cargo/workspace/
source_id.rs

1use crate::context;
2use crate::sources::registry::CRATES_IO_HTTP_INDEX;
3use crate::sources::source::Source;
4use crate::sources::{CRATES_IO_DOMAIN, CRATES_IO_INDEX, CRATES_IO_REGISTRY, DirectorySource};
5use crate::sources::{GitSource, PathSource, RegistrySource};
6use crate::util::data_structures::HashSet;
7use crate::util::interning::InternedString;
8use crate::util::{CanonicalUrl, CargoResult, GlobalContext, IntoUrl};
9use crate::workspace::GitReference;
10use crate::workspace::SourceKind;
11use anyhow::Context as _;
12use serde::de;
13use serde::ser;
14use std::cmp::{self, Ordering};
15use std::fmt::{self, Formatter};
16use std::hash::{self, Hash};
17use std::path::{Path, PathBuf};
18use std::ptr;
19use std::sync::Mutex;
20use std::sync::OnceLock;
21use tracing::trace;
22use url::Url;
23
24static SOURCE_ID_CACHE: OnceLock<Mutex<HashSet<&'static SourceIdInner>>> = OnceLock::new();
25
26/// Unique identifier for a source of packages.
27///
28/// Cargo uniquely identifies packages using [`PackageId`], a combination of the
29/// package name, version, and the code source. `SourceId` exactly represents
30/// the "code source" in `PackageId`. See [`SourceId::hash`] to learn what are
31/// taken into account for the uniqueness of a source.
32///
33/// `SourceId` is usually associated with an instance of [`Source`], which is
34/// supposed to provide a `SourceId` via [`Source::source_id`] method.
35///
36/// [`Source`]: crate::sources::source::Source
37/// [`Source::source_id`]: crate::sources::source::Source::source_id
38/// [`PackageId`]: super::PackageId
39#[derive(Clone, Copy, Eq, Debug)]
40pub struct SourceId {
41    inner: &'static SourceIdInner,
42}
43
44/// The interned version of [`SourceId`] to avoid excessive clones and borrows.
45/// Values are cached in `SOURCE_ID_CACHE` once created.
46#[derive(Eq, Clone, Debug)]
47struct SourceIdInner {
48    /// The source URL.
49    url: Url,
50    /// The canonical version of the above url. See [`CanonicalUrl`] to learn
51    /// why it is needed and how it normalizes a URL.
52    canonical_url: CanonicalUrl,
53    /// The source kind.
54    kind: SourceKind,
55    /// For example, the exact Git revision of the specified branch for a Git Source.
56    precise: Option<Precise>,
57    /// Name of the remote registry.
58    ///
59    /// WARNING: this is not always set when the name is not known,
60    /// e.g. registry coming from `--index` or Cargo.lock
61    registry_key: Option<KeyOf>,
62}
63
64#[derive(Eq, PartialEq, Clone, Debug, Hash)]
65enum Precise {
66    Locked,
67    Updated {
68        name: InternedString,
69        from: semver::Version,
70        to: semver::Version,
71    },
72    GitUrlFragment(String),
73}
74
75impl fmt::Display for Precise {
76    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
77        match self {
78            Precise::Locked => "locked".fmt(f),
79            Precise::Updated { name, from, to } => {
80                write!(f, "{name}={from}->{to}")
81            }
82            Precise::GitUrlFragment(s) => s.fmt(f),
83        }
84    }
85}
86
87/// Where the remote source key is defined.
88///
89/// The purpose of this is to provide better diagnostics for different sources of keys.
90#[derive(Debug, Clone, PartialEq, Eq)]
91enum KeyOf {
92    /// Defined in the `[registries]` table or the built-in `crates-io` key.
93    Registry(String),
94    /// Defined in the `[source]` replacement table.
95    Source(String),
96}
97
98impl SourceId {
99    /// Creates a `SourceId` object from the kind and URL.
100    ///
101    /// The canonical url will be calculated, but the precise field will not
102    fn new(kind: SourceKind, url: Url, key: Option<KeyOf>) -> CargoResult<SourceId> {
103        if kind == SourceKind::SparseRegistry {
104            // Sparse URLs are different because they store the kind prefix (sparse+)
105            // in the URL. This is because the prefix is necessary to differentiate
106            // from regular registries (git-based). The sparse+ prefix is included
107            // everywhere, including user-facing locations such as the `config.toml`
108            // file that defines the registry, or whenever Cargo displays it to the user.
