. |
| 34 | fn progress(verb: &str, done: u64, total: Option<u64>, took: Duration) -> String { |
| 35 | let rate = done as f64 / 1_048_576.0 / took.as_secs_f64().max(0.001); |
| 36 | match total.filter(|total| *total > 0) { |
| 37 | Some(total) => format!("{verb} {done} of {total} ({:.1}%), {rate:.1} MBs/sec", done as f64 * 100.0 / total as f64), |
| 38 | None => format!("{verb} {done} bytes, {rate:.1} MBs/sec"), |
| 39 | } |
| 40 | } |
| 41 | |
| 42 | fn lines(text: &str) -> Vec<String> { |
| 43 | text.lines().map(str::trim).filter(|line| !line.is_empty() && !line.starts_with('#')).map(str::to_owned).collect() |
| 44 | } |
| 45 | |
| 46 | fn quote(text: &str) -> String { |
| 47 | format!("'{}'", text.replace('\'', "'\\''")) |
| 48 | } |
| 49 | |
| 50 | impl Job { |
| 51 | fn url(&self, rest: &str) -> String { |
| 52 | format!("{}/actions/jobs/{}/{rest}", self.log.api.base, self.log.api.job) |
| 53 | } |
| 54 | |
| 55 | fn auth(&self) -> String { |
| 56 | format!("Bearer {}", self.log.api.token) |
| 57 | } |
| 58 | |
| 59 | /// Runs a shell line, logging its output; whether it succeeded. |
| 60 | fn shell(&mut self, script: &str) -> bool { |
| 61 | let mut command = Command::new("bash"); |
| 62 | command.args(["-c", script]).current_dir(&self.workspace); |
| 63 | let mut commands = Commands::default(); |
| 64 | matches!(process::run(command, Duration::from_secs(1800).min(self.remaining_time()), &mut self.log, &mut commands), Ok(Ended::Exited(0))) |
| 65 | } |
| 66 | |
| 67 | /// Reads a download to its end, handing `each` every piece, and says |
| 68 | /// how it is going every few seconds as GitHub's cache and artifact |
| 69 | /// actions do (`Received 46137344 of 523190823 (8.8%), 43.9 MBs/sec`), |
| 70 | /// so a long download is never a silent step. Its total in bytes. |
| 71 | fn receive(&mut self, reader: &mut dyn Read, total: Option<u64>, mut each: impl FnMut(&[u8]) -> std::io::Result<()>) -> Result<u64, String> { |
| 72 | let started = Instant::now(); |
| 73 | let mut told = Instant::now(); |
| 74 | let mut received = 0u64; |
| 75 | let mut buffer = vec![0u8; 256 * 1024]; |
| 76 | loop { |
| 77 | let n = reader.read(&mut buffer).map_err(|e| e.to_string())?; |
| 78 | if n == 0 { |
| 79 | break; |
| 80 | } |
| 81 | each(&buffer[..n]).map_err(|e| e.to_string())?; |
| 82 | received += n as u64; |
| 83 | if told.elapsed() >= PROGRESS_EVERY { |
| 84 | told = Instant::now(); |
| 85 | self.log.line(&progress("Received", received, total, started.elapsed())); |
| 86 | } |
| 87 | self.log.tick(); |
| 88 | } |
| 89 | Ok(received) |
| 90 | } |
| 91 | |
| 92 | /// Sends `file` in parts of `part_bytes`; each part's number and etag. |
| 93 | fn send_parts(&mut self, base: &str, upload: &str, file: &Path, part_bytes: u64) -> Result<Vec<serde_json::Value>, String> { |
| 94 | let mut reader = std::fs::File::open(file).map_err(|e| e.to_string())?; |
| 95 | let size = reader.metadata().map(|m| m.len()).unwrap_or(0); |
| 96 | let (started, mut told, mut sent) = (Instant::now(), Instant::now(), 0u64); |
| 97 | let mut parts = Vec::new(); |
| 98 | let mut buffer = vec![0u8; part_bytes as usize]; |
| 99 | for number in 1u32.. { |
| 100 | let mut filled = 0; |
| 101 | while filled < buffer.len() { |
| 102 | let read = reader.read(&mut buffer[filled..]).map_err(|e| e.to_string())?; |
| 103 | if read == 0 { |
| 104 | break; |
| 105 | } |
| 106 | filled += read; |
| 107 | } |
| 108 | if filled == 0 && number > 1 { |
| 109 | break; |
| 110 | } |
| 111 | let url = self.url(&format!("{base}/{number}?upload={}", urlencode(upload))); |
| 112 | // A part that fails is sent again, twice at most. |
| 113 | let mut tries = 0; |
| 114 | let answer = loop { |
| 115 | tries += 1; |
| 116 | match ureq::put(&url).set("authorization", &self.auth()).timeout(Duration::from_secs(600)).send_bytes(&buffer[..filled]) { |
| 117 | Ok(response) => break response.into_json::<serde_json::Value>().map_err(|e| e.to_string())?, |
| 118 | Err(ureq::Error::Status(status, response)) if status < 500 => return Err(refusal(response)), |
| 119 | Err(error) if tries >= 3 => return Err(error.to_string()), |
| 120 | Err(_) => std::thread::sleep(Duration::from_secs(2 * tries)), |
| 121 | } |
| 122 | }; |
| 123 | parts.push(serde_json::json!({ "part": number, "etag": answer["etag"] })); |
| 124 | sent += filled as u64; |
| 125 | if filled < buffer.len() { |
| 126 | break; |
| 127 | } |
| 128 | // More to send: say how far it has got, as a download does. |
| 129 | if told.elapsed() >= PROGRESS_EVERY { |
| 130 | told = Instant::now(); |
| 131 | self.log.line(&progress("Sent", sent, Some(size), started.elapsed())); |
| 132 | } |
| 133 | self.log.tick(); |
| 134 | } |
