From 36707dec17fe454f46aa6bca2cf7b9cfe459a78e Mon Sep 17 00:00:00 2001
From: Steve Dudenhoeffer <steve@stevedudenhoeffer.com>
Date: Tue, 24 Dec 2024 12:15:48 -0500
Subject: [PATCH] added useragents to go-extractor

---
 sites/useragents/cmd/useragents.go | 58 ++++++++++++++++++++++++
 sites/useragents/useragents.go     | 72 ++++++++++++++++++++++++++++++
 2 files changed, 130 insertions(+)
 create mode 100644 sites/useragents/cmd/useragents.go
 create mode 100644 sites/useragents/useragents.go

diff --git a/sites/useragents/cmd/useragents.go b/sites/useragents/cmd/useragents.go
new file mode 100644
index 0000000..cd849f6
--- /dev/null
+++ b/sites/useragents/cmd/useragents.go
@@ -0,0 +1,58 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"os"
+
+	"github.com/urfave/cli/v3"
+
+	"gitea.stevedudenhoeffer.com/steve/go-extractor/cmd/browser/pkg/browser"
+	"gitea.stevedudenhoeffer.com/steve/go-extractor/sites/useragents"
+)
+
+type UseragentsFlags []cli.Flag
+
+var Flags = UseragentsFlags{}
+
+func (f UseragentsFlags) ToConfig(_ *cli.Command) useragents.Config {
+	c := useragents.DefaultConfig
+	return c
+}
+
+func main() {
+	var flags []cli.Flag
+
+	flags = append(flags, browser.Flags...)
+	flags = append(flags, Flags...)
+
+	cli := &cli.Command{
+		Name:  "useragents",
+		Usage: "Get useragents.me information",
+		Flags: flags,
+
+		Action: func(ctx context.Context, cli *cli.Command) error {
+			b, err := browser.FromCommand(ctx, cli)
+
+			if err != nil {
+				return err
+			}
+
+			mostCommon, err := Flags.ToConfig(cli).GetMostCommonDesktopUserAgent(ctx, b)
+
+			if err != nil {
+				return err
+			}
+
+			fmt.Printf("Most Common User Agent: %+v\n", mostCommon)
+			return nil
+		},
+	}
+
+	err := cli.Run(context.Background(), os.Args)
+
+	if err != nil {
+		panic(err)
+	}
+
+}
diff --git a/sites/useragents/useragents.go b/sites/useragents/useragents.go
new file mode 100644
index 0000000..8613f00
--- /dev/null
+++ b/sites/useragents/useragents.go
@@ -0,0 +1,72 @@
+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
+}