1use crate::assets::{Asset, AssetSource};
2use crate::error::{CDResult, CargoDebError};
3use crate::listener::Listener;
4use crate::PackageConfig;
5use crate::util::pathbytes::AsUnixPathBytes;
6use std::collections::HashSet;
7use std::io::{Read, Write};
8use std::path::Path;
9use std::{fs, io};
10use tar::{EntryType, Header as TarHeader};
11
12pub(crate) struct Tarball<W: Write> {
14 added_directories: HashSet<Box<Path>>,
15 time: u64,
16 tar: tar::Builder<W>,
17}
18
19impl<W: Write> Tarball<W> {
20 pub fn new(dest: W, time: u64) -> Self {
21 Self {
22 added_directories: HashSet::new(),
23 time,
24 tar: tar::Builder::new(dest),
25 }
26 }
27
28 pub fn archive_files(mut self, package_deb: &PackageConfig, rsyncable: bool, listener: &dyn Listener) -> CDResult<W> {
30 let mut archive_data_added = 0;
31 let mut prev_is_built = false;
32 let log_display_base_dir = std::env::current_dir().unwrap_or_default();
33
34 debug_assert!(package_deb.assets.unresolved.is_empty());
35 for asset in &package_deb.assets.resolved {
36 log_asset(asset, &log_display_base_dir, listener);
37
38 if let AssetSource::Symlink(source_path) = &asset.source {
39 let link_name = fs::read_link(source_path)
40 .map_err(|e| CargoDebError::IoFile("Symlink asset", e, source_path.clone()))?;
41 self.symlink(&asset.c.target_path, &link_name)?;
42 } else {
43 let out_data = asset.source.data()?;
44 if rsyncable {
45 if archive_data_added > 1_000_000 || prev_is_built != asset.c.is_built() {
46 self.flush().map_err(|e| CargoDebError::Io(e).context("error while writing tar archive"))?;
47 archive_data_added = 0;
48 }
49 prev_is_built = asset.c.is_built();
51 archive_data_added += out_data.len();
52 }
53 self.file(&asset.c.target_path, &out_data, asset.c.chmod)?;
54 }
55 }
56
57 self.tar.into_inner().map_err(|e| CargoDebError::Io(e).context("error while finalizing tar archive"))
58 }
59
60 fn directory(&mut self, path: &Path) -> io::Result<()> {
61 let mut header = self.header_for_path(path, true)?;
62 header.set_mtime(self.time);
63 header.set_size(0);
64 header.set_mode(0o755);
65 header.set_entry_type(EntryType::Directory);
66 header.set_cksum();
67 self.tar.append(&header, &mut io::empty())
68 }
69
70 fn add_parent_directories(&mut self, path: &Path) -> CDResult<()> {
71 debug_assert!(path.is_relative());
72
73 let dirs = path.ancestors().skip(1)
74 .take_while(|&d| !self.added_directories.contains(d))
75 .filter(|&d| !d.as_os_str().is_empty())
76 .map(Box::from)
77 .collect::<Vec<_>>();
78
79 for directory in dirs.into_iter().rev() {
80 if let Err(e) = self.directory(&directory) {
81 return Err(CargoDebError::IoFile("Can't add directory to tarball", e, directory.into()));
82 }
83 self.added_directories.insert(directory);
84 }
85 Ok(())
86 }
87
88 pub(crate) fn file<P: AsRef<Path>>(&mut self, path: P, out_data: &[u8], chmod: u32) -> CDResult<()> {
89 self.file_(path.as_ref(), out_data, chmod)
90 }
91
92 fn file_(&mut self, path: &Path, out_data: &[u8], chmod: u32) -> CDResult<()> {
93 debug_assert!(path.is_relative());
94 self.add_parent_directories(path)?;
95
96 let mut header = self.header_for_path(path, false)
97 .map_err(|e| CargoDebError::IoFile("Can't set header path", e, path.into()))?;
98 header.set_mtime(self.time);
99 header.set_mode(chmod);
100 header.set_size(out_data.len() as u64);
101 header.set_cksum();
102 self.tar.append(&header, out_data)
103 .map_err(|e| CargoDebError::IoFile("Can't add file to tarball", e, path.into()))?;
104 Ok(())
105 }
106
107 pub(crate) fn symlink(&mut self, path: &Path, link_name: &Path) -> CDResult<()> {
108 debug_assert!(path.is_relative());
109 self.add_parent_directories(path.as_ref())?;
110
111 let mut header = self.header_for_path(path, false)
112 .map_err(|e| CargoDebError::IoFile("Can't set header path", e, path.into()))?;
113 header.set_mtime(self.time);