| 135 | Ok(parts) |
| 136 | } |
| 137 | |
| 138 | /// A value of the `github` context, as text. |
| 139 | fn github_value(&self, key: &str) -> String { |
| 140 | match self.contexts.get("github").and_then(|github| github.get(key)) { |
| 141 | Some(serde_json::Value::String(text)) => text.clone(), |
| 142 | Some(serde_json::Value::Null) | None => String::new(), |
| 143 | Some(other) => other.to_string(), |
| 144 | } |
| 145 | } |
| 146 | |
| 147 | fn home(&self) -> String { |
| 148 | self.base_env_value("HOME").unwrap_or_else(|| "/home/node".into()) |
| 149 | } |
| 150 | |
| 151 | /// `actions/upload-artifact`: the files `path` names, packed into one |
| 152 | /// ZIP and uploaded in parts. |
| 153 | pub(crate) fn upload_artifact(&mut self, with: &BTreeMap<String, String>) -> (bool, BTreeMap<String, String>) { |
| 154 | let name = input(with, "name").unwrap_or("artifact").to_owned(); |
| 155 | let missing = input(with, "if-no-files-found").unwrap_or("warn").to_ascii_lowercase(); |
| 156 | let checked = check_name(&name) |
| 157 | .and_then(|()| keep(with, true)) |
| 158 | .and_then(|keep| if matches!(missing.as_str(), "warn" | "error" | "ignore") { Ok(keep) } else { Err(format!("`if-no-files-found` is `{missing}`; it takes warn, error or ignore.")) }); |
| 159 | let keep = match checked { |
| 160 | Ok(keep) => keep, |
| 161 | Err(message) => { |
| 162 | self.log.line(&format!("##[error]{message}")); |
| 163 | return (false, BTreeMap::new()); |
| 164 | } |
| 165 | }; |
| 166 | let paths = lines(input(with, "path").unwrap_or_default()); |
| 167 | let found = glob::find(&paths, &self.workspace.display().to_string(), &self.home(), keep.hidden); |
| 168 | if found.files.is_empty() { |
| 169 | let message = format!("No files were found with the provided path: {}. No artifacts will be uploaded.", paths.join(", ")); |
| 170 | return match missing.as_str() { |
| 171 | "error" => { |
| 172 | self.log.line(&format!("##[error]{message}")); |
| 173 | (false, BTreeMap::new()) |
| 174 | } |
| 175 | "ignore" => { |
| 176 | self.log.line(&message); |
| 177 | (true, BTreeMap::new()) |
| 178 | } |
| 179 | _ => { |
| 180 | self.log.line(&format!("##[warning]{message}")); |
| 181 | (true, BTreeMap::new()) |
| 182 | } |
| 183 | }; |
| 184 | } |
| 185 | self.log.line(&format!("Found {} with the provided path, stored relative to {}.", count(found.files.len(), "file"), found.root)); |
| 186 | self.send_artifact(&name, &found.files, &keep) |
| 187 | } |
| 188 | |
| 189 | /// Packs `files` into a ZIP and uploads it as the artifact `name`; |
| 190 | /// whether it worked, and the outputs `upload-artifact` sets. |
| 191 | fn send_artifact(&mut self, name: &str, files: &[(String, std::path::PathBuf)], keep: &Keep) -> (bool, BTreeMap<String, String>) { |
| 192 | let archive = self.temp.join(format!("artifact-{}.zip", super::rand_id())); |
| 193 | let sent = zip::write(&archive, files, keep.level) |
| 194 | .map_err(|e| format!("The files could not be packed: {e}")) |
| 195 | .and_then(|packed| self.post_artifact(name, &archive, keep).map(|done| (packed, done))); |
| 196 | let _ = std::fs::remove_file(&archive); |
| 197 | let (packed, done) = match sent { |
| 198 | Ok(sent) => sent, |
| 199 | Err(error) => { |
| 200 | self.log.line(&format!("##[error]{error}")); |
| 201 | return (false, BTreeMap::new()); |
| 202 | } |
| 203 | }; |
| 204 | let kept = done.retention.map(|days| format!(" It is kept for {}.", count(days as usize, "day"))).unwrap_or_default(); |
| 205 | self.log.line(&format!("Uploaded artifact {name} ({}, {}).{kept}", megabytes(packed.size), count(packed.files, "file"))); |
| 206 | let url = format!( |
| 207 | "{}/{}/actions/runs/{}/artifacts/{}", |
| 208 | self.github_value("server_url").trim_end_matches('/'), |
| 209 | self.github_value("repository"), |
| 210 | self.github_value("run_id"), |
| 211 | done.id |
| 212 | ); |
| 213 | self.log.line(&format!("Artifact download URL: {url}")); |
| 214 | let mut outputs = BTreeMap::new(); |
| 215 | outputs.insert("artifact-id".into(), done.id.to_string()); |
| 216 | outputs.insert("artifact-url".into(), url); |
| 217 | outputs.insert("artifact-digest".into(), done.digest); |
| 218 | (true, outputs) |
| 219 | } |
| 220 | |
| 221 | /// Uploads the ZIP file `archive` as the artifact `name`, in parts, |
| 222 | /// giving the upload up when a part or the end fails. |
| 223 | fn post_artifact(&mut self, name: &str, archive: &Path, keep: &Keep) -> Result<Sent, String> { |
| 224 | let size = std::fs::metadata(archive).map_err(|e| e.to_string())?.len(); |
| 225 | let digest = sha256_file(archive).map_err(|e| e.to_string())?; |
| 226 | let query = format!( |