109            assert!(url.as_str().starts_with("sparse+"));
110        }
111        let source_id = SourceId::wrap(SourceIdInner {
112            kind,
113            canonical_url: CanonicalUrl::new(&url)?,
114            url,
115            precise: None,
116            registry_key: key,
117        });
118        Ok(source_id)
119    }
120
121    /// Interns the value and returns the wrapped type.
122    fn wrap(inner: SourceIdInner) -> SourceId {
123        let mut cache = SOURCE_ID_CACHE
124            .get_or_init(|| Default::default())
125            .lock()
126            .unwrap();
127        let inner = cache.get(&inner).cloned().unwrap_or_else(|| {
128            let inner = Box::leak(Box::new(inner));
129            cache.insert(inner);
130            inner
131        });
132        SourceId { inner }
133    }
134
135    fn remote_source_kind(url: &Url) -> SourceKind {
136        if url.as_str().starts_with("sparse+") {
137            SourceKind::SparseRegistry
138        } else {
139            SourceKind::Registry
140        }
141    }
142
143    /// Parses a source URL and returns the corresponding ID.
144    ///
145    /// ## Example
146    ///
147    /// ```
148    /// use cargo::workspace::SourceId;
149    /// SourceId::from_url("git+https://github.com/alexcrichton/\
150    ///                     libssh2-static-sys#80e71a3021618eb05\
151    ///                     656c58fb7c5ef5f12bc747f");
152    /// ```
153    pub fn from_url(string: &str) -> CargoResult<SourceId> {
154        let (kind, url) = string
155            .split_once('+')
156            .ok_or_else(|| anyhow::format_err!("invalid source `{}`", string))?;
157
158        match kind {
159            "git" => {
160                let mut url = url.into_url()?;
161                let reference = GitReference::from_query(url.query_pairs());
162                let precise = url.fragment().map(|s| s.to_owned());
163                url.set_fragment(None);
164                url.set_query(None);
165                Ok(SourceId::for_git(&url, reference)?.with_git_precise(precise))
166            }
167            "registry" => {
168                let url = url.into_url()?;
169                Ok(SourceId::new(SourceKind::Registry, url, None)?.with_locked_precise())
170            }
171            "sparse" => {
172                let url = string.into_url()?;
173                Ok(SourceId::new(SourceKind::SparseRegistry, url, None)?.with_locked_precise())
174            }
175            "path" => {
176                let url = url.into_url()?;
177                SourceId::new(SourceKind::Path, url, None)
178            }
179            kind => Err(anyhow::format_err!("unsupported source protocol: {}", kind)),
180        }
181    }
182
183    /// A view of the [`SourceId`] that can be `Display`ed as a URL.
184    pub fn as_url(&self) -> SourceIdAsUrl<'_> {
185        SourceIdAsUrl {
186            inner: &*self.inner,
187            encoded: false,
188        }
189    }
190
191    /// Like [`Self::as_url`] but with URL parameters encoded.
192    pub fn as_encoded_url(&self) -> SourceIdAsUrl<'_> {
193        SourceIdAsUrl {
194            inner: &*self.inner,
195            encoded: true,
196        }
197    }
198
199    /// Creates a `SourceId` from a filesystem path.
200    ///
201    /// `path`: an absolute path.
202    pub fn for_path(path: &Path) -> CargoResult<SourceId> {
203        let url = path.into_url()?;
204        SourceId::new(SourceKind::Path, url, None)
205    }
206
207    /// Creates a `SourceId` from a filesystem path.
208    ///
209    /// `path`: an absolute path.
210    pub fn for_manifest_path(manifest_path: &Path) -> CargoResult<SourceId> {
211        if crate::workspace::parser::is_embedded(manifest_path) && manifest_path.is_file() {
212            Self::for_path(manifest_path)
213        } else {
214            Self::for_path(manifest_path.parent().unwrap())
215        }
216    }
217
218    /// Creates a `SourceId` from a Git reference.
219    pub fn for_git(url: &Url, reference: GitReference) -> CargoResult<SourceId> {
220        SourceId::new(SourceKind::Git(reference), url.clone(), None)
221    }
222
223    /// Creates a `SourceId` from a remote registry URL when the registry name
224    /// cannot be determined, e.g. a user passes `--index` directly from CLI.
225    ///
226    /// Use [`SourceId::for_alt_registry`] if a name can provided, which
227    /// generates better messages for cargo.
228    pub fn for_registry(url: &Url) -> CargoResult<SourceId> {
229        let kind = Self::remote_source_kind(url);
230        SourceId::new(kind, url.to_owned(), None)
231    }
232
233    /// Creates a `SourceId` for a remote registry from the `[registries]` table or crates.io.
