uv_git/
resolver.rs

1use std::borrow::Cow;
2use std::path::PathBuf;
3use std::str::FromStr;
4use std::sync::Arc;
5
6use dashmap::DashMap;
7use dashmap::mapref::one::Ref;
8use fs_err::tokio as fs;
9use reqwest_middleware::ClientWithMiddleware;
10use tracing::debug;
11
12use uv_cache_key::{RepositoryUrl, cache_digest};
13use uv_fs::{LockedFile, LockedFileError, LockedFileMode};
14use uv_git_types::{GitHubRepository, GitOid, GitReference, GitUrl};
15use uv_static::EnvVars;
16use uv_version::version;
17
18use crate::{
19    Fetch, GitSource, Reporter,
20    rate_limit::{GITHUB_RATE_LIMIT_STATUS, is_github_rate_limited},
21};
22
23#[derive(Debug, thiserror::Error)]
24pub enum GitResolverError {
25    #[error(transparent)]
26    Io(#[from] std::io::Error),
27    #[error(transparent)]
28    LockedFile(#[from] LockedFileError),
29    #[error(transparent)]
30    Join(#[from] tokio::task::JoinError),
31    #[error("Git operation failed")]
32    Git(#[source] anyhow::Error),
33    #[error(transparent)]
34    Reqwest(#[from] reqwest::Error),
35    #[error(transparent)]
36    ReqwestMiddleware(#[from] reqwest_middleware::Error),
37}
38
39/// A resolver for Git repositories.
40#[derive(Default, Clone)]
41pub struct GitResolver(Arc<DashMap<RepositoryReference, GitOid>>);
42
43impl GitResolver {
44    /// Inserts a new [`GitOid`] for the given [`RepositoryReference`].
45    pub fn insert(&self, reference: RepositoryReference, sha: GitOid) {
46        self.0.insert(reference, sha);
47    }
48
49    /// Returns the [`GitOid`] for the given [`RepositoryReference`], if it exists.
50    fn get(&self, reference: &RepositoryReference) -> Option<Ref<'_, RepositoryReference, GitOid>> {
51        self.0.get(reference)
52    }
53
54    /// Return the [`GitOid`] for the given [`GitUrl`], if it is already known.
55    pub fn get_precise(&self, url: &GitUrl) -> Option<GitOid> {
56        // If the URL is already precise, return it.
57        if let Some(precise) = url.precise() {
58            return Some(precise);
59        }
60
61        // If we know the precise commit already, return it.
62        let reference = RepositoryReference::from(url);
63        if let Some(precise) = self.get(&reference) {
64            return Some(*precise);
65        }
66
67        None
68    }
69
70    /// Resolve a Git URL to a specific commit without performing any Git operations.
71    ///
72    /// Returns a [`GitOid`] if the URL has already been resolved (i.e., is available in the cache),
73    /// or if it can be fetched via the GitHub API. Otherwise, returns `None`.
74    pub async fn github_fast_path(
75        &self,
76        url: &GitUrl,
77        client: &ClientWithMiddleware,
78    ) -> Result<Option<GitOid>, GitResolverError> {
79        if std::env::var_os(EnvVars::UV_NO_GITHUB_FAST_PATH).is_some() {
80            return Ok(None);
81        }
82
83        // If the URL is already precise or we know the precise commit, return it.
84        if let Some(precise) = self.get_precise(url) {
85            return Ok(Some(precise));
86        }
87
88        // If the URL is a GitHub URL, attempt to resolve it via the GitHub API.
89        let Some(GitHubRepository { owner, repo }) = GitHubRepository::parse(url.repository())
90        else {
91            return Ok(None);
92        };
93
94        // Check if we're rate-limited by GitHub, before determining the Git reference
95        if GITHUB_RATE_LIMIT_STATUS.is_active() {
96            debug!("Rate-limited by GitHub. Skipping GitHub fast path attempt for: {url}");
97            return Ok(None);
98        }
99
100        // Determine the Git reference.
101        let rev = url.reference().as_rev();
102
103        let github_api_base_url = std::env::var(EnvVars::UV_GITHUB_FAST_PATH_URL)
104            .unwrap_or("https://api.github.com/repos".to_owned());
105        let github_api_url = format!("{github_api_base_url}/{owner}/{repo}/commits/{rev}");
106
107        debug!("Querying GitHub for commit at: {github_api_url}");
108        let mut request = client.get(&github_api_url);
109        request = request.header("Accept", "application/vnd.github.3.sha");
110        request = request.header(
111            "User-Agent",
112            format!("uv/{} (+https://github.com/astral-sh/uv)", version()),
113        );
114
115        let response = request.send().await?;
116        let status = response.status();
117        if !status.is_success() {
118            // Returns a 404 if the repository does not exist, and a 422 if GitHub is unable to
119            // resolve the requested rev.
120            debug!(
121                "GitHub API request failed for: {github_api_url} ({})",
122                response.status()
123            );
124
125            if is_github_rate_limited(&response) {
126                // Mark that we are being rate-limited by GitHub
127                GITHUB_RATE_LIMIT_STATUS.activate();
128            }
129
130            return Ok(None);
131        }
132
133        // Parse the response as a Git SHA.
