Support 50 more websites (#389)

internal/models/schema-recipe.go

@@ -637,6 +637,7 @@ func (i *Instructions) UnmarshalJSON(data []byte) error {
 	case string:
 		parts := strings.Split(strings.TrimSpace(x), "\n")
 		for _, s := range parts {
+			s = strings.TrimSpace(s)
 			if s != "" {
 				i.Values = append(i.Values, NewHowToStep(strings.TrimSpace(s)))
 			}

internal/scraper/aldi.go

@@ -0,0 +1,37 @@
+package scraper
+
+import (
+	"github.com/PuerkitoBio/goquery"
+	"github.com/reaper47/recipya/internal/models"
+	"github.com/reaper47/recipya/internal/utils/regex"
+	"strings"
+)
+
+func scrapeAldi(root *goquery.Document) (models.RecipeSchema, error) {
+	rs := models.NewRecipeSchema()
+	rs.Category.Value = strings.Replace(root.Find(".tab-nav--item.dropdown--list--item.m-active").Text(), "Recipes", "", 1)
+	rs.Category.Value = strings.TrimSpace(strings.Replace(rs.Category.Value, "selected", "", 1))
+
+	content := root.Find("#main-content")
+	sidebar := root.Find("#sidebar")
+
+	rs.Description.Value = getNameContent(root, "description")
+	rs.Image.Value = content.Find("img").First().AttrOr("src", "")
+	rs.Name = strings.TrimSpace(content.Find("h1").Text())
+	rs.Yield.Value = findYield(sidebar.Find("p:contains('Serves')").Text())
+	getIngredients(&rs, sidebar.Find("ul li"))
+	getInstructions(&rs, content.Find("ol li"))
+	getTime(&rs, sidebar.Find("p:contains('Prep Time')"), true)
+
+	cook := strings.TrimSpace(sidebar.Find("p:contains('Cook Time')").Text())
+	if cook != "" {
+		rs.CookTime = "PT" + regex.Digit.FindString(cook)
+		if strings.Contains(cook, "min") {
+			rs.CookTime += "M"
+		} else {
+			rs.CookTime += "H"
+		}
+	}
+
+	return rs, nil
+}

internal/scraper/alittlebityummy.go

@@ -0,0 +1,34 @@
+package scraper
+
+import (
+	"github.com/PuerkitoBio/goquery"
+	"github.com/reaper47/recipya/internal/models"
+	"strings"
+)
+
+func scrapeALittleBitYummy(root *goquery.Document) (models.RecipeSchema, error) {
+	rs, err := parseWebsite(root)
+	if err != nil {
+		return models.RecipeSchema{}, err
+	}
+
+	getInstructions(&rs, root.Find("#method ol li"))
+	rs.TotalTime = ""
+
+	if rs.Tools == nil {
+		rs.Tools = &models.Tools{}
+	}
+	root.Find("ul.recipe-equipment-list").First().Find("ul li").Each(func(_ int, sel *goquery.Selection) {
+		rs.Tools.Values = append(rs.Tools.Values, models.NewHowToTool(sel.Text()))
+	})
+
+	root.Find("div.ingredients-tab-content").First().Children().Each(func(_ int, sel *goquery.Selection) {
+		name := goquery.NodeName(sel)
+		if name == "h4" || name == "div" {
+			s := strings.Join(strings.Fields(sel.Text()), " ")
+			rs.Ingredients.Values = append(rs.Ingredients.Values, s)
+		}
+	})
+
+	return rs, nil
+}

internal/scraper/aniagotuje.go

@@ -0,0 +1,64 @@
+package scraper
+
+import (
+	"github.com/PuerkitoBio/goquery"
+	"github.com/reaper47/recipya/internal/models"
+	"github.com/reaper47/recipya/internal/utils/regex"
+	"strings"
+)
+
+func scrapeAniagotuje(root *goquery.Document) (models.RecipeSchema, error) {
+	rs := models.NewRecipeSchema()
+	rs.Description.Value = getNameContent(root, "description")
+	rs.Keywords.Values = getNameContent(root, "keywords")
+	rs.Name = getPropertyContent(root, "og:title")
+	rs.Image.Value = getPropertyContent(root, "og:image")
+
+	info := root.Find(".recipe-info")
+
+	n := info.Find("strong:contains('przygo')")
+	if len(n.Nodes) > 0 && n.Nodes[0].NextSibling != nil {
+		prep := strings.TrimSpace(n.Nodes[0].NextSibling.Data)
+		if prep != "" {
+			rs.PrepTime = "PT" + regex.Digit.FindString(prep)
+			if strings.Contains(prep, "min") {
+				rs.PrepTime += "M"
+			} else {
+				rs.PrepTime += "H"
+			}
+		}
+	}
+
+	n = info.Find("strong:contains('gotow')")
+	if len(n.Nodes) > 0 && n.Nodes[len(n.Nodes)-1].NextSibling != nil {
+		cook := strings.TrimSpace(n.Nodes[len(n.Nodes)-1].NextSibling.Data)
+		if cook != "" {
+			rs.CookTime = "PT" + regex.Digit.FindString(cook)
+			if strings.Contains(cook, "min") {
+				rs.CookTime += "M"
+			} else {
+				rs.CookTime += "H"
+			}
+		}
+	}
+
+	root.Find(".recipe-ing-list").Each(func(_ int, ul *goquery.Selection) {
+		header := strings.TrimSpace(ul.Prev().Text())
+		if header != "" {
+			rs.Ingredients.Values = append(rs.Ingredients.Values, header)
+		}
+
+		ul.Children().Each(func(_ int, li *goquery.Selection) {
+			rs.Ingredients.Values = append(rs.Ingredients.Values, strings.TrimSpace(li.Text()))
+		})
+	})
+
+	root.Find("h3").First().NextUntil("h3").Each(func(_ int, sel *goquery.Selection) {
+		s := strings.TrimSpace(sel.Text())
+		if s != "" {
+			rs.Instructions.Values = append(rs.Instructions.Values, models.NewHowToStep(s))
+		}
+	})
+
+	return rs, nil
+}

