Skip to main content

aube_lockfile/
source.rs

1use std::path::{Path, PathBuf};
2
3/// Non-registry source for a locked package.
4///
5/// When a package comes from a local path (via `file:` or `link:` in
6/// `package.json`) it doesn't have a tarball URL or integrity hash, so we
7/// record the source separately and let the linker materialize it
8/// on-the-fly.
9#[derive(Debug, Clone, PartialEq, Eq)]
10pub enum LocalSource {
11    /// `file:<dir>` — a directory on disk whose contents should be
12    /// hardlink-copied into the virtual store like a normal package.
13    /// Path is stored relative to the project root.
14    Directory(PathBuf),
15    /// `file:<tarball>` — a `.tgz` on disk, extracted into the virtual
16    /// store the same way we extract registry tarballs.
17    Tarball(PathBuf),
18    /// `link:<dir>` — a plain symlink into `node_modules/<name>`, never
19    /// materialized into the virtual store. Transitive deps are the
20    /// target's responsibility.
21    Link(PathBuf),
22    /// `portal:<dir>` — a Yarn Berry package portal. The target is a
23    /// package on disk, but unlike `link:` its dependencies are still
24    /// modeled in the lockfile graph.
25    Portal(PathBuf),
26    /// `exec:<script>` — a Yarn Berry generator script. The script is
27    /// executed at fetch time and writes the package files into a
28    /// generated build directory.
29    Exec(PathBuf),
30    /// `git+https://`, `git+ssh://`, `github:user/repo`, etc. — a
31    /// remote git repo. Cloned at fetch time and imported like a
32    /// `file:` directory. `url` is the normalized clone URL (what
33    /// gets passed to `git clone`). `committish` is the user-written
34    /// ref after `#` (branch, tag, or commit; `None` means HEAD).
35    /// `resolved` is the 40-char commit SHA that `git ls-remote`
36    /// pinned the ref to — the lockfile records this so repeat
37    /// installs reproduce bit-for-bit.
38    Git(GitSource),
39    /// `https://example.com/pkg.tgz` — a remote tarball URL. Fetched
40    /// once at resolve time so the resolver can read the enclosed
41    /// `package.json` for version + transitive deps and pin the
42    /// sha512 integrity. `integrity` stays empty on freshly-parsed
43    /// specifiers and is filled in by the resolver after download.
44    RemoteTarball(RemoteTarballSource),
45}
46
47/// A remote tarball dependency spec. See [`LocalSource::RemoteTarball`].
48#[derive(Debug, Clone, PartialEq, Eq)]
49pub struct RemoteTarballSource {
50    pub url: String,
51    pub integrity: String,
52    pub git_hosted: bool,
53}
54
55/// A git dependency spec. See [`LocalSource::Git`].
56#[derive(Debug, Clone, PartialEq, Eq)]
57pub struct GitSource {
58    pub url: String,
59    pub committish: Option<String>,
60    pub resolved: String,
61    /// SHA-512 SRI of the hosted tarball bytes when the git source was
62    /// fetched through a codeload-style archive. Plain git-clone sources
63    /// leave this unset because git object IDs verify the checkout.
64    pub integrity: Option<String>,
65    /// pnpm `&path:/sub/dir` selector — when set, only this
66    /// subdirectory of the cloned repo is treated as the package
67    /// root. Stored without leading slash so dep_path hashes are
68    /// stable regardless of whether the user wrote `path:/x` or
69    /// `path:x`.
70    pub subpath: Option<String>,
71}
72
73pub fn git_commits_match(left: &str, right: &str) -> bool {
74    if left.eq_ignore_ascii_case(right) {
75        return true;
76    }
77    let left = left.trim();
78    let right = right.trim();
79    if left.len().min(right.len()) < 7
80        || !left.bytes().all(|b| b.is_ascii_hexdigit())
81        || !right.bytes().all(|b| b.is_ascii_hexdigit())
82    {
83        return false;
84    }
85    let left = left.to_ascii_lowercase();
86    let right = right.to_ascii_lowercase();
87    (left.len() == 40 && right.len() < 40 && left.starts_with(&right))
88        || (right.len() == 40 && left.len() < 40 && right.starts_with(&left))
89}
90
91impl LocalSource {
92    /// The original path (relative to the project root) the user wrote
93    /// in `package.json`. `None` for non-path sources like git.
94    pub fn path(&self) -> Option<&Path> {
95        match self {
96            LocalSource::Directory(p)
97            | LocalSource::Tarball(p)
98            | LocalSource::Link(p)
99            | LocalSource::Portal(p)
100            | LocalSource::Exec(p) => Some(p),
101            LocalSource::Git(_) | LocalSource::RemoteTarball(_) => None,
102        }
103    }
104
105    /// The protocol kind (`"file"` / `"link"` / `"git"` / `"url"`).
106    pub fn kind_str(&self) -> &'static str {
107        match self {
108            LocalSource::Directory(_) | LocalSource::Tarball(_) => "file",
109            LocalSource::Link(_) => "link",
110            LocalSource::Portal(_) => "portal",
111            LocalSource::Exec(_) => "exec",
112            LocalSource::Git(_) => "git",
113            LocalSource::RemoteTarball(_) => "url",
114        }
115    }
116
117    /// Whether this source is pinned to immutable, globally
118    /// reproducible content and can therefore be shared across
119    /// projects inside aube's global virtual store, exactly like a
120    /// registry package.
121    ///
122    /// `Git` is pinned to a 40-char commit SHA and `RemoteTarball` to
123    /// a fetched URL (and, once resolved, an integrity hash), so two
124    /// projects that depend on the same one resolve to the same files.
125    /// `file:` / `link:` / `portal:` / `exec:` all resolve against a
126    /// path inside the depending project, so they stay per-project and
127    /// are never promoted into the shared store.
128    ///
129    /// Load-bearing for global-virtual-store correctness: a registry
130    /// package materialized into the shared store points its
131    /// dependency siblings at the hashed global path
132    /// (`virtual_store_subdir(dep_path)`). If one of those deps were a
133    /// git/tarball source that only ever landed in the per-project
134    /// `.aube/`, the sibling symlink would dangle and Node's module
135    /// walk would silently fall back to some unrelated `<name>` found
136    /// higher up the tree.
137    pub fn is_globally_shareable(&self) -> bool {
138        matches!(self, LocalSource::Git(_) | LocalSource::RemoteTarball(_))
139    }
140
141    /// The path as a POSIX-style string with forward-slash separators.
142    /// `Path::display()` and `to_string_lossy()` honor the host's
143    /// separator (backslash on Windows), which would make `dep_path`
144    /// hashes and lockfile `specifier:` strings non-portable: the
145    /// same `file:./some/dir` would render as `some\dir` on Windows
146    /// and `some/dir` on Unix, producing two different hashes for
147    /// the same logical target. Always rendering with `/` keeps
148    /// lockfiles cross-platform identical.
