go-extractor/sites/recipe/recipe_test.go

package recipe

import (
	"context"
	"testing"

	"gitea.stevedudenhoeffer.com/steve/go-extractor"
	"gitea.stevedudenhoeffer.com/steve/go-extractor/extractortest"
)

const sampleJSONLD = `{
  "@type": "Recipe",
  "name": "Chocolate Chip Cookies",
  "description": "The best chocolate chip cookies ever.",
  "author": {"@type": "Person", "name": "Jane Smith"},
  "prepTime": "PT15M",
  "cookTime": "PT10M",
  "totalTime": "PT25M",
  "recipeYield": "24 cookies",
  "recipeIngredient": [
    "2 cups flour",
    "1 cup sugar",
    "1 cup chocolate chips"
  ],
  "recipeInstructions": [
    {"@type": "HowToStep", "text": "Preheat oven to 350F."},
    {"@type": "HowToStep", "text": "Mix dry ingredients."},
    {"@type": "HowToStep", "text": "Bake for 10 minutes."}
  ],
  "image": "https://example.com/cookies.jpg",
  "nutrition": {"calories": "250 calories"},
  "aggregateRating": {"ratingValue": "4.8"}
}`

const sampleGraphJSONLD = `{
  "@graph": [
    {"@type": "WebPage", "name": "Recipe Page"},
    {
      "@type": "Recipe",
      "name": "Banana Bread",
      "author": "Bob Baker",
      "recipeIngredient": ["3 bananas", "2 cups flour"],
      "recipeInstructions": "Mix and bake at 350F for 60 minutes."
    }
  ]
}`

const sampleArrayJSONLD = `[
  {"@type": "WebSite", "name": "Cooking Blog"},
  {
    "@type": "Recipe",
    "name": "Pancakes",
    "recipeYield": ["4 servings"],
    "recipeIngredient": ["1 cup flour", "1 egg", "1 cup milk"],
    "image": ["https://example.com/pancakes.jpg"]
  }
]`

func makeRecipeDoc(jsonLD string) *extractortest.MockDocument {
	return &extractortest.MockDocument{
		URLValue: "https://www.allrecipes.com/recipe/10813/best-chocolate-chip-cookies/",
		MockNode: extractortest.MockNode{
			Children: map[string]extractor.Nodes{
				"script[type='application/ld+json']": {
					&extractortest.MockNode{TextValue: jsonLD},
				},
			},
		},
	}
}

func TestExtractRecipeFromJSONLD(t *testing.T) {
	doc := makeRecipeDoc(sampleJSONLD)

	r, err := extractRecipeFromJSONLD(doc)
	if err != nil {
		t.Fatalf("extractRecipeFromJSONLD() error: %v", err)
	}

	if r.Name != "Chocolate Chip Cookies" {
		t.Errorf("Name = %q, want %q", r.Name, "Chocolate Chip Cookies")
	}
	if r.Author != "Jane Smith" {
		t.Errorf("Author = %q, want %q", r.Author, "Jane Smith")
	}
	if r.PrepTime != "15 min" {
		t.Errorf("PrepTime = %q, want %q", r.PrepTime, "15 min")
	}
	if r.CookTime != "10 min" {
		t.Errorf("CookTime = %q, want %q", r.CookTime, "10 min")
	}
	if r.TotalTime != "25 min" {
		t.Errorf("TotalTime = %q, want %q", r.TotalTime, "25 min")
	}
	if r.Yield != "24 cookies" {
		t.Errorf("Yield = %q, want %q", r.Yield, "24 cookies")
	}
	if len(r.Ingredients) != 3 {
		t.Fatalf("len(Ingredients) = %d, want 3", len(r.Ingredients))
	}
	if r.Ingredients[0] != "2 cups flour" {
		t.Errorf("Ingredients[0] = %q, want %q", r.Ingredients[0], "2 cups flour")
	}
	if len(r.Instructions) != 3 {
		t.Fatalf("len(Instructions) = %d, want 3", len(r.Instructions))
	}
	if r.Instructions[0] != "Preheat oven to 350F." {
		t.Errorf("Instructions[0] = %q, want %q", r.Instructions[0], "Preheat oven to 350F.")
	}
	if r.ImageURL != "https://example.com/cookies.jpg" {
		t.Errorf("ImageURL = %q, want %q", r.ImageURL, "https://example.com/cookies.jpg")
	}
	if r.Calories != "250 calories" {
		t.Errorf("Calories = %q, want %q", r.Calories, "250 calories")
	}
	if r.Rating != 4.8 {
		t.Errorf("Rating = %v, want 4.8", r.Rating)
	}
}

func TestExtractRecipeFromJSONLD_Graph(t *testing.T) {
	doc := makeRecipeDoc(sampleGraphJSONLD)

	r, err := extractRecipeFromJSONLD(doc)
	if err != nil {
		t.Fatalf("extractRecipeFromJSONLD() error: %v", err)
	}

	if r.Name != "Banana Bread" {
		t.Errorf("Name = %q, want %q", r.Name, "Banana Bread")
	}
	if r.Author != "Bob Baker" {
		t.Errorf("Author = %q, want %q", r.Author, "Bob Baker")
	}
	if len(r.Ingredients) != 2 {
		t.Fatalf("len(Ingredients) = %d, want 2", len(r.Ingredients))
	}
	if len(r.Instructions) != 1 {
		t.Fatalf("len(Instructions) = %d, want 1", len(r.Instructions))
	}
	if r.Instructions[0] != "Mix and bake at 350F for 60 minutes." {
		t.Errorf("Instructions[0] = %q, want %q", r.Instructions[0], "Mix and bake at 350F for 60 minutes.")
	}
}

