]> git.proxmox.com Git - proxmox-offline-mirror.git/blame - src/mirror.rs
docs: expand/structure using medium section
[proxmox-offline-mirror.git] / src / mirror.rs
CommitLineData
9ecde319
FG
1use std::{
2 cmp::max,
3 collections::HashMap,
4 io::Read,
5 path::{Path, PathBuf},
6};
7
8b267808 8use anyhow::{bail, format_err, Error};
9ecde319 9use flate2::bufread::GzDecoder;
d035ecb5 10use nix::libc;
49997188 11use proxmox_http::{client::sync::Client, HttpClient, HttpOptions};
d035ecb5
FG
12use proxmox_sys::fs::file_get_contents;
13
14use crate::{
8b267808 15 config::{MirrorConfig, SubscriptionKey},
d035ecb5
FG
16 convert_repo_line,
17 pool::Pool,
18 types::{Snapshot, SNAPSHOT_REGEX},
19 FetchResult, Progress,
20};
9ecde319
FG
21use proxmox_apt::{
22 deb822::{
23 CheckSums, CompressionType, FileReference, FileReferenceType, PackagesFile, ReleaseFile,
24 },
25 repositories::{APTRepository, APTRepositoryPackageType},
26};
27
28use crate::helpers;
29
c598cb15
FG
30fn mirror_dir(config: &MirrorConfig) -> String {
31 format!("{}/{}", config.base_dir, config.id)
32}
33
d035ecb5 34pub(crate) fn pool(config: &MirrorConfig) -> Result<Pool, Error> {
c598cb15
FG
35 let pool_dir = format!("{}/.pool", config.base_dir);
36 Pool::open(Path::new(&mirror_dir(config)), Path::new(&pool_dir))
d035ecb5
FG
37}
38
2d13dcfc 39/// `MirrorConfig`, but some fields converted/parsed into usable types.
d035ecb5
FG
40struct ParsedMirrorConfig {
41 pub repository: APTRepository,
42 pub architectures: Vec<String>,
43 pub pool: Pool,
44 pub key: Vec<u8>,
45 pub verify: bool,
46 pub sync: bool,
8b267808 47 pub auth: Option<String>,
49997188 48 pub client: Client,
d035ecb5
FG
49}
50
51impl TryInto<ParsedMirrorConfig> for MirrorConfig {
52 type Error = anyhow::Error;
53
54 fn try_into(self) -> Result<ParsedMirrorConfig, Self::Error> {
55 let pool = pool(&self)?;
56
57 let repository = convert_repo_line(self.repository.clone())?;
58
59 let key = file_get_contents(Path::new(&self.key_path))?;
60
49997188
FG
61 let options = HttpOptions {
62 user_agent: Some("proxmox-offline-mirror 0.1".to_string()),
63 ..Default::default()
64 }; // TODO actually read version ;)
65
66 let client = Client::new(options);
8b267808 67
d035ecb5
FG
68 Ok(ParsedMirrorConfig {
69 repository,
70 architectures: self.architectures,
71 pool,
72 key,
73 verify: self.verify,
74 sync: self.sync,
8b267808 75 auth: None,
49997188 76 client,
d035ecb5
FG
77 })
78 }
79}
80
2d13dcfc 81// Helper to get absolute URL for dist-specific relative `path`.
9ecde319
FG
82fn get_dist_url(repo: &APTRepository, path: &str) -> String {
83 let dist_root = format!("{}/dists/{}", repo.uris[0], repo.suites[0]);
84
85 format!("{}/{}", dist_root, path)
86}
87
2d13dcfc 88// Helper to get dist-specific path given a `prefix` (snapshot dir) and relative `path`.
9ecde319
FG
89fn get_dist_path(repo: &APTRepository, prefix: &Path, path: &str) -> PathBuf {
90 let mut base = PathBuf::from(prefix);
91 base.push("dists");
92 base.push(&repo.suites[0]);
93 base.push(path);
94 base
95}
96
2d13dcfc 97// Helper to get generic URL given a `repo` and `path`.