149    pub fn path_posix(&self) -> String {
150        self.path()
151            .map(|p| p.to_string_lossy().replace('\\', "/"))
152            .unwrap_or_default()
153    }
154
155    /// Canonical specifier string as pnpm writes it in the `packages:`
156    /// and `snapshots:` keys (post-`<name>@` part). For `file:` /
157    /// `link:` this is `file:./vendor/foo` / `link:../sibling`. For
158    /// `git`, pnpm uses the resolved form `<url>#<commit>` (no
159    /// `git+` prefix) because the lockfile pins to the exact commit
160    /// regardless of what the user wrote. Always emits POSIX
161    /// separators so the resulting lockfile is portable.
162    pub fn specifier(&self) -> String {
163        match self {
164            LocalSource::Git(g) => match &g.subpath {
165                Some(sub) => format!("{}#{}&path:/{}", g.url, g.resolved, sub),
166                None => format!("{}#{}", g.url, g.resolved),
167            },
168            LocalSource::RemoteTarball(t) => t.url.clone(),
169            _ => format!("{}:{}", self.kind_str(), self.path_posix()),
170        }
171    }
172
173    /// Internal FS-safe dep_path used as the key in
174    /// `LockfileGraph.packages` and as the `.aube/` subdir name.
175    ///
176    /// Distinct normalized paths must map to distinct keys (otherwise the
177    /// linker would silently mix files between two local packages),
178    /// and the result must be a single filesystem component — no
179    /// `/`, `\`, `:`, or `..`. Ad-hoc character substitution trips
180    /// over cases like `../vendor` vs `__/vendor` or `a.b` vs `a_b`
181    /// collapsing to the same string, so we hash the lexically normalized
182    /// path bytes and suffix the first 16 hex chars (64 bits — more than
183    /// enough to avoid collisions inside a single project). Normalization
184    /// also makes equivalent spellings such as `./vendor` and `vendor`
185    /// share one package identity without changing their serialized
186    /// specifiers.
187    ///
188    /// The hash input is the POSIX-form path string so a checked-in
189    /// lockfile resolves to the same key regardless of which
190    /// platform ran `aube install`.
191    pub fn dep_path(&self, name: &str) -> String {
192        use sha2::{Digest, Sha256};
193        let mut hasher = Sha256::new();
194        match self {
195            LocalSource::Git(g) => {
196                hasher.update(g.url.as_bytes());
197                hasher.update(b"#");
198                hasher.update(g.resolved.as_bytes());
199                if let Some(sub) = &g.subpath {
200                    hasher.update(b"&path:/");
201                    hasher.update(sub.as_bytes());
202                }
203            }
204            LocalSource::RemoteTarball(t) => {
205                hasher.update(t.url.as_bytes());
206            }
207            LocalSource::Directory(path)
208            | LocalSource::Tarball(path)
209            | LocalSource::Link(path)
210            | LocalSource::Portal(path)
211            | LocalSource::Exec(path) => {
212                let normalized = aube_util::path::normalize_lexical(path);
213                let posix = normalized.to_string_lossy().replace('\\', "/");
214                hasher.update(posix.as_bytes());
215            }
216        }
217        let digest = hasher.finalize();
218        let short: String = digest.iter().take(8).map(|b| format!("{b:02x}")).collect();
219        format!("{name}@{}+{short}", self.kind_str())
220    }
221
222    /// Classify a user-written `file:` / `link:` specifier against the
223    /// project root. Returns `None` if `spec` isn't a local specifier.
224    /// Resolves the target path relative to `project_root`; a `file:`
225    /// target that resolves to a `.tgz` / `.tar.gz` on disk is treated
226    /// as a tarball, anything else as a directory.
227    pub fn parse(spec: &str, project_root: &Path) -> Option<Self> {
228        // Check git first so URLs like `https://host/user/repo.git`
229        // aren't swallowed by the broader bare-http tarball check
230        // below.
231        if let Some((url, committish, subpath)) = parse_git_spec(spec) {
232            // `resolved` is filled in by the resolver after running
233            // `git ls-remote`. A lockfile round-trip that never
234            // re-resolves will leave this empty, which is the sentinel
235            // the resolver checks for before calling ls-remote.
236            return Some(LocalSource::Git(GitSource {
237                url,
238                committish,
239                resolved: String::new(),
240                integrity: None,
241                subpath,
242            }));
243        }
244        // Any remaining bare `http(s)://` URL is a remote tarball.
245        // npm semantics treat *all* non-git HTTP URLs in a dependency
246        // value as tarball URLs, so services that serve tarballs from
247        // URLs without a `.tgz` extension (pkg.pr.new, GitHub
248        // codeload, etc.) classify correctly here.
249        if Self::looks_like_remote_tarball_url(spec) {
250            return Some(LocalSource::RemoteTarball(RemoteTarballSource {
251                url: spec.to_string(),
252                integrity: String::new(),
253                git_hosted: false,
254            }));
255        }
256        let (kind, rest) = if let Some(r) = spec.strip_prefix("file:") {
257            ("file", r)
258        } else if let Some(r) = spec.strip_prefix("link:") {
259            ("link", r)
260        } else if let Some(r) = spec.strip_prefix("portal:") {
261            ("portal", r)
262        } else if let Some(r) = spec.strip_prefix("exec:") {
263            return Some(LocalSource::Exec(PathBuf::from(r)));
264        } else {
265            return None;
266        };
267        let rel = PathBuf::from(rest);
268        let abs = project_root.join(&rel);
269        if kind == "link" {
270            return Some(LocalSource::Link(rel));
271        }
272        if kind == "portal" {
273            return Some(LocalSource::Portal(rel));
274        }
275        if abs.is_file() && Self::path_looks_like_tarball(&rel) {
276            return Some(LocalSource::Tarball(rel));
277        }
278        Some(LocalSource::Directory(rel))
279    }
280
281    /// Whether a specifier looks like a direct HTTP(S) URL that should
282    /// be fetched as a tarball. Per npm semantics, *any* `http://` or
283    /// `https://` URL in a dependency value is a tarball URL — services
284    /// like pkg.pr.new, GitHub codeload, and private registries with
285    /// auth-token query strings serve tarballs from URLs that don't
286    /// carry a `.tgz` extension. Git URLs must already have been
287    /// ruled out by the caller (see [`parse_git_spec`]) so a
288    /// `.git`-suffixed URL doesn't get misclassified here.
289    pub fn looks_like_remote_tarball_url(spec: &str) -> bool {
290        spec.starts_with("https://") || spec.starts_with("http://")
291    }
292
293    pub fn path_looks_like_tarball(path: &Path) -> bool {
294        let name = match path.file_name().and_then(|n| n.to_str()) {
295            Some(n) => n,
296            None => return false,
297        };
298        let lower = name.to_ascii_lowercase();
299        lower.ends_with(".tgz") || lower.ends_with(".tar.gz")
300    }
301}
302
303/// Resolve a transitive dependency's recorded spec *value* to the same
304/// `dep_path` key the lockfile parser assigns the target package, for
305/// the two content-pinned source kinds that get shared globally (git
306/// and remote tarball).
