expose a cmd for quick use / test

cmd/htmlsanitizer/main.go

@@ -0,0 +1,49 @@
+package main
+
+import (
+	"flag"
+	"io"
+	"log"
+	"net/http"
+	"os"
+	"strings"
+
+	"github.com/sym01/htmlsanitizer"
+)
+
+var (
+	srcFilePath = flag.String("src", "", "could be either source file path, or the source URL")
+)
+
+func main() {
+	flag.Parse()
+
+	if len(*srcFilePath) == 0 {
+		flag.CommandLine.Usage()
+		return
+	}
+
+	var src io.ReadCloser
+	switch {
+	case strings.HasPrefix(*srcFilePath, "http://"), strings.HasPrefix(*srcFilePath, "https://"):
+		resp, err := http.Get(*srcFilePath)
+		if err != nil {
+			log.Fatalf("unable to fetch remote content: %s", err)
+		}
+		src = resp.Body
+	default:
+		file, err := os.OpenFile(*srcFilePath, os.O_RDONLY, 0755)
+		if err != nil {
+			log.Fatalf("unable to open src file: %s", err)
+		}
+		src = file
+	}
+
+	defer src.Close()
+
+	san := htmlsanitizer.NewHTMLSanitizer()
+	writer := san.NewWriter(os.Stdout)
+	if _, err := io.Copy(writer, src); err != nil {
+		log.Printf("unable to sanitize HTML content: %s", err)
+	}
+}

sanitizer_test.go

@@ -498,6 +498,10 @@ var testCases = []struct {
 		in:  `<A HREF="javascript:document.location='http://www.google.com/'">XSS</A>`,
 		out: `<a>XSS</a>`,
 	},
+	{
+		in:  `<span>func <a class= "Documentation-source" href="https://cs.opensource.google/go/go/+/go1.21.5:src/os/path.go;l=66">RemoveAll</a> <a class="Documentation-idLink" href="#RemoveAll" aria-label="Go to RemoveAll">¶</a></span>`,
+		out: `<span>func <a class="Documentation-source" href="https://cs.opensource.google/go/go/+/go1.21.5:src/os/path.go;l=66">RemoveAll</a> <a class="Documentation-idLink" href="#RemoveAll">¶</a></span>`,
+	},
 	{
 		in: `
 <Img src = x onerror = "javascript: window.onerror = alert; throw XSS">