Refresh scraper (#460)

Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>

Author: reaper47

internal/models/schema-recipe.go

@@ -986,17 +986,67 @@ func (n *NutritionSchema) UnmarshalJSON(data []byte) error {
 
 	switch x := v.(type) {
 	case map[string]any:
-		n.Calories = regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x["calories"]), ",", "."))
-		n.Carbohydrates = regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x["carbohydrateContent"]), ",", "."))
-		n.Cholesterol = regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x["cholesterolContent"]), ",", "."))
-		n.Fat = regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x["fatContent"]), ",", "."))
-		n.SaturatedFat = regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x["saturatedFatContent"]), ",", "."))
-		n.UnsaturatedFat = regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x["unsaturatedFatContent"]), ",", "."))
-		n.TransFat = regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x["transFatContent"]), ",", "."))
-		n.Protein = regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x["proteinContent"]), ",", "."))
-		n.Sugar = regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x["sugarContent"]), ",", "."))
-		n.Sodium = regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x["sodiumContent"]), ",", "."))
-		n.Fiber = regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x["fiberContent"]), ",", "."))
+		extract := func(nutrition string) string {
+			return regex.Digit.FindString(strings.ReplaceAll(extensions.ConvertToString(x[nutrition]), ",", "."))
+		}
+
+		if _, ok := x["calories"]; ok {
+			n.Calories = extract("calories")
+		}
+
+		if _, ok := x["carbohydrateContent"]; ok {
+			n.Carbohydrates = extract("carbohydrateContent")
+		} else if _, ok = x["carbs"]; ok {
+			n.Carbohydrates = extract("carbs")
+		}
+
+		if _, ok := x["cholesterolContent"]; ok {
+			n.Cholesterol = extract("cholesterolContent")
+		}
+
+		if _, ok := x["fatContent"]; ok {
+			n.Fat = extract("fatContent")
+		} else if _, ok = x["fat"]; ok {
+			n.Fat = extract("fat")
+		}
+
+		if _, ok := x["saturatedFatContent"]; ok {
+			n.SaturatedFat = extract("saturatedFatContent")
+		} else if _, ok = x["saturatedFat"]; ok {
+			n.SaturatedFat = extract("saturatedFat")
+		}
+
+		if _, ok := x["unsaturatedFatContent"]; ok {
+			n.UnsaturatedFat = extract("unsaturatedFatContent")
+		}
+
+		if _, ok := x["transFatContent"]; ok {
+			n.TransFat = extract("transFatContent")
+		}
+
+		if _, ok := x["proteinContent"]; ok {
+			n.Protein = extract("proteinContent")
+		} else if _, ok = x["protein"]; ok {
+			n.Protein = extract("protein")
+		}
+
+		if _, ok := x["sugarContent"]; ok {
+			n.Sugar = extract("sugarContent")
+		} else if _, ok = x["sugars"]; ok {
+			n.Sugar = extract("sugars")
+		}
+
+		if _, ok := x["sodiumContent"]; ok {
+			n.Sodium = extract("sodiumContent")
+		} else if _, ok = x["salt"]; ok {
+			n.Sodium = extract("salt")
+		}
+
+		if _, ok := x["fiberContent"]; ok {
+			n.Fiber = extract("fiberContent")
+		} else if _, ok = x["fibre"]; ok {
+			n.Fiber = extract("fibre")
+		}
 
 		if val := extensions.ConvertToString(x["servingSize"]); val != "" {
 			xs := strings.Split(val, " ")

internal/scraper/alittlebityummy.go

@@ -30,5 +30,9 @@ func scrapeALittleBitYummy(root *goquery.Document) (models.RecipeSchema, error)
 		}
 	})
 
+	for i, s := range rs.Instructions.Values {
+		rs.Instructions.Values[i].Text = strings.Join(strings.Fields(s.Text), " ")
+	}
+
 	return rs, nil
 }

internal/scraper/epicurious.go

@@ -0,0 +1,48 @@
+package scraper
+
+import (
+	"github.com/PuerkitoBio/goquery"
+	"github.com/reaper47/recipya/internal/models"
+	"strings"
+)
+
+func scrapeEpicurious(root *goquery.Document) (models.RecipeSchema, error) {
+	rs, err := parseWebsite(root)
+	if err != nil {
+		return models.RecipeSchema{}, err
+	}
+
+	parseTime := func(s string) string {
+		t := "PT"
+
+		parts := strings.Split(s, " ")
+		n := len(parts)
+
+		if n == 2 {
+			t += parts[0]
+			if strings.HasPrefix(strings.ToLower(parts[1]), "min") {
+				t += "M"
+			} else {
+				t += "H"
+			}
+		} else if n >= 4 {
+			t += parts[0] + "H" + parts[2] + "M"
+		}
+
+		return t
+	}
+
+	if rs.CookTime != "" && !strings.HasPrefix(rs.CookTime, "PT") {
+		rs.CookTime = parseTime(rs.CookTime)
+	}
+
+	if rs.PrepTime != "" && !strings.HasPrefix(rs.PrepTime, "PT") {
+		rs.PrepTime = parseTime(rs.PrepTime)
+	}
+
+	if rs.TotalTime != "" && !strings.HasPrefix(rs.TotalTime, "PT") {
+		rs.TotalTime = parseTime(rs.TotalTime)
+	}
+
+	return rs, nil
+}