134        let precise = response.text().await?;
135        let precise =
136            GitOid::from_str(&precise).map_err(|err| GitResolverError::Git(err.into()))?;
137
138        // Insert the resolved URL into the in-memory cache. This ensures that subsequent fetches
139        // resolve to the same precise commit.
140        self.insert(RepositoryReference::from(url), precise);
141
142        Ok(Some(precise))
143    }
144
145    /// Fetch a remote Git repository.
146    pub async fn fetch(
147        &self,
148        url: &GitUrl,
149        disable_ssl: bool,
150        offline: bool,
151        cache: PathBuf,
152        reporter: Option<Arc<dyn Reporter>>,
153    ) -> Result<Fetch, GitResolverError> {
154        debug!("Fetching source distribution from Git: {url}");
155
156        let reference = RepositoryReference::from(url);
157
158        // If we know the precise commit already, reuse it, to ensure that all fetches within a
159        // single process are consistent.
160        let url = {
161            if let Some(precise) = self.get(&reference) {
162                Cow::Owned(url.clone().with_precise(*precise))
163            } else {
164                Cow::Borrowed(url)
165            }
166        };
167
168        // Avoid races between different processes, too.
169        let lock_dir = cache.join("locks");
170        fs::create_dir_all(&lock_dir).await?;
171        let repository_url = RepositoryUrl::new(url.repository());
172        let _lock = LockedFile::acquire(
173            lock_dir.join(cache_digest(&repository_url)),
174            LockedFileMode::Exclusive,
175            &repository_url,
176        )
177        .await?;
178
179        // Fetch the Git repository.
180        let source = if let Some(reporter) = reporter {
181            GitSource::new(url.as_ref().clone(), cache, offline).with_reporter(reporter)
182        } else {
183            GitSource::new(url.as_ref().clone(), cache, offline)
184        };
185
186        // If necessary, disable SSL.
187        let source = if disable_ssl {
188            source.dangerous()
189        } else {
190            source
191        };
192
193        let fetch = tokio::task::spawn_blocking(move || source.fetch())
194            .await?
195            .map_err(GitResolverError::Git)?;
196
197        // Insert the resolved URL into the in-memory cache. This ensures that subsequent fetches
198        // resolve to the same precise commit.
199        if let Some(precise) = fetch.git().precise() {
200            self.insert(reference, precise);
201        }
202
203        Ok(fetch)
204    }
205
206    /// Given a remote source distribution, return a precise variant, if possible.
207    ///
208    /// For example, given a Git dependency with a reference to a branch or tag, return a URL
209    /// with a precise reference to the current commit of that branch or tag.
210    ///
211    /// This method takes into account various normalizations that are independent of the Git
212    /// layer. For example: removing `#subdirectory=pkg_dir`-like fragments, and removing `git+`
213    /// prefix kinds.
214    ///
215    /// This method will only return precise URLs for URLs that have already been resolved via
216    /// [`resolve_precise`], and will return `None` for URLs that have not been resolved _or_
217    /// already have a precise reference.
218    pub fn precise(&self, url: GitUrl) -> Option<GitUrl> {
219        let reference = RepositoryReference::from(&url);
220        let precise = self.get(&reference)?;
221        Some(url.with_precise(*precise))
222    }
223
224    /// Returns `true` if the two Git URLs refer to the same precise commit.
225    pub fn same_ref(&self, a: &GitUrl, b: &GitUrl) -> bool {
226        // Convert `a` to a repository URL.
227        let a_ref = RepositoryReference::from(a);
228
229        // Convert `b` to a repository URL.
230        let b_ref = RepositoryReference::from(b);
231
232        // The URLs must refer to the same repository.
233        if a_ref.url != b_ref.url {
234            return false;
235        }
236
237        // If the URLs have the same tag, they refer to the same commit.
238        if a_ref.reference == b_ref.reference {
239            return true;
240        }
241
242        // Otherwise, the URLs must resolve to the same precise commit.
243        let Some(a_precise) = a.precise().or_else(|| self.get(&a_ref).map(|sha| *sha)) else {
244            return false;
245        };
246
247        let Some(b_precise) = b.precise().or_else(|| self.get(&b_ref).map(|sha| *sha)) else {
248            return false;
249        };
250
251        a_precise == b_precise
252    }
253}
254
255#[derive(Debug, Clone, PartialEq, Eq, Hash)]
256pub struct ResolvedRepositoryReference {
257    /// An abstract reference to a Git repository, including the URL and the commit (e.g., a branch,
258    /// tag, or revision).
259    pub reference: RepositoryReference,
260    /// The precise commit SHA of the reference.
261    pub sha: GitOid,
262}
263
264#[derive(Debug, Clone, PartialEq, Eq, Hash)]
265pub struct RepositoryReference {
266    /// The URL of the Git repository, with any query parameters and fragments removed.
267    pub url: RepositoryUrl,
268    /// The reference to the commit to use, which could be a branch, tag, or revision.
269    pub reference: GitReference,
270}
271
272impl From<&GitUrl> for RepositoryReference {
273    fn from(git: &GitUrl) -> Self {
274        Self {
275            url: RepositoryUrl::new(git.repository()),
276            reference: git.reference().clone(),
277        }
278    }
279}
uv_git/resolver.rs

uv_git/
resolver.rs