114 header.set_entry_type(EntryType::Symlink);
115 header.set_size(0);
116 header.set_mode(0o777);
117 header.set_link_name(link_name)
118 .map_err(|e| CargoDebError::IoFile("Can't set header link name", e, path.into()))?;
119 header.set_cksum();
120 self.tar.append(&header, &mut io::empty())
121 .map_err(|e| CargoDebError::IoFile("Can't add symlink to tarball", e, path.into()))?;
122 Ok(())
123 }
124
125 #[inline]
126 fn header_for_path(&mut self, path: &Path, is_dir: bool) -> io::Result<TarHeader> {
127 debug_assert!(path.is_relative());
128 let path_bytes = path.to_bytes();
129
130 let mut header = if path_bytes.len() < 98 {
131 TarHeader::new_old()
132 } else {
133 TarHeader::new_gnu()
134 };
135 self.set_header_path(&mut header, path_bytes, is_dir)?;
136 Ok(header)
137 }
138
139 #[inline(never)]
140 fn set_header_path(&mut self, header: &mut TarHeader, path_bytes: &[u8], is_dir: bool) -> io::Result<()> {
141 debug_assert!(is_dir || path_bytes.last() != Some(&b'/'));
142 let needs_slash = is_dir && path_bytes.last() != Some(&b'/');
143
144 const PREFIX: &[u8] = b"./";
145 let (prefix, path_slot) = header.as_old_mut().name.split_at_mut(PREFIX.len());
146 prefix.copy_from_slice(PREFIX);
147 let (path_slot, zero) = path_slot.split_at_mut(path_bytes.len().min(path_slot.len()));
148 path_slot.copy_from_slice(&path_bytes[..path_slot.len()]);
149 if cfg!(target_os = "windows") {
150 for b in path_slot {
151 if *b == b'\\' { *b = b'/' }
152 }
153 }
154
155 if let Some((t, rest)) = zero.split_first_mut() {
156 if !needs_slash {
157 *t = 0;
158 return Ok(());
159 }
160 if let Some(t2) = rest.first_mut() {
161 *t = b'/';
163 *t2 = 0;
164 return Ok(());
165 }
166 }
167
168 let mut header = TarHeader::new_gnu();
171 const LONG_LINK: &[u8] = b"././@LongLink\0";
172 header.as_gnu_mut().ok_or(io::ErrorKind::Other)?
173 .name[..LONG_LINK.len()].copy_from_slice(LONG_LINK);
174 header.set_mode(0o644);
175 header.set_uid(0);
176 header.set_gid(0);
177 header.set_mtime(0);
178 let suffix = b"/\0";
180 let suffix = if needs_slash { &suffix[..] } else { &suffix[1..] };
181 header.set_size((PREFIX.len() + path_bytes.len() + suffix.len()) as u64);
182 header.set_entry_type(EntryType::new(b'L'));
183 header.set_cksum();
184 self.tar.append(&header, PREFIX.chain(path_bytes).chain(suffix))
185 }
186
187 fn flush(&mut self) -> io::Result<()> {
188 self.tar.get_mut().flush()
189 }
190
191 pub fn into_inner(self) -> io::Result<W> {
192 self.tar.into_inner()
193 }
194}
195
196fn log_asset(asset: &Asset, log_display_base_dir: &Path, listener: &dyn Listener) {
197 let operation = if let AssetSource::Symlink(_) = &asset.source {
198 "Linking"
199 } else {
200 "Adding"
201 };
202 let mut log_line = format!("'{}' {}-> {}",
203 asset.processed_from.as_ref().and_then(|p| p.original_path.as_deref()).or(asset.source.path())
204 .map(|p| p.strip_prefix(log_display_base_dir).unwrap_or(p))
205 .unwrap_or_else(|| Path::new("-")).display(),
206 asset.processed_from.as_ref().map(|p| p.action).unwrap_or_default(),
207 asset.c.target_path.display()
208 );
209 if let Some(len) = asset.source.file_size() {
210 let (size, unit) = human_size(len);
211 use std::fmt::Write;
212 let _ = write!(&mut log_line, " ({size}{unit})");
213 }
214 listener.progress(operation, log_line);
215}
216
217fn human_size(len: u64) -> (u64, &'static str) {
218 if len < 1000 {
219 return (len, "B");
220 }
221 if len < 1_000_000 {
222 return (len.div_ceil(1000), "KB");
223 }
224 (len.div_ceil(1_000_000), "MB")
225}
226
227#[cfg(test)]
228mod tests {
229 use super::Tarball;
230 use std::{io::{Cursor, Read}, path::Path};
231 use tar::{Archive, EntryType};
232
233 struct ExpectedEntry<'a> {
234 path: &'a str,
235 entry_type: EntryType,
236 mode: u32,
237 check: Option<Box<dyn Fn(&mut tar::Entry<Cursor<Vec<u8>>>) + 'a>>,
238 }
239
