| 1 | //! Two-pass tarball reader for remote bundles. |
| 2 | //! |
| 3 | //! Pass one ([`scan_tarball`]) writes nothing: it rejects traversal and |
| 4 | //! absolute paths, enforces the uncompressed size cap from the headers, and |
| 5 | //! locates the single `plugin.toml` whose directory becomes the bundle root. |
| 6 | //! Pass two ([`extract_into`]) extracts only entries under that root, so a |
| 7 | //! mono-repo's symlinks elsewhere in the archive are never materialized. |
| 8 | |
| 9 | use std::fs; |
| 10 | use std::io::{Read, Write}; |
| 11 | use std::path::Path; |
| 12 | |
| 13 | use anyhow::{Context, Result}; |
| 14 | use flate2::read::GzDecoder; |
| 15 | |
| 16 | use crate::skills::install::is_safe_path; |
| 17 | |
| 18 | use super::PluginInstallError; |
| 19 | use super::stage::{StagedPlugin, fresh_staging_dir, validate_staged}; |
| 20 | |
| 21 | /// Validate a tarball and extract the `plugin.toml`-rooted subtree into a |
| 22 | /// `.staging-*` sibling of the destination. |
| 23 | pub(super) fn stage_tarball( |
| 24 | bytes: &[u8], |
| 25 | user_plugins_dir: &Path, |
| 26 | max_size: u64, |
| 27 | ) -> Result<StagedPlugin> { |
| 28 | let scan = scan_tarball(bytes, max_size)?; |
| 29 | let staged_path = fresh_staging_dir(user_plugins_dir)?; |
| 30 | let result = extract_into(&scan, bytes, &staged_path, max_size) |
| 31 | .and_then(|()| validate_staged(&staged_path)); |
| 32 | match result { |
| 33 | Ok((name, content_hash)) => Ok(StagedPlugin { |
| 34 | name, |
| 35 | staged_path, |
| 36 | content_hash, |
| 37 | }), |
| 38 | Err(error) => { |
| 39 | let _ = fs::remove_dir_all(&staged_path); |
| 40 | Err(error) |
| 41 | } |
| 42 | } |
| 43 | } |
| 44 | |
| 45 | #[derive(Debug)] |
| 46 | pub(super) struct TarballScan { |
| 47 | /// Archive-relative directory containing the single `plugin.toml` |
| 48 | /// (`""` when the manifest sits at the archive root). |
| 49 | plugin_root: String, |
| 50 | } |
| 51 | |
| 52 | /// First pass: validate entry paths, enforce the uncompressed size cap, and |
| 53 | /// locate the single `plugin.toml`. Nothing is written in this pass. |
| 54 | pub(super) fn scan_tarball(bytes: &[u8], max_size: u64) -> Result<TarballScan> { |
| 55 | let cursor = std::io::Cursor::new(bytes); |
| 56 | let gz = GzDecoder::new(cursor); |
| 57 | let mut archive = tar::Archive::new(gz); |
| 58 | |
| 59 | let mut total_size: u64 = 0; |
| 60 | let mut manifest_paths: Vec<String> = Vec::new(); |
| 61 | |
| 62 | for entry in archive |
| 63 | .entries() |
| 64 | .context("failed to read tar entries (corrupt archive?)")? |
| 65 | { |
| 66 | let entry = entry.context("failed to read tar entry")?; |
| 67 | let header = entry.header().clone(); |
| 68 | let path = entry |
| 69 | .path() |
| 70 | .context("tar entry has invalid path")? |
| 71 | .to_path_buf(); |
| 72 | let path_str = path.to_string_lossy().into_owned(); |
| 73 | if !is_safe_path(&path) { |
| 74 | return Err(PluginInstallError::PathTraversal(path_str).into()); |
| 75 | } |
| 76 | if let Ok(size) = header.size() { |
| 77 | total_size = total_size.saturating_add(size); |
| 78 | if total_size > max_size { |
| 79 | return Err(PluginInstallError::OversizedBundle { limit: max_size }.into()); |
| 80 | } |
| 81 | } |
| 82 | if header.entry_type().is_file() |
| 83 | && path |
| 84 | .file_name() |
| 85 | .is_some_and(|name| name == std::ffi::OsStr::new("plugin.toml")) |
| 86 | { |
| 87 | manifest_paths.push(path_str); |
| 88 | } |
| 89 | } |
| 90 | |
| 91 | if manifest_paths.len() != 1 { |
| 92 | return Err(PluginInstallError::PluginTomlRoots(manifest_paths.len()).into()); |
| 93 | } |
| 94 | let manifest = &manifest_paths[0]; |
| 95 | let plugin_root = manifest |
| 96 | .rsplit_once('/') |
