nice clean recipes
pear.dunkirk.sh
recipes
feat: add new recipe site
+239
-9
+7
-1
internal/extract/pipeline.go
+7
-1
internal/extract/pipeline.go
···
41
41
return &Result{Recipe: recipe}
42
42
}
43
43
44
+
if recipe, ok := schema.ExtractMicrodata(body); ok {
45
+
recipe.SourceURL = targetURL
46
+
recipe.SourceDomain = domainOf(targetURL)
47
+
return &Result{Recipe: recipe}
48
+
}
49
+
44
50
if recipe, ok := hrecipe.Extract(body); ok {
45
51
recipe.SourceURL = targetURL
46
52
recipe.SourceDomain = domainOf(targetURL)
47
53
return &Result{Recipe: recipe}
48
54
}
49
55
50
-
return &Result{Error: fmt.Errorf("no recipe found on page — tried JSON-LD and h-recipe extraction")}
56
+
return &Result{Error: fmt.Errorf("no recipe found on page — tried JSON-LD, microdata, and h-recipe extraction")}
51
57
}
52
58
53
59
func (p *Pipeline) fetch(url string) (string, error) {
+211
internal/extract/schema/microdata.go
+211
internal/extract/schema/microdata.go
···
1
+
package schema
2
+
3
+
import (
4
+
"fmt"
5
+
"strings"
6
+
7
+
"tangled.org/dunkirk.sh/pare/internal/models"
8
+
9
+
"golang.org/x/net/html"
10
+
)
11
+
12
+
func ExtractMicrodata(body string) (*models.Recipe, bool) {
13
+
doc, err := html.Parse(strings.NewReader(body))
14
+
if err != nil {
15
+
return nil, false
16
+
}
17
+
18
+
recipeNode := findMicrodataRecipe(doc)
19
+
if recipeNode == nil {
20
+
return nil, false
21
+
}
22
+
23
+
recipe := &models.Recipe{ExtractionMethod: "schema.org"}
24
+
25
+
if name := getMicrodataProp(recipeNode, "name"); name != "" {
26
+
recipe.Name = name
27
+
}
28
+
if desc := getMicrodataProp(recipeNode, "description"); desc != "" {
29
+
recipe.Description = desc
30
+
}
31
+
if img := getMicrodataImageProp(recipeNode); img != "" {
32
+
recipe.ImageURL = img
33
+
}
34
+
if y := getMicrodataProp(recipeNode, "recipeYield"); y != "" {
35
+
recipe.Yield = cleanYield(y)
36
+
}
37
+
if pt := getMicrodataProp(recipeNode, "prepTime"); pt != "" {
38
+
recipe.PrepTime = pt
39
+
}
40
+
if ct := getMicrodataProp(recipeNode, "cookTime"); ct != "" {
41
+
recipe.CookTime = ct
42
+
}
43
+
if tt := getMicrodataProp(recipeNode, "totalTime"); tt != "" {
44
+
recipe.TotalTime = tt
45
+
}
46
+
47
+
for _, ing := range getAllMicrodataProps(recipeNode, "ingredients") {
48
+
recipe.Ingredients = append(recipe.Ingredients, parseIngredient(ing))
49
+
}
50
+
// Also check "recipeIngredient" since some pages use that
51
+
for _, ing := range getAllMicrodataProps(recipeNode, "recipeIngredient") {
52
+
recipe.Ingredients = append(recipe.Ingredients, parseIngredient(ing))
53
+
}
54
+
55
+
for _, instr := range getAllMicrodataProps(recipeNode, "recipeInstructions") {
56
+
instr = strings.TrimSpace(instr)
57
+
if instr != "" {
58
+
recipe.Instructions = append(recipe.Instructions, models.Instruction{Text: instr})
59
+
}
60
+
}
61
+
62
+
if recipe.Yield != "" {
63
+
fmt.Sscanf(recipe.Yield, "%d", &recipe.Servings)
64
+
}
65
+
66
+
if recipe.Name == "" {
67
+
return nil, false
68
+
}
69
+
70
+
return recipe, true
71
+
}
72
+
73
+
func findMicrodataRecipe(n *html.Node) *html.Node {
74
+
if n.Type == html.ElementNode {
75
+
for _, attr := range n.Attr {
76
+
if attr.Key == "itemtype" {
77
+
typ := strings.TrimSpace(attr.Val)
78
+
if typ == "http://schema.org/Recipe" || typ == "https://schema.org/Recipe" {
79
+
return n
80
+
}
81
+
}
82
+
}
83
+
}
84
+
for c := n.FirstChild; c != nil; c = c.NextSibling {
85
+
if found := findMicrodataRecipe(c); found != nil {
86
+
return found
87
+
}
88
+
}
89
+
return nil
90
+
}
91
+
92
+
func getMicrodataProp(n *html.Node, prop string) string {
93
+
var result string
94
+
var f func(*html.Node)
95
+
f = func(node *html.Node) {
96
+
if result != "" {
97
+
return
98
+
}
99
+
if node.Type == html.ElementNode {
100
+
for _, attr := range node.Attr {
101
+
if attr.Key == "itemprop" && attr.Val == prop {
102
+
// For img elements, use src/content
103
+
if node.Data == "img" {
104
+
result = getAttrVal(node, "src")
105
+
if result == "" {
106
+
result = getAttrVal(node, "content")
107
+
}
108
+
} else if node.Data == "meta" {
109
+
result = getAttrVal(node, "content")
110
+
} else if node.Data == "time" {
111
+
result = getAttrVal(node, "datetime")
112
+
if result == "" {
113
+
