[pkg/ottl] Add GetXML Converter

pkg/ottl/e2e/e2e_test.go

@@ -444,6 +444,12 @@ func Test_e2e_converters(t *testing.T) {
 				tCtx.GetLogRecord().Attributes().PutInt("test", 1)
 			},
 		},
+		{
+			statement: `set(attributes["test"], GetXML("<a><b>1</b><c><b>2</b></c></a>", "/a//b"))`,
+			want: func(tCtx ottllog.TransformContext) {
+				tCtx.GetLogRecord().Attributes().PutStr("test", "<b>1</b><b>2</b>")
+			},
+		},
 		{
 			statement: `set(attributes["test"], Hex(1.0))`,
 			want: func(tCtx ottllog.TransformContext) {

pkg/ottl/ottlfuncs/README.md

@@ -418,6 +418,7 @@ Available Converters:
 - [ExtractGrokPatterns](#extractgrokpatterns)
 - [FNV](#fnv)
 - [Format](#format)
+- [GetXML](#getxml)
 - [Hex](#hex)
 - [Hour](#hour)
 - [Hours](#hours)
@@ -742,6 +743,37 @@ Examples:
 - `Format("%04d-%02d-%02d", [Year(Now()), Month(Now()), Day(Now())])`
 - `Format("%s/%s/%04d-%02d-%02d.log", [attributes["hostname"], body["program"], Year(Now()), Month(Now()), Day(Now())])`
 
+
+### GetXML
+
+`GetXML(target, xpath)`
+
+The `GetXML` Converter returns an XML string with selected elements.
+
+`target` is a Getter that returns a string. This string should be in XML format.
+If `target` is not a string, nil, or is not valid xml, `GetXML` will return an error.
+
+`xpath` is a string that specifies an [XPath](https://www.w3.org/TR/1999/REC-xpath-19991116/) expression that
+selects one or more elements. Currently, this converter only supports selecting elements.
+
+Examples:
+
+Get all elements at the root of the document with tag "a"
+
+- `GetXML(body, "/a")`
+
+Gel all elements anywhere in the document with tag "a"
+
+- `GetXML(body, "//a")`
+
+Get the first element at the root of the document with tag "a"
+
+- `GetXML(body, "/a[1]")`
+
+Get all elements in the document with tag "a" that have an attribute "b" with value "c"
+
+- `GetXML(body, "//a[@b='c']")`
+
 ### Hex
 
 `Hex(value)`

pkg/ottl/ottlfuncs/func_get_xml.go

@@ -0,0 +1,63 @@
+// Copyright The OpenTelemetry Authors
+// SPDX-License-Identifier: Apache-2.0
+
+package ottlfuncs // import "github.com/open-telemetry/opentelemetry-collector-contrib/pkg/ottl/ottlfuncs"
+
+import (
+	"context"
+	"fmt"
+
+	"github.com/antchfx/xmlquery"
+
+	"github.com/open-telemetry/opentelemetry-collector-contrib/pkg/ottl"
+)
+
+type GetXMLArguments[K any] struct {
+	Target ottl.StringGetter[K]
+	XPath  string
+}
+
+func NewGetXMLFactory[K any]() ottl.Factory[K] {
+	return ottl.NewFactory("GetXML", &GetXMLArguments[K]{}, createGetXMLFunction[K])
+}
+
+func createGetXMLFunction[K any](_ ottl.FunctionContext, oArgs ottl.Arguments) (ottl.ExprFunc[K], error) {
+	args, ok := oArgs.(*GetXMLArguments[K])
+
+	if !ok {
+		return nil, fmt.Errorf("GetXML args must be of type *GetXMLAguments[K]")
+	}
+
+	if err := validateXPath(args.XPath); err != nil {
+		return nil, err
+	}
+
+	return getXML(args.Target, args.XPath), nil
+}
+
+// getXML returns a XML formatted string that is a result of removing all matching nodes from the target XML.
+// This currently supports removal of elements, attributes, text values, comments, and CharData.
+func getXML[K any](target ottl.StringGetter[K], xPath string) ottl.ExprFunc[K] {
+	return func(ctx context.Context, tCtx K) (any, error) {
+		var doc *xmlquery.Node
+		if targetVal, err := target.Get(ctx, tCtx); err != nil {
+			return nil, err
+		} else if doc, err = parseNodesXML(targetVal); err != nil {
+			return nil, err
+		}
+
+		nodes, err := xmlquery.QueryAll(doc, xPath)
+		if err != nil {
+			return nil, err
+		}
+
+		result := &xmlquery.Node{Type: xmlquery.DocumentNode}
+		for _, n := range nodes {
+			if n.Type != xmlquery.ElementNode {
+				continue
+			}
+			xmlquery.AddChild(result, n)
+		}
+		return result.OutputXML(false), nil
+	}
+}

