nice clean recipes
pear.dunkirk.sh
recipes
feat: add better image parsing
+70
-2
+70
-2
internal/extract/schema/jsonld.go
+70
-2
internal/extract/schema/jsonld.go
···
4
4
"encoding/json"
5
5
"fmt"
6
6
"regexp"
7
+
"strconv"
7
8
"strings"
8
9
9
10
"tangled.org/dunkirk.sh/pare/internal/models"
···
25
26
if recipe.Description == "" {
26
27
recipe.Description = findMetaDescription(doc)
27
28
}
29
+
if ogImg := findMetaImage(doc); ogImg != "" {
30
+
recipe.ImageURL = ogImg
31
+
}
28
32
if recipe.ImageURL == "" || looksSmall(recipe.ImageURL) {
29
-
if ogImg := findMetaImage(doc); ogImg != "" {
30
-
recipe.ImageURL = ogImg
33
+
if largeImg := findLargestImage(doc); largeImg != "" {
34
+
recipe.ImageURL = largeImg
31
35
}
32
36
}
33
37
return recipe, true
···
258
262
}
259
263
f(n)
260
264
return result
265
+
}
266
+
267
+
type imgCandidate struct {
268
+
src string
269
+
width int
270
+
height int
271
+
}
272
+
273
+
func findLargestImage(n *html.Node) string {
274
+
var candidates []imgCandidate
275
+
var f func(*html.Node)
276
+
f = func(n *html.Node) {
277
+
if n.Type == html.ElementNode && n.Data == "img" {
278
+
var src, dataSrc string
279
+
var w, h int
280
+
for _, a := range n.Attr {
281
+
switch a.Key {
282
+
case "src":
283
+
src = a.Val
284
+
case "data-lazy-src", "data-src":
285
+
dataSrc = a.Val
286
+
case "width":
287
+
w, _ = strconv.Atoi(a.Val)
288
+
case "height":
289
+
h, _ = strconv.Atoi(a.Val)
290
+
}
291
+
}
292
+
u := dataSrc
293
+
if u == "" {
294
+
u = src
295
+
}
296
+
if u == "" || strings.HasPrefix(u, "data:") {
297
+
return
298
+
}
299
+
candidates = append(candidates, imgCandidate{src: u, width: w, height: h})
300
+
}
301
+
for c := n.FirstChild; c != nil; c = c.NextSibling {
302
+
f(c)
303
+
}
304
+
}
305
+
f(n)
306
+
if len(candidates) == 0 {
307
+
return ""
308
+
}
309
+
// Pick the largest by area (width * height), preferring ones with explicit dimensions
310
+
best := candidates[0]
311
+
bestArea := best.width * best.height
312
+
for _, c := range candidates[1:] {
313
+
area := c.width * c.height
314
+
if area > bestArea {
315
+
best = c
316
+
bestArea = area
317
+
}
318
+
}
319
+
// If no candidate had dimensions, pick the first non-small URL
320
+
if bestArea == 0 {
321
+
for _, c := range candidates {
322
+
if !looksSmall(c.src) {
323
+
return c.src
324
+
}
325
+
}
326
+
return candidates[0].src
327
+
}
328
+
return best.src
261
329
}
262
330
263
331
func extractIngredients(m map[string]interface{}) []models.Ingredient {