dovenest

dovenest/server/src/archive.rs
archive.rs
⎇
Raw
1//! Streaming archive builders (zip / tar / tar.gz / tar.zst).
2//!
3//! Everything writes directly into an `impl std::io::Write` sink — no temp
4//! files, no full in-memory buffering. The sink is typically a channel that
5//! feeds the HTTP response body, so bytes reach the client while the tree is
6//! still being walked.
7
8use std::io::{self, Write};
9use std::path::{Path, PathBuf};
10
11/// The archive formats offered for folder downloads.
12#[derive(Clone, Copy, PartialEq, Eq, Debug)]
13pub enum ArchiveFormat {
14 Zip,
15 Tar,
16 TarGz,
17 TarZst,
18}
19
20impl ArchiveFormat {
21 /// Parse the `format` query parameter. Unknown values are rejected.
22 pub fn parse(s: &str) -> Option<Self> {
23 match s {
24 "zip" => Some(Self::Zip),
25 "tar" => Some(Self::Tar),
26 "tar.gz" | "tgz" => Some(Self::TarGz),
27 "tar.zst" | "tzst" => Some(Self::TarZst),
28 _ => None,
29 }
30 }
31
32 /// The file-name suffix for the produced archive.
33 pub fn extension(self) -> &'static str {
34 match self {
35 Self::Zip => "zip",
36 Self::Tar => "tar",
37 Self::TarGz => "tar.gz",
38 Self::TarZst => "tar.zst",
39 }
40 }
41
42 /// MIME type for the produced archive.
43 pub fn mime(self) -> &'static str {
44 match self {
45 Self::Zip => "application/zip",
46 Self::Tar => "application/x-tar",
47 Self::TarGz => "application/gzip",
48 Self::TarZst => "application/zstd",
49 }
50 }
51}
52
53/// Build `dir` (top-level entry named `top_name`) as `format`, streaming into
54/// `sink`. Blocking — call from `spawn_blocking`.
55pub fn build(
56 format: ArchiveFormat,
57 dir: &Path,
58 top_name: &str,
59 sink: impl Write,
60) -> io::Result<()> {
61 match format {
62 ArchiveFormat::Tar => build_tar(dir, top_name, sink).map(|_| ()),
63 ArchiveFormat::TarGz => {
64 // Level 1: the archive is streamed, so throughput beats ratio.
65 let enc = flate2::write::GzEncoder::new(sink, flate2::Compression::new(1));
66 build_tar(dir, top_name, enc)?.finish().map(|_| ())
67 }
68 ArchiveFormat::TarZst => {
69 let enc = zstd::stream::write::Encoder::new(sink, 3)?;
70 build_tar(dir, top_name, enc)?.finish().map(|_| ())
71 }
72 ArchiveFormat::Zip => build_zip(dir, top_name, sink),
73 }
74}
75
76/// Cycle guard: a symlink loop would otherwise recurse forever. Real trees
77/// this deep are not worth archiving, so deeper levels are dropped.
78const MAX_DEPTH: usize = 64;
79
80/// Depth-first walk. Invokes `f(entry_name, abs_path, is_dir)` for the
81/// directory itself and every descendant. Directory entry names carry no
82/// trailing slash; each format appends it as needed. Deterministic order
83/// (case-insensitive name) so archives are reproducible.
84fn walk<F: FnMut(&str, &Path, bool) -> io::Result<()>>(
85 abs_dir: &Path,
86 entry_prefix: &str,
87 f: &mut F,
88) -> io::Result<()> {
89 // The canonical top directory is the containment boundary for the whole
90 // walk. Unlike browse and upload, nothing else re-checks it here.
