Split registry inference from registry validation

This is in preparation for reusing infer_registry during publishing
(which needs different validation logic).

It also changes the registry validation slightly, adding in a check
(from the publish logic) forbidding implicit source replacement. This
affects the tests (which configure a dummy registry for source
replacement), so we also weaken the checks by only erroring for registry
issues when there are actually local dependencies.

Author: jneem

src/cargo/ops/cargo_package.rs

@@ -15,7 +15,7 @@ use crate::core::{Feature, PackageIdSpecQuery, Shell, Verbosity, Workspace};
 use crate::core::{Package, PackageId, PackageSet, Resolve, SourceId};
 use crate::ops::registry::infer_registry;
 use crate::sources::registry::index::{IndexPackage, RegistryDependency};
-use crate::sources::PathSource;
+use crate::sources::{PathSource, CRATES_IO_REGISTRY};
 use crate::util::cache_lock::CacheLockMode;
 use crate::util::context::JobsConfig;
 use crate::util::errors::CargoResult;
@@ -24,7 +24,7 @@ use crate::util::{
     self, human_readable_bytes, restricted_names, FileLock, Filesystem, GlobalContext, Graph,
 };
 use crate::{drop_println, ops};
-use anyhow::Context as _;
+use anyhow::{bail, Context as _};
 use cargo_util::paths;
 use flate2::read::GzDecoder;
 use flate2::{Compression, GzBuilder};
@@ -33,6 +33,8 @@ use tar::{Archive, Builder, EntryType, Header, HeaderMode};
 use tracing::debug;
 use unicase::Ascii as UncasedAscii;
 
+use super::RegistryOrIndex;
+
 #[derive(Clone)]
 pub struct PackageOpts<'gctx> {
     pub gctx: &'gctx GlobalContext,
@@ -173,6 +175,43 @@ fn create_package(
     return Ok(dst);
 }
 
+/// Determine which registry the packages are for.
+///
+/// The registry only affects the built packages if there are dependencies within the
+/// packages that we're packaging: if we're packaging foo-bin and foo-lib, and foo-bin
+/// depends on foo-lib, then the foo-lib entry in foo-bin's lockfile will depend on the
+/// registry that we're building packages for.
+fn get_registry(
+    gctx: &GlobalContext,
+    pkgs: &[&Package],
+    reg_or_index: Option<RegistryOrIndex>,
+) -> CargoResult<SourceId> {
+    let reg_or_index = match reg_or_index.clone() {
+        Some(r) => Some(r),
+        None => infer_registry(pkgs)?,
+    };
+
+    // Validate the registry against the packages' allow-lists.
+    let reg = reg_or_index
+        .clone()
+        .unwrap_or_else(|| RegistryOrIndex::Registry(CRATES_IO_REGISTRY.to_owned()));
+    if let RegistryOrIndex::Registry(reg_name) = reg {
+        for pkg in pkgs {
+            if let Some(allowed) = pkg.publish().as_ref() {
+                if !allowed.iter().any(|a| a == &reg_name) {
+                    bail!(
+                        "`{}` cannot be packaged.\n\
+                         The registry `{}` is not listed in the `package.publish` value in Cargo.toml.",
+                        pkg.name(),
+                        reg_name
+                    );
+                }
+            }
+        }
+    }
+    Ok(ops::registry::get_source_id(gctx, reg_or_index.as_ref())?.replacement)
+}
+
 /// Packages an entire workspace.
 ///
 /// Returns the generated package files. If `opts.list` is true, skips
@@ -196,19 +235,34 @@ pub fn package(ws: &Workspace<'_>, opts: &PackageOpts<'_>) -> CargoResult<Vec<Fi
         // below, and will be validated during the verification step.
     }
 
