Fix an issue pull-through caching (wasn't saving all metadata)

pulp_rust/app/models.py

@@ -3,6 +3,9 @@
 from logging import getLogger
 
 from django.db import models
+from django_lifecycle import hook, AFTER_CREATE
+
+from pulp_rust.app.utils import extract_cargo_toml, extract_dependencies
 
 from pulpcore.plugin.models import (
     Content,
@@ -106,15 +109,32 @@ def init_from_artifact_and_relative_path(artifact, relative_path):
         Create an unsaved RustContent from a downloaded .crate artifact.
 
         Called by pulpcore's content handler during pull-through caching.
-        Only populates name, version, and checksum -- dependency and feature
-        metadata is served from the upstream sparse index via the proxy.
+        Extracts full metadata (dependencies, features, etc.) from the
+        Cargo.toml inside the .crate tarball.
         """
         crate_name, version = _parse_crate_relative_path(relative_path)
-        return RustContent(
+        cargo_toml = extract_cargo_toml(artifact.file.path, crate_name, version)
+
+        content = RustContent(
             name=crate_name,
             vers=version,
             cksum=artifact.sha256,
+            features=cargo_toml.get("features", {}),
+            links=cargo_toml.get("package", {}).get("links"),
+            rust_version=cargo_toml.get("package", {}).get("rust-version"),
         )
+        # Store parsed dep data for the AFTER_CREATE hook to consume
+        content._parsed_deps = extract_dependencies(cargo_toml)
+        return content
+
+    @hook(AFTER_CREATE)
+    def _create_dependencies_from_parsed_data(self):
+        """Create RustDependency records from data parsed during pull-through."""
+        parsed_deps = getattr(self, "_parsed_deps", None)
+        if parsed_deps:
+            RustDependency.objects.bulk_create(
+                [RustDependency(content=self, **dep) for dep in parsed_deps]
+            )
 
     class Meta:
         default_related_name = "%(app_label)s_%(model_name)s"

pulp_rust/app/utils.py

@@ -0,0 +1,90 @@
+import tarfile
+
+try:
+    import tomllib
+except ModuleNotFoundError:
+    import tomli as tomllib
+
+
+def extract_cargo_toml(crate_path, crate_name, version):
+    """Extract and parse Cargo.toml from a .crate tarball."""
+    expected_path = f"{crate_name}-{version}/Cargo.toml"
+    with tarfile.open(crate_path, "r:gz") as tar:
+        cargo_toml_file = tar.extractfile(expected_path)
+        if cargo_toml_file is None:
+            raise FileNotFoundError(f"No Cargo.toml found in {crate_path} at {expected_path}")
+        return tomllib.load(cargo_toml_file)
+
+
+def _normalize_req(version_str):
+    """Normalize a Cargo version requirement to its explicit form.
+
+    In Cargo.toml, a bare version like "1.0" is shorthand for "^1.0".
+    The index format uses the explicit form with the comparator prefix.
+    """
+    if not version_str or version_str == "*":
+        return version_str
+    # Already has a comparator prefix
+    if version_str[0] in ("^", "~", "=", ">", "<"):
+        return version_str
+    return f"^{version_str}"
+
+
+def parse_dep(name, spec, kind="normal", target=None):
+    """Convert a single Cargo.toml dependency entry to index format."""
+    if isinstance(spec, str):
+        # Simple form: dep = "1.0"
+        return {
+            "name": name,
+            "req": _normalize_req(spec),
+            "features": [],
+            "optional": False,
+            "default_features": True,
+            "target": target,
+            "kind": kind,
+            "registry": None,
+            "package": None,
+        }
+
+    # Table form: dep = { version = "1.0", optional = true, ... }
+    dep = {
+        "name": name,
+        "req": _normalize_req(spec.get("version", "*")),
+        "features": spec.get("features", []),
+        "optional": spec.get("optional", False),
+        "default_features": spec.get("default-features", True),
+        "target": target,
+        "kind": kind,
+        "registry": spec.get("registry"),
+        "package": None,
+    }
+    # If the dep was renamed, "name" in the index is the alias (the key),
+    # and "package" is the real crate name
+    if "package" in spec:
+        dep["package"] = spec["package"]
+    return dep
+
+
+def extract_dependencies(cargo_toml):
+    """Extract all dependencies from a parsed Cargo.toml into index format."""
+    deps = []
+
+    for name, spec in cargo_toml.get("dependencies", {}).items():
+        deps.append(parse_dep(name, spec, kind="normal"))
+
+    for name, spec in cargo_toml.get("dev-dependencies", {}).items():
+        deps.append(parse_dep(name, spec, kind="dev"))
+
+    for name, spec in cargo_toml.get("build-dependencies", {}).items():
+        deps.append(parse_dep(name, spec, kind="build"))
+
+    # Platform-specific dependencies: [target.'cfg(...)'.dependencies]
+    for target, target_deps in cargo_toml.get("target", {}).items():
+        for name, spec in target_deps.get("dependencies", {}).items():
+            deps.append(parse_dep(name, spec, kind="normal", target=target))
+        for name, spec in target_deps.get("dev-dependencies", {}).items():
+            deps.append(parse_dep(name, spec, kind="dev", target=target))
+        for name, spec in target_deps.get("build-dependencies", {}).items():
+            deps.append(parse_dep(name, spec, kind="build", target=target))
+
+    return deps

