go-extractor/sites/useragents/useragents.go

73 lines
1.3 KiB
Go
Raw Normal View History

2024-12-24 12:15:48 -05:00
package useragents
import (
"context"
"encoding/json"
"fmt"
"io"
"gitea.stevedudenhoeffer.com/steve/go-extractor"
)
type Config struct{}
var DefaultConfig = Config{}
func deferClose(cl io.Closer) {
if cl != nil {
_ = cl.Close()
}
}
func GetMostCommonDesktopUserAgent(ctx context.Context, b extractor.Browser) (string, error) {
return DefaultConfig.GetMostCommonDesktopUserAgent(ctx, b)
}
func (c Config) GetMostCommonDesktopUserAgent(ctx context.Context, b extractor.Browser) (string, error) {
doc, err := b.Open(ctx, "https://www.useragents.me/", extractor.OpenPageOptions{})
if err != nil {
return "", fmt.Errorf("failed to open useragents.me: %w", err)
}
defer deferClose(doc)
s := doc.Select("#most-common-desktop-useragents-json-csv > div:nth-child(1) > textarea:nth-child(4)")
text := ""
for _, el := range s {
t, err := el.Content()
if err != nil {
return "", fmt.Errorf("failed to get text: %w", err)
}
text += t
}
data := []map[string]any{}
fmt.Println("text", text)
err = json.Unmarshal([]byte(text), &data)
if err != nil {
return "", err
}
highestAgent := ""
highestPct := 0.0
for _, agent := range data {
pct, ok := agent["pct"].(float64)
if !ok {
continue
}
if pct > highestPct {
highestPct = pct
highestAgent = agent["ua"].(string)
}
}
return highestAgent, nil
}