91 let base = abs_dir.canonicalize()?;
92 walk_in(&base, &base, entry_prefix, 0, f)
93}
94
95fn walk_in<F: FnMut(&str, &Path, bool) -> io::Result<()>>(
96 base: &Path,
97 abs_dir: &Path,
98 entry_prefix: &str,
99 depth: usize,
100 f: &mut F,
101) -> io::Result<()> {
102 f(entry_prefix, abs_dir, true)?;
103 if depth >= MAX_DEPTH {
104 tracing::warn!(path = %abs_dir.display(), "archive: depth limit reached, subtree skipped");
105 return Ok(());
106 }
107 let rd = std::fs::read_dir(abs_dir)?;
108 let mut children: Vec<(String, PathBuf, bool)> = Vec::new();
109 for e in rd.flatten() {
110 let name = e.file_name().to_string_lossy().into_owned();
111 // Resolve symlinks: a link inside the tree may point outside it, and
112 // its contents must not end up in the archive. One bad entry is
113 // skipped instead of failing the whole download.
114 let Ok(p) = e.path().canonicalize() else {
115 tracing::warn!(path = %e.path().display(), "archive: unreadable entry skipped");
116 continue;
117 };
118 if !crate::fs::is_within_or_eq(base, &p) {
119 tracing::warn!(path = %e.path().display(), "archive: entry outside the archive root skipped");
120 continue;
121 }
122 let is_dir = p.is_dir();
123 children.push((name, p, is_dir));
124 }
125 children.sort_by_key(|c| c.0.to_lowercase());
126 for (name, p, is_dir) in children {
127 let child = format!("{entry_prefix}/{name}");
128 if is_dir {
129 walk_in(base, &p, &child, depth + 1, f)?;
130 } else {
131 f(&child, &p, false)?;
132 }
133 }
134 Ok(())
135}
136
137// ---------------------------------------------------------------------------
138// tar
139// ---------------------------------------------------------------------------
140
141fn tar_add<W: Write>(
142 tar: &mut tar::Builder<W>,
143 entry: &str,
144 abs: &Path,
145 is_dir: bool,
146) -> io::Result<()> {
147 let mut header = tar::Header::new_gnu();
148 let meta = std::fs::metadata(abs)?;
149 header.set_mode(if is_dir { 0o755 } else { 0o644 });
150 header.set_mtime(crate::fs::mtime_secs(&meta).unwrap_or(0).max(0) as u64);
151 if is_dir {
152 header.set_entry_type(tar::EntryType::Directory);
153 header.set_size(0);
154 tar.append_data(&mut header, format!("{entry}/"), io::empty())
155 } else {
156 header.set_entry_type(tar::EntryType::Regular);
157 header.set_size(meta.len());
158 tar.append_data(&mut header, entry, std::fs::File::open(abs)?)
159 }
160}
161
162/// Write the tar stream into `sink` and hand `sink` back, so a caller that
163/// wrapped it in a compressor can finish that compressor.
164fn build_tar<W: Write>(dir: &Path, top: &str, sink: W) -> io::Result<W> {
165 let mut tar = tar::Builder::new(sink);
166 walk(dir, top, &mut |entry: &str, abs: &Path, is_dir: bool| {
167 tar_add(&mut tar, entry, abs, is_dir)
168 })?;
169 tar.finish()?;
170 tar.into_inner()
171}
172
173// ---------------------------------------------------------------------------
174// zip
175// ---------------------------------------------------------------------------
176
177fn build_zip<W: Write>(dir: &Path, top: &str, sink: W) -> io::Result<()> {
178 // Streaming mode: no `Seek` needed, entries use data descriptors.