| 227 | "artifacts/uploads?name={}&size={size}&retention_days={}&overwrite={}&format=zip", |
| 228 | urlencode(name), |
| 229 | keep.retention, |
| 230 | keep.overwrite |
| 231 | ); |
| 232 | let failed = |e: ureq::Error| match e { |
| 233 | ureq::Error::Status(_, response) => format!("The artifact could not be uploaded: {}", refusal(response)), |
| 234 | other => format!("The artifact could not be uploaded: {other}"), |
| 235 | }; |
| 236 | let started = ureq::post(&self.url(&query)) |
| 237 | .set("authorization", &self.auth()) |
| 238 | .timeout(Duration::from_secs(60)) |
| 239 | .call() |
| 240 | .map_err(failed)? |
| 241 | .into_json::<serde_json::Value>() |
| 242 | .map_err(|e| e.to_string())?; |
| 243 | let id = number(&started["id"]).ok_or("g1t did not say which artifact the upload is")?; |
| 244 | let upload = started["upload"].as_str().unwrap_or_default().to_owned(); |
| 245 | let part_bytes = started["part_bytes"].as_u64().filter(|n| *n > 0).unwrap_or(32 * 1024 * 1024); |
| 246 | let retention = started["retention_days"].as_u64(); |
| 247 | if let Some(applied) = retention |
| 248 | && keep.retention != 0 |
| 249 | && applied != u64::from(keep.retention) |
| 250 | { |
| 251 | self.log.line(&format!( |
| 252 | "##[notice]The artifact is kept for {}, not the {} asked for: the most this repository keeps artifacts.", |
| 253 | count(applied as usize, "day"), |
| 254 | keep.retention |
| 255 | )); |
| 256 | } |
| 257 | let base = format!("artifacts/uploads/{id}"); |
| 258 | let give_up = |job: &Job| { |
| 259 | let _ = ureq::delete(&job.url(&format!("{base}?upload={}", urlencode(&upload)))).set("authorization", &job.auth()).call(); |
| 260 | }; |
| 261 | let parts = match self.send_parts(&base, &upload, archive, part_bytes) { |
| 262 | Ok(parts) => parts, |
| 263 | Err(error) => { |
| 264 | give_up(self); |
| 265 | return Err(format!("The artifact could not be uploaded: {error}")); |
| 266 | } |
| 267 | }; |
| 268 | let done = ureq::post(&self.url(&format!("{base}/complete?upload={}", urlencode(&upload)))) |
| 269 | .set("authorization", &self.auth()) |
| 270 | .timeout(Duration::from_secs(120)) |
| 271 | .send_json(serde_json::json!({ "size": size, "parts": parts, "digest": format!("sha256:{digest}") })); |
| 272 | let done = match done { |
| 273 | Ok(response) => response.into_json::<serde_json::Value>().unwrap_or_default(), |
| 274 | Err(error) => { |
| 275 | give_up(self); |
| 276 | return Err(failed(error)); |
| 277 | } |
| 278 | }; |
| 279 | Ok(Sent { id: number(&done["id"]).unwrap_or(id), digest, retention: retention.map(|d| d as u32).or((keep.retention != 0).then_some(keep.retention)) }) |
| 280 | } |
| 281 | |
| 282 | /// The artifacts of this run, or of the run `run_id` of this |
| 283 | /// repository; one per name, the newest. |
| 284 | fn list_artifacts(&mut self, run_id: Option<&str>) -> Result<Vec<Listed>, String> { |
| 285 | let rest = match run_id { |
| 286 | Some(run) => format!("artifacts?run_id={}", urlencode(run)), |
| 287 | None => "artifacts".to_owned(), |
| 288 | }; |
| 289 | let listed = ureq::get(&self.url(&rest)) |
| 290 | .set("authorization", &self.auth()) |
| 291 | .timeout(Duration::from_secs(60)) |
| 292 | .call() |
| 293 | .map_err(|e| match e { |
| 294 | ureq::Error::Status(_, response) => refusal(response), |
| 295 | other => other.to_string(), |
| 296 | })? |
| 297 | .into_json::<Vec<serde_json::Value>>() |
| 298 | .map_err(|e| e.to_string())?; |
| 299 | Ok(newest(listed.iter().filter_map(Listed::from_json).collect())) |
| 300 | } |
| 301 | |
| 302 | /// Downloads an artifact to a file, as it comes, and unpacks it into |
| 303 | /// `into`. |
| 304 | fn fetch_artifact(&mut self, artifact: &Listed, into: &Path) -> Result<(), String> { |
| 305 | let file = self.temp.join(format!("download-{}.zip", super::rand_id())); |
| 306 | let fetched = self.fetch_to(artifact, &file).and_then(|format| { |
| 307 | if format == "tgz" { |
| 308 | std::fs::create_dir_all(into).map_err(|e| e.to_string())?; |
| 309 | let script = format!("tar{} -xzf {} -C {}", tar_local(), quote(&file.display().to_string()), quote(&into.display().to_string())); |
| 310 | if self.shell(&script) { Ok(()) } else { Err("it could not be unpacked".into()) } |
| 311 | } else { |
| 312 | zip::extract(&file, into).map(|_| ()).map_err(|e| format!("it could not be unpacked: {e}")) |
| 313 | } |
| 314 | }); |
| 315 | let _ = std::fs::remove_file(&file); |
| 316 | fetched |
| 317 | } |
| 318 | |
| 319 | /// Downloads an artifact into `file`, checking its digest; how it is |
| 320 | /// packed, `zip` or `tgz`. |
| 321 | fn fetch_to(&mut self, artifact: &Listed, file: &Path) -> Result<String, String> { |