pulp_rust/app/views.py

@@ -178,19 +178,21 @@ def _build_index_response(crate_versions):
         for crate_version in crate_versions:
             deps = []
             for dep in crate_version.dependencies.all():
-                deps.append(
-                    {
-                        "name": dep.name,
-                        "req": dep.req,
-                        "features": dep.features,
-                        "optional": dep.optional,
-                        "default_features": dep.default_features,
-                        "target": dep.target,
-                        "kind": dep.kind,
-                        "registry": dep.registry,
-                        "package": dep.package,
-                    }
-                )
+                dep_obj = {
+                    "name": dep.name,
+                    "req": dep.req,
+                    "features": dep.features,
+                    "optional": dep.optional,
+                    "default_features": dep.default_features,
+                    "target": dep.target,
+                    "kind": dep.kind,
+                }
+                # crates.io omits these keys when not set
+                if dep.registry is not None:
+                    dep_obj["registry"] = dep.registry
+                if dep.package is not None:
+                    dep_obj["package"] = dep.package
+                deps.append(dep_obj)
 
             version_obj = {
                 "name": crate_version.name,

pulp_rust/tests/functional/api/test_cargo_api.py

@@ -6,9 +6,7 @@
 import pytest
 from aiohttp.client_exceptions import ClientResponseError
 
-from pulp_rust.tests.functional.utils import download_file
-
-CRATES_IO_URL = "sparse+https://index.crates.io/"
+from pulp_rust.tests.functional.utils import CRATES_IO_URL, download_file
 
 
 def test_config_json(

pulp_rust/tests/functional/api/test_download_content.py

@@ -7,7 +7,7 @@
 import hashlib
 from urllib.parse import urljoin
 
-from pulp_rust.tests.functional.utils import download_file
+from pulp_rust.tests.functional.utils import CRATES_IO_URL, download_file
 
 
 def test_download_content(
@@ -27,7 +27,7 @@ def test_download_content(
     3. Verify that the content was automatically added to the repository.
     4. Remove the remote and verify the content is still served from cache.
     """
-    remote = rust_remote_factory(url="sparse+https://index.crates.io/")
+    remote = rust_remote_factory(url=CRATES_IO_URL)
     repository = rust_repo_factory(remote=remote.pulp_href)
     distribution = rust_distribution_factory(
         remote=remote.pulp_href, repository=repository.pulp_href