307///
308/// pnpm records a git / remote-tarball dependency inside a snapshot's
309/// `dependencies:` map by its *resolved spec* — `<url>#<sha>` for git,
310/// the tarball URL for remote tarballs (e.g. request-promise-core lists
311/// `request: https://github.com/request/request.git#<sha>`). The parser,
312/// however, keys the package itself under [`LocalSource::dep_path`] — the
313/// short `name@git+<hash>` / `name@url+<hash>` form. A naive
314/// `format!("{name}@{value}")` lookup therefore points at a key that was
315/// never inserted into the graph, so:
316///
317/// * the linker's sibling symlink dangles (Node resolves the wrong
318///   `<name>` or none — the request-promise-core crash), and
319/// * the graph hasher skips the child entirely, so neither its content
320///   fingerprint nor its build/engine taint cascades into the parent's
321///   global-virtual-store hash.
322///
323/// Mirror `pnpm::read::push_direct`'s keying so the resolved value lands
324/// on the exact `dep_path` the package was materialized under. Returns
325/// `None` for every other value (plain semver, `file:`, `link:`, npm
326/// aliases, …) so callers keep the verbatim `name@value` key those
327/// already resolve correctly with.
328pub fn shared_local_dep_path(dep_name: &str, dep_value: &str) -> Option<String> {
329    // pnpm appends a `(peer@ver)` suffix to some spec values; the parser
330    // strips it before classifying the source, so strip it here too.
331    //
332    // This MUST stay byte-for-byte identical to `pnpm::read::push_direct`'s
333    // `classify_version` (`info.version.split('(').next()`), which is what
334    // produced the `dep_path` keys in `graph.packages` we're matching
335    // against. A "smarter" strip (e.g. only a trailing `(peer@…)` via
336    // rfind) would *desync* the two: any value with a non-peer `(` would
337    // hash differently here than the key the parser inserted, silently
338    // re-skipping that child in the linker and graph hasher. If the
339    // first-`(` truncation is ever wrong for a real spec, fix it in
340    // `push_direct` and here together — never in isolation.
341    let classify = dep_value.split('(').next().unwrap_or(dep_value);
342    match LocalSource::parse(classify, Path::new("")) {
343        Some(LocalSource::Git(mut git)) => {
344            // Snapshot specs carry the pinned commit after `#`, which
345            // `parse` records as `committish` rather than `resolved`. The
346            // package was keyed with that commit promoted to `resolved`
347            // (see `push_direct`), so promote it here too — otherwise the
348            // `url#resolved` hash diverges from the package's dep_path.
349            if git.resolved.is_empty() {
350                git.resolved = git.committish.take()?;
351            }
352            Some(LocalSource::Git(git).dep_path(dep_name))
353        }
354        Some(tarball @ LocalSource::RemoteTarball(_)) => Some(tarball.dep_path(dep_name)),
355        _ => None,
356    }
357}
358
359/// Resolve a dependency edge `(name, tail)` to the graph key of the child
360/// package node, honoring every reader's storage convention. Returns the
361/// first candidate that satisfies `contains` (the caller's "is this a real
362/// package key?" predicate), or `None` when the edge points outside the
363/// graph (a pruned optional, an unresolved peer, a `link:` target, …).
364///
365/// Three conventions coexist because the readers disagree on what a
366/// dependency *value* holds, and a graph walker that only knows one of
367/// them silently drops the others:
368///   1. `tail` verbatim — npm/yarn/bun store the full dep_path as the
369///      value (`"foo@1.2.3"`).
370///   2. `name@tail` — the pnpm reader stores only the tail (`"1.2.3"`),
371///      so the key is the name re-joined to it.
372///   3. [`shared_local_dep_path`] — git / remote-tarball deps store the
373///      resolved URL as the tail, but the node is keyed under the short
374///      `name@git+<hash>` / `name@url+<hash>` form. The linker's
375///      `materialize` already bridges the edge this way; reachability /
376///      marking walkers that skip it prune the entire git/tarball subtree
377///      (a content-pinned git/tarball child and everything under it
378///      vanishes from the walk once the node is keyed canonically).
379pub fn resolve_dep_edge(name: &str, tail: &str, contains: impl Fn(&str) -> bool) -> Option<String> {
380    if contains(tail) {
381        return Some(tail.to_string());
382    }
383    let rejoined = format!("{name}@{tail}");
384    if contains(&rejoined) {
385        return Some(rejoined);
386    }
387    shared_local_dep_path(name, tail).filter(|key| contains(key))
388}
389
390/// Parse a git dependency specifier into `(clone_url, committish)`.
391///
392/// Recognized forms:
393/// - `git+https://host/user/repo.git[#ref]`
394/// - `git+ssh://git@host/user/repo.git[#ref]`
395/// - `git://host/user/repo.git[#ref]`
396/// - `https://host/user/repo.git[#ref]` (only when ending in `.git`)
397/// - `user@host:path[.git][#ref]` (scp-form, only for github.com / gitlab.com /
398///   bitbucket.org — matches pnpm 11 behavior, where unknown SCP hosts are
399///   treated as local paths) → `ssh://user@host/path[.git]`
400/// - `github:user/repo[#ref]` → `https://github.com/user/repo.git`
401/// - `gitlab:user/repo[#ref]` → `https://gitlab.com/user/repo.git`
402/// - `bitbucket:user/repo[#ref]` → `https://bitbucket.org/user/repo.git`
403/// - `user/repo[#ref]` (bare GitHub shorthand, npm/pnpm compat)
404///   → `https://github.com/user/repo.git`
405///
406/// Returns `None` for any specifier that doesn't look like a git URL,
407/// so the caller can fall through to other protocol parsers.
408pub fn parse_git_spec(spec: &str) -> Option<(String, Option<String>, Option<String>)> {
409    let (body, committish, subpath) = match spec.find('#') {
410        Some(idx) => {
411            let (c, s) = parse_git_fragment(&spec[idx + 1..]);
412            (&spec[..idx], c, s)
413        }
414        None => (spec, None, None),
415    };
416    let is_bare_transport = body.starts_with("https://")
417        || body.starts_with("http://")
418        || body.starts_with("ssh://")
419        || body.starts_with("file://");
420    let url = if let Some(rest) = body.strip_prefix("git+") {
421        // `git+` explicitly tags the URL as git, so the `.git`
422        // suffix is optional (GitHub/GitLab accept both forms).