234    pub fn for_alt_registry(url: &Url, key: &str) -> CargoResult<SourceId> {
235        let kind = Self::remote_source_kind(url);
236        let key = KeyOf::Registry(key.into());
237        SourceId::new(kind, url.to_owned(), Some(key))
238    }
239
240    /// Creates a `SourceId` for a remote registry from the `[source]` replacement table.
241    pub fn for_source_replacement_registry(url: &Url, key: &str) -> CargoResult<SourceId> {
242        let kind = Self::remote_source_kind(url);
243        let key = KeyOf::Source(key.into());
244        SourceId::new(kind, url.to_owned(), Some(key))
245    }
246
247    /// Creates a `SourceId` from a local registry path.
248    pub fn for_local_registry(path: &Path) -> CargoResult<SourceId> {
249        let url = path.into_url()?;
250        SourceId::new(SourceKind::LocalRegistry, url, None)
251    }
252
253    /// Creates a `SourceId` from a directory path.
254    pub fn for_directory(path: &Path) -> CargoResult<SourceId> {
255        let url = path.into_url()?;
256        SourceId::new(SourceKind::Directory, url, None)
257    }
258
259    /// Returns the `SourceId` corresponding to the main repository.
260    ///
261    /// This is the main cargo registry by default, but it can be overridden in
262    /// a `.cargo/config.toml`.
263    pub fn crates_io(gctx: &GlobalContext) -> CargoResult<SourceId> {
264        gctx.crates_io_source_id()
265    }
266
267    /// Returns the `SourceId` corresponding to the main repository, using the
268    /// sparse HTTP index if allowed.
269    pub fn crates_io_maybe_sparse_http(gctx: &GlobalContext) -> CargoResult<SourceId> {
270        if Self::crates_io_is_sparse(gctx)? {
271            gctx.check_registry_index_not_set()?;
272            let url = CRATES_IO_HTTP_INDEX.into_url().unwrap();
273            let key = KeyOf::Registry(CRATES_IO_REGISTRY.into());
274            SourceId::new(SourceKind::SparseRegistry, url, Some(key))
275        } else {
276            Self::crates_io(gctx)
277        }
278    }
279
280    /// Returns whether to access crates.io over the sparse protocol.
281    pub fn crates_io_is_sparse(gctx: &GlobalContext) -> CargoResult<bool> {
282        let proto: Option<context::Value<String>> = gctx.get("registries.crates-io.protocol")?;
283        let is_sparse = match proto.as_ref().map(|v| v.val.as_str()) {
284            Some("sparse") => true,
285            Some("git") => false,
286            Some(unknown) => anyhow::bail!(
287                "unsupported registry protocol `{unknown}` (defined in {})",
288                proto.as_ref().unwrap().definition
289            ),
290            None => true,
291        };
292        Ok(is_sparse)
293    }
294
295    /// Gets the `SourceId` associated with given name of the remote registry.
296    pub fn alt_registry(gctx: &GlobalContext, key: &str) -> CargoResult<SourceId> {
297        if key == CRATES_IO_REGISTRY {
298            return Self::crates_io(gctx);
299        }
300        let url = gctx.get_registry_index(key)?;
301        Self::for_alt_registry(&url, key)
302    }
303
304    /// Gets this source URL.
305    pub fn url(&self) -> &Url {
306        &self.inner.url
307    }
308
309    /// Gets the canonical URL of this source, used for internal comparison
310    /// purposes.
311    pub fn canonical_url(&self) -> &CanonicalUrl {
312        &self.inner.canonical_url
313    }
314
315    /// Displays the text "crates.io index" for Cargo shell status output.
316    pub fn display_index(self) -> String {
317        if self.is_crates_io() {
318            format!("{} index", CRATES_IO_DOMAIN)
319        } else {
320            format!("`{}` index", self.display_registry_name())
321        }
322    }
323
324    /// Displays the name of a registry if it has one. Otherwise just the URL.
325    pub fn display_registry_name(self) -> String {
326        if let Some(key) = self.inner.registry_key.as_ref().map(|k| k.key()) {
327            key.into()
328        } else if self.has_precise() {
329            // We remove `precise` here to retrieve an permissive version of
330            // `SourceIdInner`, which may contain the registry name.
331            self.without_precise().display_registry_name()
332        } else {
333            url_display(self.url())
334        }
335    }
336
337    /// Gets the name of the remote registry as defined in the `[registries]` table,
338    /// or the built-in `crates-io` key.
339    pub fn alt_registry_key(&self) -> Option<&str> {
340        self.inner.registry_key.as_ref()?.alternative_registry()
341    }
342
343    /// Returns `true` if this source is from a filesystem path.