| 322 | let response = match ureq::get(&self.url(&format!("artifacts/{}/download", artifact.id))) |
| 323 | .set("authorization", &self.auth()) |
| 324 | .timeout(Duration::from_secs(4 * 3600)) |
| 325 | .call() |
| 326 | { |
| 327 | Ok(response) => response, |
| 328 | Err(ureq::Error::Status(404, _)) => return Err("it is gone: it expired or was deleted".into()), |
| 329 | Err(ureq::Error::Status(_, response)) => return Err(refusal(response)), |
| 330 | Err(error) => return Err(error.to_string()), |
| 331 | }; |
| 332 | let format = response.header("x-g1t-format").map(str::to_owned).unwrap_or_else(|| artifact.format.clone()); |
| 333 | let total = response.header("content-length").and_then(|n| n.parse::<u64>().ok()).or(Some(artifact.size)); |
| 334 | let mut reader = response.into_reader(); |
| 335 | let mut out = std::io::BufWriter::new(std::fs::File::create(file).map_err(|e| e.to_string())?); |
| 336 | let mut hasher = Sha256::new(); |
| 337 | self.receive(&mut reader, total, |bytes| { |
| 338 | hasher.update(bytes); |
| 339 | out.write_all(bytes) |
| 340 | })?; |
| 341 | out.flush().map_err(|e| e.to_string())?; |
| 342 | let got = hex::encode(hasher.finalize()); |
| 343 | if let Some(expected) = artifact.digest.as_deref().map(|d| d.trim_start_matches("sha256:")) |
| 344 | && !expected.is_empty() |
| 345 | && !expected.eq_ignore_ascii_case(&got) |
| 346 | { |
| 347 | self.log.line(&format!("##[warning]Artifact {} does not match its digest (sha256:{got}, not sha256:{expected}): it may have been changed since it was uploaded.", artifact.name)); |
| 348 | } |
| 349 | Ok(format) |
| 350 | } |
| 351 | |
| 352 | /// Where artifacts are downloaded: `path`, under the workspace unless |
| 353 | /// absolute, `~` the home folder. |
| 354 | fn download_dir(&self, path: Option<&str>) -> std::path::PathBuf { |
| 355 | match path { |
| 356 | None => self.workspace.clone(), |
| 357 | Some(path) => { |
| 358 | let home = self.home(); |
| 359 | let path = if path == "~" { |
| 360 | home |
| 361 | } else if let Some(rest) = path.strip_prefix("~/") { |
| 362 | format!("{}/{rest}", home.trim_end_matches('/')) |
| 363 | } else { |
| 364 | path.to_owned() |
| 365 | }; |
| 366 | self.workspace.join(path) |
| 367 | } |
| 368 | } |
| 369 | } |
| 370 | |
| 371 | /// `actions/download-artifact`: one artifact by name straight into |
| 372 | /// `path`, or the run's artifacts (by `pattern` or `artifact-ids`) each |
| 373 | /// into a folder of its name, or all into `path` with `merge-multiple`. |
| 374 | pub(crate) fn download_artifact(&mut self, with: &BTreeMap<String, String>) -> (bool, BTreeMap<String, String>) { |
| 375 | let fail = |job: &mut Job, message: &str| { |
| 376 | job.log.line(&format!("##[error]{message}")); |
| 377 | (false, BTreeMap::new()) |
| 378 | }; |
| 379 | let name = input(with, "name"); |
| 380 | let ids = match artifact_ids(input(with, "artifact-ids").unwrap_or_default()) { |
| 381 | Ok(ids) => ids, |
| 382 | Err(message) => return fail(self, &message), |
| 383 | }; |
| 384 | if name.is_some() && !ids.is_empty() { |
| 385 | return fail(self, "Inputs 'name' and 'artifact-ids' cannot be used together. Please specify only one."); |
| 386 | } |
| 387 | let merge = match flag(with, "merge-multiple", false) { |
| 388 | Ok(merge) => merge, |
| 389 | Err(message) => return fail(self, &message), |
| 390 | }; |
| 391 | let dest = self.download_dir(input(with, "path")); |
| 392 | // Another run's artifacts, as v4 has it: with a token and a run id. |
| 393 | let run_id = match (input(with, "github-token"), input(with, "run-id")) { |
| 394 | (Some(_), Some(run)) => { |
| 395 | let own = self.github_value("repository"); |
| 396 | let repository = input(with, "repository").unwrap_or(&own); |
| 397 | if !repository.eq_ignore_ascii_case(&own) { |
| 398 | return fail(self, &format!("g1t downloads artifacts from other runs of the same repository only; {repository} is not this run's repository, {own}.")); |
| 399 | } |
| 400 | Some(run.to_owned()) |
| 401 | } |
| 402 | _ => None, |
| 403 | }; |
| 404 | let listed = match self.list_artifacts(run_id.as_deref()) { |
| 405 | Ok(listed) => listed, |
| 406 | Err(error) => return fail(self, &format!("The artifacts could not be listed: {error}")), |
| 407 | }; |
| 408 | let chosen: Vec<Listed> = if let Some(name) = name { |
| 409 | match listed.into_iter().find(|a| a.name == name) { |
| 410 | Some(artifact) => vec![artifact], |
| 411 | None => return fail(self, &format!("Unable to download artifact(s): Artifact not found for name: {name}")), |
| 412 | } |
| 413 | } else if !ids.is_empty() { |
| 414 | let chosen: Vec<Listed> = listed.into_iter().filter(|a| ids.contains(&a.id)).collect(); |
| 415 | if chosen.is_empty() { |
