feat: Add PackageRecord::validate function (#911)

.github/workflows/python-bindings.yml

@@ -46,4 +46,4 @@ jobs:
       - name: Run tests
         run: |
           cd py-rattler
-          pixi run -e test test
+          pixi run -e test test --color=yes

crates/rattler_conda_types/src/lib.rs

@@ -50,7 +50,7 @@ pub use repo_data::{
     compute_package_url,
     patches::{PackageRecordPatch, PatchInstructions, RepoDataPatch},
     sharded::{Shard, ShardedRepodata, ShardedSubdirInfo},
-    ChannelInfo, ConvertSubdirError, PackageRecord, RepoData,
+    ChannelInfo, ConvertSubdirError, PackageRecord, RepoData, ValidatePackageRecordsError,
 };
 pub use repo_data_record::RepoDataRecord;
 pub use run_export::RunExportKind;

crates/rattler_conda_types/src/repo_data/mod.rs

@@ -19,14 +19,16 @@ use serde_with::{serde_as, skip_serializing_none, OneOrMany};
 use thiserror::Error;
 use url::Url;
 
-use crate::utils::serde::sort_map_alphabetically;
 use crate::utils::url::add_trailing_slash;
 use crate::{
     build_spec::BuildNumber,
     package::{IndexJson, RunExportsJson},
     utils::serde::DeserializeFromStrUnchecked,
     Channel, NoArchType, PackageName, PackageUrl, Platform, RepoDataRecord, VersionWithSource,
 };
+use crate::{
+    utils::serde::sort_map_alphabetically, MatchSpec, Matches, ParseMatchSpecError, ParseStrictness,
+};
 
 /// [`RepoData`] is an index of package binaries available on in a subdirectory
 /// of a Conda channel.
@@ -275,6 +277,12 @@ pub fn compute_package_url(
         .expect("failed to join base_url and filename")
 }
 
+impl AsRef<PackageRecord> for PackageRecord {
+    fn as_ref(&self) -> &PackageRecord {
+        self
+    }
+}
+
 impl PackageRecord {
     /// A simple helper method that constructs a `PackageRecord` with the bare
     /// minimum values.
@@ -315,6 +323,74 @@ impl PackageRecord {
     pub fn sort_topologically<T: AsRef<PackageRecord> + Clone>(records: Vec<T>) -> Vec<T> {
         topological_sort::sort_topologically(records)
     }
+
+    /// Validate that the given package records are valid w.r.t. 'depends' and 'constrains'.
+    /// This function will return Ok(()) if all records form a valid environment, i.e., all dependencies
+    /// of each package are satisfied by the other packages in the list.
+    /// If there is a dependency that is not satisfied, this function will return an error.
+    pub fn validate<T: AsRef<PackageRecord>>(
+        records: Vec<T>,
+    ) -> Result<(), ValidatePackageRecordsError> {
+        for package in records.iter() {
+            let package = package.as_ref();
+            // First we check if all dependencies are in the environment.
+            for dep in package.depends.iter() {
+                // We ignore virtual packages, e.g. `__unix`.
+                if dep.starts_with("__") {
+                    continue;
+                }
+                let dep_spec = MatchSpec::from_str(dep, ParseStrictness::Lenient)?;
+                if !records.iter().any(|p| dep_spec.matches(p.as_ref())) {
+                    return Err(ValidatePackageRecordsError::DependencyNotInEnvironment {
+                        package: package.to_owned(),
+                        dependency: dep.to_string(),
+                    });
+                }
+            }
+
+            // Then we check if all constraints are satisfied.
+            for constraint in package.constrains.iter() {
+                let constraint_spec = MatchSpec::from_str(constraint, ParseStrictness::Lenient)?;
+                let matching_package = records
+                    .iter()
+                    .find(|record| Some(record.as_ref().name.clone()) == constraint_spec.name);
+                if matching_package.is_some_and(|p| !constraint_spec.matches(p.as_ref())) {
+                    return Err(ValidatePackageRecordsError::PackageConstraintNotSatisfied {
+                        package: package.to_owned(),
+                        constraint: constraint.to_owned(),
+                        violating_package: matching_package.unwrap().as_ref().to_owned(),
+                    });
+                }
+            }
+        }
+        Ok(())
+    }
+}
+
+/// An error when validating package records.
+#[derive(Debug, Error)]
+pub enum ValidatePackageRecordsError {
+    /// A package is not present in the environment.
+    #[error("package '{package}' has dependency '{dependency}', which is not in the environment")]
+    DependencyNotInEnvironment {
+        /// The package containing the unmet dependency.
+        package: PackageRecord,
+        /// The dependency that is not in the environment.
+        dependency: String,
+    },
+    /// A package constraint is not met in the environment.
+    #[error("package '{package}' has constraint '{constraint}', which is not satisfied by '{violating_package}' in the environment")]
+    PackageConstraintNotSatisfied {
+        /// The package containing the unmet constraint.
+        package: PackageRecord,
+        /// The constraint that is violated.
+        constraint: String,
+        /// The corresponding package that violates the constraint.
+        violating_package: PackageRecord,
+    },
+    /// Failed to parse a matchspec.
+    #[error(transparent)]
+    ParseMatchSpec(#[from] ParseMatchSpecError),
 }
 