344    pub fn is_path(self) -> bool {
345        self.inner.kind == SourceKind::Path
346    }
347
348    /// Returns the local path if this is a path dependency.
349    pub fn local_path(self) -> Option<PathBuf> {
350        if self.inner.kind != SourceKind::Path {
351            return None;
352        }
353
354        Some(self.inner.url.to_file_path().unwrap())
355    }
356
357    pub fn kind(&self) -> &SourceKind {
358        &self.inner.kind
359    }
360
361    /// Returns `true` if this source is from a registry (either local or not).
362    pub fn is_registry(self) -> bool {
363        matches!(
364            self.inner.kind,
365            SourceKind::Registry | SourceKind::SparseRegistry | SourceKind::LocalRegistry
366        )
367    }
368
369    /// Returns `true` if this source is from a sparse registry.
370    pub fn is_sparse(self) -> bool {
371        matches!(self.inner.kind, SourceKind::SparseRegistry)
372    }
373
374    /// Returns `true` if this source is a "remote" registry.
375    ///
376    /// "remote" may also mean a file URL to a git index, so it is not
377    /// necessarily "remote". This just means it is not `local-registry`.
378    pub fn is_remote_registry(self) -> bool {
379        matches!(
380            self.inner.kind,
381            SourceKind::Registry | SourceKind::SparseRegistry
382        )
383    }
384
385    /// Returns `true` if this source from a Git repository.
386    pub fn is_git(self) -> bool {
387        matches!(self.inner.kind, SourceKind::Git(_))
388    }
389
390    /// Creates an implementation of `Source` corresponding to this ID.
391    pub fn load<'a>(self, gctx: &'a GlobalContext) -> CargoResult<Box<dyn Source + 'a>> {
392        trace!("loading SourceId; {}", self);
393        match self.inner.kind {
394            SourceKind::Git(..) => Ok(Box::new(GitSource::new(self, gctx)?)),
395            SourceKind::Path => {
396                let path = self
397                    .inner
398                    .url
399                    .to_file_path()
400                    .expect("path sources cannot be remote");
401                if crate::workspace::parser::is_embedded(&path) && path.is_file() {
402                    anyhow::bail!("single file packages cannot be used as dependencies")
403                }
404                Ok(Box::new(PathSource::new(&path, self, gctx)))
405            }
406            SourceKind::Registry | SourceKind::SparseRegistry => {
407                Ok(Box::new(RegistrySource::remote(self, gctx)?))
408            }
409            SourceKind::LocalRegistry => {
410                let path = self
411                    .inner
412                    .url
413                    .to_file_path()
414                    .expect("path sources cannot be remote");
415                Ok(Box::new(RegistrySource::local(self, &path, gctx)))
416            }
417            SourceKind::Directory => {
418                let path = self
419                    .inner
420                    .url
421                    .to_file_path()
422                    .expect("path sources cannot be remote");
423                Ok(Box::new(DirectorySource::new(&path, self, gctx)))
424            }
425        }
426    }
427
428    /// Gets the Git reference if this is a git source, otherwise `None`.
429    pub fn git_reference(self) -> Option<&'static GitReference> {
430        match self.inner.kind {
431            SourceKind::Git(ref s) => Some(s),
432            _ => None,
433        }
434    }
435
436    /// Check if the precise data field has bean set
437    pub fn has_precise(self) -> bool {
438        self.inner.precise.is_some()
439    }
440
441    /// Check if the precise data field has bean set to "locked"
442    pub fn has_locked_precise(self) -> bool {
443        self.inner.precise == Some(Precise::Locked)
444    }
445
446    /// Check if two sources have the same precise data field
447    pub fn has_same_precise_as(self, other: Self) -> bool {
448        self.inner.precise == other.inner.precise
449    }
450
451    /// Check if the precise data field stores information for this `name`
452    /// from a call to [`SourceId::with_precise_registry_version`].
453    ///
454    /// If so return the version currently in the lock file and the version to be updated to.
455    pub fn precise_registry_version(
456        self,
457        pkg: &str,
458    ) -> Option<(&semver::Version, &semver::Version)> {
459        match &self.inner.precise {
460            Some(Precise::Updated { name, from, to }) if name == pkg => Some((from, to)),
461            _ => None,
462        }
463    }
464
465    pub fn precise_git_fragment(self) -> Option<&'static str> {
466        match &self.inner.precise {
467            Some(Precise::GitUrlFragment(s)) => Some(&s),
468            _ => None,
469        }
470    }
471
472    /// Creates a new `SourceId` from this source with the given `precise`.
