[processor/attributes] Add support to filter on log body

CHANGELOG.md

@@ -11,6 +11,7 @@
 - Add `make crosslink` target to ensure replace statements are included in `go.mod` for all transitive dependencies within repository (#8822)
 - `filestorageextension`: Change bbolt DB settings for better performance (#9004)
 - `jaegerremotesamplingextension`: Add local and remote sampling stores (#8818)
+- `attributesprocessor`: Add support to filter on log body (#8996)
 
 ### 🛑 Breaking changes 🛑
 

internal/coreinternal/processor/filterconfig/config.go

@@ -95,6 +95,10 @@ type MatchProperties struct {
 	// Deprecated: the Name field is removed from the log data model.
 	LogNames []string `mapstructure:"log_names"`
 
+	// LogBodies is a list of strings that the LogRecord's body field must match
+	// against.
+	LogBodies []string `mapstructure:"log_bodies"`
+
 	// MetricNames is a list of strings to match metric name against.
 	// A match occurs if metric name matches at least one item in the list.
 	// This field is optional.
@@ -123,6 +127,10 @@ func (mp *MatchProperties) ValidateForSpans() error {
 		return errors.New("log_names should not be specified for trace spans")
 	}
 
+	if len(mp.LogBodies) > 0 {
+		return errors.New("log_bodies should not be specified for trace spans")
+	}
+
 	if len(mp.Services) == 0 && len(mp.SpanNames) == 0 && len(mp.Attributes) == 0 &&
 		len(mp.Libraries) == 0 && len(mp.Resources) == 0 {
 		return errors.New(`at least one of "services", "span_names", "attributes", "libraries" or "resources" field must be specified`)
@@ -137,8 +145,8 @@ func (mp *MatchProperties) ValidateForLogs() error {
 		return errors.New("neither services nor span_names should be specified for log records")
 	}
 
-	if len(mp.Attributes) == 0 && len(mp.Libraries) == 0 && len(mp.Resources) == 0 {
-		return errors.New(`at least one of "attributes", "libraries" or "resources" field must be specified`)
+	if len(mp.Attributes) == 0 && len(mp.Libraries) == 0 && len(mp.Resources) == 0 && len(mp.LogBodies) == 0 {
+		return errors.New(`at least one of "attributes", "libraries", "resources" or "log_bodies" field must be specified`)
 	}
 
 	return nil

internal/coreinternal/processor/filterlog/filterlog.go

@@ -38,6 +38,9 @@ type propertiesMatcher struct {
 
 	// log names to compare to.
 	nameFilters filterset.FilterSet
+
+	// log bodies to compare to.
+	bodyFilters filterset.FilterSet
 }
 
 // NewMatcher creates a LogRecord Matcher that matches based on the given MatchProperties.
@@ -62,20 +65,33 @@ func NewMatcher(mp *filterconfig.MatchProperties) (Matcher, error) {
 			return nil, fmt.Errorf("error creating log record name filters: %v", err)
 		}
 	}
+	var bodyFS filterset.FilterSet
+	if len(mp.LogBodies) > 0 {
+		bodyFS, err = filterset.CreateFilterSet(mp.LogBodies, &mp.Config)
+		if err != nil {
+			return nil, fmt.Errorf("error creating log record body filters: %v", err)
+		}
+	}
 
 	return &propertiesMatcher{
 		PropertiesMatcher: rm,
 		nameFilters:       nameFS,
+		bodyFilters:       bodyFS,
 	}, nil
 }
 
 // MatchLogRecord matches a log record to a set of properties.
 // There are 3 sets of properties to match against.
 // The log record names are matched, if specified.
+// The log record bodies are matched, if specified.
 // The attributes are then checked, if specified.
 // At least one of log record names or attributes must be specified. It is
 // supported to have more than one of these specified, and all specified must
 // evaluate to true for a match to occur.
 func (mp *propertiesMatcher) MatchLogRecord(lr pdata.LogRecord, resource pdata.Resource, library pdata.InstrumentationScope) bool {
+	if lr.Body().Type() == pdata.ValueTypeString && mp.bodyFilters != nil && mp.bodyFilters.Matches(lr.Body().StringVal()) {
+		return true
+	}
+
 	return mp.PropertiesMatcher.Match(lr.Attributes(), resource, library)
 }