func TestExtractRecipeFromJSONLD_Array(t *testing.T) {
	doc := makeRecipeDoc(sampleArrayJSONLD)

	r, err := extractRecipeFromJSONLD(doc)
	if err != nil {
		t.Fatalf("extractRecipeFromJSONLD() error: %v", err)
	}

	if r.Name != "Pancakes" {
		t.Errorf("Name = %q, want %q", r.Name, "Pancakes")
	}
	if r.Yield != "4 servings" {
		t.Errorf("Yield = %q, want %q", r.Yield, "4 servings")
	}
	if r.ImageURL != "https://example.com/pancakes.jpg" {
		t.Errorf("ImageURL = %q, want %q", r.ImageURL, "https://example.com/pancakes.jpg")
	}
}

func TestExtractRecipeFromJSONLD_NoRecipe(t *testing.T) {
	doc := makeRecipeDoc(`{"@type": "WebPage", "name": "Not a recipe"}`)

	_, err := extractRecipeFromJSONLD(doc)
	if err == nil {
		t.Error("expected error for non-Recipe JSON-LD")
	}
}

func TestExtractRecipeFromDOM(t *testing.T) {
	doc := &extractortest.MockDocument{
		MockNode: extractortest.MockNode{
			Children: map[string]extractor.Nodes{
				"h1": {
					&extractortest.MockNode{TextValue: "Grandma's Apple Pie"},
				},
				"div.recipe-summary p": {
					&extractortest.MockNode{TextValue: "A classic apple pie recipe."},
				},
				"li.ingredient": {
					&extractortest.MockNode{TextValue: "6 apples"},
					&extractortest.MockNode{TextValue: "1 cup sugar"},
					&extractortest.MockNode{TextValue: "2 pie crusts"},
				},
				"li.instruction": {
					&extractortest.MockNode{TextValue: "Peel and slice apples."},
					&extractortest.MockNode{TextValue: "Fill pie crust and bake."},
				},
			},
		},
	}

	r := extractRecipeFromDOM(doc)

	if r.Name != "Grandma's Apple Pie" {
		t.Errorf("Name = %q, want %q", r.Name, "Grandma's Apple Pie")
	}
	if r.Description != "A classic apple pie recipe." {
		t.Errorf("Description = %q, want %q", r.Description, "A classic apple pie recipe.")
	}
	if len(r.Ingredients) != 3 {
		t.Fatalf("len(Ingredients) = %d, want 3", len(r.Ingredients))
	}
	if len(r.Instructions) != 2 {
		t.Fatalf("len(Instructions) = %d, want 2", len(r.Instructions))
	}
}

func TestExtractRecipe_MockBrowser(t *testing.T) {
	doc := makeRecipeDoc(sampleJSONLD)

	browser := &extractortest.MockBrowser{
		Documents: map[string]*extractortest.MockDocument{
			"https://www.allrecipes.com/recipe/10813/best-chocolate-chip-cookies/": doc,
		},
	}

	r, err := DefaultConfig.ExtractRecipe(
		context.Background(),
		browser,
		"https://www.allrecipes.com/recipe/10813/best-chocolate-chip-cookies/",
	)
	if err != nil {
		t.Fatalf("ExtractRecipe() error: %v", err)
	}

	if r.Name != "Chocolate Chip Cookies" {
		t.Errorf("Name = %q, want %q", r.Name, "Chocolate Chip Cookies")
	}
	if r.SourceURL != "https://www.allrecipes.com/recipe/10813/best-chocolate-chip-cookies/" {
		t.Errorf("SourceURL = %q, want recipe URL", r.SourceURL)
	}
}

func TestExtractRecipe_FallbackToDOM(t *testing.T) {
	doc := &extractortest.MockDocument{
		URLValue: "https://example.com/recipe",
		MockNode: extractortest.MockNode{
			Children: map[string]extractor.Nodes{
				"h1": {
					&extractortest.MockNode{TextValue: "Simple Recipe"},
				},
				"li.ingredient": {
					&extractortest.MockNode{TextValue: "1 cup flour"},
				},
			},
		},
	}

	browser := &extractortest.MockBrowser{
		Documents: map[string]*extractortest.MockDocument{
			"https://example.com/recipe": doc,
		},
	}

	r, err := DefaultConfig.ExtractRecipe(context.Background(), browser, "https://example.com/recipe")
	if err != nil {
		t.Fatalf("ExtractRecipe() error: %v", err)
	}

	if r.Name != "Simple Recipe" {
		t.Errorf("Name = %q, want %q", r.Name, "Simple Recipe")
	}
	if len(r.Ingredients) != 1 {
		t.Fatalf("len(Ingredients) = %d, want 1", len(r.Ingredients))
	}
}

func TestFormatDuration(t *testing.T) {
	tests := []struct {
		input string
		want  string
	}{
		{"PT15M", "15 min"},
		{"PT1H30M", "1 hr 30 min"},
		{"PT10M", "10 min"},
		{"PT2H", "2 hr"},
		{"PT45S", "45 sec"},
		{"PT1H15M30S", "1 hr 15 min 30 sec"},
		{"", ""},
	}

	for _, tt := range tests {
		got := formatDuration(tt.input)
		if got != tt.want {
			t.Errorf("formatDuration(%q) = %q, want %q", tt.input, got, tt.want)
		}
	}
}

func TestExtractRecipeFromDOM_Empty(t *testing.T) {
	doc := &extractortest.MockDocument{
		MockNode: extractortest.MockNode{
			Children: map[string]extractor.Nodes{},
		},
	}

	r := extractRecipeFromDOM(doc)
	if r.Name != "" {
		t.Errorf("expected empty name, got %q", r.Name)
	}
}