 /// An error that can occur when parsing a platform from a string.
@@ -331,7 +407,7 @@ pub enum ConvertSubdirError {
     /// Platform key is empty
     #[error("platform key is empty in index.json")]
     PlatformEmpty,
-    /// Arc key is empty
+    /// Arch key is empty
     #[error("arch key is empty in index.json")]
     ArchEmpty,
 }
@@ -437,7 +513,7 @@ mod test {
 
     use crate::{
         repo_data::{compute_package_url, determine_subdir},
-        Channel, ChannelConfig, RepoData,
+        Channel, ChannelConfig, PackageRecord, RepoData,
     };
 
     // isl-0.12.2-1.tar.bz2
@@ -554,4 +630,46 @@ mod test {
         let data_path = test_data_path.join(path);
         RepoData::from_path(data_path).unwrap()
     }
+
+    #[test]
+    fn test_validate() {
+        // load test data
+        let test_data_path = dunce::canonicalize(
+            std::path::Path::new(env!("CARGO_MANIFEST_DIR")).join("../../test-data"),
+        )
+        .unwrap();
+        let data_path = test_data_path.join("channels/dummy/linux-64/repodata.json");
+        let repodata = RepoData::from_path(&data_path).unwrap();
+
+        let package_depends_only_virtual_package = repodata
+            .packages
+            .get("baz-1.0-unix_py36h1af98f8_2.tar.bz2")
+            .unwrap();
+        let package_depends = repodata.packages.get("foobar-2.0-bla_1.tar.bz2").unwrap();
+        let package_constrains = repodata
+            .packages
+            .get("foo-3.0.2-py36h1af98f8_3.conda")
+            .unwrap();
+        let package_bors_1 = repodata.packages.get("bors-1.2.1-bla_1.tar.bz2").unwrap();
+        let package_bors_2 = repodata.packages.get("bors-2.1-bla_1.tar.bz2").unwrap();
+
+        assert!(PackageRecord::validate(vec![package_depends_only_virtual_package]).is_ok());
+        for packages in [vec![package_depends], vec![package_depends, package_bors_2]] {
+            let result = PackageRecord::validate(packages);
+            assert!(result.is_err());
+            assert!(result.err().unwrap().to_string().contains(
+                "package 'foobar=2.0=bla_1' has dependency 'bors <2.0', which is not in the environment"
+            ));
+        }
+
+        assert!(PackageRecord::validate(vec![package_depends, package_bors_1]).is_ok());
+        assert!(PackageRecord::validate(vec![package_constrains]).is_ok());
+        assert!(PackageRecord::validate(vec![package_constrains, package_bors_1]).is_ok());
+
+        let result = PackageRecord::validate(vec![package_constrains, package_bors_2]);
+        assert!(result.is_err());
+        assert!(result.err().unwrap().to_string().contains(
+            "package 'foo=3.0.2=py36h1af98f8_3' has constraint 'bors <2.0', which is not satisfied by 'bors=2.1=bla_1' in the environment"
+        ));
+    }
 }

crates/rattler_conda_types/src/repo_data/snapshots/rattler_conda_types__repo_data__test__base_url_packages.snap

