Add strip_html

cmd/redpanda-connect/main.go

@@ -12,8 +12,9 @@ import (
 )
 
 var (
-	Version   string
-	DateBuilt string
+	Version    string
+	DateBuilt  string
+	BinaryName string = "redpanda-connect"
 )
 
 func redpandaTopLevelConfigField() *service.ConfigField {
@@ -26,6 +27,7 @@ func main() {
 	service.RunCLI(
 		context.Background(),
 		service.CLIOptSetVersion(Version, DateBuilt),
+		service.CLIOptSetBinaryName(BinaryName),
 		service.CLIOptSetProductName("Redpanda Connect"),
 		service.CLIOptSetDocumentationURL("https://docs.redpanda.com/redpanda-connect"),
 		service.CLIOptSetMainSchemaFrom(func() *service.ConfigSchema {

internal/impl/html/bloblang.go

@@ -0,0 +1,55 @@
+package html
+
+import (
+	"fmt"
+
+	"github.com/microcosm-cc/bluemonday"
+	"github.com/redpanda-data/benthos/v4/public/bloblang"
+)
+
+func init() {
+	stripHTMLSpec := bloblang.NewPluginSpec().
+		Category("String Manipulation").
+		Description(`Attempts to remove all HTML tags from a target string.`).
+		Example("", `root.stripped = this.value.strip_html()`,
+			[2]string{
+				`{"value":"<p>the plain <strong>old text</strong></p>"}`,
+				`{"stripped":"the plain old text"}`,
+			}).
+		Example("It's also possible to provide an explicit list of element types to preserve in the output.",
+			`root.stripped = this.value.strip_html(["article"])`,
+			[2]string{
+				`{"value":"<article><p>the plain <strong>old text</strong></p></article>"}`,
+				`{"stripped":"<article>the plain old text</article>"}`,
+			}).
+		Param(bloblang.NewAnyParam("preserve").Description("An optional array of element types to preserve in the output.").Optional())
+
+	if err := bloblang.RegisterMethodV2(
+		"strip_html", stripHTMLSpec,
+		func(args *bloblang.ParsedParams) (bloblang.Method, error) {
+			p := bluemonday.NewPolicy()
+
+			var tags []any
+			if rawArgs := args.AsSlice(); len(rawArgs) > 0 {
+				tags, _ = rawArgs[0].([]any)
+			}
+
+			if len(tags) > 0 {
+				tagStrs := make([]string, len(tags))
+				for i, ele := range tags {
+					var ok bool
+					if tagStrs[i], ok = ele.(string); !ok {
+						return nil, fmt.Errorf("invalid arg at index %v: expected string, got %T", i, ele)
+					}
+				}
+				p = p.AllowElements(tagStrs...)
+			}
+
+			return bloblang.StringMethod(func(s string) (any, error) {
+				return p.Sanitize(s), nil
+			}), nil
+		},
+	); err != nil {
+		panic(err)
+	}
+}

internal/impl/html/bloblang_test.go

@@ -0,0 +1,36 @@
+package html
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/redpanda-data/benthos/v4/public/bloblang"
+)
+
+func TestStripHTMLNoArgs(t *testing.T) {
+	e, err := bloblang.Parse(`root = this.strip_html()`)
+	require.NoError(t, err)
+
+	res, err := e.Query(`<div>meow</div>`)
+	require.NoError(t, err)
+
+	assert.Equal(t, "meow", res)
+}
+
+func TestStripHTMLWithArgs(t *testing.T) {
+	e, err := bloblang.Parse(`root = this.strip_html(["strong","h1"])`)
+	require.NoError(t, err)
+
+	res, err := e.Query(`<div>
+  <h1>meow</h1>
+  <p>hello world this is <strong>some</strong> text.
+</div>`)
+	require.NoError(t, err)
+
+	assert.Equal(t, `
+  <h1>meow</h1>
+  hello world this is <strong>some</strong> text.
+`, res)
+}

public/components/pure/extended/package.go

@@ -13,6 +13,7 @@ import (
 	_ "github.com/redpanda-data/benthos/v4/public/components/pure/extended"
 
 	_ "github.com/redpanda-data/connect/v4/internal/impl/awk"
+	_ "github.com/redpanda-data/connect/v4/internal/impl/html"
 	_ "github.com/redpanda-data/connect/v4/internal/impl/jsonpath"
 	_ "github.com/redpanda-data/connect/v4/internal/impl/lang"
 	_ "github.com/redpanda-data/connect/v4/internal/impl/msgpack"