+    let deps = local_deps(pkgs.iter().map(|(p, f)| ((*p).clone(), f.clone())));
     let just_pkgs: Vec<_> = pkgs.iter().map(|p| p.0).collect();
-    let publish_reg = infer_registry(ws.gctx(), &just_pkgs, opts.reg_or_index.clone())?;
-    debug!("packaging for registry {publish_reg}");
+
+    let sid = match get_registry(ws.gctx(), &just_pkgs, opts.reg_or_index.clone()) {
+        Ok(sid) => {
+            debug!("packaging for registry {}", sid);
+            Some(sid)
+        }
+        Err(e) => {
+            if deps.has_no_dependencies() && opts.reg_or_index.is_none() {
+                // The publish registry doesn't matter unless there are local dependencies,
+                // so ignore any errors if we don't need it. If they explicitly passed a registry
+                // on the CLI, we check it no matter what.
+                None
+            } else {
+                return Err(e);
+            }
+        }
+    };
 
     let mut local_reg = if ws.gctx().cli_unstable().package_workspace {
         let reg_dir = ws.target_dir().join("package").join("tmp-registry");
-        Some(TmpRegistry::new(ws.gctx(), reg_dir, publish_reg)?)
+        sid.map(|sid| TmpRegistry::new(ws.gctx(), reg_dir, sid))
+            .transpose()?
     } else {
         None
     };
 
-    let deps = local_deps(pkgs.iter().map(|(p, f)| ((*p).clone(), f.clone())));
-
     // Packages need to be created in dependency order, because dependencies must
     // be added to our local overlay before we can create lockfiles that depend on them.
     let sorted_pkgs = deps.sort();
@@ -262,6 +316,12 @@ impl LocalDependencies {
             .map(|name| self.packages[&name].clone())
             .collect()
     }
+
+    pub fn has_no_dependencies(&self) -> bool {
+        self.graph
+            .iter()
+            .all(|node| self.graph.edges(node).next().is_none())
+    }
 }
 
 /// Build just the part of the dependency graph that's between the given packages,

src/cargo/ops/registry/mod.rs

@@ -21,7 +21,7 @@ use url::Url;
 
 use crate::core::{Package, PackageId, SourceId};
 use crate::sources::source::Source;
-use crate::sources::{RegistrySource, SourceConfigMap, CRATES_IO_REGISTRY};
+use crate::sources::{RegistrySource, SourceConfigMap};
 use crate::util::auth;
 use crate::util::cache_lock::CacheLockMode;
 use crate::util::context::{GlobalContext, PathAndArgs};
@@ -191,7 +191,7 @@ fn registry(
 ///
 /// The return value is a pair of `SourceId`s: The first may be a built-in replacement of
 /// crates.io (such as index.crates.io), while the second is always the original source.
-fn get_source_id(
+pub(crate) fn get_source_id(
     gctx: &GlobalContext,
     reg_or_index: Option<&RegistryOrIndex>,
 ) -> CargoResult<RegistrySourceIds> {
@@ -324,87 +324,40 @@ pub(crate) struct RegistrySourceIds {
 }
 