@@ -1,6 +1,6 @@
 ---
 source: crates/rattler_conda_types/src/repo_data/mod.rs
-assertion_line: 538
+assertion_line: 594
 expression: file_urls
 ---
 - channels/dummy/linux-64/issue_717-2.1-bla_1.tar.bz2
@@ -9,6 +9,7 @@ expression: file_urls
 - channels/dummy/linux-64/foo-3.0.2-py36h1af98f8_1.conda
 - channels/dummy/linux-64/foo-3.0.2-py36h1af98f8_1.tar.bz2
 - channels/dummy/linux-64/foo-4.0.2-py36h1af98f8_2.tar.bz2
+- channels/dummy/linux-64/foo-3.0.2-py36h1af98f8_3.conda
 - channels/dummy/linux-64/bors-1.2.1-bla_1.tar.bz2
 - channels/dummy/linux-64/baz-1.0-unix_py36h1af98f8_2.tar.bz2
 - channels/dummy/linux-64/bors-2.1-bla_1.tar.bz2
@@ -17,5 +18,5 @@ expression: file_urls
 - channels/dummy/linux-64/bors-1.1-bla_1.tar.bz2
 - channels/dummy/linux-64/baz-2.0-unix_py36h1af98f8_2.tar.bz2
 - channels/dummy/linux-64/foobar-2.1-bla_1.tar.bz2
-- channels/dummy/linux-64/cuda-version-12.5-hd4f0392_3.conda
 - channels/dummy/linux-64/bors-2.0-bla_1.tar.bz2
+- channels/dummy/linux-64/cuda-version-12.5-hd4f0392_3.conda

crates/rattler_conda_types/src/repo_data/snapshots/rattler_conda_types__repo_data__test__serialize_packages-2.snap

@@ -184,6 +184,23 @@ expression: json
       "timestamp": 1715610974,
       "version": "3.0.2"
     },
+    "foo-3.0.2-py36h1af98f8_3.conda": {
+      "build": "py36h1af98f8_3",
+      "build_number": 3,
+      "constrains": [
+        "bors <2.0"
+      ],
+      "depends": [],
+      "license": "MIT",
+      "license_family": "MIT",
+      "md5": "fb731d9290f0bcbf3a054665f33ec94f",
+      "name": "foo",
+      "sha256": "67a63bec3fd3205170eaad532d487595b8aaceb9814d13c6858d7bac3ef24cd4",
+      "size": 414494,
+      "subdir": "linux-64",
+      "timestamp": 1715610974,
+      "version": "3.0.2"
+    },
     "foo-4.0.2-py36h1af98f8_2.tar.bz2": {
       "build": "py36h1af98f8_2",
       "build_number": 1,

crates/rattler_conda_types/src/repo_data/snapshots/rattler_conda_types__repo_data__test__serialize_packages.snap

@@ -167,6 +167,21 @@ packages:
     subdir: linux-64
     timestamp: 1715610974
     version: 3.0.2
+  foo-3.0.2-py36h1af98f8_3.conda:
+    build: py36h1af98f8_3
+    build_number: 3
+    constrains:
+      - bors <2.0
+    depends: []
+    license: MIT
+    license_family: MIT
+    md5: fb731d9290f0bcbf3a054665f33ec94f
+    name: foo
+    sha256: 67a63bec3fd3205170eaad532d487595b8aaceb9814d13c6858d7bac3ef24cd4
+    size: 414494
+    subdir: linux-64
+    timestamp: 1715610974
+    version: 3.0.2
   foo-4.0.2-py36h1af98f8_2.tar.bz2:
     build: py36h1af98f8_2
     build_number: 1

crates/rattler_solve/tests/backends.rs

@@ -346,17 +346,17 @@ macro_rules! solver_backend_tests {
             assert_eq!(1, pkgs.len());
             let info = &pkgs[0];
 
-            assert_eq!("foo-3.0.2-py36h1af98f8_2.conda", info.file_name);
+            assert_eq!("foo-3.0.2-py36h1af98f8_3.conda", info.file_name);
             assert_eq!(
-                "https://conda.anaconda.org/conda-forge/linux-64/foo-3.0.2-py36h1af98f8_2.conda",
+                "https://conda.anaconda.org/conda-forge/linux-64/foo-3.0.2-py36h1af98f8_3.conda",
                 info.url.to_string()
             );
             assert_eq!("https://conda.anaconda.org/conda-forge/", info.channel);
             assert_eq!("foo", info.package_record.name.as_normalized());
             assert_eq!("linux-64", info.package_record.subdir);
             assert_eq!("3.0.2", info.package_record.version.to_string());
-            assert_eq!("py36h1af98f8_2", info.package_record.build);
-            assert_eq!(2, info.package_record.build_number);
+            assert_eq!("py36h1af98f8_3", info.package_record.build);
+            assert_eq!(3, info.package_record.build_number);
             assert_eq!(
                 rattler_digest::parse_digest_from_hex::<rattler_digest::Sha256>(
                     "67a63bec3fd3205170eaad532d487595b8aaceb9814d13c6858d7bac3ef24cd4"
@@ -665,17 +665,17 @@ mod libsolv_c {
         assert_eq!(1, pkgs.len());
         let info = &pkgs[0];
 
