nice clean recipes
pear.dunkirk.sh
recipes
feat: grab largest image
+72
-10
+72
-10
internal/extract/schema/jsonld.go
+72
-10
internal/extract/schema/jsonld.go
···
25
25
if recipe.Description == "" {
26
26
recipe.Description = findMetaDescription(doc)
27
27
}
28
+
if recipe.ImageURL == "" || looksSmall(recipe.ImageURL) {
29
+
if ogImg := findMetaImage(doc); ogImg != "" {
30
+
recipe.ImageURL = ogImg
31
+
}
32
+
}
28
33
return recipe, true
29
34
}
30
35
}
···
179
184
}
180
185
181
186
func extractImage(m map[string]interface{}) string {
182
-
img := m["image"]
187
+
var urls []string
188
+
collectImageURLs(m["image"], &urls)
189
+
if best := pickBestImage(urls); best != "" {
190
+
return best
191
+
}
192
+
return ""
193
+
}
194
+
195
+
func collectImageURLs(img interface{}, urls *[]string) {
183
196
switch v := img.(type) {
184
197
case string:
185
-
return v
198
+
*urls = append(*urls, v)
186
199
case map[string]interface{}:
187
-
return strVal(v, "url")
200
+
if u := strVal(v, "url"); u != "" {
201
+
*urls = append(*urls, u)
202
+
}
203
+
if u := strVal(v, "contentUrl"); u != "" {
204
+
*urls = append(*urls, u)
205
+
}
188
206
case []interface{}:
189
-
if len(v) > 0 {
190
-
switch first := v[0].(type) {
191
-
case string:
192
-
return first
193
-
case map[string]interface{}:
194
-
return strVal(first, "url")
207
+
for _, item := range v {
208
+
collectImageURLs(item, urls)
209
+
}
210
+
}
211
+
}
212
+
213
+
func pickBestImage(urls []string) string {
214
+
if len(urls) == 0 {
215
+
return ""
216
+
}
217
+
// Prefer URLs that don't look like thumbnails
218
+
for _, u := range urls {
219
+
if !looksSmall(u) {
220
+
return u
221
+
}
222
+
}
223
+
return urls[0]
224
+
}
225
+
226
+
var smallImageRe = regexp.MustCompile(`[-_]sm\b|[-_]thumb(?:nail)?\b|[-_]small\b|[-_]\d{2,3}x\d{2,3}\b|[-_]\d{2,3}w\b`)
227
+
228
+
func looksSmall(u string) bool {
229
+
return smallImageRe.MatchString(u)
230
+
}
231
+
232
+
func findMetaImage(n *html.Node) string {
233
+
var result string
234
+
var f func(*html.Node)
235
+
f = func(n *html.Node) {
236
+
if n.Type == html.ElementNode && n.Data == "meta" {
237
+
prop := ""
238
+
content := ""
239
+
for _, a := range n.Attr {
240
+
if a.Key == "property" {
241
+
prop = a.Val
242
+
}
243
+
if a.Key == "content" {
244
+
content = a.Val
245
+
}
246
+
}
247
+
if prop == "og:image" && content != "" {
248
+
result = content
249
+
return
250
+
}
251
+
}
252
+
for c := n.FirstChild; c != nil; c = c.NextSibling {
253
+
f(c)
254
+
if result != "" {
255
+
return
195
256
}
196
257
}
197
258
}
198
-
return ""
259
+
f(n)
260
+
return result
199
261
}
200
262
201
263
func extractIngredients(m map[string]interface{}) []models.Ingredient {