internal/coreinternal/processor/filterlog/filterlog_test.go

@@ -40,14 +40,15 @@ func TestLogRecord_validateMatchesConfiguration_InvalidConfig(t *testing.T) {
 		{
 			name:        "empty_property",
 			property:    filterconfig.MatchProperties{},
-			errorString: "at least one of \"attributes\", \"libraries\" or \"resources\" field must be specified",
+			errorString: `at least one of "attributes", "libraries", "resources" or "log_bodies" field must be specified`,
 		},
 		{
 			name: "empty_log_names_and_attributes",
 			property: filterconfig.MatchProperties{
-				LogNames: []string{},
+				LogNames:  []string{},
+				LogBodies: []string{},
 			},
-			errorString: "at least one of \"attributes\", \"libraries\" or \"resources\" field must be specified",
+			errorString: `at least one of "attributes", "libraries", "resources" or "log_bodies" field must be specified`,
 		},
 		{
 			name: "span_properties",
@@ -149,10 +150,18 @@ func TestLogRecord_Matching_True(t *testing.T) {
 				},
 			},
 		},
+		{
+			name: "log_body_regexp_match",
+			properties: &filterconfig.MatchProperties{
+				Config:    *createConfig(filterset.Regexp),
+				LogBodies: []string{"AUTH.*"},
+			},
+		},
 	}
 
 	lr := pdata.NewLogRecord()
 	lr.Attributes().InsertString("abc", "def")
+	lr.Body().SetStringVal("AUTHENTICATION FAILED")
 
 	for _, tc := range testcases {
 		t.Run(tc.name, func(t *testing.T) {

processor/attributesprocessor/README.md

@@ -166,13 +166,13 @@ if the input data should be included or excluded from the processor. To configur
 this option, under `include` and/or `exclude` at least `match_type` and one of the following
 is required:
 - For spans, one of `services`, `span_names`, `attributes`, `resources`, or `libraries` must be specified
-with a non-empty value for a valid configuration. The `log_names`, `expressions`, `resource_attributes` and
+with a non-empty value for a valid configuration. The `log_names`, `log_bodies`, `expressions`, `resource_attributes` and
 `metric_names` fields are invalid.
-- For logs, one of `log_names`, `attributes`, `resources`, or `libraries` must be specified with a
+- For logs, one of `log_names`, `log_bodies`, `attributes`, `resources`, or `libraries` must be specified with a
 non-empty value for a valid configuration. The `span_names`, `metric_names`, `expressions`, `resource_attributes`,
 and `services` fields are invalid.
 - For metrics, one of `metric_names`, `resources` must be specified
-with a valid non-empty value for a valid configuration. The `span_names`, `log_names`, and
+with a valid non-empty value for a valid configuration. The `span_names`, `log_names`, `log_bodies` and
 `services` fields are invalid.
 
 
@@ -218,6 +218,11 @@ attributes:
       # This is an optional field.
       log_names: [<item1>, ..., <itemN>]
 
+      # The log body must match at least one of the items.
+      # Currently only string body types are supported.
+      # This is an optional field.
+      log_bodies: [<item1>, ..., <itemN>]
+
       # The metric name must match at least one of the items.
       # This is an optional field.
       metric_names: [<item1>, ..., <itemN>]

processor/attributesprocessor/testdata/config.yaml

@@ -307,6 +307,24 @@ processors:
         action: update
         value: "SELECT * FROM USERS [obfuscated]"
 
+
+  # The following demonstrates how to process logs that have a body that match regexp
+  # patterns. This processor will remove "token" attribute and will obfuscate "password"
+  # attribute in spans where body matches "AUTH.*".
+  attributes/log_body_regexp:
+    # Specifies the span properties that must exist for the processor to be applied.
+    include:
+      # match_type defines that "services" is an array of regexp-es.
+      match_type: regexp
+      # The span service name must match "auth.*" pattern.
+      log_bodies: ["AUTH.*"]
+    actions:
+      - key: password
+        action: update
+        value: "obfuscated"
+      - key: token
+        action: delete
+
 receivers:
   nop: