fix(uv): correct GitHub archive URLs for git ref sources (#1238)

### Changes are visible to end-users: no

### Test plan

- Covered by existing test cases
- New test cases added

Author: jbedard

e2e/cases/MODULE.bazel

@@ -123,6 +123,7 @@ include("//uv-deps-650:setup.MODULE.bazel")
 include("//uv-data-purelib:setup.MODULE.bazel")
 include("//uv-dup-lock-records:setup.MODULE.bazel")
 include("//uv-duplicate-whl-labels:setup.MODULE.bazel")
+include("//uv-git-source:setup.MODULE.bazel")
 include("//uv-include-group:setup.MODULE.bazel")
 include("//uv-incompatible-wheel-source:setup.MODULE.bazel")
 include("//uv-invalid-build-overrides:setup.MODULE.bazel")

e2e/cases/uv-git-source/BUILD.bazel

@@ -0,0 +1,17 @@
+load("@aspect_rules_py//py:defs.bzl", "py_test")
+
+# Runtime coverage for git-sourced lockfile packages. six is locked with
+# `source = { git = "https://github.com/benjaminp/six?tag=1.17.0#<commit>" }`,
+# exercising parse_git_url on a real uv-generated URL, the GitHub
+# git-to-http_archive conversion in collect_sdists, and the source build of
+# the resulting archive.
+py_test(
+    name = "test",
+    srcs = ["__test__.py"],
+    dep_group = "uv_git_source",
+    main = "__test__.py",
+    python_version = "3.11",
+    deps = [
+        "@pypi_git_source//six",
+    ],
+)

e2e/cases/uv-git-source/__test__.py

@@ -0,0 +1,6 @@
+import six
+
+assert six.__version__ == "1.17.0", six.__version__
+assert six.ensure_str(b"git-source") == "git-source"
+
+print("six", six.__version__, "imported from git source")

e2e/cases/uv-git-source/pyproject.toml

@@ -0,0 +1,18 @@
+[project]
+name = "uv_git_source"
+version = "0.0.0"
+requires-python = ">=3.11"
+dependencies = [
+    # `build` and `setuptools` surface in the project repo as deps of the
+    # pep517_whl build_tool that builds six from its git source tree.
+    "build",
+    "setuptools",
+    "six",
+]
+
+# six is resolved from git rather than a registry, exercising the
+# `source = { git = ... }` lockfile pathway: parse_git_url on the
+# uv-generated URL (query params + pinned commit fragment) and the
+# git-to-http_archive conversion for GitHub remotes.
+[tool.uv.sources]
+six = { git = "https://github.com/benjaminp/six", tag = "1.17.0" }

e2e/cases/uv-git-source/setup.MODULE.bazel

@@ -0,0 +1,23 @@
+"""Coverage for git-sourced lockfile packages: six resolves from a
+`source = { git = ... }` entry (uv-generated URL with query params and a
+pinned commit fragment), which is converted to an http_archive of the
+GitHub tarball and built from source. Uses its own hub to keep the git
+source isolated from the shared `pypi` hub.
+"""
+
+uv = use_extension("@aspect_rules_py//uv:extensions.bzl", "uv")
+uv.declare_hub(hub_name = "pypi_git_source")
+uv.project(
+    default_build_dependencies = [
+        "build",
+        "setuptools",
+    ],
+    hub_name = "pypi_git_source",
+    lock = "//uv-git-source:uv.lock",
+    pyproject = "//uv-git-source:pyproject.toml",
+)
+use_repo(
+    uv,
+    "pypi_git_source",
+    "sdist_build__uv_git_source__six__1_17_0",
+)

e2e/cases/uv-git-source/uv.lock

@@ -1,10 +1,13 @@
 load("@bazel_lib//:bzl_library.bzl", "bzl_library")
 load(":test_defs.bzl", "defs_test_suite")
+load(":test_git_utils.bzl", "git_utils_test_suite")
 load(":test_graph_utils.bzl", "graph_utils_test_suite")
 load(":test_projectfile.bzl", "projectfile_test_suite")
 
 defs_test_suite()
 
+git_utils_test_suite()
+
 graph_utils_test_suite()
 
 projectfile_test_suite()
@@ -84,6 +87,16 @@ bzl_library(
     ],
 )
 
+bzl_library(
+    name = "test_git_utils",
+    srcs = ["test_git_utils.bzl"],
+    visibility = ["//uv:__subpackages__"],
+    deps = [
+        ":git_utils",
+        "@bazel_skylib//lib:unittest",
+    ],
+)
+
 bzl_library(
     name = "test_graph_utils",
     srcs = ["test_graph_utils.bzl"],

uv/private/extension/BUILD.bazel

@@ -3,19 +3,19 @@ Utilities for parsing git URLs and converting them to http_archive.
 """
 
 def ensure_ref(maybe_ref):
-    """Ensures a git ref starts with "ref/".
+    """Ensures a git ref starts with "refs/".
 
     Args:
         maybe_ref: The git ref string.
 
     Returns:
-        The git ref string, prefixed with "ref/" if it is not already.
+        The git ref string, prefixed with "refs/" if it is not already.
     """
     if maybe_ref == None:
         return None
 
-    if not maybe_ref.startswith("ref/"):
-        return "ref/" + maybe_ref
+    if not maybe_ref.startswith("refs/"):
+        return "refs/" + maybe_ref
 
     return maybe_ref
 
@@ -98,7 +98,7 @@ def try_git_to_http_archive(git_cfg):
                 "url": url,
             }
         elif "ref" in git_cfg:
-            url = "{}/archive/{}.tar.gz".format(url, git_cfg["tag"])
+            url = "{}/archive/{}.tar.gz".format(url, git_cfg["ref"])
             return {
                 "url": url,
             }

uv/private/extension/git_utils.bzl

@@ -0,0 +1,141 @@
+load("@bazel_skylib//lib:unittest.bzl", "asserts", "unittest")
+load(":git_utils.bzl", "ensure_ref", "parse_git_url", "try_git_to_http_archive")
+
+def _ensure_ref_test_impl(ctx):
+    env = unittest.begin(ctx)
+    asserts.equals(env, None, ensure_ref(None))
+    asserts.equals(env, "refs/heads/main", ensure_ref("heads/main"))
+    asserts.equals(env, "refs/pull/64/head", ensure_ref("pull/64/head"))
+
+    # An already fully-qualified ref must not be prefixed again.
+    asserts.equals(env, "refs/pull/64/head", ensure_ref("refs/pull/64/head"))
+    asserts.equals(env, "refs/tags/v1.0.0", ensure_ref("refs/tags/v1.0.0"))
+    return unittest.end(env)
+
+ensure_ref_test = unittest.make(_ensure_ref_test_impl)
+
+def _parse_git_url_fragment_commit_test_impl(ctx):
+    env = unittest.begin(ctx)
+    result = parse_git_url("https://github.com/user/repo.git#c7076a0c6e34d7b2fa4e0ecd7ba4b8e9d3d9e0f1")
+    asserts.equals(env, {
+        "remote": "https://github.com/user/repo.git",
+        "commit": "c7076a0c6e34d7b2fa4e0ecd7ba4b8e9d3d9e0f1",
+    }, result)
+    return unittest.end(env)
+
+parse_git_url_fragment_commit_test = unittest.make(_parse_git_url_fragment_commit_test_impl)
+
+def _parse_git_url_query_commit_test_impl(ctx):
+    env = unittest.begin(ctx)
+    result = parse_git_url("https://github.com/user/repo.git?commit=c7076a0c6e34d7b2fa4e0ecd7ba4b8e9d3d9e0f1")
+    asserts.equals(env, {
+        "remote": "https://github.com/user/repo.git",
+        "commit": "c7076a0c6e34d7b2fa4e0ecd7ba4b8e9d3d9e0f1",
+    }, result)
+    return unittest.end(env)
+
+parse_git_url_query_commit_test = unittest.make(_parse_git_url_query_commit_test_impl)
+
+def _parse_git_url_query_ref_test_impl(ctx):
+    env = unittest.begin(ctx)
+    result = parse_git_url("https://github.com/user/repo.git?ref=refs%2Fpull%2F64%2Fhead")
+    asserts.equals(env, {
+        "remote": "https://github.com/user/repo.git",
+        "ref": "refs/pull/64/head",
+    }, result)
+    return unittest.end(env)
+
+parse_git_url_query_ref_test = unittest.make(_parse_git_url_query_ref_test_impl)
+
+def _parse_git_url_fragment_wins_over_query_test_impl(ctx):
+    env = unittest.begin(ctx)
+    result = parse_git_url("https://github.com/user/repo.git?ref=refs%2Fheads%2Fmain#c7076a0c6e34d7b2fa4e0ecd7ba4b8e9d3d9e0f1")
+    asserts.equals(env, {
+        "remote": "https://github.com/user/repo.git",
+        "commit": "c7076a0c6e34d7b2fa4e0ecd7ba4b8e9d3d9e0f1",
+    }, result)
+    return unittest.end(env)
+
+parse_git_url_fragment_wins_over_query_test = unittest.make(_parse_git_url_fragment_wins_over_query_test_impl)
+
+def _parse_git_url_bare_remote_test_impl(ctx):
+    env = unittest.begin(ctx)
+    result = parse_git_url("https://github.com/user/repo.git")
+    asserts.equals(env, {"remote": "https://github.com/user/repo.git"}, result)
+    return unittest.end(env)
+
+parse_git_url_bare_remote_test = unittest.make(_parse_git_url_bare_remote_test_impl)
+
+def _git_to_http_archive_commit_test_impl(ctx):
+    env = unittest.begin(ctx)
+    result = try_git_to_http_archive({
+        "remote": "git+https://github.com/user/repo.git",
+        "commit": "c7076a0c6e34d7b2fa4e0ecd7ba4b8e9d3d9e0f1",
+    })
+    asserts.equals(env, {
+        "url": "https://github.com/user/repo/archive/c7076a0c6e34d7b2fa4e0ecd7ba4b8e9d3d9e0f1.tar.gz",
+    }, result)
+    return unittest.end(env)
+
+git_to_http_archive_commit_test = unittest.make(_git_to_http_archive_commit_test_impl)
+
+def _git_to_http_archive_ref_test_impl(ctx):
+    env = unittest.begin(ctx)
+
+    # Regression test: this branch used to read git_cfg["tag"], a key
+    # parse_git_url never produces.
+    result = try_git_to_http_archive({
+        "remote": "https://github.com/user/repo",
+        "ref": "refs/pull/64/head",
+    })
+    asserts.equals(env, {
+        "url": "https://github.com/user/repo/archive/refs/pull/64/head.tar.gz",
+    }, result)
+    return unittest.end(env)
+
+git_to_http_archive_ref_test = unittest.make(_git_to_http_archive_ref_test_impl)
+
+def _git_to_http_archive_parsed_ref_url_test_impl(ctx):
+    env = unittest.begin(ctx)
+    result = try_git_to_http_archive(parse_git_url("git+https://github.com/user/repo.git?ref=refs%2Ftags%2Fv1.0.0"))
+    asserts.equals(env, {
+        "url": "https://github.com/user/repo/archive/refs/tags/v1.0.0.tar.gz",
+    }, result)
+    return unittest.end(env)
+
+git_to_http_archive_parsed_ref_url_test = unittest.make(_git_to_http_archive_parsed_ref_url_test_impl)
+
+def _git_to_http_archive_non_github_test_impl(ctx):
+    env = unittest.begin(ctx)
+    result = try_git_to_http_archive({
+        "remote": "https://gitlab.com/user/repo.git",
+        "commit": "c7076a0c6e34d7b2fa4e0ecd7ba4b8e9d3d9e0f1",
+    })
+    asserts.equals(env, None, result)
+    return unittest.end(env)
+
+git_to_http_archive_non_github_test = unittest.make(_git_to_http_archive_non_github_test_impl)
+
+def _git_to_http_archive_no_rev_test_impl(ctx):
+    env = unittest.begin(ctx)
+    result = try_git_to_http_archive({"remote": "https://github.com/user/repo.git"})
+    asserts.equals(env, None, result)
+    return unittest.end(env)
+
+git_to_http_archive_no_rev_test = unittest.make(_git_to_http_archive_no_rev_test_impl)
+
+def git_utils_test_suite():
+    unittest.suite(
+        "git_utils_tests",
+        ensure_ref_test,
+        parse_git_url_fragment_commit_test,
+        parse_git_url_query_commit_test,
+        parse_git_url_query_ref_test,
+        parse_git_url_fragment_wins_over_query_test,
+        parse_git_url_bare_remote_test,
+        git_to_http_archive_commit_test,
+        git_to_http_archive_ref_test,
+        git_to_http_archive_parsed_ref_url_test,
+        git_to_http_archive_non_github_test,
+        git_to_http_archive_no_rev_test,
+    )