New option in filestream: include_files && check after symlink is…

… resolved (#25080)

## What does this PR do?

This PR adds support for a new feature in `filestream` input named `include_files`. This option is the counterpart of `exclude_files`. It expects a list of regexes and it only includes files that match the regexes.

The PR also adds one more check after symlinks are resolved, the original file is tested again with the regexes in `include_files` and `exclude_files`.

## Why is it important?

`exclude_files` was not enough for disallowing files.

(cherry picked from commit dfb98b2)

filebeat/_meta/config/filebeat.inputs.reference.yml.tmpl

@@ -281,6 +281,10 @@ filebeat.inputs:
   # are matching any regular expression from the list. By default, no files are dropped.
   #prospector.scanner.exclude_files: ['.gz$']
 
+  # Include files. A list of regular expressions to match. Filebeat keeps only the files that
+  # are matching any regular expression from the list. By default, no files are dropped.
+  #prospector.scanner.include_files: ['/var/log/.*']
+
   # Expand "**" patterns into regular glob patterns.
   #prospector.scanner.recursive_glob: true
 

filebeat/filebeat.reference.yml

@@ -688,6 +688,10 @@ filebeat.inputs:
   # are matching any regular expression from the list. By default, no files are dropped.
   #prospector.scanner.exclude_files: ['.gz$']
 
+  # Include files. A list of regular expressions to match. Filebeat keeps only the files that
+  # are matching any regular expression from the list. By default, no files are dropped.
+  #prospector.scanner.include_files: ['/var/log/.*']
+
   # Expand "**" patterns into regular glob patterns.
   #prospector.scanner.recursive_glob: true
 

filebeat/input/filestream/fswatch.go

@@ -51,6 +51,7 @@ type watcherFactory func(paths []string, cfg *common.Config) (loginp.FSWatcher,
 type fileScanner struct {
 	paths         []string
 	excludedFiles []match.Matcher
+	includedFiles []match.Matcher
 	symlinks      bool
 
 	log *logp.Logger
@@ -234,6 +235,7 @@ func (w *fileWatcher) GetFiles() map[string]os.FileInfo {
 
 type fileScannerConfig struct {
 	ExcludedFiles []match.Matcher `config:"exclude_files"`
+	IncludedFiles []match.Matcher `config:"include_files"`
 	Symlinks      bool            `config:"symlinks"`
 	RecursiveGlob bool            `config:"recursive_glob"`
 }
@@ -249,6 +251,7 @@ func newFileScanner(paths []string, cfg fileScannerConfig) (loginp.FSScanner, er
 	fs := fileScanner{
 		paths:         paths,
 		excludedFiles: cfg.ExcludedFiles,
+		includedFiles: cfg.IncludedFiles,
 		symlinks:      cfg.Symlinks,
 		log:           logp.NewLogger(scannerName),
 	}
@@ -337,7 +340,7 @@ func (s *fileScanner) GetFiles() map[string]os.FileInfo {
 }
 
 func (s *fileScanner) shouldSkipFile(file string) bool {
-	if s.isFileExcluded(file) {
+	if s.isFileExcluded(file) || !s.isFileIncluded(file) {
 		s.log.Debugf("Exclude file: %s", file)
 		return true
 	}
@@ -359,6 +362,18 @@ func (s *fileScanner) shouldSkipFile(file string) bool {
 		return true
 	}
 
+	originalFile, err := filepath.EvalSymlinks(file)
+	if err != nil {
+		s.log.Debugf("finding path to original file has failed %s: %+v", file, err)
+		return true
+	}
+	// Check if original file is included to make sure we are not reading from
+	// unwanted files.
+	if s.isFileExcluded(originalFile) || !s.isFileIncluded(originalFile) {
+		s.log.Debugf("Exclude original file: %s", file)
+		return true
+	}
+
 	return false
 }
 
@@ -384,6 +399,13 @@ func (s *fileScanner) isFileExcluded(file string) bool {
 	return len(s.excludedFiles) > 0 && s.matchAny(s.excludedFiles, file)
 }
 
+func (s *fileScanner) isFileIncluded(file string) bool {
+	if len(s.includedFiles) == 0 {
+		return true
+	}
+	return s.matchAny(s.includedFiles, file)
+}
+
 // matchAny checks if the text matches any of the regular expressions
 func (s *fileScanner) matchAny(matchers []match.Matcher, text string) bool {
 	for _, m := range matchers {