423        rest.to_string()
424    } else if body.starts_with("git://") {
425        body.to_string()
426    } else if let Some(scp) = parse_scp_url(body) {
427        scp
428    } else if let Some(path) = body.strip_prefix("github:") {
429        format!("https://github.com/{path}.git")
430    } else if let Some(path) = body.strip_prefix("gitlab:") {
431        format!("https://gitlab.com/{path}.git")
432    } else if let Some(path) = body.strip_prefix("bitbucket:") {
433        format!("https://bitbucket.org/{path}.git")
434    } else if is_bare_transport && body.ends_with(".git") {
435        body.to_string()
436    } else if is_bare_transport
437        && committish
438            .as_deref()
439            .is_some_and(|c| c.len() == 40 && c.chars().all(|ch| ch.is_ascii_hexdigit()))
440    {
441        // Lockfile round-trip form: `specifier()` writes the stored
442        // URL verbatim plus `#<sha>`. URLs that dropped the `git+`
443        // prefix (and happen to lack `.git`) are disambiguated from
444        // plain tarball URLs by the 40-hex committish suffix.
445        body.to_string()
446    } else if is_bare_github_shorthand(body) {
447        // npm/pnpm bare GitHub shorthand: `user/repo` expands to
448        // `github:user/repo`. Placed last so all explicit URL/scheme
449        // forms above shadow it.
450        format!("https://github.com/{body}.git")
451    } else {
452        return None;
453    };
454    Some((url, committish, subpath))
455}
456
457/// `user/repo` — a single `/`, both segments non-empty, ASCII
458/// alphanumeric + `_.-` only, owner doesn't start with `.` so
459/// single-component relative paths (`./repo`, `../repo`) are rejected.
460/// Excludes scoped npm names (`@scope/pkg`) and file paths. Other
461/// URL/SCP forms are ruled out by placement order in `parse_git_spec`.
462fn is_bare_github_shorthand(body: &str) -> bool {
463    let Some((owner, repo)) = body.split_once('/') else {
464        return false;
465    };
466    !owner.is_empty()
467        && !owner.starts_with('.')
468        && !repo.is_empty()
469        && !repo.contains('/')
470        && owner
471            .bytes()
472            .all(|b| b.is_ascii_alphanumeric() || matches!(b, b'_' | b'.' | b'-'))
473        && repo
474            .bytes()
475            .all(|b| b.is_ascii_alphanumeric() || matches!(b, b'_' | b'.' | b'-'))
476}
477
478/// A git URL that maps to one of the three "hosted" providers npm /
479/// pnpm both special-case (github / gitlab / bitbucket). For these
480/// hosts a public read can be served as a flat HTTPS tarball over
481/// `codeload.github.com` (or each host's equivalent), bypassing `git`
482/// entirely. The lockfile's stored URL is canonical-identity only —
483/// pnpm and npm both re-derive the fetch URL from `(host, owner,
484/// repo)` on every install rather than dialing whatever scheme
485/// happens to be in `resolved:`.
486#[derive(Debug, Clone, PartialEq, Eq)]
487pub struct HostedGit {
488    pub host: HostedGitHost,
489    pub owner: String,
490    pub repo: String,
491}
492
493#[derive(Debug, Clone, Copy, PartialEq, Eq)]
494pub enum HostedGitHost {
495    GitHub,
496    GitLab,
497    Bitbucket,
498}
499
500impl HostedGit {
501    /// `https://github.com/<owner>/<repo>.git` — the form `git fetch`
502    /// can dial without an SSH key. Used as the runtime fetch URL when
503    /// the lockfile's stored URL is `git+ssh://git@…` (npm canonical
504    /// identity) but the actual install host has no SSH configured.
505    pub fn https_url(&self) -> String {
506        let host = self.host.host_domain();
507        format!("https://{host}/{}/{}.git", self.owner, self.repo)
508    }
509
510    /// `https://codeload.github.com/<owner>/<repo>/tar.gz/<sha>` (or
511    /// each host's equivalent) — a flat HTTPS tarball at the given
512    /// commit. Returns `None` unless `committish` is a 40-char hex
513    /// SHA, since the codeload path can't be verified after extraction
514    /// without `.git/` metadata. Branch / tag names round-trip through
515    /// `git ls-remote` to get pinned to a SHA first.
516    pub fn tarball_url(&self, committish: &str) -> Option<String> {
517        if committish.len() != 40 || !committish.chars().all(|c| c.is_ascii_hexdigit()) {
518            return None;
519        }
520        let sha = committish.to_ascii_lowercase();
521        Some(match self.host {
522            HostedGitHost::GitHub => format!(
523                "https://codeload.github.com/{}/{}/tar.gz/{sha}",
524                self.owner, self.repo
525            ),
526            HostedGitHost::GitLab => format!(
527                "https://gitlab.com/{}/{}/-/archive/{sha}/{}-{sha}.tar.gz",
528                self.owner, self.repo, self.repo
529            ),
530            HostedGitHost::Bitbucket => format!(
531                "https://bitbucket.org/{}/{}/get/{sha}.tar.gz",
532                self.owner, self.repo
533            ),
534        })
535    }
536}
537
538impl HostedGitHost {
539    fn from_domain(domain: &str) -> Option<Self> {
540        match domain {
541            "github.com" => Some(HostedGitHost::GitHub),
542            "gitlab.com" => Some(HostedGitHost::GitLab),
543            "bitbucket.org" => Some(HostedGitHost::Bitbucket),
544            _ => None,
545        }
546    }
547
548    pub fn host_domain(self) -> &'static str {
549        match self {
550            HostedGitHost::GitHub => "github.com",
551            HostedGitHost::GitLab => "gitlab.com",
552            HostedGitHost::Bitbucket => "bitbucket.org",
553        }
554    }
555}
556
557/// Parse a clone URL — in any form `parse_git_spec` accepts as input
558/// or produces as output — into its `(host, owner, repo)` components,
559/// when the host is one of the three providers npm / pnpm route
560/// through HTTPS tarballs. Returns `None` for any other host (including
561/// self-hosted GitLab / Gitea / Bitbucket Data Center): those still
562/// need a real `git clone` because no codeload-style HTTP archive is
563/// available.
564///
565/// Accepts:
566/// - `https://github.com/owner/repo[.git]`
567/// - `git+https://github.com/owner/repo[.git]`
568/// - `git://github.com/owner/repo[.git]`
569/// - `ssh://git@github.com/owner/repo[.git]`
570/// - `git+ssh://git@github.com/owner/repo[.git]` (npm canonical lockfile form)
571/// - `git@github.com:owner/repo[.git]` (scp shorthand, in case a caller
572///   parses raw lockfile fields without going through `parse_git_spec`)
573pub fn parse_hosted_git(url: &str) -> Option<HostedGit> {
574    let body = url.strip_prefix("git+").unwrap_or(url);
575    let after_scheme = if let Some(rest) = body.strip_prefix("https://") {
576        rest
577    } else if let Some(rest) = body.strip_prefix("http://") {
578        rest
579    } else if let Some(rest) = body.strip_prefix("ssh://") {
580        rest
581    } else if let Some(rest) = body.strip_prefix("git://") {
582        rest
583    } else {
584        // scp shorthand `user@host:path` — not produced by parse_git_spec
585        // but accepted defensively in case a raw lockfile string ever
586        // bypasses it.