| 416 | return fail(self, "Unable to download artifact(s): None of the provided artifact IDs were found."); |
| 417 | } |
| 418 | if chosen.len() < ids.len() { |
| 419 | self.log.line(&format!("##[warning]Could only find {} of {} artifact IDs.", chosen.len(), ids.len())); |
| 420 | } |
| 421 | chosen |
| 422 | } else { |
| 423 | match input(with, "pattern") { |
| 424 | Some(pattern) => listed.into_iter().filter(|a| glob::matches_part(pattern, &a.name)).collect(), |
| 425 | None => listed, |
| 426 | } |
| 427 | }; |
| 428 | let mut outputs = BTreeMap::new(); |
| 429 | outputs.insert("download-path".into(), dest.display().to_string()); |
| 430 | if chosen.is_empty() { |
| 431 | match input(with, "pattern") { |
| 432 | Some(pattern) => self.log.line(&format!("No artifacts matched the pattern {pattern}; nothing was downloaded.")), |
| 433 | None => self.log.line("The run has no artifacts; nothing was downloaded."), |
| 434 | } |
| 435 | return (true, outputs); |
| 436 | } |
| 437 | let targets = download_targets(&dest, &chosen, name.is_some(), !ids.is_empty(), merge); |
| 438 | for (artifact, target) in chosen.iter().zip(targets) { |
| 439 | let Some(target) = target else { |
| 440 | return fail(self, &format!("The artifact {} cannot be downloaded into a folder of its name.", artifact.name)); |
| 441 | }; |
| 442 | if let Err(error) = self.fetch_artifact(artifact, &target) { |
| 443 | return fail(self, &format!("The artifact {} could not be downloaded: {error}", artifact.name)); |
| 444 | } |
| 445 | self.log.line(&format!("Downloaded artifact {} ({}) into {}", artifact.name, megabytes(artifact.size), target.display())); |
| 446 | } |
| 447 | if chosen.len() > 1 { |
| 448 | self.log.line(&format!("Downloaded {}.", count(chosen.len(), "artifact"))); |
| 449 | } |
| 450 | (true, outputs) |
| 451 | } |
| 452 | |
| 453 | /// `actions/upload-artifact/merge`: the run's artifacts matching |
| 454 | /// `pattern`, downloaded together and uploaded again as one. |
| 455 | pub(crate) fn merge_artifacts(&mut self, with: &BTreeMap<String, String>) -> (bool, BTreeMap<String, String>) { |
| 456 | let fail = |job: &mut Job, message: &str| { |
| 457 | job.log.line(&format!("##[error]{message}")); |
| 458 | (false, BTreeMap::new()) |
| 459 | }; |
| 460 | let name = input(with, "name").unwrap_or("merged-artifacts").to_owned(); |
| 461 | let pattern = input(with, "pattern").unwrap_or("*").to_owned(); |
| 462 | let options = check_name(&name).and_then(|()| { |
| 463 | let keep = keep(with, false)?; |
| 464 | Ok((keep, flag(with, "separate-directories", false)?, flag(with, "delete-merged", false)?)) |
| 465 | }); |
| 466 | let (keep, separate, delete) = match options { |
| 467 | Ok(options) => options, |
| 468 | Err(message) => return fail(self, &message), |
| 469 | }; |
| 470 | let listed = match self.list_artifacts(None) { |
| 471 | Ok(listed) => listed, |
| 472 | Err(error) => return fail(self, &format!("The artifacts could not be listed: {error}")), |
| 473 | }; |
| 474 | let chosen: Vec<Listed> = listed.into_iter().filter(|a| glob::matches_part(&pattern, &a.name)).collect(); |
| 475 | if chosen.is_empty() { |
| 476 | return fail(self, &format!("No artifacts found matching pattern '{pattern}'.")); |
| 477 | } |
| 478 | self.log.line(&format!("Merging {}: {}", count(chosen.len(), "artifact"), chosen.iter().map(|a| a.name.as_str()).collect::<Vec<_>>().join(", "))); |
| 479 | let folder = self.temp.join(format!("merge-{}", super::rand_id())); |
| 480 | let targets = download_targets(&folder, &chosen, false, false, !separate); |
| 481 | let mut result = None; |
| 482 | for (artifact, target) in chosen.iter().zip(targets) { |
| 483 | let Some(target) = target else { |
| 484 | result = Some(fail(self, &format!("The artifact {} cannot be merged into a folder of its name.", artifact.name))); |
| 485 | break; |
| 486 | }; |
| 487 | if let Err(error) = self.fetch_artifact(artifact, &target) { |
| 488 | result = Some(fail(self, &format!("The artifact {} could not be downloaded: {error}", artifact.name))); |
| 489 | break; |
| 490 | } |
| 491 | } |
| 492 | let result = result.unwrap_or_else(|| { |
| 493 | let root = folder.display().to_string(); |
| 494 | let found = glob::find(std::slice::from_ref(&root), &root, &self.home(), keep.hidden); |
| 495 | if found.files.is_empty() { |
| 496 | fail(self, "The artifacts matched hold no files to merge.") |
| 497 | } else { |
| 498 | self.send_artifact(&name, &found.files, &keep) |
| 499 | } |
| 500 | }); |
| 501 | let _ = std::fs::remove_dir_all(&folder); |
| 502 | if result.0 && delete { |
| 503 | for artifact in &chosen { |