filebeat/input/filestream/fswatch_test.go

@@ -52,6 +52,7 @@ func TestFileScanner(t *testing.T) {
 	testCases := map[string]struct {
 		paths         []string
 		excludedFiles []match.Matcher
+		includedFiles []match.Matcher
 		symlinks      bool
 		expectedFiles []string
 	}{
@@ -66,6 +67,13 @@ func TestFileScanner(t *testing.T) {
 			},
 			expectedFiles: []string{includedFilePath},
 		},
+		"only include included_files": {
+			paths: []string{excludedFilePath, includedFilePath},
+			includedFiles: []match.Matcher{
+				match.MustCompile(includedFileName),
+			},
+			expectedFiles: []string{includedFilePath},
+		},
 		"skip directories": {
 			paths:         []string{filepath.Join(tmpDir, directoryPath)},
 			expectedFiles: []string{},
@@ -78,6 +86,7 @@ func TestFileScanner(t *testing.T) {
 		t.Run(name, func(t *testing.T) {
 			cfg := fileScannerConfig{
 				ExcludedFiles: test.excludedFiles,
+				IncludedFiles: test.includedFiles,
 				Symlinks:      test.symlinks,
 				RecursiveGlob: false,
 			}

filebeat/input/filestream/fswatch_test_non_windows.go

@@ -24,6 +24,7 @@ import (
 	"io/ioutil"
 	"os"
 	"path/filepath"
+	"strconv"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
@@ -44,13 +45,14 @@ func TestFileScannerSymlinks(t *testing.T) {
 	testCases := map[string]struct {
 		paths         []string
 		excludedFiles []match.Matcher
+		includedFiles []match.Matcher
 		symlinks      bool
 		expectedFiles []string
 	}{
 		// covers test_input.py/test_skip_symlinks
 		"skip symlinks": {
 			paths: []string{
-				filepath.Join(tmpDir, "symlink_to_included_file"),
+				filepath.Join(tmpDir, "symlink_to_0"),
 				filepath.Join(tmpDir, "included_file"),
 			},
 			symlinks: false,
@@ -60,22 +62,37 @@ func TestFileScannerSymlinks(t *testing.T) {
 		},
 		"return a file once if symlinks are enabled": {
 			paths: []string{
-				filepath.Join(tmpDir, "symlink_to_included_file"),
+				filepath.Join(tmpDir, "symlink_to_0"),
 				filepath.Join(tmpDir, "included_file"),
 			},
 			symlinks: true,
 			expectedFiles: []string{
 				mustAbsPath(filepath.Join(tmpDir, "included_file")),
 			},
 		},
+		"do not return symlink if original file is not allowed": {
+			paths: []string{
+				filepath.Join(tmpDir, "symlink_to_1"),
+				filepath.Join(tmpDir, "included_file"),
+			},
+			excludedFiles: []match.Matcher{
+				match.MustCompile("original_" + excludedFileName),
+			},
+			symlinks: true,
+			expectedFiles: []string{
+				mustAbsPath(filepath.Join(tmpDir, "included_file")),
+			},
+		},
 	}
 
-	err := os.Symlink(
-		mustAbsPath(filepath.Join(tmpDir, "included_file")),
-		mustAbsPath(filepath.Join(tmpDir, "symlink_to_included_file")),
-	)
-	if err != nil {
-		t.Fatal(err)
+	for i, filename := range []string{"included_file", "excluded_file"} {
+		err := os.Symlink(
+			mustAbsPath(filepath.Join(tmpDir, "original_"+filename)),
+			mustAbsPath(filepath.Join(tmpDir, "symlink_to_"+strconv.Itoa(i))),
+		)
+		if err != nil {
+			t.Fatal(err)
+		}
 	}
 
 	for name, test := range testCases {
@@ -84,6 +101,7 @@ func TestFileScannerSymlinks(t *testing.T) {
 		t.Run(name, func(t *testing.T) {
 			cfg := fileScannerConfig{
 				ExcludedFiles: test.excludedFiles,
+				IncludedFiles: test.includedFiles,
 				Symlinks:      true,
 				RecursiveGlob: false,
 			}
@@ -150,3 +168,11 @@ func TestFileWatcherRenamedFile(t *testing.T) {
 	assert.Equal(t, testPath, evt.OldPath)
 	assert.Equal(t, renamedPath, evt.NewPath)
 }
+
+func mustAbsPath(filename string) string {
+	abspath, err := filepath.Abs(filename)
+	if err != nil {
+		panic(err)
+	}
+	return abspath
+}

x-pack/filebeat/filebeat.reference.yml

@@ -2694,6 +2694,10 @@ filebeat.inputs:
   # are matching any regular expression from the list. By default, no files are dropped.
   #prospector.scanner.exclude_files: ['.gz$']
 
+  # Include files. A list of regular expressions to match. Filebeat keeps only the files that
+  # are matching any regular expression from the list. By default, no files are dropped.
+  #prospector.scanner.include_files: ['/var/log/.*']
+
   # Expand "**" patterns into regular glob patterns.
   #prospector.scanner.recursive_glob: true