Fetch the codegen test corpus at test time

The codegen tests read the shared test corpus (schemas, PCL programs,
types.json) from tests/testdata/codegen in pulumi/pulumi, previously
reached through the submodule. At 11MB across ~2,250 files the corpus is
too large to vendor, so `make test_codegen` now fetches it into the
gitignored pulumi-language-dotnet/codegen/testdata/upstream directory
via a sparse clone of the version pinned in go.mod. A version stamp
skips the download when the corpus is already current. The script uses
cp instead of rsync so it also runs under Git Bash on Windows.

Author: iwahbe

.gitignore

@@ -461,4 +461,6 @@ pulumi-resource-testprovider
 
 # Ignore user-provided go.work files.
 /go.work
-/go.work.sum
+/go.work.sum
+# Codegen test corpus fetched from pulumi/pulumi by scripts/get_codegen_testdata.sh.
+pulumi-language-dotnet/codegen/testdata/upstream/

Makefile

@@ -135,8 +135,13 @@ test_fast: test_sdk test_sdk_automation test_automation_codegen
 test_conformance: build
 	cd pulumi-language-dotnet && gotestsum -- $(GO_TEST_FILTER_FLAG) --timeout 60m .
 
+## Fetches the codegen test corpus from the pinned pulumi/pulumi version.
+.PHONY: codegen_testdata
+codegen_testdata:
+	./scripts/get_codegen_testdata.sh
+
 .PHONY: test_codegen
-test_codegen: build
+test_codegen: build codegen_testdata
 	cd pulumi-language-dotnet && gotestsum -- $(GO_TEST_FILTER_FLAG) --timeout 60m ./codegen/...
 
 .PHONY: test_integration

pulumi-language-dotnet/AGENTS.md

@@ -9,6 +9,7 @@ Go language host for Pulumi .NET. Invoked by the Pulumi CLI to build, run, and g
 - `codegen/gen.go` — code generation from Pulumi schemas to C#
 - `codegen/gen_program.go` — PCL to C# transpilation
 - `codegen/testdata/` — golden files for codegen tests
+- `codegen/testdata/upstream/` — gitignored codegen test corpus (schemas, PCL programs) fetched from the pinned pulumi/pulumi version by `make codegen_testdata`
 - `testdata/` — conformance test fixtures (projects and SDKs)
 - `version/` — version string injected at build time via `-ldflags`
 

pulumi-language-dotnet/codegen/gen_program_test.go

@@ -54,7 +54,7 @@ func TestGenerateProgram(t *testing.T) {
 		GenProgram: GenerateProgram,
 		TestCases:  test.PulumiPulumiProgramTests,
 
-		InputDirectory:  filepath.Join("..", "..", "pulumi", "tests", "testdata", "codegen"),
+		InputDirectory:  filepath.Join("testdata", "upstream", "codegen"),
 		ResultDirectory: "testdata",
 	})
 }
@@ -73,7 +73,7 @@ func TestGenerateProgramYAML(t *testing.T) {
 		GenProgram: GenerateProgram,
 		TestCases:  test.PulumiPulumiYAMLProgramTests,
 
-		InputDirectory:  filepath.Join("..", "..", "pulumi", "tests", "testdata", "codegen"),
+		InputDirectory:  filepath.Join("testdata", "upstream", "codegen"),
 		ResultDirectory: "testdata",
 	})
 }
@@ -123,7 +123,7 @@ resource "test-organization" "tfe:index/organization:Organization" {
 	program, diags, err := parseAndBindProgram(t,
 		source,
 		"main.pp",
-		filepath.Join("..", "..", "pulumi", "tests", "testdata", "codegen", "parameterized-schemas"))
+		filepath.Join("testdata", "upstream", "codegen", "parameterized-schemas"))
 
 	require.NoError(t, err)
 	require.False(t, diags.HasErrors(), "unexpected diags: %v", diags)
@@ -189,7 +189,7 @@ resource "test-organization" "tfe:index/organization:Organization" {
 	program, diags, err := parseAndBindProgram(t,
 		source,
 		"main.pp",
-		filepath.Join("..", "..", "pulumi", "tests", "testdata", "codegen", "parameterized-schemas"))
+		filepath.Join("testdata", "upstream", "codegen", "parameterized-schemas"))
 
 	require.NoError(t, err)
 	require.False(t, diags.HasErrors(), "unexpected diags: %v", diags)

pulumi-language-dotnet/codegen/gen_test.go

@@ -67,7 +67,7 @@ func TestGeneratePackage(t *testing.T) {
 		},
 		TestCases: filterTests(),
 
-		InputDir:  filepath.Join("..", "..", "pulumi", "tests", "testdata", "codegen"),
+		InputDir:  filepath.Join("testdata", "upstream", "codegen"),
 		ResultDir: "testdata",
 	})
 }
@@ -160,5 +160,5 @@ func TestGenerateTypeNames(t *testing.T) {
 		return func(t schema.Type) string {
 			return root.typeString(t, "", false, false, false)
 		}
-	}, filepath.FromSlash("../../pulumi/tests/testdata/codegen"))
+	}, filepath.FromSlash("testdata/upstream/codegen"))
 }

scripts/get_codegen_testdata.sh

@@ -0,0 +1,37 @@
+#!/usr/bin/env bash
+#
+# Fetches the shared codegen test corpus (tests/testdata/codegen) from
+# pulumi/pulumi at the version pinned in pulumi-language-dotnet/go.mod into
+# pulumi-language-dotnet/codegen/testdata/upstream (gitignored). The download
+# is skipped when the corpus is already present at the pinned version.
+#
+# Uses cp instead of rsync because this also runs under Git Bash on Windows.
+
+set -euo pipefail
+
+cd "$(dirname "$0")/.."
+
+VERSION=$(sed -n 's/^.*github\.com\/pulumi\/pulumi\/sdk\/v3 v//p' pulumi-language-dotnet/go.mod)
+if [ -z "$VERSION" ]; then
+    echo "error: could not determine the pulumi version from pulumi-language-dotnet/go.mod" >&2
+    exit 1
+fi
+
+DEST=pulumi-language-dotnet/codegen/testdata/upstream
+if [ -f "$DEST/.version" ] && [ "$(cat "$DEST/.version")" = "v$VERSION" ]; then
+    exit 0
+fi
+
+echo "Fetching the codegen test corpus from pulumi/pulumi v$VERSION"
+
+tmp=$(mktemp -d)
+trap 'rm -rf "$tmp"' EXIT
+
+git -c advice.detachedHead=false clone --quiet --depth 1 --branch "v$VERSION" --filter=blob:none --sparse \
+    https://github.com/pulumi/pulumi.git "$tmp"
+git -C "$tmp" sparse-checkout set --no-cone tests/testdata/codegen
+
+rm -rf "$DEST"
+mkdir -p "$DEST"
+cp -R "$tmp/tests/testdata/codegen" "$DEST/codegen"
+echo "v$VERSION" > "$DEST/.version"