refactor(logformat): replace parser switch with registry

2 files changed, 140 insertions, 25 deletions

diff --git a/internal/mapr/logformat/parser.go b/internal/mapr/logformat/parser.go
index 37d7a63..b6ed87d 100644
--- a/internal/mapr/logformat/parser.go
+++ b/internal/mapr/logformat/parser.go
@@ -3,6 +3,8 @@ package logformat
 import (
 	"errors"
 	"fmt"
+	"strings"
+	"sync"
 	"time"
 
 	"github.com/mimecast/dtail/internal/config"
@@ -18,6 +20,62 @@ type Parser interface {
 	MakeFields(string) (map[string]string, error)
 }
 
+// ParserFactory builds a Parser for a specific log format.
+type ParserFactory func(hostname, timeZoneName string, timeZoneOffset int) (Parser, error)
+
+var parserFactories = make(map[string]ParserFactory)
+var parserFactoriesMu sync.RWMutex
+
+func init() {
+	registerBuiltInParsers()
+}
+
+// RegisterParser registers or replaces a parser factory for a log format name.
+func RegisterParser(logFormatName string, factory ParserFactory) error {
+	name := strings.TrimSpace(logFormatName)
+	if name == "" {
+		return errors.New("log format name cannot be empty")
+	}
+	if factory == nil {
+		return errors.New("parser factory cannot be nil")
+	}
+
+	parserFactoriesMu.Lock()
+	defer parserFactoriesMu.Unlock()
+	parserFactories[name] = factory
+	return nil
+}
+
+func getParserFactory(logFormatName string) (ParserFactory, bool) {
+	parserFactoriesMu.RLock()
+	defer parserFactoriesMu.RUnlock()
+	factory, found := parserFactories[logFormatName]
+	return factory, found
+}
+
+func registerBuiltInParsers() {
+	mustRegisterParser("generic", wrapParserFactory(newGenericParser))
+	mustRegisterParser("generickv", wrapParserFactory(newGenericKVParser))
+	mustRegisterParser("csv", wrapParserFactory(newCSVParser))
+	mustRegisterParser("mimecast", wrapParserFactory(newMimecastParser))
+	mustRegisterParser("mimecastgeneric", wrapParserFactory(newMimecastGenericParser))
+	mustRegisterParser("default", wrapParserFactory(newDefaultParser))
+	mustRegisterParser("custom1", wrapParserFactory(newCustom1Parser))
+	mustRegisterParser("custom2", wrapParserFactory(newCustom2Parser))
+}
+
+func mustRegisterParser(logFormatName string, factory ParserFactory) {
+	if err := RegisterParser(logFormatName, factory); err != nil {
+		panic(err)
+	}
+}
+
+func wrapParserFactory[T Parser](factory func(string, string, int) (T, error)) ParserFactory {
+	return func(hostname, timeZoneName string, timeZoneOffset int) (Parser, error) {
+		return factory(hostname, timeZoneName, timeZoneOffset)
+	}
+}
+
 // NewParser returns a new log parser.
 func NewParser(logFormatName string, query *mapr.Query) (Parser, error) {
 	hostname, err := config.Hostname()
@@ -27,30 +85,19 @@ func NewParser(logFormatName string, query *mapr.Query) (Parser, error) {
 	now := time.Now()
 	timeZoneName, timeZoneOffset := now.Zone()
 