 /// If this set of packages has an unambiguous publish registry, find it.
-pub(crate) fn infer_registry(
-    gctx: &GlobalContext,
-    pkgs: &[&Package],
-    reg_or_index: Option<RegistryOrIndex>,
-) -> CargoResult<SourceId> {
-    let reg_or_index = match reg_or_index {
-        Some(r) => r,
-        None => {
-            if pkgs[1..].iter().all(|p| p.publish() == pkgs[0].publish()) {
-                // If all packages have the same publish settings, we take that as the default.
-                match pkgs[0].publish().as_deref() {
-                    Some([unique_pkg_reg]) => RegistryOrIndex::Registry(unique_pkg_reg.to_owned()),
-                    None | Some([]) => RegistryOrIndex::Registry(CRATES_IO_REGISTRY.to_owned()),
-                    Some([reg, ..]) if pkgs.len() == 1 => {
-                        // For backwards compatibility, avoid erroring if there's only one package.
-                        // The registry doesn't affect packaging in this case.
-                        RegistryOrIndex::Registry(reg.to_owned())
-                    }
-                    Some(regs) => {
-                        let mut regs: Vec<_> = regs.iter().map(|s| format!("\"{}\"", s)).collect();
-                        regs.sort();
-                        regs.dedup();
-                        // unwrap: the match block ensures that there's more than one reg.
-                        let (last_reg, regs) = regs.split_last().unwrap();
-                        bail!(
-                            "--registry is required to disambiguate between {} or {} registries",
-                            regs.join(", "),
-                            last_reg
-                        )
-                    }
-                }
-            } else {
-                let common_regs = pkgs
-                    .iter()
-                    // `None` means "all registries", so drop them instead of including them
-                    // in the intersection.
-                    .filter_map(|p| p.publish().as_deref())
-                    .map(|p| p.iter().collect::<HashSet<_>>())
-                    .reduce(|xs, ys| xs.intersection(&ys).cloned().collect())
-                    .unwrap_or_default();
-                if common_regs.is_empty() {
-                    bail!("conflicts between `package.publish` fields in the selected packages");
-                } else {
-                    bail!(
-                        "--registry is required because not all `package.publish` settings agree",
-                    );
-                }
+pub(crate) fn infer_registry(pkgs: &[&Package]) -> CargoResult<Option<RegistryOrIndex>> {
+    if pkgs[1..].iter().all(|p| p.publish() == pkgs[0].publish()) {
+        // If all packages have the same publish settings, we take that as the default.
+        match pkgs[0].publish().as_deref() {
+            Some([unique_pkg_reg]) => {
+                Ok(Some(RegistryOrIndex::Registry(unique_pkg_reg.to_owned())))
             }
-        }
-    };
-
-    // Validate the registry against the packages' allow-lists. For backwards compatibility, we
-    // skip this if only a single package is being published (because in that case the registry
-    // doesn't affect the packaging step).
-    if pkgs.len() > 1 {
-        if let RegistryOrIndex::Registry(reg_name) = &reg_or_index {
-            for pkg in pkgs {
-                if let Some(allowed) = pkg.publish().as_ref() {
-                    if !allowed.iter().any(|a| a == reg_name) {
-                        bail!(
-                        "`{}` cannot be packaged.\n\
-                         The registry `{}` is not listed in the `package.publish` value in Cargo.toml.",
-                        pkg.name(),
-                        reg_name
-                    );
-                    }
-                }
+            None | Some([]) => Ok(None),
+            Some(regs) => {
+                let mut regs: Vec<_> = regs.iter().map(|s| format!("\"{}\"", s)).collect();
+                regs.sort();
+                regs.dedup();
+                // unwrap: the match block ensures that there's more than one reg.
+                let (last_reg, regs) = regs.split_last().unwrap();
+                bail!(
+                    "--registry is required to disambiguate between {} or {} registries",
+                    regs.join(", "),
+                    last_reg
+                )
             }
         }
+    } else {
+        let common_regs = pkgs
+            .iter()
+            // `None` means "all registries", so drop them instead of including them
+            // in the intersection.
+            .filter_map(|p| p.publish().as_deref())
+            .map(|p| p.iter().collect::<HashSet<_>>())
+            .reduce(|xs, ys| xs.intersection(&ys).cloned().collect())
+            .unwrap_or_default();
+        if common_regs.is_empty() {
+            bail!("conflicts between `package.publish` fields in the selected packages");
+        } else {
+            bail!("--registry is required because not all `package.publish` settings agree",);
+        }
     }
-
-    let sid = match reg_or_index {
-        RegistryOrIndex::Index(url) => SourceId::for_registry(&url)?,
-        RegistryOrIndex::Registry(reg) if reg == CRATES_IO_REGISTRY => SourceId::crates_io(gctx)?,
-        RegistryOrIndex::Registry(reg) => SourceId::alt_registry(gctx, &reg)?,
-    };
-
-    // Load source replacements that are built-in to Cargo.
-    let sid = SourceConfigMap::empty(gctx)?
-        .load(sid, &HashSet::new())?
-        .replaced_source_id();
-
-    Ok(sid)
 }

tests/testsuite/package.rs

@@ -5999,7 +5999,8 @@ fn registry_not_in_publish_list() {
         .masquerade_as_nightly_cargo(&["package-workspace"])
         .with_status(101)
         .with_stderr_data(str![[r#"
-[ERROR] registry index was not found in any configuration: `alternative`
+[ERROR] `foo` cannot be packaged.
+The registry `alternative` is not listed in the `package.publish` value in Cargo.toml.
 
 "#]])
         .run();