473    pub fn with_git_precise(self, fragment: Option<String>) -> SourceId {
474        self.with_precise(&fragment.map(|f| Precise::GitUrlFragment(f)))
475    }
476
477    /// Creates a new `SourceId` from this source without a `precise`.
478    pub fn without_precise(self) -> SourceId {
479        self.with_precise(&None)
480    }
481
482    /// Creates a new `SourceId` from this source without a `precise`.
483    pub fn with_locked_precise(self) -> SourceId {
484        self.with_precise(&Some(Precise::Locked))
485    }
486
487    /// Creates a new `SourceId` from this source with the `precise` from some other `SourceId`.
488    pub fn with_precise_from(self, v: Self) -> SourceId {
489        self.with_precise(&v.inner.precise)
490    }
491
492    fn with_precise(self, precise: &Option<Precise>) -> SourceId {
493        if &self.inner.precise == precise {
494            self
495        } else {
496            SourceId::wrap(SourceIdInner {
497                precise: precise.clone(),
498                ..(*self.inner).clone()
499            })
500        }
501    }
502
503    /// When updating a lock file on a version using `cargo update --precise`
504    /// the requested version is stored in the precise field.
505    /// On a registry dependency we also need to keep track of the package that
506    /// should be updated and even which of the versions should be updated.
507    /// All of this gets encoded in the precise field using this method.
508    /// The data can be read with [`SourceId::precise_registry_version`]
509    pub fn with_precise_registry_version(
510        self,
511        name: InternedString,
512        version: semver::Version,
513        precise: &str,
514    ) -> CargoResult<SourceId> {
515        let precise = semver::Version::parse(precise).with_context(|| {
516            if let Some(stripped) = precise.strip_prefix("v") {
517                return format!(
518                    "the version provided, `{precise}` is not a \
519                    valid SemVer version\n\n\
520                    help: try changing the version to `{stripped}`",
521                );
522            }
523            format!("invalid version format for precise version `{precise}`")
524        })?;
525
526        Ok(SourceId::wrap(SourceIdInner {
527            precise: Some(Precise::Updated {
528                name,
529                from: version,
530                to: precise,
531            }),
532            ..(*self.inner).clone()
533        }))
534    }
535
536    /// Returns `true` if the remote registry is the standard <https://crates.io>.
537    pub fn is_crates_io(self) -> bool {
538        match self.inner.kind {
539            SourceKind::Registry | SourceKind::SparseRegistry => {}
540            _ => return false,
541        }
542        let url = self.inner.url.as_str();
543        url == CRATES_IO_INDEX || url == CRATES_IO_HTTP_INDEX || is_overridden_crates_io_url(url)
544    }
545
546    /// Hashes `self` to be used in the name of some Cargo folders, so shouldn't vary.
547    ///
548    /// For git and url, `as_str` gives the serialisation of a url (which has a spec) and so
549    /// insulates against possible changes in how the url crate does hashing.
550    ///
551    /// For paths, remove the workspace prefix so the same source will give the
552    /// same hash in different locations, helping reproducible builds.
553    pub fn stable_hash<S: hash::Hasher>(self, workspace: &Path, into: &mut S) {
554        if self.is_path() {
555            if let Ok(p) = self
556                .inner
557                .url
558                .to_file_path()
559                .unwrap()
560                .strip_prefix(workspace)
561            {
562                self.inner.kind.hash(into);
563                p.to_str().unwrap().hash(into);
564                return;
565            }
566        }
567        self.inner.kind.hash(into);
568        match self.inner.kind {
569            SourceKind::Git(_) => (&self).inner.canonical_url.hash(into),
570            _ => (&self).inner.url.as_str().hash(into),
571        }
572    }
573
574    pub fn full_eq(self, other: SourceId) -> bool {
575        ptr::eq(self.inner, other.inner)
576    }
577
578    pub fn full_hash<S: hash::Hasher>(self, into: &mut S) {
579        ptr::NonNull::from(self.inner).hash(into)
580    }
581}
582
583impl PartialEq for SourceId {
584    fn eq(&self, other: &SourceId) -> bool {
585        self.cmp(other) == Ordering::Equal
586    }
587}
588
589impl PartialOrd for SourceId {
590    fn partial_cmp(&self, other: &SourceId) -> Option<Ordering> {
591        Some(self.cmp(other))
592    }
593}
594
595// Custom comparison defined as source kind and canonical URL equality,
596// ignoring the `precise` and `name` fields.