587        let scp_path = parse_scp_url(body)?;
588        return parse_hosted_git(&scp_path);
589    };
590    // Strip optional `user@` (always `git@` for hosted forms).
591    let host_and_path = match after_scheme.split_once('@') {
592        Some((_, rest)) => rest,
593        None => after_scheme,
594    };
595    let (host, path) = host_and_path.split_once('/')?;
596    let host = HostedGitHost::from_domain(host)?;
597    // Take exactly two path segments: owner and repo. Anything beyond
598    // (subgroup-style GitLab paths) doesn't have a stable HTTPS tarball
599    // form on the three providers we care about, so refuse and let the
600    // caller fall back to clone.
601    let mut segs = path.splitn(3, '/');
602    let owner = segs.next()?;
603    let repo = segs.next()?;
604    if owner.is_empty() || repo.is_empty() || segs.next().is_some() {
605        return None;
606    }
607    let repo = repo
608        .strip_suffix(".git")
609        .unwrap_or(repo)
610        .trim_end_matches('/');
611    if repo.is_empty() {
612        return None;
613    }
614    Some(HostedGit {
615        host,
616        owner: owner.to_string(),
617        repo: repo.to_string(),
618    })
619}
620
621fn parse_scp_url(body: &str) -> Option<String> {
622    if body.contains("://") {
623        return None;
624    }
625    let colon = body.find(':')?;
626    let before = &body[..colon];
627    let path = &body[colon + 1..];
628    if before.is_empty() || path.is_empty() {
629        return None;
630    }
631    if path.starts_with('/') {
632        return None;
633    }
634    let at = before.find('@')?;
635    let user = &before[..at];
636    let host = &before[at + 1..];
637    if user.is_empty() || host.is_empty() || host.contains('/') || host.contains('@') {
638        return None;
639    }
640    // pnpm 11 only resolves SCP-form as hosted Git for the three known
641    // providers; other hosts (e.g. `git@example.com:foo/bar.git`) are
642    // treated as local paths, and `host:path` without a user errors.
643    if !matches!(host, "github.com" | "gitlab.com" | "bitbucket.org") {
644        return None;
645    }
646    Some(format!("ssh://{user}@{host}/{path}"))
647}
648
649/// Normalize git URL fragments used by npm-compatible lockfiles.
650///
651/// Plain git accepts `#<ref>`, while npm and Yarn Berry also write
652/// key/value fragments such as `#commit=<sha>` for pinned git deps.
653/// Downstream code passes this value directly to `git ls-remote` and
654/// `git checkout`, so strip the selector key here and keep only the
655/// actual ref name or SHA.
656pub(crate) fn normalize_git_fragment(fragment: &str) -> Option<String> {
657    parse_git_fragment(fragment).0
658}
659
660/// Parse a git URL fragment into `(committish, subpath)`. Handles the
661/// pnpm/hosted-git-info form `<ref>&path:/sub/dir` (the `path:` key
662/// uses a colon, not `=`, by historical convention) as well as the
663/// `key=value` form npm/Yarn Berry write. Unknown selectors are
664/// ignored. Subpath is returned without leading slash so the caller
665/// can join it with a clone dir without tripping the absolute-path
666/// branch of `Path::join`.
667pub(crate) fn parse_git_fragment(fragment: &str) -> (Option<String>, Option<String>) {
668    if fragment.is_empty() {
669        return (None, None);
670    }
671
672    let mut fallback: Option<&str> = None;
673    let mut preferred: Option<&str> = None;
674    let mut subpath: Option<String> = None;
675    for part in fragment.split('&') {
676        if part.is_empty() {
677            continue;
678        }
679        // Try `key=value` first; fall back to `key:value` only for
680        // the small set of selectors we actually handle below. A tag
681        // name with a colon (e.g. `release:2026-01`) is left alone —
682        // and `semver:^1.0.0` stays as a literal ref so `ls-remote`
683        // surfaces an explicit error rather than silently HEAD-ing.
684        let split = part.split_once('=').or_else(|| {
685            part.split_once(':')
686                .filter(|(k, _)| matches!(*k, "commit" | "tag" | "head" | "branch" | "path"))
687        });
688        let (key, value) = split.unwrap_or(("", part));
689        if value.is_empty() {
690            continue;
691        }
692        match key {
693            "commit" => {
694                preferred.get_or_insert(value);
695            }
696            "tag" | "head" | "branch" => {
697                fallback.get_or_insert(value);
698            }
699            "path" => {
700                // Strip leading slashes (pnpm writes `path:/sub`) and
701                // reject any `..` / `.` component. Without this, a
702                // crafted spec like `&path:/../../etc` would let the
703                // resolver and installer escape the clone dir and
704                // import an arbitrary host directory into the store.
705                if subpath.is_some() {
706                    // First-wins, matching the other selectors above.
707                    continue;
708                }
709                let trimmed = value.trim_start_matches('/');
710                if trimmed.is_empty() {
711                    continue;
712                }
713                if trimmed
714                    .split('/')
715                    .any(|c| c.is_empty() || c == "." || c == "..")
716                {
717                    continue;
718                }
719                subpath = Some(trimmed.to_string());
720            }
721            "" => {
722                fallback.get_or_insert(value);
723            }
724            _ => {}
725        }
726    }
727
728    (preferred.or(fallback).map(ToString::to_string), subpath)
729}
730
731#[cfg(test)]
732mod tests {
733    use super::*;
734
735    #[test]
736    fn matches_https_tgz() {
737        assert!(LocalSource::looks_like_remote_tarball_url(
738            "https://example.com/pkg-1.0.0.tgz"
739        ));
740    }
741
742    #[test]
743    fn matches_http_tar_gz() {
744        assert!(LocalSource::looks_like_remote_tarball_url(
745            "http://example.com/pkg-1.0.0.tar.gz"
746        ));
747    }
748
749    #[test]
750    fn strips_fragment_before_suffix_check() {
751        assert!(LocalSource::looks_like_remote_tarball_url(
752            "https://example.com/pkg-1.0.0.tgz#sha512-abc"
753        ));
754    }
755
756    #[test]
757    fn strips_query_string_before_suffix_check() {
758        // Auth-token URLs from private registries (JFrog, Nexus,
759        // CodeArtifact, …) routinely trail `?token=…` after the
760        // filename. Must still classify as a tarball URL.
761        assert!(LocalSource::looks_like_remote_tarball_url(
762            "https://registry.example.com/pkg/-/pkg-1.0.0.tgz?token=abc"
763        ));
764        assert!(LocalSource::looks_like_remote_tarball_url(
765            "https://example.com/pkg-1.0.0.tar.gz?v=2&signed=1"
766        ));
767    }
768
769    #[test]
770    fn matches_bare_http_url_without_tarball_suffix() {
771        // pkg.pr.new serves tarballs from URLs without a `.tgz`
772        // extension; npm treats all non-git http(s) URLs as tarball
773        // URLs, so these must classify as remote tarballs.