internal/scraper/antilliaans-eten.go

@@ -0,0 +1,34 @@
+package scraper
+
+import (
+	"github.com/PuerkitoBio/goquery"
+	"github.com/reaper47/recipya/internal/models"
+	"strings"
+)
+
+func scrapeAntilliaansEten(root *goquery.Document) (models.RecipeSchema, error) {
+	rs := models.NewRecipeSchema()
+	rs.Description.Value = getPropertyContent(root, "og:description")
+	rs.DateModified = getPropertyContent(root, "article:modified_time")
+	rs.DatePublished = getPropertyContent(root, "article:published_time")
+	rs.Image.Value = root.Find(".post-image img").AttrOr("src", "")
+	rs.Name = strings.TrimSpace(root.Find("h1.post-title").Text())
+	rs.PrepTime = root.Find(`time[itemprop="prepTime"]`).AttrOr("datetime", "")
+	rs.PrepTime = strings.Replace(rs.PrepTime, "0H", "", 1)
+
+	rs.CookTime = root.Find(`time[itemprop="cookTime"]`).AttrOr("datetime", "")
+	rs.CookTime = strings.Replace(rs.CookTime, "0H", "", 1)
+
+	ul := root.Find("h2:contains('Ingrediënten')").First().Next()
+	getIngredients(&rs, ul.Children())
+	getInstructions(&rs, ul.NextUntil("hr"))
+
+	nodes := root.Find(`meta[property="article:tag"]`)
+	xk := make([]string, 0, len(nodes.Nodes))
+	nodes.Each(func(_ int, sel *goquery.Selection) {
+		xk = append(xk, sel.AttrOr("content", ""))
+	})
+	rs.Keywords.Values = strings.Join(xk, ",")
+
+	return rs, nil
+}

internal/scraper/brianlagerstrom.go

@@ -9,7 +9,6 @@ import (
 
 func scrapeBrianLagerstrom(root *goquery.Document) (models.RecipeSchema, error) {
 	rs := models.NewRecipeSchema()
-
 	rs.Description.Value = getItempropContent(root, "description")
 	rs.Image.Value = getItempropContent(root, "thumbnailUrl")
 
@@ -79,6 +78,28 @@ func scrapeBrianLagerstrom(root *goquery.Document) (models.RecipeSchema, error)
 		}...)
 	}
 
+	if len(rs.Ingredients.Values) == 0 {
+		root.Find("p:contains('▪')").Each(func(_ int, sel *goquery.Selection) {
+			s := strings.TrimPrefix(sel.Text(), "▪")
+			rs.Ingredients.Values = append(rs.Ingredients.Values, strings.TrimSpace(s))
+		})
+	}
+
+	if len(rs.Instructions.Values) == 0 {
+		root.Find("p").Each(func(_ int, sel *goquery.Selection) {
+			s := strings.TrimSpace(sel.Text())
+			if s == "" {
+				return
+			}
+
+			dotIndex := strings.Index(s, ".")
+			if dotIndex != -1 && dotIndex < 3 {
+				_, after, _ := strings.Cut(s, ".")
+				rs.Instructions.Values = append(rs.Instructions.Values, models.NewHowToStep(after))
+			}
+		})
+	}
+
 	node := root.Find("p:contains('portion')").First()
 	if node != nil {
 		rs.Yield.Value = findYield(node.Text())