597impl Ord for SourceId {
598    fn cmp(&self, other: &SourceId) -> Ordering {
599        // If our interior pointers are to the exact same `SourceIdInner` then
600        // we're guaranteed to be equal.
601        if ptr::eq(self.inner, other.inner) {
602            return Ordering::Equal;
603        }
604
605        // Sort first based on `kind`, deferring to the URL comparison if
606        // the kinds are equal.
607        let ord_kind = self.inner.kind.cmp(&other.inner.kind);
608        ord_kind.then_with(|| self.inner.canonical_url.cmp(&other.inner.canonical_url))
609    }
610}
611
612impl ser::Serialize for SourceId {
613    fn serialize<S>(&self, s: S) -> Result<S::Ok, S::Error>
614    where
615        S: ser::Serializer,
616    {
617        if self.is_path() {
618            None::<String>.serialize(s)
619        } else {
620            s.collect_str(&self.as_url())
621        }
622    }
623}
624
625impl<'de> de::Deserialize<'de> for SourceId {
626    fn deserialize<D>(d: D) -> Result<SourceId, D::Error>
627    where
628        D: de::Deserializer<'de>,
629    {
630        let string = String::deserialize(d)?;
631        SourceId::from_url(&string).map_err(de::Error::custom)
632    }
633}
634
635fn url_display(url: &Url) -> String {
636    if url.scheme() == "file" {
637        if let Ok(path) = url.to_file_path() {
638            if let Some(path_str) = path.to_str() {
639                return path_str.to_string();
640            }
641        }
642    }
643
644    url.as_str().to_string()
645}
646
647impl fmt::Display for SourceId {
648    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
649        match self.inner.kind {
650            SourceKind::Git(ref reference) => {
651                // Don't replace the URL display for git references,
652                // because those are kind of expected to be URLs.
653                write!(f, "{}", self.inner.url)?;
654                if let Some(pretty) = reference.pretty_ref(true) {
655                    write!(f, "?{}", pretty)?;
656                }
657
658                if let Some(s) = &self.inner.precise {
659                    let s = s.to_string();
660                    let len = cmp::min(s.len(), 8);
661                    write!(f, "#{}", &s[..len])?;
662                }
663                Ok(())
664            }
665            SourceKind::Path => write!(f, "{}", url_display(&self.inner.url)),
666            SourceKind::Registry | SourceKind::SparseRegistry => {
667                write!(f, "registry `{}`", self.display_registry_name())
668            }
669            SourceKind::LocalRegistry => write!(f, "registry `{}`", url_display(&self.inner.url)),
670            SourceKind::Directory => write!(f, "dir {}", url_display(&self.inner.url)),
671        }
672    }
673}
674
675impl Hash for SourceId {
676    fn hash<S: hash::Hasher>(&self, into: &mut S) {
677        self.inner.kind.hash(into);
678        self.inner.canonical_url.hash(into);
679    }
680}
681
682/// The hash of `SourceIdInner` is used to retrieve its interned value from
683/// `SOURCE_ID_CACHE`. We only care about fields that make `SourceIdInner`
684/// unique. Optional fields not affecting the uniqueness must be excluded,
685/// such as [`registry_key`]. That's why this is not derived.
686///
687/// [`registry_key`]: SourceIdInner::registry_key
688impl Hash for SourceIdInner {
689    fn hash<S: hash::Hasher>(&self, into: &mut S) {
690        self.kind.hash(into);
691        self.precise.hash(into);
692        self.canonical_url.hash(into);
693    }
694}
695
696/// This implementation must be synced with [`SourceIdInner::hash`].
697impl PartialEq for SourceIdInner {
698    fn eq(&self, other: &Self) -> bool {
699        self.kind == other.kind
700            && self.precise == other.precise
701            && self.canonical_url == other.canonical_url
702    }
703}
704
705/// A `Display`able view into a `SourceId` that will write it as a url
706pub struct SourceIdAsUrl<'a> {
707    inner: &'a SourceIdInner,
708    encoded: bool,
709}
710
711impl<'a> fmt::Display for SourceIdAsUrl<'a> {
712    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
713        if let Some(protocol) = self.inner.kind.protocol() {
714            write!(f, "{protocol}+")?;
715        }
716        write!(f, "{}", self.inner.url)?;
717        if let SourceIdInner {
718            kind: SourceKind::Git(ref reference),
719            ref precise,
720            ..
721        } = *self.inner
722        {
723            if let Some(pretty) = reference.pretty_ref(self.encoded) {
724                write!(f, "?{}", pretty)?;
725            }
726            if let Some(precise) = precise.as_ref() {
727                write!(f, "#{}", precise)?;
728            }
729        }
730        Ok(())
731    }
732}
733
734impl KeyOf {
735    /// Gets the underlying key.
