Dissect: Support trimming captured values and other configuration opt…

…ions (elastic#19464) (elastic#19685)

This adds 4 new configuration parameters for the dissect processor:

- ignore_failure, so that it doesn't return an error when the tokenizer doesn't match the input. This is useful for chaining multiple processors together.
- overwrite_keys, so that existing keys are overwritten if they already exist.
- trim_values, to enable trimming blank space (left,(leading) right(trailing) or all) in captured values.
- trim_chars is the set of characters that are trimmed by the above option (default is the ascii space character).

The default values for these new flags have been chosen so that the default behavior of the processor is unchanged.

(cherry picked from commit a57e390)

CHANGELOG.next.asciidoc

@@ -385,6 +385,9 @@ field. You can revert this change by configuring tags for the module and omittin
 - Upgrade k8s.io/client-go and k8s keystore tests. {pull}18817[18817]
 - Add support for multiple sets of hints on autodiscover {pull}18883[18883]
 - Add a configurable delay between retries when an app metadata cannot be retrieved by `add_cloudfoundry_metadata`. {pull}19181[19181]
+- Add the `ignore_failure` configuration option to the dissect processor. {pull}19464[19464]
+- Add the `overwrite_keys` configuration option to the dissect processor. {pull}19464[19464]
+- Add support to trim captured values in the dissect processor. {pull}19464[19464]
 
 *Auditbeat*
 

libbeat/processors/dissect/config.go

@@ -17,15 +17,35 @@
 
 package dissect
 
+import (
+	"strings"
+
+	"github.com/pkg/errors"
+)
+
+type trimMode byte
+
+const (
+	trimModeNone trimMode = iota
+	trimModeRight
+	trimModeLeft
+	trimModeAll = trimModeRight | trimModeLeft
+)
+
 type config struct {
-	Tokenizer    *tokenizer `config:"tokenizer" validate:"required"`
-	Field        string     `config:"field"`
-	TargetPrefix string     `config:"target_prefix"`
+	Tokenizer     *tokenizer `config:"tokenizer" validate:"required"`
+	Field         string     `config:"field"`
+	TargetPrefix  string     `config:"target_prefix"`
+	IgnoreFailure bool       `config:"ignore_failure"`
+	OverwriteKeys bool       `config:"overwrite_keys"`
+	TrimValues    trimMode   `config:"trim_values"`
+	TrimChars     string     `config:"trim_chars"`
 }
 
 var defaultConfig = config{
 	Field:        "message",
 	TargetPrefix: "dissect",
+	TrimChars:    " ",
 }
 
 // tokenizer add validation at the unpack level for this specific field.
@@ -40,3 +60,20 @@ func (t *tokenizer) Unpack(v string) error {
 	*t = *d
 	return nil
 }
+
+// Unpack the trim mode from a string.
+func (tm *trimMode) Unpack(v string) error {
+	switch strings.ToLower(v) {
+	case "", "none":
+		*tm = trimModeNone
+	case "left":
+		*tm = trimModeLeft
+	case "right":
+		*tm = trimModeRight
+	case "all", "both":
+		*tm = trimModeAll
+	default:
+		return errors.Errorf("unsupported value %s. Must be one of [none, left, right, all]", v)
+	}
+	return nil
+}

libbeat/processors/dissect/config_test.go

@@ -40,6 +40,7 @@ func TestConfig(t *testing.T) {
 		if !assert.NoError(t, err) {
 			return
 		}
+		assert.Equal(t, trimModeNone, cfg.TrimValues)
 	})
 
 	t.Run("invalid", func(t *testing.T) {
@@ -100,4 +101,39 @@ func TestConfig(t *testing.T) {
 			return
 		}
 	})
+
+	t.Run("with wrong trim_mode", func(t *testing.T) {
+		c, err := common.NewConfigFrom(map[string]interface{}{
+			"tokenizer":   "hello %{what}",
+			"field":       "message",
+			"trim_values": "bananas",
+		})
+		if !assert.NoError(t, err) {
+			return
+		}
+
+		cfg := config{}
+		err = c.Unpack(&cfg)
+		if !assert.Error(t, err) {
+			return
+		}
+	})
+
+	t.Run("with valid trim_mode", func(t *testing.T) {
+		c, err := common.NewConfigFrom(map[string]interface{}{
+			"tokenizer":   "hello %{what}",
+			"field":       "message",
+			"trim_values": "all",
+		})
+		if !assert.NoError(t, err) {
+			return
+		}
+
+		cfg := config{}
+		err = c.Unpack(&cfg)
+		if !assert.NoError(t, err) {
+			return
+		}
+		assert.Equal(t, trimModeAll, cfg.TrimValues)
+	})
 }