| 504 | match ureq::delete(&self.url(&format!("artifacts/{}", artifact.id))).set("authorization", &self.auth()).timeout(Duration::from_secs(60)).call() { |
| 505 | Ok(_) => self.log.line(&format!("Deleted artifact {}.", artifact.name)), |
| 506 | Err(ureq::Error::Status(_, response)) => self.log.line(&format!("##[warning]Artifact {} could not be deleted: {}", artifact.name, refusal(response))), |
| 507 | Err(error) => self.log.line(&format!("##[warning]Artifact {} could not be deleted: {error}", artifact.name)), |
| 508 | } |
| 509 | } |
| 510 | } |
| 511 | result |
| 512 | } |
| 513 | |
| 514 | } |
| 515 | |
| 516 | /// An input, trimmed; `None` when empty. |
| 517 | fn input<'a>(with: &'a BTreeMap<String, String>, key: &str) -> Option<&'a str> { |
| 518 | with.get(key).map(|v| v.trim()).filter(|v| !v.is_empty()) |
| 519 | } |
| 520 | |
| 521 | /// A true-or-false input, as `core.getBooleanInput` reads it. |
| 522 | fn flag(with: &BTreeMap<String, String>, key: &str, default: bool) -> Result<bool, String> { |
| 523 | match input(with, key) { |
| 524 | None => Ok(default), |
| 525 | Some("true" | "True" | "TRUE") => Ok(true), |
| 526 | Some("false" | "False" | "FALSE") => Ok(false), |
| 527 | Some(other) => Err(format!("`{key}` is `{other}`; it takes true or false.")), |
| 528 | } |
| 529 | } |
| 530 | |
| 531 | /// `1 file`, `3 files`. |
| 532 | fn count(n: usize, what: &str) -> String { |
| 533 | if n == 1 { format!("1 {what}") } else { format!("{n} {what}s") } |
| 534 | } |
| 535 | |
| 536 | /// An id that came as a number or as text. |
| 537 | fn number(value: &serde_json::Value) -> Option<u64> { |
| 538 | value.as_u64().or_else(|| value.as_str().and_then(|s| s.parse().ok())) |
| 539 | } |
| 540 | |
| 541 | /// Whether a name is one GitHub takes for an artifact: not empty, at most |
| 542 | /// 256 characters, none of `" : < > | * ? \ /` or a line break. |
| 543 | fn check_name(name: &str) -> Result<(), String> { |
| 544 | if name.is_empty() { |
| 545 | return Err("The artifact needs a name.".into()); |
| 546 | } |
| 547 | if name.chars().count() > 256 { |
| 548 | return Err(format!("The artifact name `{name}` is longer than 256 characters.")); |
| 549 | } |
| 550 | if let Some(bad) = name.chars().find(|c| matches!(c, '"' | ':' | '<' | '>' | '|' | '*' | '?' | '\r' | '\n' | '\\' | '/')) { |
| 551 | let shown = match bad { |
| 552 | '\r' => "a carriage return".to_owned(), |
| 553 | '\n' => "a line break".to_owned(), |
| 554 | c => format!("`{c}`"), |
| 555 | }; |
| 556 | return Err(format!("The artifact name `{name}` is not valid: it contains {shown}. A name may not contain \" : < > | * ? \\ / or line breaks.")); |
| 557 | } |
| 558 | Ok(()) |
| 559 | } |
| 560 | |
| 561 | /// How an artifact is packed and kept, from the inputs `upload-artifact` |
| 562 | /// and its merge share. |
| 563 | #[derive(Debug, PartialEq)] |
| 564 | struct Keep { |
| 565 | /// Days; 0 for the repository's own setting. |
| 566 | retention: u32, |
| 567 | /// 0 (stored) to 9. |
| 568 | level: u32, |
| 569 | overwrite: bool, |
| 570 | /// Whether files and folders whose names start with `.` are taken. |
| 571 | hidden: bool, |
| 572 | } |
| 573 | |
| 574 | fn keep(with: &BTreeMap<String, String>, takes_overwrite: bool) -> Result<Keep, String> { |
| 575 | let retention = match input(with, "retention-days") { |
| 576 | None => 0, |
| 577 | Some(text) => match text.parse::<u32>() { |
| 578 | Ok(days @ 0..=90) => days, |
| 579 | _ => return Err(format!("`retention-days` is `{text}`; it takes a number of days from 1 to 90, or nothing for the repository's setting.")), |
| 580 | }, |
| 581 | }; |
| 582 | let level = match input(with, "compression-level") { |
| 583 | None => 6, |
| 584 | Some(text) => match text.parse::<u32>() { |
| 585 | Ok(level @ 0..=9) => level, |
| 586 | _ => return Err(format!("`compression-level` is `{text}`; it takes 0 (no compression) to 9.")), |
| 587 | }, |
| 588 | }; |
| 589 | Ok(Keep { |
| 590 | retention, |
| 591 | level, |
| 592 | overwrite: takes_overwrite && flag(with, "overwrite", false)?, |
| 593 | hidden: flag(with, "include-hidden-files", false)?, |
| 594 | }) |
| 595 | } |
| 596 | |
| 597 | /// What finishing an upload gave: the artifact's id, the ZIP's SHA-256 in |
| 598 | /// hex, and the days it is kept, when known. |
| 599 | struct Sent { |
| 600 | id: u64, |
| 601 | digest: String, |
| 602 | retention: Option<u32>, |
| 603 | } |
| 604 | |
| 605 | /// An artifact as g1t lists it. |
| 606 | #[derive(Debug, Clone, PartialEq)] |
| 607 | struct Listed { |
| 608 | id: u64, |
| 609 | name: String, |
| 610 | size: u64, |
| 611 | digest: Option<String>, |
| 612 | /// `zip`, or `tgz` for an artifact an older runner stored. |
| 613 | format: String, |
| 614 | } |