736    fn key(&self) -> &str {
737        match self {
738            KeyOf::Registry(k) | KeyOf::Source(k) => k,
739        }
740    }
741
742    /// Gets the key if it's from an alternative registry.
743    fn alternative_registry(&self) -> Option<&str> {
744        match self {
745            KeyOf::Registry(k) => Some(k),
746            _ => None,
747        }
748    }
749}
750
751#[cfg(test)]
752mod tests {
753    use super::{GitReference, SourceId, SourceKind};
754    use crate::util::{GlobalContext, IntoUrl};
755
756    #[test]
757    fn github_sources_equal() {
758        let loc = "https://github.com/foo/bar".into_url().unwrap();
759        let default = SourceKind::Git(GitReference::DefaultBranch);
760        let s1 = SourceId::new(default.clone(), loc, None).unwrap();
761
762        let loc = "git://github.com/foo/bar".into_url().unwrap();
763        let s2 = SourceId::new(default, loc.clone(), None).unwrap();
764
765        assert_eq!(s1, s2);
766
767        let foo = SourceKind::Git(GitReference::Branch("foo".to_string()));
768        let s3 = SourceId::new(foo, loc, None).unwrap();
769        assert_ne!(s1, s3);
770    }
771
772    // This is a test that the hash of the `SourceId` for crates.io is a well-known
773    // value.
774    //
775    // Note that the hash value matches what the crates.io source id has hashed
776    // since Rust 1.84.0. We strive to keep this value the same across
777    // versions of Cargo because changing it means that users will need to
778    // redownload the index and all crates they use when using a new Cargo version.
779    //
780    // This isn't to say that this hash can *never* change, only that when changing
781    // this it should be explicitly done. If this hash changes accidentally and
782    // you're able to restore the hash to its original value, please do so!
783    // Otherwise please just leave a comment in your PR as to why the hash value is
784    // changing and why the old value can't be easily preserved.
785    // If it takes an ugly hack to restore it,
786    // then leave a link here so we can remove the hack next time we change the hash.
787    //
788    // Hacks to remove next time the hash changes:
789    // - (fill in your code here)
790    //
791    // The hash value should be stable across platforms, and doesn't depend on
792    // endianness and bit-width. One caveat is that absolute paths on Windows
793    // are inherently different than on Unix-like platforms. Unless we omit or
794    // strip the prefix components (e.g. `C:`), there is not way to have a true
795    // cross-platform stable hash for absolute paths.
796    #[test]
797    fn test_stable_hash() {
798        use std::hash::Hasher;
799        use std::path::Path;
800
801        use snapbox::IntoData as _;
802        use snapbox::assert_data_eq;
803        use snapbox::str;
804
805        use crate::util::StableHasher;
806        use crate::util::hex::short_hash;
807
808        #[cfg(not(windows))]
809        let ws_root = Path::new("/tmp/ws");
810        #[cfg(windows)]
811        let ws_root = Path::new(r"C:\\tmp\ws");
812
813        let gen_hash = |source_id: SourceId| {
814            let mut hasher = StableHasher::new();
815            source_id.stable_hash(ws_root, &mut hasher);
816            Hasher::finish(&hasher).to_string()
817        };
818
819        let source_id = SourceId::crates_io(&GlobalContext::default().unwrap()).unwrap();
820        assert_data_eq!(gen_hash(source_id), str!["7062945687441624357"].raw());
821        assert_data_eq!(short_hash(&source_id), str!["25cdd57fae9f0462"].raw());
822
823        let url = "https://my-crates.io".into_url().unwrap();
824        let source_id = SourceId::for_registry(&url).unwrap();
825        assert_data_eq!(gen_hash(source_id), str!["8310250053664888498"].raw());
826        assert_data_eq!(short_hash(&source_id), str!["b2d65deb64f05373"].raw());
827
828        let url = "https://your-crates.io".into_url().unwrap();
829        let source_id = SourceId::for_alt_registry(&url, "alt").unwrap();
830        assert_data_eq!(gen_hash(source_id), str!["14149534903000258933"].raw());
831        assert_data_eq!(short_hash(&source_id), str!["755952de063f5dc4"].raw());
832
833        let url = "sparse+https://my-crates.io".into_url().unwrap();