-        assert_eq!("foo-3.0.2-py36h1af98f8_2.conda", info.file_name);
+        assert_eq!("foo-3.0.2-py36h1af98f8_3.conda", info.file_name);
         assert_eq!(
-            "https://conda.anaconda.org/conda-forge/linux-64/foo-3.0.2-py36h1af98f8_2.conda",
+            "https://conda.anaconda.org/conda-forge/linux-64/foo-3.0.2-py36h1af98f8_3.conda",
             info.url.to_string()
         );
         assert_eq!("https://conda.anaconda.org/conda-forge/", info.channel);
         assert_eq!("foo", info.package_record.name.as_normalized());
         assert_eq!("linux-64", info.package_record.subdir);
         assert_eq!("3.0.2", info.package_record.version.to_string());
-        assert_eq!("py36h1af98f8_2", info.package_record.build);
-        assert_eq!(2, info.package_record.build_number);
+        assert_eq!("py36h1af98f8_3", info.package_record.build);
+        assert_eq!(3, info.package_record.build_number);
         assert_eq!(
             rattler_digest::parse_digest_from_hex::<rattler_digest::Sha256>(
                 "67a63bec3fd3205170eaad532d487595b8aaceb9814d13c6858d7bac3ef24cd4"
@@ -781,7 +781,7 @@ mod resolvo {
             Version::from_str("3.0.2").unwrap()
         );
         assert_eq!(
-            result[0].package_record.build_number, 2,
+            result[0].package_record.build_number, 3,
             "expected the highest build number"
         );
     }

py-rattler/rattler/exceptions.py

@@ -20,6 +20,7 @@
         EnvironmentCreationError,
         ExtractError,
         GatewayError,
+        ValidatePackageRecordsException,
     )
 except ImportError:
     # They are only redefined for documentation purposes
@@ -85,6 +86,9 @@ class ExtractError(Exception):  # type: ignore[no-redef]
     class GatewayError(Exception):  # type: ignore[no-redef]
         """An error that can occur when querying the repodata gateway."""
 
+    class ValidatePackageRecordsException(Exception):  # type: ignore[no-redef]
+        """An error when validating package records."""
+
 
 __all__ = [
     "ActivationError",
@@ -107,4 +111,5 @@ class GatewayError(Exception):  # type: ignore[no-redef]
     "EnvironmentCreationError",
     "ExtractError",
     "GatewayError",
+    "ValidatePackageRecordsException",
 ]

py-rattler/rattler/repo_data/package_record.py

@@ -119,6 +119,37 @@ def to_graph(records: List[PackageRecord]) -> nx.DiGraph:  # type: ignore[type-a
 
         return graph
 
+    @staticmethod
+    def validate(records: List[PackageRecord]) -> None:
+        """
+        Validate that the given package records are valid w.r.t. 'depends' and 'constrains'.
+
+        This function will return nothing if all records form a valid environment, i.e., all dependencies
+        of each package are satisfied by the other packages in the list.
+        If there is a dependency that is not satisfied, this function will raise an exception.
+
+        Examples
+        --------
+        ```python
+        >>> from os import listdir
+        >>> from os.path import isfile, join
+        >>> from rattler import PrefixRecord
+        >>> from rattler.exceptions import ValidatePackageRecordsException
+        >>> records = [
+        ...     PrefixRecord.from_path(join("../test-data/conda-meta/", f))
+        ...     for f in sorted(listdir("../test-data/conda-meta"))
+        ...     if isfile(join("../test-data/conda-meta", f))
+        ... ]
+        >>> try:
+        ...     PackageRecord.validate(records)
+        ... except ValidatePackageRecordsException as e:
+        ...     print(e)
+        package 'libsqlite=3.40.0=hcfcfb64_0' has dependency 'ucrt >=10.0.20348.0', which is not in the environment
+        >>>
+        ```
+        """
+        return PyRecord.validate(records)
+
     @classmethod
     def _from_py_record(cls, py_record: PyRecord) -> PackageRecord:
         """