pkg/ottl/ottlfuncs/func_get_xml_test.go

@@ -0,0 +1,144 @@
+// Copyright The OpenTelemetry Authors
+// SPDX-License-Identifier: Apache-2.0
+
+package ottlfuncs // import "github.com/open-telemetry/opentelemetry-collector-contrib/pkg/ottl/ottlfuncs"
+
+import (
+	"context"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+
+	"github.com/open-telemetry/opentelemetry-collector-contrib/pkg/ottl"
+)
+
+func Test_GetXML(t *testing.T) {
+	tests := []struct {
+		name     string
+		document string
+		xPath    string
+		want     string
+	}{
+		{
+			name:     "get single element",
+			document: `<a><b/></a>`,
+			xPath:    "/a/b",
+			want:     `<b></b>`,
+		},
+		{
+			name:     "get single complex element",
+			document: `<a foo="bar"><b>hello</b></a>`,
+			xPath:    "/a",
+			want:     `<a foo="bar"><b>hello</b></a>`,
+		},
+		{
+			name:     "get uniform elements from same parent",
+			document: `<a><b>hello</b><b>world</b></a>`,
+			xPath:    "/a/b",
+			want:     `<b>hello</b><b>world</b>`,
+		},
+		{
+			name:     "get nonuniform elements from same parent",
+			document: `<a><b>hello</b><b><c>world</c></b><d/></a>`,
+			xPath:    "/a/*",
+			want:     `<b>hello</b><b><c>world</c></b><d></d>`,
+		},
+		{
+			name:     "get elements from various places",
+			document: `<a><x>1</x><b><x>2</x></b><d><e><f><x>3</x></f></e></d></a>`,
+			xPath:    "/a//x",
+			want:     `<x>1</x><x>2</x><x>3</x>`,
+		},
+		{
+			name:     "get filtered elements from various places",
+			document: `<a><x env="prod">1</x><b><x env="dev">2</x></b><d><e><f><x env="prod">3</x></f></e></d></a>`,
+			xPath:    "/a//x[@env='prod']",
+			want:     `<x env="prod">1</x><x env="prod">3</x>`,
+		},
+		{
+			name:     "ignore empty",
+			document: ``,
+			xPath:    "/",
+			want:     ``,
+		},
+		{
+			name:     "ignore declaration",
+			document: `<?xml version="1.0" encoding="UTF-8"?><a></a>`,
+			xPath:    "/*",
+			want:     `<a></a>`,
+		},
+		{
+			name:     "ignore comments",
+			document: `<!-- comment --><a></a><!-- comment -->`,
+			xPath:    "/*",
+			want:     `<a></a>`,
+		},
+		{
+			name:     "ignore attribute selection",
+			document: `<a foo="bar"></a>`,
+			xPath:    "/@foo",
+			want:     ``,
+		},
+		{
+			name:     "ignore text selection",
+			document: `<a>hello</a>`,
+			xPath:    "/a/text()",
+			want:     ``,
+		},
+		{
+			name:     "ignore chardata selection",
+			document: `<a><![CDATA[hello]]></a>`,
+			xPath:    "/a/text()",
+			want:     ``,
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			factory := NewGetXMLFactory[any]()
+			exprFunc, err := factory.CreateFunction(
+				ottl.FunctionContext{},
+				&GetXMLArguments[any]{
+					Target: ottl.StandardStringGetter[any]{
+						Getter: func(_ context.Context, _ any) (any, error) {
+							return tt.document, nil
+						},
+					},
+					XPath: tt.xPath,
+				})
+			assert.NoError(t, err)
+
+			result, err := exprFunc(context.Background(), nil)
+			assert.NoError(t, err)
+			assert.Equal(t, tt.want, result)
+		})
+	}
+}
+
+func TestCreateGetXMLFunc(t *testing.T) {
+	factory := NewGetXMLFactory[any]()
+	fCtx := ottl.FunctionContext{}
+
+	// Invalid arg type
+	exprFunc, err := factory.CreateFunction(fCtx, nil)
+	assert.Error(t, err)
+	assert.Nil(t, exprFunc)
+
+	// Invalid XPath should error on function creation
+	exprFunc, err = factory.CreateFunction(
+		fCtx, &GetXMLArguments[any]{
+			XPath: "!",
+		})
+	assert.Error(t, err)
+	assert.Nil(t, exprFunc)
+
+	// Invalid XML should error on function execution
+	exprFunc, err = factory.CreateFunction(
+		fCtx, &GetXMLArguments[any]{
+			Target: invalidXMLGetter(),
+			XPath:  "/",
+		})
+	assert.NoError(t, err)
+	assert.NotNil(t, exprFunc)
+	_, err = exprFunc(context.Background(), nil)
+	assert.Error(t, err)
+}

pkg/ottl/ottlfuncs/func_remove_xml.go

@@ -82,7 +82,7 @@ func parseNodesXML(targetVal string) (*xmlquery.Node, error) {
 	if err != nil {
 		return nil, fmt.Errorf("parse xml: %w", err)
 	}
-	if !preserveDeclearation {
+	if !preserveDeclearation && top.FirstChild != nil {
 		xmlquery.RemoveFromTree(top.FirstChild)
 	}
 	return top, nil

pkg/ottl/ottlfuncs/functions.go

@@ -46,6 +46,7 @@ func converters[K any]() []ottl.Factory[K] {
 		NewExtractPatternsFactory[K](),
 		NewExtractGrokPatternsFactory[K](),
 		NewFnvFactory[K](),
+		NewGetXMLFactory[K](),
 		NewHourFactory[K](),
 		NewHoursFactory[K](),
 		NewIntFactory[K](),