package llmjson

import "testing"

func TestUnmarshalToleratesEllipsisAndCommas(t *testing.T) {
	raw := []byte(`{
  "nodes": [
    {"label": "a", "url": "https://example.com/a"},
    ...,
    {"label": "b"},
  ],
  "edges": [...]
}`)
	var out struct {
		Nodes []struct {
			Label string `json:"label"`
			URL   string `json:"url"`
		} `json:"nodes"`
		Edges []any `json:"edges"`
	}
	if err := Unmarshal(raw, &out); err != nil {
		t.Fatalf("Unmarshal error: %v", err)
	}
	if len(out.Nodes) != 2 {
		t.Fatalf("nodes = %d, want 2", len(out.Nodes))
	}
	if out.Nodes[0].URL != "https://example.com/a" {
		t.Fatalf("url mutated: %q", out.Nodes[0].URL)
	}
}

func TestSanitizeReturnsNilWhenValid(t *testing.T) {
	if got := Sanitize([]byte(`{"a":1}`)); got != nil {
		t.Fatalf("expected nil for clean json, got %q", got)
	}
}

func TestSanitizeEscapesNewlinesInsideStrings(t *testing.T) {
	raw := []byte(`{"examples":"第一行
第二行","identity":"觀察者"}`)
	cleaned := Sanitize(raw)
	if cleaned == nil {
		t.Fatal("expected sanitized output")
	}
	var out struct {
		Examples string `json:"examples"`
		Identity string `json:"identity"`
	}
	if err := Unmarshal(raw, &out); err != nil {
		t.Fatalf("Unmarshal error: %v", err)
	}
	if out.Examples != "第一行\n第二行" {
		t.Fatalf("examples = %q", out.Examples)
	}
	if out.Identity != "觀察者" {
		t.Fatalf("identity = %q", out.Identity)
	}
}

func TestSanitizeKeepsDotsInsideStrings(t *testing.T) {
	raw := []byte(`{"note":"see ... and https://a.b/c", "list":[1, ...]}`)
	cleaned := Sanitize(raw)
	if cleaned == nil {
		t.Fatal("expected sanitized output")
	}
	var out struct {
		Note string `json:"note"`
		List []int  `json:"list"`
	}
	if err := Unmarshal(raw, &out); err != nil {
		t.Fatalf("Unmarshal error: %v", err)
	}
	if out.Note != "see ... and https://a.b/c" {
		t.Fatalf("string content mutated: %q", out.Note)
	}
	if len(out.List) != 1 || out.List[0] != 1 {
		t.Fatalf("list = %v, want [1]", out.List)
	}
}