| 615 | |
| 616 | impl Listed { |
| 617 | fn from_json(value: &serde_json::Value) -> Option<Listed> { |
| 618 | Some(Listed { |
| 619 | id: number(&value["id"])?, |
| 620 | name: value["name"].as_str()?.to_owned(), |
| 621 | size: value["size"].as_u64().unwrap_or(0), |
| 622 | digest: value["digest"].as_str().map(str::to_owned), |
| 623 | format: value["format"].as_str().unwrap_or("zip").to_owned(), |
| 624 | }) |
| 625 | } |
| 626 | } |
| 627 | |
| 628 | /// One artifact per name, the newest (highest id), in name order. |
| 629 | fn newest(listed: Vec<Listed>) -> Vec<Listed> { |
| 630 | let mut by_name: BTreeMap<String, Listed> = BTreeMap::new(); |
| 631 | for artifact in listed { |
| 632 | if by_name.get(&artifact.name).is_none_or(|kept| kept.id < artifact.id) { |
| 633 | by_name.insert(artifact.name.clone(), artifact); |
| 634 | } |
| 635 | } |
| 636 | by_name.into_values().collect() |
| 637 | } |
| 638 | |
| 639 | /// `artifact-ids`: numbers, separated by commas. |
| 640 | fn artifact_ids(text: &str) -> Result<Vec<u64>, String> { |
| 641 | text.split(',') |
| 642 | .map(str::trim) |
| 643 | .filter(|id| !id.is_empty()) |
| 644 | .map(|id| id.parse::<u64>().map_err(|_| format!("`artifact-ids` takes artifact ids, numbers separated by commas; `{id}` is not one."))) |
| 645 | .collect() |
| 646 | } |
| 647 | |
| 648 | /// The folder each artifact is unpacked into: `dest` itself for one |
| 649 | /// artifact by name, for `merge-multiple`, and for a single artifact by |
| 650 | /// id; otherwise a folder of the artifact's name in `dest`. `None` for a |
| 651 | /// name that cannot be a folder. |
| 652 | fn download_targets(dest: &Path, chosen: &[Listed], by_name: bool, by_ids: bool, merge: bool) -> Vec<Option<std::path::PathBuf>> { |
| 653 | let straight = by_name || merge || (by_ids && chosen.len() == 1); |
| 654 | chosen |
| 655 | .iter() |
| 656 | .map(|artifact| { |
| 657 | if straight { |
| 658 | Some(dest.to_path_buf()) |
| 659 | } else if artifact.name.is_empty() || artifact.name == "." || artifact.name == ".." || artifact.name.contains(['/', '\\']) { |
| 660 | None |
| 661 | } else { |
| 662 | Some(dest.join(&artifact.name)) |
| 663 | } |
| 664 | }) |
| 665 | .collect() |
| 666 | } |
| 667 | |
| 668 | /// The SHA-256 of a file, in hex, read as it comes. |
| 669 | fn sha256_file(path: &Path) -> std::io::Result<String> { |
| 670 | let mut file = std::fs::File::open(path)?; |
| 671 | let mut hasher = Sha256::new(); |
| 672 | let mut buffer = vec![0u8; 256 * 1024]; |
| 673 | loop { |
| 674 | let n = file.read(&mut buffer)?; |
| 675 | if n == 0 { |
| 676 | break; |
| 677 | } |
| 678 | hasher.update(&buffer[..n]); |
| 679 | } |
| 680 | Ok(hex::encode(hasher.finalize())) |
| 681 | } |
| 682 | |
| 683 | /// The message of a refused request, from its JSON body. |
| 684 | fn refusal(response: ureq::Response) -> String { |
| 685 | let status = response.status(); |
| 686 | let body: serde_json::Value = response.into_json().unwrap_or_default(); |
| 687 | body["error"]["message"].as_str().map_or_else(|| format!("g1t answered {status}"), str::to_owned) |
| 688 | } |
| 689 | |
| 690 | fn megabytes(bytes: u64) -> String { |
| 691 | if bytes < 1_048_576 { format!("{} KB", bytes.div_ceil(1024)) } else { format!("{:.1} MB", bytes as f64 / 1_048_576.0) } |
| 692 | } |
| 693 | |
| 694 | /// GNU tar on Windows reads `C:...` as a host and a file; `--force-local` |
| 695 | /// makes every path a local one. Elsewhere the flag is left off, since |
| 696 | /// bsdtar (macOS) doesn't know it. |
| 697 | pub(crate) fn tar_local() -> &'static str { |
| 698 | if cfg!(windows) { |
| 699 | " --force-local" |
| 700 | } else { |
| 701 | "" |
| 702 | } |
| 703 | } |
| 704 | |
| 705 | fn urlencode(text: &str) -> String { |
| 706 | let mut out = String::new(); |
| 707 | for byte in text.bytes() { |
| 708 | if byte.is_ascii_alphanumeric() || matches!(byte, b'-' | b'_' | b'.' | b'~') { |
| 709 | out.push(byte as char); |
| 710 | } else { |
| 711 | out.push_str(&format!("%{byte:02X}")); |
| 712 | } |
| 713 | } |
| 714 | out |
| 715 | } |
| 716 | |
| 717 | #[cfg(test)] |
| 718 | mod tests { |
| 719 | use super::*; |
| 720 | |
| 721 | #[test] |
| 722 | fn transfers_say_how_far_they_have_got_as_githubs_do() { |
| 723 | assert_eq!(progress("Received", 46_137_344, Some(523_190_823), Duration::from_secs(1)), "Received 46137344 of 523190823 (8.8%), 44.0 MBs/sec"); |
| 724 | assert_eq!(progress("Sent", 1_048_576, None, Duration::from_secs(2)), "Sent 1048576 bytes, 0.5 MBs/sec"); |
| 725 | assert_eq!(progress("Received", 10, Some(0), Duration::ZERO), "Received 10 bytes, 0.0 MBs/sec"); |
| 726 | } |
| 727 | |
| 728 | #[test] |
