Initial commit

internal/classifier/parse_test.go

@@ -0,0 +1,156 @@
+package classifier
+
+import (
+	"testing"
+)
+
+func TestParseTier1Response_Clean(t *testing.T) {
+	raw := `{"channel": "tech", "platforms": [], "duplicate_of": null, "rationale": "GPU launch story"}`
+	result, err := ParseTier1Response(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.Channel != "tech" {
+		t.Errorf("channel = %q, want tech", result.Channel)
+	}
+	if len(result.Platforms) != 0 {
+		t.Errorf("platforms = %v, want empty", result.Platforms)
+	}
+	if result.DuplicateOf != nil {
+		t.Errorf("duplicate_of = %v, want nil", result.DuplicateOf)
+	}
+}
+
+func TestParseTier1Response_WithPlatforms(t *testing.T) {
+	raw := `{"channel":"gaming","platforms":["nintendo-switch","pc"],"duplicate_of":null,"rationale":"Switch 2 launch"}`
+	result, err := ParseTier1Response(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.Channel != "gaming" {
+		t.Errorf("channel = %q", result.Channel)
+	}
+	if len(result.Platforms) != 2 {
+		t.Fatalf("platforms = %v, want 2", result.Platforms)
+	}
+	if result.Platforms[0] != "nintendo-switch" || result.Platforms[1] != "pc" {
+		t.Errorf("platforms = %v", result.Platforms)
+	}
+}
+
+func TestParseTier1Response_WithDuplicate(t *testing.T) {
+	raw := `{"channel":"politics","platforms":[],"duplicate_of":"guid-abc-123","rationale":"same event"}`
+	result, err := ParseTier1Response(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.DuplicateOf == nil || *result.DuplicateOf != "guid-abc-123" {
+		t.Errorf("duplicate_of = %v", result.DuplicateOf)
+	}
+}
+
+func TestParseTier1Response_InvalidChannel(t *testing.T) {
+	raw := `{"channel":"sports","platforms":[],"duplicate_of":null,"rationale":"test"}`
+	_, err := ParseTier1Response(raw)
+	if err == nil {
+		t.Error("expected error for invalid channel")
+	}
+}
+
+func TestParseTier1Response_FiltersInvalidPlatforms(t *testing.T) {
+	raw := `{"channel":"gaming","platforms":["nintendo-switch","fake-platform","pc","also-fake"],"duplicate_of":null,"rationale":"test"}`
+	result, err := ParseTier1Response(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(result.Platforms) != 2 {
+		t.Fatalf("platforms = %v, want [nintendo-switch pc]", result.Platforms)
+	}
+}
+
+func TestParseTier2Response_PostTrue(t *testing.T) {
+	raw := `{"post":true,"channel":"politics","platforms":[],"duplicate_of":null,"rationale":"major world event"}`
+	result, err := ParseTier2Response(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !result.Post {
+		t.Error("post = false, want true")
+	}
+	if result.Channel != "politics" {
+		t.Errorf("channel = %q", result.Channel)
+	}
+}
+
+func TestParseTier2Response_PostFalse(t *testing.T) {
+	raw := `{"post":false,"platforms":[],"duplicate_of":null,"rationale":"local news"}`
+	result, err := ParseTier2Response(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.Post {
+		t.Error("post = true, want false")
+	}
+}
+
+func TestRepairJSON_MarkdownFences(t *testing.T) {
+	raw := "```json\n{\"channel\": \"tech\", \"platforms\": [], \"duplicate_of\": null, \"rationale\": \"test\"}\n```"
+	result, err := ParseTier1Response(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.Channel != "tech" {
+		t.Errorf("channel = %q", result.Channel)
+	}
+}
+
+func TestRepairJSON_ThinkBlocks(t *testing.T) {
+	raw := "<think>I need to classify this as tech because it's about GPUs</think>\n{\"channel\": \"tech\", \"platforms\": [], \"duplicate_of\": null, \"rationale\": \"GPU story\"}"
+	result, err := ParseTier1Response(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.Channel != "tech" {
+		t.Errorf("channel = %q", result.Channel)
+	}
+}
+
+func TestRepairJSON_TrailingCommas(t *testing.T) {
+	raw := `{"channel": "politics", "platforms": [], "duplicate_of": null, "rationale": "test",}`
+	result, err := ParseTier1Response(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.Channel != "politics" {
+		t.Errorf("channel = %q", result.Channel)
+	}
+}
+
+func TestRepairJSON_TrailingCommaInArray(t *testing.T) {
+	raw := `{"channel": "gaming", "platforms": ["pc", "xbox",], "duplicate_of": null, "rationale": "test"}`
+	result, err := ParseTier1Response(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(result.Platforms) != 2 {
+		t.Errorf("platforms = %v", result.Platforms)
+	}
+}
+
+func TestRepairJSON_CombinedIssues(t *testing.T) {
+	raw := "<think>reasoning here</think>\n```json\n{\"channel\": \"tech\", \"platforms\": [],\"duplicate_of\": null, \"rationale\": \"test\",}\n```"
+	result, err := ParseTier1Response(raw)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.Channel != "tech" {
+		t.Errorf("channel = %q", result.Channel)
+	}
+}
+
+func TestRepairJSON_GarbageInput(t *testing.T) {
+	_, err := ParseTier1Response("I think this should go to tech because reasons")
+	if err == nil {
+		t.Error("expected error for garbage input")
+	}
+}