9ecde319
FG
98fn get_repo_url(repo: &APTRepository, path: &str) -> String {
99 format!("{}/{}", repo.uris[0], path)
100}
101
2d13dcfc
FG
102/// Helper to fetch file from URI and optionally verify the responses checksum.
103///
104/// Only fetches and returns data, doesn't store anything anywhere.
9ecde319 105fn fetch_repo_file(
49997188 106 client: &Client,
9ecde319 107 uri: &str,
d7e210ac 108 max_size: usize,
9ecde319 109 checksums: Option<&CheckSums>,
8b267808 110 auth: Option<&str>,
9ecde319
FG
111) -> Result<FetchResult, Error> {
112 println!("-> GET '{}'..", uri);
113
49997188
FG
114 let headers = if let Some(auth) = auth {
115 let mut map = HashMap::new();
116 map.insert("Authorization".to_string(), auth.to_string());
117 Some(map)
8b267808 118 } else {
49997188 119 None
8b267808
FG
120 };
121
49997188 122 let response = client.get(uri, headers.as_ref())?;
9ecde319 123
49997188 124 let reader: Box<dyn Read> = response.into_body();
d7e210ac 125 let mut reader = reader.take(max_size as u64);
9ecde319 126 let mut data = Vec::new();
49997188 127 reader.read_to_end(&mut data)?;
9ecde319
FG
128
129 if let Some(checksums) = checksums {
130 checksums.verify(&data)?;
131 }
132
133 Ok(FetchResult {
49997188 134 fetched: data.len(),
9ecde319 135 data,
9ecde319
FG
136 })
137}
138
2d13dcfc
FG
139/// Helper to fetch InRelease (`detached` == false) or Release/Release.gpg (`detached` == true) files from repository.
140///
141/// Verifies the contained/detached signature, stores all fetched files under `prefix`, and returns the verified raw release file data.
9ecde319
FG
142fn fetch_release(
143 config: &ParsedMirrorConfig,
144 prefix: &Path,
145 detached: bool,
146) -> Result<FetchResult, Error> {
147 let (name, fetched, sig) = if detached {
148 println!("Fetching Release/Release.gpg files");
8b267808 149 let sig = fetch_repo_file(
49997188 150 &config.client,
8b267808 151 &get_dist_url(&config.repository, "Release.gpg"),
d7e210ac 152 1024 * 1024,
8b267808
FG
153 None,
154 config.auth.as_deref(),
155 )?;
9ecde319 156 let mut fetched = fetch_repo_file(
49997188 157 &config.client,
9ecde319 158 &get_dist_url(&config.repository, "Release"),
d7e210ac 159 256 * 1024 * 1024,
9ecde319 160 None,
8b267808 161 config.auth.as_deref(),
9ecde319
FG
162 )?;
163 fetched.fetched += sig.fetched;
164 ("Release(.gpg)", fetched, Some(sig.data()))
165 } else {
166 println!("Fetching InRelease file");
167 let fetched = fetch_repo_file(
49997188 168 &config.client,
9ecde319 169 &get_dist_url(&config.repository, "InRelease"),
d7e210ac 170 256 * 1024 * 1024,
9ecde319 171 None,
8b267808 172 config.auth.as_deref(),
9ecde319
FG
173 )?;
174 ("InRelease", fetched, None)
175 };
176
177 println!("Verifying '{name}' signature using provided repository key..");
178 let content = fetched.data_ref();
179 let verified = helpers::verify_signature(content, &config.key, sig.as_deref())?;
180 println!("Success");
181
182 let sha512 = Some(openssl::sha::sha512(content));
183 let csums = CheckSums {
184 sha512,
185 ..Default::default()
186 };
187
188 let locked = &config.pool.lock()?;
189
190 if !locked.contains(&csums) {
d035ecb5 191 locked.add_file(content, &csums, config.sync)?;
9ecde319
FG
192 }
193
194 if detached {
195 locked.link_file(
196 &csums,
197 Path::new(&get_dist_path(&config.repository, prefix, "Release")),
198 )?;
199 let sig = sig.unwrap();
200 let sha512 = Some(openssl::sha::sha512(&sig));
201 let csums = CheckSums {
202 sha512,
203 ..Default::default()
204 };
205 if !locked.contains(&csums) {
d035ecb5 206 locked.add_file(&sig, &csums, config.sync)?;
9ecde319
FG
207 }
208 locked.link_file(
209 &csums,
210 Path::new(&get_dist_path(&config.repository, prefix, "Release.gpg")),
211 )?;
212 } else {
213 locked.link_file(
214 &csums,
215 Path::new(&get_dist_path(&config.repository, prefix, "InRelease")),
216 )?;
217 }
218
219 Ok(FetchResult {
220 data: verified,
221 fetched: fetched.fetched,
222 })
223}
224
2d13dcfc
FG
225/// Helper to fetch an index file referenced by a `ReleaseFile`.