libbeat/processors/dissect/dissect.go

@@ -33,8 +33,9 @@ type position struct {
 // Dissector is a tokenizer based on the Dissect syntax as defined at:
 // https://www.elastic.co/guide/en/logstash/current/plugins-filters-dissect.html
 type Dissector struct {
-	raw    string
-	parser *parser
+	raw     string
+	parser  *parser
+	trimmer trimmer
 }
 
 // Dissect takes the raw string and will use the defined tokenizer to return a map with the
@@ -57,7 +58,12 @@ func (d *Dissector) Dissect(s string) (Map, error) {
 	if len(positions) == 0 {
 		return nil, errParsingFailure
 	}
-
+	if d.trimmer != nil {
+		for idx, pos := range positions {
+			pos.start, pos.end = d.trimmer.Trim(s, pos.start, pos.end)
+			positions[idx] = pos
+		}
+	}
 	return d.resolve(s, positions), nil
 }
 

libbeat/processors/dissect/docs/dissect.asciidoc

@@ -25,7 +25,29 @@ The `dissect` processor has the following configuration settings:
 `target_prefix`:: (Optional) The name of the field where the values will be extracted. When an empty
 string is defined, the processor will create the keys at the root of the event. Default is
 `dissect`. When the target key already exists in the event, the processor won't replace it and log
-an error; you need to either drop or rename the key before using dissect.
+an error; you need to either drop or rename the key before using dissect, or
+enable the `overwrite_keys` flag.
+
+`ignore_failure`:: (Optional) Flag to control whether the processor returns an error if the
+tokenizer fails to match the message field. If set to true, the processor will silently restore
+the original event, allowing execution of subsequent processors (if any). If set to false
+(default), the processor will log an error, preventing execution of other processors.
+
+`overwrite_keys`:: (Optional) When set to true, the processor will overwrite
+existing keys in the event. The default is false, which causes the processor
+to fail when a key already exists.
+
+`trim_values`:: (Optional) Enables the trimming of the extracted values. Useful
+to remove leading and/or trailing spaces. Possible values are:
+- `none`: (default) no trimming is performed.
+- `left`: values are trimmed on the left (leading).
+- `right`: values are trimmed on the right (trailing).
+- `all`: values are trimmed for leading and trailing.
+
+`trim_chars`:: (Optional) Set of characters to trim from values, when trimming
+is enabled. The default is to trim the space character (`" "`). To trim multiple
+characters, simply set it to a string containing all characters to trim. For example,
+`trim_chars: " \t"` will trim spaces and/or tabs.
 
 For tokenization to be successful, all keys must be found and extracted, if one of them cannot be
 found an error will be logged and no modification is done on the original event.

libbeat/processors/dissect/processor.go

@@ -46,6 +46,14 @@ func NewProcessor(c *common.Config) (processors.Processor, error) {
 	if err != nil {
 		return nil, err
 	}
+	if config.TrimValues != trimModeNone {
+		config.Tokenizer.trimmer, err = newTrimmer(config.TrimChars,
+			config.TrimValues&trimModeLeft != 0,
+			config.TrimValues&trimModeRight != 0)
+		if err != nil {
+			return nil, err
+		}
+	}
 	p := &processor{config: config}
 
 	return p, nil
@@ -72,7 +80,9 @@ func (p *processor) Run(event *beat.Event) (*beat.Event, error) {
 		); err != nil {
 			return event, errors.Wrap(err, "cannot add new flag the event")
 		}
-
+		if p.config.IgnoreFailure {
+			return event, nil
+		}
 		return event, err
 	}
 
@@ -94,7 +104,7 @@ func (p *processor) mapper(event *beat.Event, m common.MapStr) (*beat.Event, err
 	var prefixKey string
 	for k, v := range m {
 		prefixKey = prefix + k
-		if _, err := event.GetValue(prefixKey); err == common.ErrKeyNotFound {
+		if _, err := event.GetValue(prefixKey); err == common.ErrKeyNotFound || p.config.OverwriteKeys {
 			event.PutValue(prefixKey, v)
 		} else {
 			event.Fields = copy