internal/scraper/finedininglovers.go

@@ -0,0 +1,32 @@
+package scraper
+
+import (
+	"github.com/PuerkitoBio/goquery"
+	"github.com/reaper47/recipya/internal/models"
+	"strings"
+)
+
+func scrapeFineDiningLovers(root *goquery.Document) (models.RecipeSchema, error) {
+	rs, err := parseWebsite(root)
+	if err != nil {
+		return models.RecipeSchema{}, err
+	}
+
+	nodes := root.Find(".field--name-field-srh-ingredients p")
+	rs.Ingredients.Values = make([]string, 0, nodes.Length())
+	nodes.Each(func(_ int, sel *goquery.Selection) {
+		rs.Ingredients.Values = append(rs.Ingredients.Values, strings.TrimSpace(sel.Text()))
+	})
+
+	nodes = root.Find(".paragraph--type-srh-step")
+	rs.Instructions.Values = make([]models.HowToItem, 0, nodes.Length())
+	nodes.Each(func(_ int, sel *goquery.Selection) {
+		s := strings.TrimSpace(sel.Text())
+		s = strings.ReplaceAll(s, "  ", "")
+		s = strings.ReplaceAll(s, "\u00a0", "\n")
+		s = strings.ReplaceAll(s, "\n\n\n", "\n\n")
+		rs.Instructions.Values = append(rs.Instructions.Values, models.NewHowToStep(s))
+	})
+
+	return rs, nil
+}

internal/scraper/foodbag.go

@@ -13,7 +13,7 @@ func scrapeGazoakleychef(root *goquery.Document) (models.RecipeSchema, error) {
 
 	rs.DateModified = getPropertyContent(root, "article:modified_time")
 	rs.Image.Value = getPropertyContent(root, "og:image")
-	rs.Name = root.Find(".entry-title").Text()
+	rs.Name = root.Find(".entry-header .entry-title").First().Text()
 
 	root.Find(".entry-quick-info div.row div").Each(func(_ int, sel *goquery.Selection) {
 		c := strings.TrimSpace(sel.Text())

internal/scraper/gazoakleychef.go

@@ -7,6 +7,7 @@ import (
 	"github.com/reaper47/recipya/internal/models"
 	"io"
 	"net/http"
+	"slices"
 	"strconv"
 	"strings"
 )
@@ -159,7 +160,7 @@ func (s *Scraper) scrapeGousto(rawURL string) (models.RecipeSchema, error) {
 
 	entry := g.Data.Entry
 	rs := models.NewRecipeSchema()
-	rs.Name = entry.Title
+	rs.Name = strings.TrimSpace(entry.Title)
 
 	if len(entry.Categories) > 0 {
 		cat := strings.ToLower(entry.Categories[len(entry.Categories)-1].Title)
@@ -174,7 +175,7 @@ func (s *Scraper) scrapeGousto(rawURL string) (models.RecipeSchema, error) {
 		rs.Image.Value = entry.Media.Images[len(entry.Media.Images)-1].Image
 	}
 
-	rs.Description.Value = entry.Description
+	rs.Description.Value = strings.TrimSpace(entry.Description)
 
 	if entry.PrepTimes.For2 > 0 {
 		rs.PrepTime = "PT" + strconv.Itoa(entry.PrepTimes.For2) + "M"
@@ -189,6 +190,9 @@ func (s *Scraper) scrapeGousto(rawURL string) (models.RecipeSchema, error) {
 	for _, basic := range entry.Basics {
 		rs.Ingredients.Values = append(rs.Ingredients.Values, basic.Title)
 	}
+	rs.Ingredients.Values = slices.DeleteFunc(rs.Ingredients.Values, func(s string) bool {
+		return s == ""
+	})
 
 	rs.Instructions.Values = make([]models.HowToItem, 0, len(entry.CookingInstructions))
 	for _, ins := range entry.CookingInstructions {

internal/scraper/gousto.go

@@ -53,7 +53,7 @@ func scrapeInstantPot(root *goquery.Document) (models.RecipeSchema, error) {
 		rs.PrepTime = prep
 	}
 
-	getIngredients(&rs, root.Find(".article__ingredients").First().Find("li"))
+	getIngredients(&rs, root.Find(".article__ingredients").Last().Find("li"))
 	getInstructions(&rs, root.Find(".article__instructions").First().Find("li"))
 
 	return rs, nil