226///
227/// Since these usually come in compressed and uncompressed form, with the latter often not actually existing in the source repository as file, this fetches and if necessary decompresses to obtain a copy of the uncompressed data.
228/// Will skip fetching if both references are already available with the expected checksum in the pool, in which case they will just be re-linked under the new path.
229///
230/// Returns the uncompressed data.
9ecde319
FG
231fn fetch_index_file(
232 config: &ParsedMirrorConfig,
233 prefix: &Path,
234 reference: &FileReference,
235 uncompressed: &FileReference,
236) -> Result<FetchResult, Error> {
237 let url = get_dist_url(&config.repository, &reference.path);
238 let path = get_dist_path(&config.repository, prefix, &reference.path);
239 let uncompressed_path = get_dist_path(&config.repository, prefix, &uncompressed.path);
240
241 if config.pool.contains(&reference.checksums) && config.pool.contains(&uncompressed.checksums) {
242 let data = config
243 .pool
244 .get_contents(&uncompressed.checksums, config.verify)?;
245
246 // Ensure they're linked at current path
247 config.pool.lock()?.link_file(&reference.checksums, &path)?;
248 config
249 .pool
250 .lock()?
251 .link_file(&uncompressed.checksums, &uncompressed_path)?;
252 return Ok(FetchResult { data, fetched: 0 });
253 }
254
d7e210ac
FG
255 let res = fetch_plain_file(
256 config,
257 &url,
258 &path,
259 reference.size,
260 &reference.checksums,
261 true,
262 )?;
9ecde319
FG
263
264 let mut buf = Vec::new();
265 let raw = res.data_ref();
266
267 let decompressed = match reference.file_type.compression() {
268 None => raw,
269 Some(CompressionType::Gzip) => {
270 let mut gz = GzDecoder::new(raw);
271 gz.read_to_end(&mut buf)?;
272 &buf[..]
273 }
274 Some(CompressionType::Bzip2) => {
275 let mut bz = bzip2::read::BzDecoder::new(raw);
276 bz.read_to_end(&mut buf)?;
277 &buf[..]
278 }
279 Some(CompressionType::Lzma) | Some(CompressionType::Xz) => {
280 let mut xz = xz2::read::XzDecoder::new(raw);
281 xz.read_to_end(&mut buf)?;
282 &buf[..]
283 }
284 };
285
286 let locked = &config.pool.lock()?;
287 if !locked.contains(&uncompressed.checksums) {
d035ecb5 288 locked.add_file(decompressed, &uncompressed.checksums, config.sync)?;
9ecde319
FG
289 }
290
291 // Ensure it's linked at current path
292 locked.link_file(&uncompressed.checksums, &uncompressed_path)?;
293
294 Ok(FetchResult {
295 data: decompressed.to_owned(),
296 fetched: res.fetched,
297 })
298}
299
2d13dcfc
FG
300/// Helper to fetch arbitrary files like binary packages.