774        assert!(LocalSource::looks_like_remote_tarball_url(
775            "https://pkg.pr.new/lunariajs/lunaria/@lunariajs/core@904b935"
776        ));
777        assert!(LocalSource::looks_like_remote_tarball_url(
778            "https://codeload.github.com/user/repo/tar.gz/main"
779        ));
780    }
781
782    #[test]
783    fn git_commits_match_only_allows_full_sha_prefix_pairs() {
784        let full = "abcdef0123456789abcdef0123456789abcdef01";
785        assert!(git_commits_match(full, "abcdef0"));
786        assert!(git_commits_match("abcdef0", full));
787        assert!(git_commits_match(full, full));
788        assert!(!git_commits_match("abcdef0", "abcdef012"));
789        assert!(!git_commits_match(full, "abcdef1"));
790        assert!(!git_commits_match("main", full));
791    }
792
793    #[test]
794    fn rejects_non_http_schemes() {
795        assert!(!LocalSource::looks_like_remote_tarball_url(
796            "ftp://example.com/pkg.tgz"
797        ));
798        assert!(!LocalSource::looks_like_remote_tarball_url(
799            "git://example.com/repo.git"
800        ));
801    }
802
803    #[test]
804    fn parse_classifies_bare_http_url_as_remote_tarball() {
805        use std::path::Path;
806        let parsed = LocalSource::parse(
807            "https://pkg.pr.new/lunariajs/lunaria/@lunariajs/core@904b935",
808            Path::new(""),
809        );
810        assert!(matches!(parsed, Some(LocalSource::RemoteTarball(_))));
811    }
812
813    #[test]
814    fn parse_prefers_git_over_tarball_for_dot_git_url() {
815        use std::path::Path;
816        let parsed = LocalSource::parse("https://github.com/user/repo.git", Path::new(""));
817        assert!(matches!(parsed, Some(LocalSource::Git(_))));
818    }
819
820    #[test]
821    fn parse_classifies_exec_as_local_source() {
822        let parsed = LocalSource::parse("exec:./scripts/generate.js", Path::new(""));
823        assert_eq!(
824            parsed,
825            Some(LocalSource::Exec(PathBuf::from("./scripts/generate.js")))
826        );
827    }
828
829    #[test]
830    fn git_plus_https_without_dot_git_roundtrips_via_lockfile_form() {
831        // Initial parse: `git+https://…/repo` (no `.git`).
832        let (url, committish, subpath) = parse_git_spec("git+https://host/user/repo").unwrap();
833        assert_eq!(url, "https://host/user/repo");
834        assert_eq!(committish, None);
835        assert_eq!(subpath, None);
836
837        // After resolving, the serializer writes `<url>#<sha>` into
838        // the lockfile's importer `version:` field.
839        let sha = "abcdef0123456789abcdef0123456789abcdef01";
840        let source = LocalSource::Git(GitSource {
841            url: url.clone(),
842            committish: None,
843            resolved: sha.to_string(),
844            integrity: None,
845            subpath: None,
846        });
847        let lockfile_version = source.specifier();
848        assert_eq!(lockfile_version, format!("https://host/user/repo#{sha}"));
849
850        // Re-parse must recognize the bare URL because the 40-hex
851        // committish suffix unambiguously tags it as git.
852        let (round_url, round_committish, round_subpath) =
853            parse_git_spec(&lockfile_version).unwrap();
854        assert_eq!(round_url, "https://host/user/repo");
855        assert_eq!(round_committish.as_deref(), Some(sha));
856        assert_eq!(round_subpath, None);
857    }
858
859    #[test]
860    fn bare_https_without_dot_git_and_no_committish_is_not_git() {
861        // A plain `https://…` URL with no `.git` and no SHA could be
862        // anything (including a tarball); don't claim it.
863        assert!(parse_git_spec("https://example.com/pkg").is_none());
864    }
865
866    #[test]
867    fn github_shorthand_expands_and_roundtrips() {
868        let (url, _, _) = parse_git_spec("github:user/repo").unwrap();
869        assert_eq!(url, "https://github.com/user/repo.git");
870    }
871
872    #[test]
873    fn bare_user_repo_expands_to_github() {
874        let (url, committish, subpath) = parse_git_spec("kevva/is-negative").unwrap();
875        assert_eq!(url, "https://github.com/kevva/is-negative.git");
876        assert!(committish.is_none());
877        assert!(subpath.is_none());
878    }
879
880    #[test]
881    fn bare_user_repo_with_committish_preserved() {
882        let (url, committish, _) = parse_git_spec("kevva/is-negative#v1.0.0").unwrap();
883        assert_eq!(url, "https://github.com/kevva/is-negative.git");
884        assert_eq!(committish.as_deref(), Some("v1.0.0"));
885    }
886
887    #[test]
888    fn bare_scope_pkg_is_not_git_shorthand() {
889        // npm-style `@scope/pkg` is a registry name, not a GitHub shorthand.
890        assert!(parse_git_spec("@types/node").is_none());
891    }
892
893    #[test]
894    fn bare_relative_path_is_not_git_shorthand() {
895        // Single-component relative paths split as owner=".", owner="..",
896        // so owner-starts-with-`.` is the load-bearing guard here.
897        assert!(parse_git_spec("./repo").is_none());
898        assert!(parse_git_spec("../repo").is_none());
899        // Multi-component relative paths additionally fail the
900        // single-`/`-only guard.
901        assert!(parse_git_spec("./local/path").is_none());
902        assert!(parse_git_spec("../local/path").is_none());
903    }
904
905    #[test]
906    fn bare_path_with_extra_slashes_is_not_git_shorthand() {
907        // Real GitHub shorthand is exactly `user/repo` — anything with a
908        // second `/` is a path, not a shorthand.
909        assert!(parse_git_spec("path/with/slashes/extra").is_none());
910    }
911
912    #[test]
913    fn bare_scp_form_unknown_host_is_not_github_shorthand() {
914        // `user@host:repo.git` is scp form (handled or rejected above);
915        // the bare-shorthand branch must not pick it up.
916        assert!(parse_git_spec("user@host:repo.git").is_none());
917    }
918
919    #[test]
920    fn scp_form_recognized() {
921        let (url, committish, _) =
922            parse_git_spec("git@github.com:EthanHenrickson/math-mcp.git").unwrap();
923        assert_eq!(url, "ssh://git@github.com/EthanHenrickson/math-mcp.git");
924        assert!(committish.is_none());
925    }
926
927    #[test]
928    fn scp_form_with_ref_recognized() {
929        let (url, committish, _) =
930            parse_git_spec("git@github.com:EthanHenrickson/math-mcp.git#0.1.5").unwrap();
931        assert_eq!(url, "ssh://git@github.com/EthanHenrickson/math-mcp.git");
932        assert_eq!(committish.as_deref(), Some("0.1.5"));
933    }
934
935    #[test]
936    fn scp_form_bitbucket_recognized() {
937        let (url, _, _) = parse_git_spec("git@bitbucket.org:pnpmjs/git-resolver.git").unwrap();
938        assert_eq!(url, "ssh://git@bitbucket.org/pnpmjs/git-resolver.git");
939    }
940
941    #[test]
942    fn scp_form_unknown_host_rejected() {
943        // pnpm 11 treats `user@unknown-host:path` as a local path, not Git.