| 729 | fn keys_are_encoded_and_names_checked() { |
| 730 | assert_eq!(urlencode("Linux-node-abc/1 2"), "Linux-node-abc%2F1%202"); |
| 731 | assert!(check_name("coverage report").is_ok()); |
| 732 | assert!(check_name("dist-linux_x64.1 (debug)").is_ok()); |
| 733 | assert!(check_name("ünïcødé").is_ok()); |
| 734 | assert!(check_name(&"a".repeat(256)).is_ok()); |
| 735 | assert!(check_name(&"a".repeat(257)).is_err()); |
| 736 | assert!(check_name("").is_err()); |
| 737 | for bad in ["a/b", "a\\b", "a:b", "a*b", "a?b", "a\"b", "a<b", "a>b", "a|b", "a\nb", "a\rb"] { |
| 738 | assert!(check_name(bad).is_err(), "{bad}"); |
| 739 | } |
| 740 | assert_eq!(lines("dist/\n\n# note\n coverage \n"), ["dist/", "coverage"]); |
| 741 | } |
| 742 | |
| 743 | fn with(pairs: &[(&str, &str)]) -> BTreeMap<String, String> { |
| 744 | pairs.iter().map(|(k, v)| (k.to_string(), v.to_string())).collect() |
| 745 | } |
| 746 | |
| 747 | #[test] |
| 748 | fn upload_inputs_are_read_as_v4_reads_them() { |
| 749 | assert_eq!(keep(&with(&[]), true).unwrap(), Keep { retention: 0, level: 6, overwrite: false, hidden: false }); |
| 750 | let set = with(&[("retention-days", "14"), ("compression-level", "0"), ("overwrite", "true"), ("include-hidden-files", "True")]); |
| 751 | assert_eq!(keep(&set, true).unwrap(), Keep { retention: 14, level: 0, overwrite: true, hidden: true }); |
| 752 | // The merge takes no `overwrite`. |
| 753 | assert!(!keep(&set, false).unwrap().overwrite); |
| 754 | assert_eq!(keep(&with(&[("retention-days", "0")]), true).unwrap().retention, 0); |
| 755 | assert!(keep(&with(&[("retention-days", "91")]), true).is_err()); |
| 756 | assert!(keep(&with(&[("retention-days", "-1")]), true).is_err()); |
| 757 | assert!(keep(&with(&[("retention-days", "two")]), true).is_err()); |
| 758 | assert!(keep(&with(&[("compression-level", "10")]), true).is_err()); |
| 759 | assert!(keep(&with(&[("overwrite", "yes")]), true).is_err()); |
| 760 | assert_eq!(artifact_ids(" 12, 34 ,,").unwrap(), [12, 34]); |
| 761 | assert!(artifact_ids("12,abc").is_err()); |
| 762 | assert_eq!(count(1, "file"), "1 file"); |
| 763 | assert_eq!(count(37, "file"), "37 files"); |
| 764 | } |
| 765 | |
| 766 | fn listed(id: u64, name: &str) -> Listed { |
| 767 | Listed { id, name: name.into(), size: 0, digest: None, format: "zip".into() } |
| 768 | } |
| 769 | |
| 770 | #[test] |
| 771 | fn listings_read_and_keep_the_newest_of_a_name() { |
| 772 | let json = serde_json::json!([ |
| 773 | { "id": 1, "name": "dist", "size": 10, "digest": null, "format": "tgz", "created_at": "x", "expires_at": "y" }, |
| 774 | { "id": "3", "name": "dist", "size": 30, "digest": "sha256:ab", "format": "zip" }, |
| 775 | { "id": 2, "name": "logs", "size": 5 }, |
| 776 | { "name": "no id" } |
| 777 | ]); |
| 778 | let all: Vec<Listed> = json.as_array().unwrap().iter().filter_map(Listed::from_json).collect(); |
| 779 | assert_eq!(all.len(), 3); |
| 780 | assert_eq!(all[0].format, "tgz"); |
| 781 | let kept = newest(all); |
| 782 | assert_eq!(kept.iter().map(|a| (a.id, a.name.as_str())).collect::<Vec<_>>(), [(3, "dist"), (2, "logs")]); |
| 783 | assert_eq!(kept[0].digest.as_deref(), Some("sha256:ab")); |
| 784 | } |
| 785 | |
| 786 | #[test] |
| 787 | fn downloads_land_where_v4_puts_them() { |
| 788 | let dest = Path::new("/w/out"); |
| 789 | let one = [listed(1, "dist")]; |
| 790 | let two = [listed(1, "dist"), listed(2, "logs")]; |
| 791 | let at = |targets: Vec<Option<std::path::PathBuf>>| targets.into_iter().map(|t| t.unwrap()).collect::<Vec<_>>(); |
| 792 | // By name: straight into `path`. |
| 793 | assert_eq!(at(download_targets(dest, &one, true, false, false)), [dest.to_path_buf()]); |
| 794 | // Every artifact, or by pattern: a folder each. |
| 795 | assert_eq!(at(download_targets(dest, &two, false, false, false)), [dest.join("dist"), dest.join("logs")]); |
| 796 | assert_eq!(at(download_targets(dest, &one, false, false, false)), [dest.join("dist")]); |
| 797 | // merge-multiple: all into `path`. |
| 798 | assert_eq!(at(download_targets(dest, &two, false, false, true)), [dest.to_path_buf(), dest.to_path_buf()]); |
| 799 | // By id: one straight into `path`, several a folder each. |
| 800 | assert_eq!(at(download_targets(dest, &one, false, true, false)), [dest.to_path_buf()]); |
| 801 | assert_eq!(at(download_targets(dest, &two, false, true, false)), [dest.join("dist"), dest.join("logs")]); |
| 802 | // A name that is no folder is refused. |
| 803 | assert!(download_targets(dest, &[listed(1, ".."), listed(2, "x")], false, false, false)[0].is_none()); |
| 804 | assert!(glob::matches_part("dist-*", "dist-linux")); |
| 805 | assert!(!glob::matches_part("dist-*", "logs")); |
| 806 | } |
| 807 | } |