301///
302/// Will skip fetching if matching file already exists locally, in which case it will just be re-linked under the new path.
303///
304/// If need_data is false and the mirror config is set to skip verification, reading the file's content will be skipped as well if fetching was skipped.
9ecde319
FG
305fn fetch_plain_file(
306 config: &ParsedMirrorConfig,
307 url: &str,
308 file: &Path,
d7e210ac 309 max_size: usize,
9ecde319
FG
310 checksums: &CheckSums,
311 need_data: bool,
312) -> Result<FetchResult, Error> {
313 let locked = &config.pool.lock()?;
314 let res = if locked.contains(checksums) {
315 if need_data || config.verify {
316 locked
317 .get_contents(checksums, config.verify)
318 .map(|data| FetchResult { data, fetched: 0 })?
319 } else {
320 // performance optimization for .deb files if verify is false
321 // we never need the file contents and they make up the bulk of a repo
322 FetchResult {
323 data: vec![],
324 fetched: 0,
325 }
326 }
327 } else {
8b267808 328 let fetched = fetch_repo_file(
49997188 329 &config.client,
8b267808 330 url,
d7e210ac 331 max_size,
8b267808
FG
332 Some(checksums),
333 config.auth.as_deref(),
334 )?;
9ecde319
FG
335 locked.add_file(fetched.data_ref(), checksums, config.verify)?;
336 fetched
337 };
338
339 // Ensure it's linked at current path
340 locked.link_file(checksums, file)?;
341
342 Ok(res)
343}
344
2d13dcfc 345/// Initialize a new mirror (by creating the corresponding pool).
d035ecb5 346pub fn init(config: &MirrorConfig) -> Result<(), Error> {
c598cb15
FG
347 let pool_dir = format!("{}/.pool", config.base_dir);
348
349 let dir = format!("{}/{}", config.base_dir, config.id);
350
351 Pool::create(Path::new(&dir), Path::new(&pool_dir))?;
d035ecb5
FG
352 Ok(())
353}
354
c598cb15 355/// Destroy a mirror (by destroying the corresponding pool's link dir followed by GC).
d035ecb5
FG
356pub fn destroy(config: &MirrorConfig) -> Result<(), Error> {
357 let pool: Pool = pool(config)?;
358 pool.lock()?.destroy()?;
359
360 Ok(())
361}
362
2d13dcfc 363/// List snapshots
d035ecb5
FG
364pub fn list_snapshots(config: &MirrorConfig) -> Result<Vec<Snapshot>, Error> {
365 let _pool: Pool = pool(config)?;
366
367 let mut list: Vec<Snapshot> = vec![];
368
c598cb15
FG
369 let dir = mirror_dir(config);
370
371 let path = Path::new(&dir);
d035ecb5
FG
372
373 proxmox_sys::fs::scandir(
374 libc::AT_FDCWD,
375 path,
376 &SNAPSHOT_REGEX,
377 |_l2_fd, snapshot, file_type| {
378 if file_type != nix::dir::Type::Directory {
379 return Ok(());
380 }
381
382 list.push(snapshot.parse()?);
383
384 Ok(())
385 },
386 )?;
387
45aa8bea
FG
388 list.sort_unstable();
389
d035ecb5
FG
390 Ok(list)
391}
392
2d13dcfc
FG
393/// Create a new snapshot of the remote repository, fetching and storing files as needed.
394///
395/// Operates in three phases:
396/// - Fetch and verify release files
397/// - Fetch referenced indices according to config
398/// - Fetch binary packages referenced by package indices
399///
400/// Files will be linked in a temporary directory and only renamed to the final, valid snapshot directory at the end. In case of error, leftover `XXX.tmp` directories at the top level of `base_dir` can be safely removed once the next snapshot was successfully created, as they only contain hardlinks.