179 let mut zip = zip::write::ZipWriter::new_stream(sink);
180 let mut add = |entry: &str, abs: &Path, is_dir: bool| -> io::Result<()> {
181 let opts = zip::write::SimpleFileOptions::default();
182 if is_dir {
183 zip.add_directory(format!("{entry}/"), opts)?;
184 } else {
185 zip.start_file(entry, opts)?;
186 io::copy(&mut std::fs::File::open(abs)?, &mut zip)?;
187 }
188 Ok(())
189 };
190 walk(dir, top, &mut add)?;
191 zip.finish()?;
192 Ok(())
193}
194
195// ---------------------------------------------------------------------------
196// Tests
197// ---------------------------------------------------------------------------
198
199#[cfg(test)]
200mod tests {
201 use super::*;
202 use std::collections::BTreeMap;
203 use std::io::Read as _;
204
205 fn sample_dir() -> (tempfile::TempDir, PathBuf) {
206 let tmp = tempfile::tempdir().unwrap();
207 let dir = tmp.path().to_path_buf();
208 std::fs::write(dir.join("alpha.txt"), "alpha content").unwrap();
209 std::fs::create_dir_all(dir.join("sub/deep")).unwrap();
210 std::fs::create_dir(dir.join("empty-dir")).unwrap();
211 std::fs::write(dir.join("sub/beta.txt"), "beta").unwrap();
212 std::fs::write(
213 dir.join("sub/deep/gamma.bin"),
214 (0u8..=255).collect::<Vec<_>>(),
215 )
216 .unwrap();
217 (tmp, dir)
218 }
219
220 fn build_to_mem(fmt: ArchiveFormat, dir: &Path) -> Vec<u8> {
221 let mut out: Vec<u8> = Vec::new();
222 build(fmt, dir, "top", &mut out).unwrap();
223 out
224 }
225
226 #[test]
227 fn format_parsing() {
228 assert_eq!(ArchiveFormat::parse("zip"), Some(ArchiveFormat::Zip));
229 assert_eq!(ArchiveFormat::parse("tar"), Some(ArchiveFormat::Tar));
230 assert_eq!(ArchiveFormat::parse("tar.gz"), Some(ArchiveFormat::TarGz));
231 assert_eq!(ArchiveFormat::parse("tgz"), Some(ArchiveFormat::TarGz));
232 assert_eq!(ArchiveFormat::parse("tar.zst"), Some(ArchiveFormat::TarZst));
233 assert_eq!(ArchiveFormat::parse("tzst"), Some(ArchiveFormat::TarZst));
234 for bad in [
235 "", "ZIP", "gzip", "rar", "7z", "tar.bz2", "tar.xz", "tar.zstx", "tar.gz ",
236 ] {
237 assert_eq!(ArchiveFormat::parse(bad), None, "{bad:?}");
238 }
239 }
240
241 #[test]
242 fn format_metadata() {
243 assert_eq!(ArchiveFormat::Zip.extension(), "zip");
244 assert_eq!(ArchiveFormat::Zip.mime(), "application/zip");
245 assert_eq!(ArchiveFormat::Tar.extension(), "tar");
246 assert_eq!(ArchiveFormat::Tar.mime(), "application/x-tar");
247 assert_eq!(ArchiveFormat::TarGz.extension(), "tar.gz");
248 assert_eq!(ArchiveFormat::TarGz.mime(), "application/gzip");
249 assert_eq!(ArchiveFormat::TarZst.extension(), "tar.zst");
250 assert_eq!(ArchiveFormat::TarZst.mime(), "application/zstd");
251 }
252
253 /// Read a tar stream into a name → content map (files only).