internal/scraper/britishbakels.go

@@ -0,0 +1,95 @@
+package scraper
+
+import (
+	"github.com/PuerkitoBio/goquery"
+	"github.com/reaper47/recipya/internal/models"
+	"strings"
+)
+
+func scrapeBritishBakels(root *goquery.Document) (models.RecipeSchema, error) {
+	rs := models.NewRecipeSchema()
+	rs.Description.Value = getNameContent(root, "description")
+	rs.DateModified = getPropertyContent(root, "article:modified_time")
+	rs.Image.Value = getPropertyContent(root, "og:image")
+	rs.Name = strings.TrimSpace(root.Find(".content-header ").Text())
+	rs.Category.Value = root.Find("h4:contains('Category')").Next().Text()
+
+	var (
+		isHeader   bool
+		unit       string
+		ingredient string
+		numRight   int
+	)
+
+	root.Find("div.card:contains('Ingredients')").Find("[id^=tab-ingredients] div").Each(func(_ int, sel *goquery.Selection) {
+		class := sel.AttrOr("class", "")
+		if !strings.Contains(class, "text-xs-") {
+			return
+		}
+
+		s := strings.TrimSpace(sel.Text())
+		if strings.HasPrefix(s, "Group") || s == "" {
+			rs.Ingredients.Values = append(rs.Ingredients.Values, s)
+			return
+		}
+
+		if strings.HasPrefix(s, "Ingredient") {
+			isHeader = true
+			return
+		}
+
+		isLeft := strings.Contains(class, "text-xs-left")
+		if isLeft {
+			numRight = 0
+		} else {
+			numRight += 1
+		}
+
+		if numRight == 2 {
+			return
+		}
+
+		if isHeader {
+			if unit == "" && !isLeft {
+				unit = s
+				return
+			} else if isLeft {
+				isHeader = false
+			} else {
+				return
+			}
+		}
+
+		if isLeft {
+			ingredient = s
+		}
+
+		if numRight == 1 {
+			rs.Ingredients.Values = append(rs.Ingredients.Values, s+" "+unit+" "+ingredient)
+		}
+	})
+
+	content := root.Find("div.card:contains('Method')")
+	nodes := content.Find("ol")
+	if nodes.Length() == 0 {
+		text := strings.TrimSpace(content.Find(".card-body").Text())
+		for _, s := range strings.Split(text, "\n") {
+			rs.Instructions.Values = append(rs.Instructions.Values, models.NewHowToStep(s))
+		}
+	} else {
+		nodes.Each(func(_ int, sel *goquery.Selection) {
+			name := strings.TrimSpace(sel.Prev().Text())
+			var h models.HowToItem
+			if name != "" {
+				h.Name = name
+			}
+
+			sel.Children().Each(func(_ int, li *goquery.Selection) {
+				h.Text = strings.TrimSpace(li.Text())
+				rs.Instructions.Values = append(rs.Instructions.Values, h)
+			})
+		})
+	}
+
+	return rs, nil
+}

internal/scraper/chetnamakan.go

@@ -0,0 +1,38 @@
+package scraper
+
+import (
+	"github.com/PuerkitoBio/goquery"
+	"github.com/reaper47/recipya/internal/models"
+	"strings"
+)
+
+func scrapeChetnamakan(root *goquery.Document) (models.RecipeSchema, error) {
+	rs := models.NewRecipeSchema()
+	rs.Description.Value = getNameContent(root, "description")
+	rs.Image.Value = getPropertyContent(root, "og:image")
+	rs.DatePublished = getPropertyContent(root, "article:published_time")
+	rs.DateModified = getPropertyContent(root, "article:modified_time")
+	rs.Name = root.Find(`h1[itemprop=headline]`).Text()
+
+	node := root.Find("strong:contains('Ingredients')")
+	if node != nil {
+		for c := node.Nodes[0].NextSibling; c != nil; c = c.NextSibling {
+			s := strings.TrimSpace(c.Data)
+			if s != "br" {
+				rs.Ingredients.Values = append(rs.Ingredients.Values, s)
+			}
+		}
+	}
+
+	node = root.Find("strong:contains('Method')")
+	if node != nil {
+		for c := node.Nodes[0].NextSibling; c != nil; c = c.NextSibling {
+			s := strings.TrimSpace(c.Data)
+			if s != "br" {
+				rs.Instructions.Values = append(rs.Instructions.Values, models.NewHowToStep(strings.TrimPrefix(s, "–")))
+			}
+		}
+	}
+
+	return rs, nil
+}

internal/scraper/chinesecookingdemystified.go

@@ -1,6 +1,7 @@
 package scraper
 
 import (
+	"encoding/json"
 	"github.com/PuerkitoBio/goquery"
 	"github.com/reaper47/recipya/internal/models"
 	"github.com/reaper47/recipya/internal/utils/regex"
@@ -31,5 +32,21 @@ func scrapeChinesecookingdemystified(root *goquery.Document) (models.RecipeSchem
 
 	getInstructions(&rs, root.Find("h3:contains('Process')").NextUntil(":not(p)"))
 
+	video := root.Find(".youtube-wrap").AttrOr("data-attrs", "")
+	if video != "" {
+		var m map[string]string
+		err = json.Unmarshal([]byte(video), &m)
+		if err == nil {
+			v, ok := m["videoId"]
+			if ok {
+				rs.Video.Values = append(rs.Video.Values, models.VideoObject{
+					AtType:   "VideoObject",
+					EmbedURL: "https://youtube.com/embed/" + v,
+					IsIFrame: true,
+				})
+			}
+		}
+	}
+
 	return rs, nil
 }