8b267808
FG
401pub fn create_snapshot(
402 config: MirrorConfig,
403 snapshot: &Snapshot,
404 subscription: Option<SubscriptionKey>,
405) -> Result<(), Error> {
406 let auth = if let Some(product) = &config.use_subscription {
407 match subscription {
408 None => {
409 bail!(
410 "Mirror {} requires a subscription key, but none given.",
411 config.id
412 );
413 }
414 Some(key) if key.product() == *product => {
415 let base64 = base64::encode(format!("{}:{}", key.key, key.server_id));
416 Some(format!("basic {base64}"))
417 }
418 Some(key) => {
419 bail!(
420 "Repository product type '{}' and key product type '{}' don't match.",
421 product,
422 key.product()
423 );
424 }
425 }
426 } else {
427 None
428 };
429
430 let mut config: ParsedMirrorConfig = config.try_into()?;
431 config.auth = auth;
9ecde319
FG
432
433 let prefix = format!("{snapshot}.tmp");
434 let prefix = Path::new(&prefix);
435
436 let mut total_progress = Progress::new();
437
438 let parse_release = |res: FetchResult, name: &str| -> Result<ReleaseFile, Error> {
439 println!("Parsing {name}..");
440 let parsed: ReleaseFile = res.data[..].try_into()?;
441 println!(
442 "'{name}' file has {} referenced files..",
443 parsed.files.len()
444 );
445 Ok(parsed)
446 };
447
448 // we want both on-disk for compat reasons
449 let res = fetch_release(&config, prefix, true)?;
450 total_progress.update(&res);
451 let _release = parse_release(res, "Release")?;
452
453 let res = fetch_release(&config, prefix, false)?;
454 total_progress.update(&res);
455 let release = parse_release(res, "InRelease")?;
456
457 let mut per_component = HashMap::new();
458 let mut others = Vec::new();
459 let binary = &config
460 .repository
461 .types
462 .contains(&APTRepositoryPackageType::Deb);
463 let source = &config
464 .repository
465 .types
466 .contains(&APTRepositoryPackageType::DebSrc);
467
468 for (basename, references) in &release.files {
469 let reference = references.first();
470 let reference = if let Some(reference) = reference {
471 reference.clone()
472 } else {
473 continue;
474 };
475 let skip_components = !&config.repository.components.contains(&reference.component);
476
477 let skip = skip_components
478 || match &reference.file_type {
479 FileReferenceType::Ignored => true,
480 FileReferenceType::PDiff => true, // would require fetching the patches as well
481 FileReferenceType::Contents(arch, _)
482 | FileReferenceType::ContentsUdeb(arch, _)
483 | FileReferenceType::Packages(arch, _)
484 | FileReferenceType::PseudoRelease(Some(arch)) => {
485 !binary || !config.architectures.contains(arch)
486 }
487 FileReferenceType::Sources(_) => !source,
488 _ => false,
489 };
490 if skip {
491 println!("Skipping {}", reference.path);
492 others.push(reference);
493 } else {
494 let list = per_component
495 .entry(reference.component)
496 .or_insert_with(Vec::new);
497 list.push(basename);
498 }
499 }
500 println!();
501
502 let mut indices_size = 0_usize;
503 let mut total_count = 0;
504
505 for (component, references) in &per_component {
506 println!("Component '{component}'");
507
508 let mut component_indices_size = 0;
509
510 for basename in references {
511 for reference in release.files.get(*basename).unwrap() {
512 println!("\t{:?}: {:?}", reference.path, reference.file_type);
513 component_indices_size += reference.size;
514 }
515 }
516 indices_size += component_indices_size;
517
518 let component_count = references.len();
519 total_count += component_count;
520
521 println!("Component references count: {component_count}");
522 println!("Component indices size: {component_indices_size}");
523 if references.is_empty() {
524 println!("\tNo references found..");
525 }
526 }
527 println!("Total indices count: {total_count}");