944        assert!(parse_git_spec("git@example.com:org/repo.git").is_none());
945        assert!(parse_git_spec("alice@host.example.com:org/repo.git").is_none());
946    }
947
948    #[test]
949    fn scp_form_without_user_rejected() {
950        // pnpm 11 errors on bare `host:path` as unsupported.
951        assert!(parse_git_spec("github.com:user/repo.git").is_none());
952    }
953
954    #[test]
955    fn commit_selector_fragment_normalizes_to_sha() {
956        let sha = "abcdef0123456789abcdef0123456789abcdef01";
957        let (url, committish, _) =
958            parse_git_spec(&format!("https://host/user/repo.git#commit={sha}")).unwrap();
959        assert_eq!(url, "https://host/user/repo.git");
960        assert_eq!(committish.as_deref(), Some(sha));
961    }
962
963    #[test]
964    fn named_selector_fragment_normalizes_to_ref() {
965        let (url, committish, _) = parse_git_spec("git+https://host/user/repo#tag=v1.2.3").unwrap();
966        assert_eq!(url, "https://host/user/repo");
967        assert_eq!(committish.as_deref(), Some("v1.2.3"));
968    }
969
970    #[test]
971    fn pnpm_path_subpath_extracted_from_fragment() {
972        // pnpm syntax: `<url>#<ref>&path:/<subdir>` selects a
973        // subdirectory of the cloned repo as the package root.
974        let (url, committish, subpath) =
975            parse_git_spec("github:org/dep#v0.1.4&path:/packages/special").unwrap();
976        assert_eq!(url, "https://github.com/org/dep.git");
977        assert_eq!(committish.as_deref(), Some("v0.1.4"));
978        assert_eq!(subpath.as_deref(), Some("packages/special"));
979    }
980
981    #[test]
982    fn path_subpath_roundtrips_via_specifier() {
983        let sha = "abcdef0123456789abcdef0123456789abcdef01";
984        let source = LocalSource::Git(GitSource {
985            url: "https://github.com/org/dep.git".to_string(),
986            committish: None,
987            resolved: sha.to_string(),
988            integrity: None,
989            subpath: Some("packages/special".to_string()),
990        });
991        let spec = source.specifier();
992        assert_eq!(
993            spec,
994            format!("https://github.com/org/dep.git#{sha}&path:/packages/special")
995        );
996        let (url, committish, subpath) = parse_git_spec(&spec).unwrap();
997        assert_eq!(url, "https://github.com/org/dep.git");
998        assert_eq!(committish.as_deref(), Some(sha));
999        assert_eq!(subpath.as_deref(), Some("packages/special"));
1000    }
1001
1002    #[test]
1003    fn parse_hosted_git_recognizes_canonical_forms() {
1004        // All these point at the same (github.com, owner, repo) tuple
1005        // and must map to the same HostedGit so the runtime fetch URL
1006        // doesn't depend on which scheme the lockfile happens to record.
1007        let canonical = HostedGit {
1008            host: HostedGitHost::GitHub,
1009            owner: "owner".to_string(),
1010            repo: "repo".to_string(),
1011        };
1012        for spec in [
1013            "https://github.com/owner/repo.git",
1014            "https://github.com/owner/repo",
1015            "http://github.com/owner/repo.git",
1016            "git+https://github.com/owner/repo.git",
1017            "git+https://github.com/owner/repo",
1018            "git://github.com/owner/repo.git",
1019            "ssh://git@github.com/owner/repo.git",
1020            "git+ssh://git@github.com/owner/repo.git",
1021            "git@github.com:owner/repo.git",
1022        ] {
1023            assert_eq!(
1024                parse_hosted_git(spec).as_ref(),
1025                Some(&canonical),
1026                "spec {spec} should map to canonical HostedGit",
1027            );
1028        }
1029    }
1030
1031    #[test]
1032    fn parse_hosted_git_returns_none_for_non_hosted() {
1033        // Self-hosted GitLab / Gitea / arbitrary hosts: no codeload
1034        // template, so the codeload fast path doesn't apply.
1035        for spec in [
1036            "https://example.com/owner/repo.git",
1037            "ssh://git@gitea.internal/owner/repo.git",
1038            "git+ssh://git@gitlab.example.com/group/sub/repo.git",
1039            "https://github.com/owner/repo/sub",
1040            "https://github.com/owner",
1041        ] {
1042            assert!(
1043                parse_hosted_git(spec).is_none(),
1044                "spec {spec} must not match a hosted provider",
1045            );
1046        }
1047    }
1048
1049    #[test]
1050    fn hosted_tarball_url_only_for_full_sha() {
1051        let g = HostedGit {
1052            host: HostedGitHost::GitHub,
1053            owner: "o".to_string(),
1054            repo: "r".to_string(),
1055        };
1056        let sha = "abcdef0123456789abcdef0123456789abcdef01";
1057        assert_eq!(
1058            g.tarball_url(sha).as_deref(),
1059            Some("https://codeload.github.com/o/r/tar.gz/abcdef0123456789abcdef0123456789abcdef01"),
1060        );
1061        // Branch / tag / abbreviated SHA don't take the fast path —
1062        // codeload accepts them but the wrapper-dir name varies and
1063        // we can't verify a non-SHA committish post-extraction.
1064        assert!(g.tarball_url("main").is_none());
1065        assert!(g.tarball_url("v1.2.3").is_none());
1066        assert!(g.tarball_url("abcdef0").is_none());
1067    }
1068
1069    #[test]
1070    fn hosted_tarball_url_per_provider() {
1071        let sha = "abcdef0123456789abcdef0123456789abcdef01";
1072        let gitlab = HostedGit {
1073            host: HostedGitHost::GitLab,
1074            owner: "g".to_string(),
1075            repo: "r".to_string(),
1076        }
1077        .tarball_url(sha)
1078        .unwrap();
1079        assert!(gitlab.starts_with("https://gitlab.com/g/r/-/archive/"));
1080        assert!(gitlab.ends_with("/r-abcdef0123456789abcdef0123456789abcdef01.tar.gz"));
1081        let bitbucket = HostedGit {
1082            host: HostedGitHost::Bitbucket,
1083            owner: "g".to_string(),
1084            repo: "r".to_string(),
1085        }
1086        .tarball_url(sha)
1087        .unwrap();
1088        assert_eq!(
1089            bitbucket,
1090            "https://bitbucket.org/g/r/get/abcdef0123456789abcdef0123456789abcdef01.tar.gz",
1091        );
1092    }
1093
1094    #[test]
1095    fn hosted_https_url_normalizes() {
1096        let g = parse_hosted_git("git+ssh://git@github.com/owner/repo.git").unwrap();
1097        assert_eq!(g.https_url(), "https://github.com/owner/repo.git");
1098    }
1099
1100    #[test]
1101    fn path_traversal_components_in_subpath_are_rejected() {
1102        // `..` and `.` components would let a crafted spec escape the
1103        // clone dir at install time. The parser drops them so the
1104        // resolver/installer never see a traversal-laden subpath.