-	// Extend this for adding more log formats!
-	switch logFormatName {
-	case "generic":
-		return newGenericParser(hostname, timeZoneName, timeZoneOffset)
-	case "generickv":
-		return newGenericKVParser(hostname, timeZoneName, timeZoneOffset)
-	case "csv":
-		return newCSVParser(hostname, timeZoneName, timeZoneOffset)
-	case "mimecast":
-		return newMimecastParser(hostname, timeZoneName, timeZoneOffset)
-	case "mimecastgeneric":
-		return newMimecastGenericParser(hostname, timeZoneName, timeZoneOffset)
-	case "default":
-		return newDefaultParser(hostname, timeZoneName, timeZoneOffset)
-	case "custom1":
-		return newCustom1Parser(hostname, timeZoneName, timeZoneOffset)
-	case "custom2":
-		return newCustom2Parser(hostname, timeZoneName, timeZoneOffset)
-	default:
-		p, err := newDefaultParser(hostname, timeZoneName, timeZoneOffset)
-		if err != nil {
-			return p, fmt.Errorf("No '%s' mapr log format and problem creating default one: %v",
-				logFormatName, err)
-		}
-		return p, fmt.Errorf("No '%s' mapr log format", logFormatName)
+	if parserFactory, found := getParserFactory(logFormatName); found {
+		return parserFactory(hostname, timeZoneName, timeZoneOffset)
+	}
+
+	defaultFactory, found := getParserFactory("default")
+	if !found {
+		return nil, fmt.Errorf("No '%s' mapr log format and no default parser registered", logFormatName)
+	}
+
+	p, err := defaultFactory(hostname, timeZoneName, timeZoneOffset)
+	if err != nil {
+		return p, fmt.Errorf("No '%s' mapr log format and problem creating default one: %v",
+			logFormatName, err)
 	}
+	return p, fmt.Errorf("No '%s' mapr log format", logFormatName)
 }
diff --git a/internal/mapr/logformat/parser_test.go b/internal/mapr/logformat/parser_test.go
new file mode 100644
index 0000000..4a4c31e
--- /dev/null
+++ b/internal/mapr/logformat/parser_test.go
@@ -0,0 +1,68 @@
+package logformat
+
+import (
+	"strings"
+	"testing"
+)
+
+type testParser struct{}
+
+func (p *testParser) MakeFields(maprLine string) (map[string]string, error) {
+	return map[string]string{"line": maprLine}, nil
+}
+
+func TestRegisterParserValidation(t *testing.T) {
+	if err := RegisterParser("", wrapParserFactory(newDefaultParser)); err == nil {
+		t.Errorf("Expected error when registering parser with empty name")
+	}
+
+	if err := RegisterParser("test-nil-factory", nil); err == nil {
+		t.Errorf("Expected error when registering parser with nil factory")
+	}
+}
+
+func TestNewParserUsesRegistry(t *testing.T) {
+	const parserName = "unit-test-registry-parser"
+
+	if err := RegisterParser(parserName, func(string, string, int) (Parser, error) {
+		return &testParser{}, nil
+	}); err != nil {
+		t.Fatalf("Unable to register parser: %s", err.Error())
+	}
+
+	parser, err := NewParser(parserName, nil)
+	if err != nil {
+		t.Fatalf("Unable to create parser from registry: %s", err.Error())
+	}
+
+	fields, err := parser.MakeFields("hello")
+	if err != nil {
+		t.Fatalf("Unable to parse line: %s", err.Error())
+	}
+	if fields["line"] != "hello" {
+		t.Errorf("Expected custom parser output, got '%s'", fields["line"])
+	}
+}
+
+func TestNewParserFallbackToDefault(t *testing.T) {
+	parser, err := NewParser("missing-parser-format", nil)
+	if err == nil {
+		t.Fatalf("Expected NewParser to return error for missing parser format")
+	}
+	if !strings.Contains(err.Error(), "No 'missing-parser-format' mapr log format") {
+		t.Errorf("Unexpected error message: %s", err.Error())
+	}
+	if parser == nil {
+		t.Fatalf("Expected default parser fallback when format is missing")
+	}
+
+	fields, parseErr := parser.MakeFields(
+		"INFO|20211002-072342|1|parser_test.go:0|8|14|7|0.21|471h0m21s|MAPREDUCE:STATS|foo=bar",
+	)
+	if parseErr != nil {
+		t.Fatalf("Fallback parser failed to parse line: %s", parseErr.Error())
+	}
+	if val, ok := fields["$severity"]; !ok || val != "INFO" {
+		t.Errorf("Fallback parser did not behave like default parser")
+	}
+}