528 println!("Total indices size: {indices_size}");
529
530 if !others.is_empty() {
531 println!("Skipped {} references", others.len());
532 }
533 println!();
534
535 let mut packages_size = 0_usize;
536 let mut packages_indices = HashMap::new();
537 for (component, references) in per_component {
538 println!("\nFetching indices for component '{component}'");
539 let mut component_deb_size = 0;
540 let mut fetch_progress = Progress::new();
541
542 for basename in references {
543 println!("\tFetching '{basename}'..");
544 let files = release.files.get(basename).unwrap();
545 let uncompressed_ref = files
546 .iter()
547 .find(|reference| reference.path == *basename)
548 .ok_or_else(|| format_err!("Found derived reference without base reference."))?;
549 let mut package_index_data = None;
550
551 for reference in files {
552 // if both compressed and uncompressed are referenced, the uncompressed file may not exist on the server
553 if reference == uncompressed_ref && files.len() > 1 {
554 continue;
555 }
556
557 // this will ensure the uncompressed file will be written locally
558 let res = fetch_index_file(&config, prefix, reference, uncompressed_ref)?;
559 fetch_progress.update(&res);
560
561 if package_index_data.is_none() && reference.file_type.is_package_index() {
562 package_index_data = Some(res.data());
563 }
564 }
565 if let Some(data) = package_index_data {
566 let packages: PackagesFile = data[..].try_into()?;
567 let size: usize = packages.files.iter().map(|p| p.size).sum();
568 println!("\t{} packages totalling {size}", packages.files.len());
569 component_deb_size += size;
570
571 packages_indices.entry(basename).or_insert(packages);
572 }
573 println!("Progress: {fetch_progress}");
574 }
575 println!("Total deb size for component: {component_deb_size}");
576 packages_size += component_deb_size;
577 total_progress += fetch_progress;
578 }
579 println!("Total deb size: {packages_size}");
580
581 println!("\nFetching packages..");
582 for (basename, references) in packages_indices {
583 let total_files = references.files.len();
584 if total_files == 0 {
585 println!("\n{basename} - no files, skipping.");
586 continue;
587 } else {
588 println!("\n{basename} - {total_files} total file(s)");
589 }
590
591 let mut fetch_progress = Progress::new();
592 for package in references.files {
593 let mut full_path = PathBuf::from(prefix);
594 full_path.push(&package.file);
595 let res = fetch_plain_file(
596 &config,
597 &get_repo_url(&config.repository, &package.file),
598 &full_path,
d7e210ac 599 package.size,
9ecde319
FG
600 &package.checksums,
601 false,
602 )?;
603 fetch_progress.update(&res);
604 if fetch_progress.file_count() % (max(total_files / 100, 1)) == 0 {
605 println!("\tProgress: {fetch_progress}");
606 }
607 }
608 println!("\tProgress: {fetch_progress}");
609 total_progress += fetch_progress;
610 }
611
612 println!("\nStats: {total_progress}");
613
614 println!("Rotating temp. snapshot in-place: {prefix:?} -> \"{snapshot}\"");
615 let locked = config.pool.lock()?;
616 locked.rename(prefix, Path::new(&format!("{snapshot}")))?;
617
618 Ok(())
619}
d035ecb5 620
2d13dcfc 621/// Remove a snapshot by removing the corresponding snapshot directory. To actually free up space, a garbage collection needs to be run afterwards.
d035ecb5
FG
622pub fn remove_snapshot(config: &MirrorConfig, snapshot: &Snapshot) -> Result<(), Error> {
623 let pool: Pool = pool(config)?;
624 let path = pool.get_path(Path::new(&snapshot.to_string()))?;
625
626 pool.lock()?.remove_dir(&path)
627}
628
2d13dcfc 629/// Run a garbage collection on the underlying pool.
d035ecb5
FG
630pub fn gc(config: &MirrorConfig) -> Result<(usize, u64), Error> {
631 let pool: Pool = pool(config)?;
632
633 pool.lock()?.gc()
634}