240 impl<'a> ExpectedEntry<'a> {
241 fn with_check<F>(mut self, check: F) -> Self
242 where F: Fn(&mut tar::Entry<Cursor<Vec<u8>>>) + 'a
243 {
244 self.check = Some(Box::new(check));
245 self
246 }
247 }
248
249 fn expected_entry(path: &str, entry_type: EntryType, mode: u32) -> ExpectedEntry<'_> {
250 ExpectedEntry { path, entry_type, mode, check: None }
251 }
252
253 fn check_tarball_content(tarball: Vec<u8>, expected_entries: &[ExpectedEntry]) {
254 let cursor = Cursor::new(tarball);
255 let mut archive = Archive::new(cursor);
256 let mut entries = archive.entries().unwrap();
257 let mut expected_entries = expected_entries.iter();
258 loop {
259 let (entry_result, expected_entry) = match (entries.next(), expected_entries.next()) {
260 (Some(entry_result), Some(expected_entry)) => (entry_result, expected_entry),
261 (None, None) => break,
262 _ => panic!("mismatched number of entries"),
263 };
264 let mut entry = entry_result.unwrap();
265 let path = entry.path().unwrap().to_string_lossy().to_string();
266 let entry_type = entry.header().entry_type();
267 let mode = entry.header().mode().unwrap();
268 let mtime = entry.header().mtime().unwrap();
269 assert_eq!(path.strip_prefix("./").unwrap(), expected_entry.path);
270 assert_eq!(entry_type, expected_entry.entry_type);
271 assert_eq!(mode, expected_entry.mode);
272 assert_eq!(mtime, 1234567890);
273 if let Some(check) = &expected_entry.check {
274 check(&mut entry);
275 }
276 }
277 }
278
279 #[test]
280 fn basic() {
281 let buffer = Vec::new();
282 let mut tarball = Tarball::new(buffer, 1234567890);
283 let file_content = b"Hello, world!";
284 tarball.file("test/file.txt", file_content, 0o644).unwrap();
285 let script_content = b"#!/bin/bash\necho 'test'";
286 tarball.file("usr/bin/script", script_content, 0o755).unwrap();
287 tarball.symlink(Path::new("usr/bin/link"), Path::new("script")).unwrap();
288
289 let buffer = tarball.into_inner().unwrap();
290 check_tarball_content(buffer, &[
291 expected_entry("test/", EntryType::Directory, 0o755),
292 expected_entry("test/file.txt", EntryType::Regular, 0o644).with_check(|entry| {
293 let mut content = Vec::new();
294 entry.read_to_end(&mut content).unwrap();
295 assert_eq!(content, file_content);
296 }),
297 expected_entry("usr/", EntryType::Directory, 0o755),
298 expected_entry("usr/bin/", EntryType::Directory, 0o755),
299 expected_entry("usr/bin/script", EntryType::Regular, 0o755).with_check(|entry| {
300 let mut content = Vec::new();
301 entry.read_to_end(&mut content).unwrap();
302 assert_eq!(content, script_content);
303 }),
304 expected_entry("usr/bin/link", EntryType::Symlink, 0o777).with_check(|entry| {
305 let link_name = entry.header().link_name().unwrap().unwrap();
306 assert_eq!(link_name.to_string_lossy(), "script");
307 }),
308 ]);
309 }
310
311 #[test]
312 fn long_path() {
313 let buffer = Vec::new();
314 let mut tarball = Tarball::new(buffer, 1234567890);
315
316 tarball.file("a.txt", b"start", 0o644).unwrap();
317 let level = "long/";
318 let deep_path = level.repeat(25) + "file.txt";
319 tarball.file(&deep_path, b"long path", 0o644).unwrap();
320 let long_filename = "very_".repeat(25) + "long_filename.txt";
321 tarball.file(&long_filename, b"long filename", 0o644).unwrap();
322 tarball.file("b.txt", b"end", 0o644).unwrap();
323 let buffer = tarball.into_inner().unwrap();
324
325 let mut expected_entries = vec![expected_entry("a.txt", EntryType::Regular, 0o644)];
326 expected_entries.extend((1..=25).map(|i| expected_entry(&deep_path[..i * level.len()], EntryType::Directory, 0o755)));
327 expected_entries.extend([
328 expected_entry(&deep_path, EntryType::Regular, 0o644),
329 expected_entry(&long_filename, EntryType::Regular, 0o644),
330 expected_entry("b.txt", EntryType::Regular, 0o644),
331 ]);
332 check_tarball_content(buffer, &expected_entries);
333 }
334}