nice clean recipes
pear.dunkirk.sh
recipes
feat: add meta parsing
+35
+35
internal/extract/schema/jsonld.go
+35
internal/extract/schema/jsonld.go
···
22
22
recipe := parseJSONLD(node)
23
23
if recipe != nil {
24
24
recipe.ExtractionMethod = "schema.org"
25
+
if recipe.Description == "" {
26
+
recipe.Description = findMetaDescription(doc)
27
+
}
25
28
return recipe, true
26
29
}
27
30
}
28
31
29
32
return nil, false
33
+
}
34
+
35
+
func findMetaDescription(n *html.Node) string {
36
+
var f func(*html.Node) string
37
+
f = func(n *html.Node) string {
38
+
if n.Type == html.ElementNode && n.Data == "meta" {
39
+
name := ""
40
+
prop := ""
41
+
content := ""
42
+
for _, a := range n.Attr {
43
+
if a.Key == "name" {
44
+
name = a.Val
45
+
}
46
+
if a.Key == "property" {
47
+
prop = a.Val
48
+
}
49
+
if a.Key == "content" {
50
+
content = a.Val
51
+
}
52
+
}
53
+
if name == "description" || prop == "og:description" {
54
+
return content
55
+
}
56
+
}
57
+
for c := n.FirstChild; c != nil; c = c.NextSibling {
58
+
if desc := f(c); desc != "" {
59
+
return desc
60
+
}
61
+
}
62
+
return ""
63
+
}
64
+
return f(n)
30
65
}
31
66
32
67
func findJSONLDScripts(n *html.Node) []string {