254 fn tar_map<R: std::io::Read>(r: R) -> BTreeMap<String, Vec<u8>> {
255 let mut map = BTreeMap::new();
256 for entry in tar::Archive::new(r).entries().unwrap() {
257 let mut e = entry.unwrap();
258 if !e.header().entry_type().is_file() {
259 continue;
260 }
261 let name = e.path().unwrap().to_string_lossy().into_owned();
262 let mut buf = Vec::new();
263 e.read_to_end(&mut buf).unwrap();
264 map.insert(name, buf);
265 }
266 map
267 }
268 fn expected_map() -> BTreeMap<String, Vec<u8>> {
269 let mut m = BTreeMap::new();
270 m.insert("top/alpha.txt".to_string(), b"alpha content".to_vec());
271 m.insert("top/sub/beta.txt".to_string(), b"beta".to_vec());
272 m.insert("top/sub/deep/gamma.bin".to_string(), (0u8..=255).collect());
273 m
274 }
275
276 #[test]
277 fn zip_round_trip() {
278 let (_tmp, dir) = sample_dir();
279 let bytes = build_to_mem(ArchiveFormat::Zip, &dir);
280 let mut zip = zip::ZipArchive::new(std::io::Cursor::new(bytes)).unwrap();
281
282 let mut map = BTreeMap::new();
283 for i in 0..zip.len() {
284 let mut f = zip.by_index(i).unwrap();
285 let name = f.name().unwrap().to_string();
286 if name.ends_with('/') {
287 continue; // directory entry
288 }
289 let mut buf = Vec::new();
290 f.read_to_end(&mut buf).unwrap();
291 map.insert(name, buf);
292 }
293 assert_eq!(map, expected_map());
294
295 // Directory entries are present and the order is deterministic.
296 let names: Vec<String> = zip.file_names().map(|n| n.unwrap().to_string()).collect();
297 let has = |n: &str| names.iter().any(|x| x == n);
298 assert!(has("top/"));
299 assert!(has("top/sub/"));
300 assert!(has("top/sub/deep/"));
301 assert!(has("top/empty-dir/"));
302 let mut sorted = names.clone();
303 sorted.sort_unstable();
304 assert_eq!(names, sorted);
305 }
306
307 #[test]
308 fn tar_round_trip() {
309 let (_tmp, dir) = sample_dir();
310 let bytes = build_to_mem(ArchiveFormat::Tar, &dir);
311 let map = tar_map(std::io::Cursor::new(bytes));
312 assert_eq!(map, expected_map());
313 }
314
315 #[test]
316 fn tar_gz_round_trip() {
317 let (_tmp, dir) = sample_dir();
318 let bytes = build_to_mem(ArchiveFormat::TarGz, &dir);
319 let gz = flate2::read::GzDecoder::new(std::io::Cursor::new(bytes));
320 let map = tar_map(gz);
321 assert_eq!(map, expected_map());
322 }
323
324 #[test]
325 fn tar_zst_round_trip() {
326 let (_tmp, dir) = sample_dir();
327 let bytes = build_to_mem(ArchiveFormat::TarZst, &dir);
328 let dec = zstd::stream::read::Decoder::new(std::io::Cursor::new(bytes)).unwrap();
329 let map = tar_map(dec);
330 assert_eq!(map, expected_map());
331 }
332
333 #[test]
334 fn walk_is_sorted_case_insensitively() {
335 let tmp = tempfile::tempdir().unwrap();
336 let dir = tmp.path();
337 for name in ["Zeta", "alpha", "Beta", "a.txt", "B.txt"] {
338 if name.ends_with(".txt") {
339 std::fs::write(dir.join(name), name).unwrap();
340 } else {
341 std::fs::create_dir(dir.join(name)).unwrap();
342 }
343 }
344 let mut order = Vec::new();
345 walk(dir, "top", &mut |name, _p, _is_dir| {
346 order.push(name.to_string());
347 Ok(())
348 })
349 .unwrap();
350 assert_eq!(
351 order,
352 vec![
353 "top",
354 "top/a.txt",
355 "top/alpha",
356 "top/B.txt",
357 "top/Beta",
358 "top/Zeta"
359 ]
360 );
361 }
362
363 #[test]
364 fn build_fails_on_missing_dir() {
365 let mut out = Vec::new();
366 let r = build(
367 ArchiveFormat::Zip,
368 Path::new("/nonexistent-dovenest-test-dir"),
369 "top",
370 &mut out,
371 );
372 assert!(r.is_err());
373 // The tar path fails too.
374 let mut out = Vec::new();
375 let r = build(
376 ArchiveFormat::Tar,
377 Path::new("/nonexistent-dovenest-test-dir"),
378 "top",
379 &mut out,
380 );
381 assert!(r.is_err());
382 }
383}
384