result = textContent(node)
114
+
}
115
+
} else if node.Data == "link" {
116
+
result = getAttrVal(node, "href")
117
+
} else {
118
+
result = textContent(node)
119
+
}
120
+
return
121
+
}
122
+
}
123
+
}
124
+
for c := node.FirstChild; c != nil; c = c.NextSibling {
125
+
f(c)
126
+
}
127
+
}
128
+
f(n)
129
+
return strings.TrimSpace(result)
130
+
}
131
+
132
+
func getMicrodataImageProp(n *html.Node) string {
133
+
var f func(*html.Node) string
134
+
f = func(node *html.Node) string {
135
+
if node.Type == html.ElementNode {
136
+
for _, attr := range node.Attr {
137
+
if attr.Key == "itemprop" && (attr.Val == "image") {
138
+
if node.Data == "img" {
139
+
if src := getAttrVal(node, "src"); src != "" {
140
+
return src
141
+
}
142
+
}
143
+
if href := getAttrVal(node, "href"); href != "" {
144
+
return href
145
+
}
146
+
if content := getAttrVal(node, "content"); content != "" {
147
+
return content
148
+
}
149
+
}
150
+
}
151
+
}
152
+
for c := node.FirstChild; c != nil; c = c.NextSibling {
153
+
if found := f(c); found != "" {
154
+
return found
155
+
}
156
+
}
157
+
return ""
158
+
}
159
+
return f(n)
160
+
}
161
+
162
+
func getAllMicrodataProps(n *html.Node, prop string) []string {
163
+
var results []string
164
+
var f func(*html.Node)
165
+
f = func(node *html.Node) {
166
+
if node.Type == html.ElementNode {
167
+
for _, attr := range node.Attr {
168
+
if attr.Key == "itemprop" && attr.Val == prop {
169
+
text := ""
170
+
if node.Data == "img" {
171
+
text = getAttrVal(node, "src")
172
+
} else if node.Data == "meta" {
173
+
text = getAttrVal(node, "content")
174
+
} else {
175
+
text = textContent(node)
176
+
}
177
+
text = strings.TrimSpace(text)
178
+
if text != "" {
179
+
results = append(results, text)
180
+
}
181
+
return
182
+
}
183
+
}
184
+
}
185
+
for c := node.FirstChild; c != nil; c = c.NextSibling {
186
+
f(c)
187
+
}
188
+
}
189
+
f(n)
190
+
return results
191
+
}
192
+
193
+
func getAttrVal(n *html.Node, key string) string {
194
+
for _, attr := range n.Attr {
195
+
if attr.Key == key {
196
+
return attr.Val
197
+
}
198
+
}
199
+
return ""
200
+
}
201
+
202
+
func textContent(n *html.Node) string {
203
+
if n.Type == html.TextNode {
204
+
return n.Data
205
+
}
206
+
var sb strings.Builder
207
+
for c := n.FirstChild; c != nil; c = c.NextSibling {
208
+
sb.WriteString(textContent(c))
209
+
}
210
+
return strings.TrimSpace(sb.String())
211
+
}
+3
main.go
+3
main.go
+16
-6
ui/static/style.css
+16
-6
ui/static/style.css
···
128
128
font-size:0.8rem;
129
129
margin-top:0.35rem;
130
130
font-family:'Poppins',system-ui,sans-serif;
131
+
overflow:hidden;
132
+
text-overflow:ellipsis;
133
+
white-space:nowrap;
131
134
}
132
135
.recipe-source a svg{
133
136
margin-right:0.25rem;
···
304
307
.btn-secondary:hover{border-color:var(--accent);text-decoration:none}
305
308
306
309
.error-box{
307
-
background:var(--error-bg);
308
-
color:var(--error-text);
309
-
padding:1.5rem;
310
-
border-radius:var(--radius);
311
310
text-align:center;
311
+
padding:3rem 1.5rem;
312
312
}
313
-
.error-box h3{margin-bottom:0.5rem}
314
-
.error-box p{font-size:0.9rem;margin-bottom:1rem}
313
+
.error-box h3{
314
+
font-family:'Poppins',system-ui,sans-serif;
315
+
font-size:1.1rem;
316
+
font-weight:600;
317
+
color:var(--text);
318
+
margin-bottom:0.5rem;
319
+
}
320
+
.error-box p{
321
+
font-size:0.9rem;
322
+
color:var(--text-muted);
323
+
margin-bottom:1.25rem;
324
+
}
315
325
316
326
.bookmarklet{
317
327
margin-top:2rem;
+1
-1
ui/templates/index.html
+1
-1
ui/templates/index.html
···
4
4
<head>
5
5
<meta charset="utf-8">
6
6
<meta name="viewport" content="width=device-width,initial-scale=1">
7
-
<title>pare</title>
7
+
<title>pare: nice recipes</title>
8
8
<link rel="icon" type="image/svg+xml" href="/static/favicon.svg">
9
9
<link rel="stylesheet" href="/static/style.css">
10
10
</head>
+1
-1
ui/templates/recipe.html
+1
-1
ui/templates/recipe.html
···
4
4
<head>
5
5
<meta charset="utf-8">
6
6
<meta name="viewport" content="width=device-width,initial-scale=1">
7
-
<title>pare - {{.Recipe.Name}}</title>
7
+
<title>{{.Recipe.Name}}</title>
8
8
<link rel="icon" type="image/svg+xml" href="/static/favicon.svg">
9
9
<link rel="stylesheet" href="/static/style.css">
10
10
</head>