834        let source_id = SourceId::for_registry(&url).unwrap();
835        assert_data_eq!(gen_hash(source_id), str!["16249512552851930162"].raw());
836        assert_data_eq!(short_hash(&source_id), str!["327cfdbd92dd81e1"].raw());
837
838        let url = "sparse+https://your-crates.io".into_url().unwrap();
839        let source_id = SourceId::for_alt_registry(&url, "alt").unwrap();
840        assert_data_eq!(gen_hash(source_id), str!["6156697384053352292"].raw());
841        assert_data_eq!(short_hash(&source_id), str!["64a713b6a6fb7055"].raw());
842
843        let url = "file:///tmp/ws/crate".into_url().unwrap();
844        let source_id = SourceId::for_git(&url, GitReference::DefaultBranch).unwrap();
845        assert_data_eq!(gen_hash(source_id), str!["473480029881867801"].raw());
846        assert_data_eq!(short_hash(&source_id), str!["199e591d94239206"].raw());
847
848        let path = &ws_root.join("crate");
849        let source_id = SourceId::for_local_registry(path).unwrap();
850        #[cfg(not(windows))]
851        {
852            assert_data_eq!(gen_hash(source_id), str!["11515846423845066584"].raw());
853            assert_data_eq!(short_hash(&source_id), str!["58d73c154f81d09f"].raw());
854        }
855        #[cfg(windows)]
856        {
857            assert_data_eq!(gen_hash(source_id), str!["6146331155906064276"].raw());
858            assert_data_eq!(short_hash(&source_id), str!["946fb2239f274c55"].raw());
859        }
860
861        let source_id = SourceId::for_path(path).unwrap();
862        assert_data_eq!(gen_hash(source_id), str!["215644081443634269"].raw());
863        #[cfg(not(windows))]
864        assert_data_eq!(short_hash(&source_id), str!["64bace89c92b101f"].raw());
865        #[cfg(windows)]
866        assert_data_eq!(short_hash(&source_id), str!["01e1e6c391813fb6"].raw());
867
868        let source_id = SourceId::for_directory(path).unwrap();
869        #[cfg(not(windows))]
870        {
871            assert_data_eq!(gen_hash(source_id), str!["6127590343904940368"].raw());
872            assert_data_eq!(short_hash(&source_id), str!["505191d1f3920955"].raw());
873        }
874        #[cfg(windows)]
875        {
876            assert_data_eq!(gen_hash(source_id), str!["10423446877655960172"].raw());
877            assert_data_eq!(short_hash(&source_id), str!["6c8ad69db585a790"].raw());
878        }
879    }
880
881    #[test]
882    fn serde_roundtrip() {
883        let url = "sparse+https://my-crates.io/".into_url().unwrap();
884        let source_id = SourceId::for_registry(&url).unwrap();
885        let formatted = format!("{}", source_id.as_url());
886        let deserialized = SourceId::from_url(&formatted).unwrap();
887        assert_eq!(formatted, "sparse+https://my-crates.io/");
888        assert_eq!(source_id, deserialized);
889    }
890
891    #[test]
892    fn gitrefs_roundtrip() {
893        let base = "https://host/path".into_url().unwrap();
894        let branch = GitReference::Branch("*-._+20%30 Z/z#foo=bar&zap[]?to\\()'\"".to_string());
895        let s1 = SourceId::for_git(&base, branch).unwrap();
896        let ser1 = format!("{}", s1.as_encoded_url());
897        let s2 = SourceId::from_url(&ser1).expect("Failed to deserialize");
898        let ser2 = format!("{}", s2.as_encoded_url());
899        // Serializing twice should yield the same result
900        assert_eq!(ser1, ser2, "Serialized forms don't match");
901        // SourceId serializing the same should have the same semantics
902        // This used to not be the case (# was ambiguous)
903        assert_eq!(s1, s2, "SourceId doesn't round-trip");
904        // Freeze the format to match an x-www-form-urlencoded query string
905        // https://url.spec.whatwg.org/#application/x-www-form-urlencoded
906        assert_eq!(
907            ser1,
908            "git+https://host/path?branch=*-._%2B20%2530+Z%2Fz%23foo%3Dbar%26zap%5B%5D%3Fto%5C%28%29%27%22"
909        );
910    }
911}
912
913/// Check if `url` equals to the overridden crates.io URL.
914#[expect(
915    clippy::disallowed_methods,
916    reason = "testing only, no reason for config support"
917)]
918fn is_overridden_crates_io_url(url: &str) -> bool {
919    std::env::var("__CARGO_TEST_CRATES_IO_URL_DO_NOT_USE_THIS").map_or(false, |v| v == url)
920}
cargo/workspace/source_id.rs

cargo/workspace/
source_id.rs
