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};
10use std::time::UNIX_EPOCH;
11
12/// The archive formats offered for folder downloads.
13#[derive(Clone, Copy, PartialEq, Eq, Debug)]
14pub enum ArchiveFormat {
15 Zip,
16 Tar,
17 TarGz,
18 TarZst,
19}
20
21impl ArchiveFormat {
22 /// Parse the `format` query parameter. Unknown values are rejected.
23 pub fn parse(s: &str) -> Option<Self> {
24 match s {
25 "zip" => Some(Self::Zip),
26 "tar" => Some(Self::Tar),
27 "tar.gz" | "tgz" => Some(Self::TarGz),
28 "tar.zst" | "tzst" => Some(Self::TarZst),
29 _ => None,
30 }
31 }
32
33 /// The file-name suffix for the produced archive.
34 pub fn extension(self) -> &'static str {
35 match self {
36 Self::Zip => "zip",
37 Self::Tar => "tar",
38 Self::TarGz => "tar.gz",
39 Self::TarZst => "tar.zst",
40 }
41 }
42
43 /// MIME type for the produced archive.
44 pub fn mime(self) -> &'static str {
45 match self {
46 Self::Zip => "application/zip",
47 Self::Tar => "application/x-tar",
48 Self::TarGz => "application/gzip",
49 Self::TarZst => "application/zstd",
50 }
51 }
52}
53
54/// Build `dir` (top-level entry named `top_name`) as `format`, streaming into
55/// `sink`. Blocking — call from `spawn_blocking`.
56pub fn build(
57 format: ArchiveFormat,
58 dir: &Path,
59 top_name: &str,
60 sink: impl Write,
61) -> io::Result<()> {
62 match format {
63 ArchiveFormat::Tar => build_tar(dir, top_name, sink),
64 ArchiveFormat::TarGz => build_tar_gz(dir, top_name, sink),
65 ArchiveFormat::TarZst => build_tar_zst(dir, top_name, sink),
66 ArchiveFormat::Zip => build_zip(dir, top_name, sink),
67 }
68}
69
70fn mtime_secs(p: &Path) -> u64 {
71 std::fs::metadata(p)
72 .and_then(|m| m.modified())
73 .ok()
74 .and_then(|t| t.duration_since(UNIX_EPOCH).ok())
75 .map(|d| d.as_secs())
76 .unwrap_or(0)
77}
78
79/// Depth-first walk. Invokes `f(entry_name, abs_path, is_dir)` for the
80/// directory itself and every descendant. Directory entry names carry no
81/// trailing slash; each format appends it as needed. Deterministic order
82/// (case-insensitive name) so archives are reproducible.
83fn walk<F: FnMut(&str, &Path, bool) -> io::Result<()>>(
84 abs_dir: &Path,
85 entry_prefix: &str,
86 f: &mut F,
87) -> io::Result<()> {
88 f(entry_prefix, abs_dir, true)?;
89 let rd = std::fs::read_dir(abs_dir)?;
90 let mut children: Vec<(String, PathBuf, bool)> = Vec::new();
91 for e in rd.flatten() {
92 let name = e.file_name().to_string_lossy().into_owned();
93 let p = e.path();
94 // Follows symlinks (a broken link shows up as an empty file).
95 let is_dir = p.is_dir();
96 children.push((name, p, is_dir));
97 }
98 children.sort_by(|a, b| a.0.to_lowercase().cmp(&b.0.to_lowercase()));
99 for (name, p, is_dir) in children {
100 let child = format!("{entry_prefix}/{name}");
101 if is_dir {
102 walk(&p, &child, f)?;
103 } else {
104 f(&child, &p, false)?;
105 }
106 }
107 Ok(())
108}
109
110// ---------------------------------------------------------------------------
111// tar
112// ---------------------------------------------------------------------------
113
114fn tar_add<W: Write>(
115 tar: &mut tar::Builder<W>,
116 entry: &str,
117 abs: &Path,
118 is_dir: bool,
119) -> io::Result<()> {
120 let mut header = tar::Header::new_gnu();
121 let meta = std::fs::metadata(abs)?;
122 header.set_mode(if is_dir { 0o755 } else { 0o644 });
123 header.set_mtime(mtime_secs(abs));
124 let name = if is_dir {
125 format!("{entry}/")
126 } else {
127 entry.to_string()
128 };
129 if is_dir {
130 header.set_entry_type(tar::EntryType::Directory);
131 header.set_size(0);
132 tar.append_data(&mut header, name, io::empty())?;
133 } else {
134 header.set_entry_type(tar::EntryType::Regular);
135 header.set_size(meta.len());
136 let f = std::fs::File::open(abs)?;
137 tar.append_data(&mut header, name, f)?;
138 }
139 Ok(())
140}
141
142fn build_tar<W: Write>(dir: &Path, top: &str, sink: W) -> io::Result<()> {
143 let mut tar = tar::Builder::new(sink);
144 let mut add = |entry: &str, abs: &Path, is_dir: bool| -> io::Result<()> {
145 tar_add(&mut tar, entry, abs, is_dir)
146 };
147 walk(dir, top, &mut add)?;
148 tar.finish()?;
149 Ok(())
150}
151
152fn build_tar_gz<W: Write>(dir: &Path, top: &str, sink: W) -> io::Result<()> {
153 let enc = flate2::write::GzEncoder::new(sink, flate2::Compression::default());
154 let mut tar = tar::Builder::new(enc);
155 let mut add = |entry: &str, abs: &Path, is_dir: bool| -> io::Result<()> {
156 tar_add(&mut tar, entry, abs, is_dir)
157 };
158 walk(dir, top, &mut add)?;
159 let enc = tar.into_inner()?;
160 enc.finish()?;
161 Ok(())
162}
163
164fn build_tar_zst<W: Write>(dir: &Path, top: &str, sink: W) -> io::Result<()> {
165 let enc = zstd::stream::write::Encoder::new(sink, 3)?;
166 let mut tar = tar::Builder::new(enc);
167 let mut add = |entry: &str, abs: &Path, is_dir: bool| -> io::Result<()> {
168 tar_add(&mut tar, entry, abs, is_dir)
169 };
170 walk(dir, top, &mut add)?;
171 let enc = tar.into_inner()?;
172 enc.finish()?;
173 Ok(())
174}
175
176// ---------------------------------------------------------------------------
177// zip
178// ---------------------------------------------------------------------------
179
180fn build_zip<W: Write>(dir: &Path, top: &str, sink: W) -> io::Result<()> {
181 // Streaming mode: no `Seek` needed, entries use data descriptors.
182 let mut zip = zip::write::ZipWriter::new_stream(sink);
183 let mut add = |entry: &str, abs: &Path, is_dir: bool| -> io::Result<()> {
184 let opts = zip::write::SimpleFileOptions::default();
185 let name = if is_dir {
186 format!("{entry}/")
187 } else {
188 entry.to_string()
189 };
190 if is_dir {
191 zip.add_directory(&name, opts)?;
192 } else {
193 zip.start_file(&name, opts)?;
194 let mut f = std::fs::File::open(abs)?;
195 io::copy(&mut f, &mut zip)?;
196 }
197 Ok(())
198 };
199 walk(dir, top, &mut add)?;
200 zip.finish()?;
201 Ok(())
202}
203