making code more generic

csv-parser.go

@@ -1,5 +1,8 @@
 package csvparser
 
+import (
+	"strings"
+)
 type EncloserId int
 
 const (
@@ -7,18 +10,91 @@ const (
 	SingleQuotes
 	RoundBrackets
 	SquareBrackets
-
+	CurlyBrackets
 )
 
+type Encloser struct {
+	Open  rune
+	Close rune
+}
+
+var EnclosersRunes = map[EncloserId]Encloser{
+	DoubleQuotes:   Encloser{'"', '"'},
+	SingleQuotes:   Encloser{'\'', '\''},
+	RoundBrackets:  Encloser{'(', ')'},
+	SquareBrackets: Encloser{'[', ']'},
+	CurlyBrackets:  Encloser{'{', '}'},
+}
 
 type CsvParser struct {
-	enclosers []EncloserId
+	Enclosers []EncloserId
-	delimiter  rune
+	Delimiter rune
+	DelimiterString string
 }
 
 func (parser *CsvParser) Init() {
-	parser.enclosers = append(parser.enclosers, DoubleQuotes, SquareBrackets)
+	parser.Enclosers = []EncloserId{DoubleQuotes, SquareBrackets}
-	parser.delimiter = ' '
+	parser.Delimiter = ' '
+	parser.DelimiterString = string(parser.Delimiter)
 }
-func (parser *CsvParser) Parse(line string) {
+func (parser *CsvParser) Parse(line string) error {
+	for len(line) > 0 {
+		line = strings.TrimLeft(line, parser.DelimiterString)
+		line = strings.TrimRight(line, parser.DelimiterString)
+
+		if len(line) == 0 {
+			break
+		}
+		// Search for an encloser
+		for id := range parser.Enclosers {
+			if line[0] == EnclosersRunes[id].Open {
+				endChar := EnclosersRunes[id].Close
+				endOfFieldFound := false
+				l := len(line)
+				for i := 1; i < l; i++ {
+					if line[i] == endChar && line[i-1] != '\\' {
+						fields = append(fields, line[1:i])
+						line = line[i+1:]
+						endOfFieldFound = true
+						break
+					}
+				}
+				if !endOfFieldFound {
+					return nil, errors.New("#ERR: bad format")
+				}
+				break;
+			}
+		}
+		if line[0] == '"' || line[0] == '[' {
+			var endChar byte = '"'
+			if line[0] == '[' {
+				endChar = ']'
+			}
+			endOfFieldFound := false
+			l := len(line)
+			for i := 1; i < l; i++ {
+				if line[i] == endChar && line[i-1] != '\\' {
+					fields = append(fields, line[1:i])
+					line = line[i+1:]
+					endOfFieldFound = true
+					break
+				}
+			}
+			if !endOfFieldFound {
+				return nil, errors.New("#ERR: bad format")
+			}
+		} else {
+			nextSpace := strings.IndexAny(line, " \t")
+			if nextSpace != -1 {
+				fields = append(fields, line[:nextSpace])
+				line = line[nextSpace:]
+			} else {
+				fields = append(fields, line)
+				line = ""
+				break
+			}
+		}
+	}
+	return fields, nil
+
 }