| 97 | .map(|(dir, _)| dir.to_string()) |
| 98 | .unwrap_or_default(); |
| 99 | Ok(TarballScan { plugin_root }) |
| 100 | } |
| 101 | |
| 102 | /// Second pass: extract only entries under the scanned bundle root. |
| 103 | fn extract_into(scan: &TarballScan, bytes: &[u8], dest: &Path, max_size: u64) -> Result<()> { |
| 104 | let cursor = std::io::Cursor::new(bytes); |
| 105 | let gz = GzDecoder::new(cursor); |
| 106 | let mut archive = tar::Archive::new(gz); |
| 107 | let mut total_size: u64 = 0; |
| 108 | |
| 109 | for entry in archive |
| 110 | .entries() |
| 111 | .context("failed to read tar entries (corrupt archive?)")? |
| 112 | { |
| 113 | let mut entry = entry.context("failed to read tar entry")?; |
| 114 | let header = entry.header().clone(); |
| 115 | let entry_type = header.entry_type(); |
| 116 | let path = entry |
| 117 | .path() |
| 118 | .context("tar entry has invalid path")? |
| 119 | .to_path_buf(); |
| 120 | let path_str = path.to_string_lossy().into_owned(); |
| 121 | if !is_safe_path(&path) { |
| 122 | return Err(PluginInstallError::PathTraversal(path_str).into()); |
| 123 | } |
| 124 | |
| 125 | // Keep only the bundle subtree. Entries outside it (including any |
| 126 | // symlinks a mono-repo ships elsewhere) are ignored, never extracted. |
| 127 | let stripped = if scan.plugin_root.is_empty() { |
| 128 | path_str.clone() |
| 129 | } else if path_str == scan.plugin_root { |
| 130 | String::new() |
| 131 | } else if let Some(rest) = path_str.strip_prefix(&format!("{}/", scan.plugin_root)) { |
| 132 | rest.to_string() |
| 133 | } else { |
| 134 | continue; |
| 135 | }; |
| 136 | if stripped.is_empty() { |
| 137 | // The bundle root directory itself — the staging dir already exists. |
| 138 | continue; |
| 139 | } |
| 140 | // Defense-in-depth: re-validate the stripped path. |
| 141 | let stripped_path = Path::new(&stripped); |
| 142 | if !is_safe_path(stripped_path) { |
| 143 | return Err(PluginInstallError::PathTraversal(stripped).into()); |
| 144 | } |
| 145 | if entry_type.is_symlink() || entry_type.is_hard_link() { |
| 146 | return Err(PluginInstallError::SymlinkRejected.into()); |
| 147 | } |
| 148 | |
| 149 | let target = dest.join(stripped_path); |
| 150 | // Final paranoia check: the composed target must stay under dest. |
| 151 | let target_components: Vec<_> = target.components().collect(); |
| 152 | let dest_components: Vec<_> = dest.components().collect(); |
| 153 | if !target_components.starts_with(dest_components.as_slice()) { |
| 154 | return Err(PluginInstallError::PathTraversal(stripped).into()); |
| 155 | } |
| 156 | |
| 157 | if entry_type.is_dir() { |
| 158 | fs::create_dir_all(&target) |
| 159 | .with_context(|| format!("failed to create dir {}", target.display()))?; |
| 160 | continue; |
| 161 | } |
| 162 | if entry_type.is_file() { |
| 163 | if let Some(parent) = target.parent() { |
| 164 | fs::create_dir_all(parent) |
| 165 | .with_context(|| format!("failed to create dir {}", parent.display()))?; |
| 166 | } |
| 167 | let mut buf = Vec::new(); |
| 168 | entry |
| 169 | .read_to_end(&mut buf) |
| 170 | .with_context(|| format!("failed to read {}", path.display()))?; |
| 171 | total_size = total_size.saturating_add(buf.len() as u64); |
| 172 | if total_size > max_size { |
| 173 | return Err(PluginInstallError::OversizedBundle { limit: max_size }.into()); |
| 174 | } |
| 175 | let mut out = fs::OpenOptions::new() |
| 176 | .create_new(true) |
| 177 | .write(true) |
| 178 | .open(&target) |
| 179 | .with_context(|| format!("failed to create {}", target.display()))?; |
| 180 | out.write_all(&buf) |
| 181 | .with_context(|| format!("failed to write {}", target.display()))?; |
| 182 | } |
| 183 | } |
| 184 | Ok(()) |
| 185 | } |
| 186 |