1105        let cases = [
1106            "github:org/dep#main&path:/../../etc",
1107            "github:org/dep#main&path:/packages/../../../etc",
1108            "github:org/dep#main&path:/./packages/foo",
1109            "github:org/dep#main&path:/packages//foo",
1110        ];
1111        for spec in cases {
1112            let (_, _, subpath) = parse_git_spec(spec).unwrap();
1113            assert_eq!(subpath, None, "spec should drop subpath: {spec}");
1114        }
1115    }
1116
1117    #[test]
1118    fn dep_path_distinguishes_subpaths_under_same_commit() {
1119        // Two packages from the same repo+commit but different
1120        // subdirs must hash to distinct dep_paths so the linker
1121        // doesn't collapse them.
1122        let sha = "abcdef0123456789abcdef0123456789abcdef01";
1123        let a = LocalSource::Git(GitSource {
1124            url: "https://example.com/r.git".to_string(),
1125            committish: None,
1126            resolved: sha.to_string(),
1127            integrity: None,
1128            subpath: Some("packages/a".to_string()),
1129        });
1130        let b = LocalSource::Git(GitSource {
1131            url: "https://example.com/r.git".to_string(),
1132            committish: None,
1133            resolved: sha.to_string(),
1134            integrity: None,
1135            subpath: Some("packages/b".to_string()),
1136        });
1137        assert_ne!(a.dep_path("dep"), b.dep_path("dep"));
1138    }
1139
1140    #[test]
1141    fn dep_path_normalizes_equivalent_local_paths() {
1142        let root = LocalSource::Directory(PathBuf::from("./injected/lib-b"));
1143        let transitive = LocalSource::Directory(PathBuf::from("injected/lib-a/../lib-b"));
1144
1145        assert_eq!(root.dep_path("lib-b"), transitive.dep_path("lib-b"));
1146        assert_eq!(root.specifier(), "file:./injected/lib-b");
1147    }
1148
1149    const SHARED_SHA: &str = "0123456789abcdef0123456789abcdef01234567";
1150
1151    /// The dep_path the lockfile parser keys a git package under, given
1152    /// its normalized clone URL and pinned commit.
1153    fn git_key(url: &str, resolved: &str) -> String {
1154        LocalSource::Git(GitSource {
1155            url: url.to_string(),
1156            committish: None,
1157            resolved: resolved.to_string(),
1158            integrity: None,
1159            subpath: None,
1160        })
1161        .dep_path("request")
1162    }
1163
1164    /// The dep_path the lockfile parser keys a remote-tarball package
1165    /// under, given its fetch URL.
1166    fn tarball_key(url: &str) -> String {
1167        LocalSource::RemoteTarball(RemoteTarballSource {
1168            url: url.to_string(),
1169            integrity: String::new(),
1170            git_hosted: false,
1171        })
1172        .dep_path("request")
1173    }
1174
1175    #[test]
1176    fn shared_github_shorthand_maps_to_git_dep_path() {
1177        // A dependent records its git `request` via the `github:` spec,
1178        // but the package is keyed under the hashed `git+` dep_path. The
1179        // sibling symlink / hasher lookup must use that same key or it
1180        // dangles / silently skips the child.
1181        let got = shared_local_dep_path("request", &format!("github:request/request#{SHARED_SHA}"))
1182            .expect("github: spec is a shareable local source");
1183        assert_eq!(
1184            got,
1185            git_key("https://github.com/request/request.git", SHARED_SHA)
1186        );
1187        assert!(got.starts_with("request@git+"), "unexpected key: {got}");
1188    }
1189
1190    #[test]
1191    fn shared_git_url_and_shorthand_converge() {
1192        // Whether the dependent recorded the shorthand or the resolved
1193        // `<url>.git#<sha>` form, both must canonicalize to one key.
1194        let from_shorthand =
1195            shared_local_dep_path("request", &format!("github:request/request#{SHARED_SHA}"))
1196                .unwrap();
1197        let from_url = shared_local_dep_path(
1198            "request",
1199            &format!("https://github.com/request/request.git#{SHARED_SHA}"),
1200        )
1201        .unwrap();
1202        assert_eq!(from_shorthand, from_url);
1203    }
1204
1205    #[test]
1206    fn shared_missing_resolved_is_promoted_from_committish() {
1207        // A lockfile round-trip that never re-resolved leaves `resolved`
1208        // empty and only carries `#<committish>`; the helper must promote
1209        // it so the hash matches the package's `<url>#<sha>` key.
1210        let got = shared_local_dep_path(
1211            "request",
1212            &format!("https://github.com/request/request.git#{SHARED_SHA}"),
1213        )
1214        .unwrap();
1215        assert_eq!(
1216            got,
1217            git_key("https://github.com/request/request.git", SHARED_SHA)
1218        );
1219    }
1220
1221    #[test]
1222    fn shared_codeload_tarball_maps_to_url_dep_path() {
1223        // The exact form pnpm records for a `github:` dep that resolves to
1224        // a codeload archive. This is the case that crashed
1225        // request-promise-core under the global virtual store.
1226        let url = format!("https://codeload.github.com/request/request/tar.gz/{SHARED_SHA}");
1227        let got = shared_local_dep_path("request", &url).unwrap();
1228        assert_eq!(got, tarball_key(&url));
1229        assert!(got.starts_with("request@url+"), "unexpected key: {got}");
1230    }
1231
1232    #[test]
1233    fn shared_strips_peer_suffix_before_classifying() {
1234        let url = format!("https://codeload.github.com/request/request/tar.gz/{SHARED_SHA}");
1235        let with_peer = format!("{url}(typescript@5.8.3)");
1236        assert_eq!(
1237            shared_local_dep_path("request", &with_peer),
1238            shared_local_dep_path("request", &url),
1239        );
1240    }
1241
1242    #[test]
1243    fn shared_returns_none_for_non_shareable_specs() {
1244        for value in [
1245            "4.18.1",
1246            "^1.2.3",
1247            "link:../sibling",
1248            "file:./vendor/x",
1249            "npm:lodash@4.18.1",
1250        ] {
1251            assert!(
1252                shared_local_dep_path("dep", value).is_none(),
1253                "{value:?} must not be treated as a shareable local source",
1254            );
1255        }
1256    }
1257}