Enable the multiline flag in line splitter regex by default

operator/builtin/input/file/file.go

@@ -165,35 +165,32 @@ func lookupEncoding(enc string) (encoding.Encoding, error) {
 
 // getSplitFunc will return the split function associated the configured mode.
 func (c InputConfig) getSplitFunc(encoding encoding.Encoding) (bufio.SplitFunc, error) {
-	var splitFunc bufio.SplitFunc
 	if c.Multiline == nil {
-		var err error
-		splitFunc, err = NewNewlineSplitFunc(encoding)
+		return NewNewlineSplitFunc(encoding)
+	}
+	endPattern := c.Multiline.LineEndPattern
+	startPattern := c.Multiline.LineStartPattern
+
+	switch {
+	case endPattern != "" && startPattern != "":
+		return nil, fmt.Errorf("only one of line_start_pattern or line_end_pattern can be set")
+	case endPattern == "" && startPattern == "":
+		return nil, fmt.Errorf("one of line_start_pattern or line_end_pattern must be set")
+	case endPattern != "":
+		re, err := regexp.Compile("(?m)" + c.Multiline.LineEndPattern)
 		if err != nil {
-			return nil, err
+			return nil, fmt.Errorf("compile line end regex: %s", err)
 		}
-	} else {
-		definedLineEndPattern := c.Multiline.LineEndPattern != ""
-		definedLineStartPattern := c.Multiline.LineStartPattern != ""
-
-		switch {
-		case definedLineEndPattern == definedLineStartPattern:
-			return nil, fmt.Errorf("if multiline is configured, exactly one of line_start_pattern or line_end_pattern must be set")
-		case definedLineEndPattern:
-			re, err := regexp.Compile(c.Multiline.LineEndPattern)
-			if err != nil {
-				return nil, fmt.Errorf("compile line end regex: %s", err)
-			}
-			splitFunc = NewLineEndSplitFunc(re)
-		case definedLineStartPattern:
-			re, err := regexp.Compile(c.Multiline.LineStartPattern)
-			if err != nil {
-				return nil, fmt.Errorf("compile line start regex: %s", err)
-			}
-			splitFunc = NewLineStartSplitFunc(re)
+		return NewLineEndSplitFunc(re), nil
+	case startPattern != "":
+		re, err := regexp.Compile("(?m)" + c.Multiline.LineStartPattern)
+		if err != nil {
+			return nil, fmt.Errorf("compile line start regex: %s", err)
 		}
+		return NewLineStartSplitFunc(re), nil
+	default:
+		return nil, fmt.Errorf("unreachable")
 	}
-	return splitFunc, nil
 }
 
 // InputOperator is an operator that monitors files for entries

operator/builtin/input/file/line_splitter_test.go

@@ -58,6 +58,15 @@ func TestLineStartSplitFunc(t *testing.T) {
 				`LOGSTART 234 log2 `,
 			},
 		},
+		{
+			Name:    "TwoLogsLineStart",
+			Pattern: `^LOGSTART \d+ `,
+			Raw:     []byte("LOGSTART 123 LOGSTART 345 log1\nLOGSTART 234 log2\nLOGSTART 345 foo"),
+			ExpectedTokenized: []string{
+				"LOGSTART 123 LOGSTART 345 log1\n",
+				"LOGSTART 234 log2\n",
+			},
+		},
 		{
 			Name:              "NoMatches",
 			Pattern:           `LOGSTART \d+ `,
@@ -114,8 +123,12 @@ func TestLineStartSplitFunc(t *testing.T) {
 	}
 
 	for _, tc := range testCases {
-		re := regexp.MustCompile(tc.Pattern)
-		splitFunc := NewLineStartSplitFunc(re)
+		cfg := NewInputConfig("")
+		cfg.Multiline = &MultilineConfig{
+			LineStartPattern: tc.Pattern,
+		}
+		splitFunc, err := cfg.getSplitFunc(unicode.UTF8)
+		require.NoError(t, err)
 		t.Run(tc.Name, tc.RunFunc(splitFunc))
 	}
 
@@ -158,6 +171,15 @@ func TestLineEndSplitFunc(t *testing.T) {
 				`log2 LOGEND 234`,
 			},
 		},
+		{
+			Name:    "TwoLogsLineEndSimple",
+			Pattern: `LOGEND$`,
+			Raw:     []byte("log1 LOGEND LOGEND\nlog2 LOGEND\n"),
+			ExpectedTokenized: []string{
+				"log1 LOGEND LOGEND",
+				"\nlog2 LOGEND",
+			},
+		},
 		{
 			Name:              "NoMatches",
 			Pattern:           `LOGEND \d+`,
@@ -210,8 +232,12 @@ func TestLineEndSplitFunc(t *testing.T) {
 	}
 
 	for _, tc := range testCases {
-		re := regexp.MustCompile(tc.Pattern)
-		splitFunc := NewLineEndSplitFunc(re)
+		cfg := NewInputConfig("")
+		cfg.Multiline = &MultilineConfig{
+			LineEndPattern: tc.Pattern,
+		}
+		splitFunc, err := cfg.getSplitFunc(unicode.UTF8)
+		require.NoError(t, err)
 		t.Run(tc.Name, tc.RunFunc(splitFunc))
 	}
 }