csv-parser/csv-parser.go

96 lines
2.0 KiB
Go

package csvparser
import (
"strings"
"errors"
)
type EncloserId int
const (
None EncloserId = iota
DoubleQuotes
SingleQuotes
RoundBrackets
SquareBrackets
CurlyBrackets
)
type Encloser struct {
Open byte
Close byte
}
var EnclosersRunes = map[EncloserId]Encloser{
DoubleQuotes: Encloser{'"', '"'},
SingleQuotes: Encloser{'\'', '\''},
RoundBrackets: Encloser{'(', ')'},
SquareBrackets: Encloser{'[', ']'},
CurlyBrackets: Encloser{'{', '}'},
}
type Parser struct {
Enclosers []EncloserId
Delimiter string
Fields []string
Line string
}
func (parser *Parser) Init() {
parser.Enclosers = []EncloserId{DoubleQuotes, SquareBrackets}
parser.Delimiter = " \t"
}
func (parser *Parser) ExtractEnclosedFieldValue(endChar byte) error {
lineLen := len(parser.Line)
for i := 1; i < lineLen; i++ {
if parser.Line[i] == endChar && parser.Line[i-1] != '\\' {
parser.Fields = append(parser.Fields, parser.Line[0:i])
parser.Line = parser.Line[i+1:]
return nil
}
}
return errors.New("Encloser close not found")
}
func (parser *Parser) Parse(CsvLine string) error {
var err error = nil
parser.Fields = make([]string,0)
parser.Line = CsvLine
for len(parser.Line) > 0 {
parser.Line = strings.TrimLeft(parser.Line, parser.Delimiter)
parser.Line = strings.TrimRight(parser.Line, parser.Delimiter)
if len(parser.Line) == 0 {
break
}
// Search for an encloser
encloserId := None
for _, id := range parser.Enclosers {
if parser.Line[0] == EnclosersRunes[id].Open {
encloserId = id
break
}
}
if encloserId != None {
parser.Line = parser.Line[1:]
err = parser.ExtractEnclosedFieldValue(EnclosersRunes[encloserId].Close)
if err != nil {
return err
}
} else {
nextSpace := strings.IndexAny(parser.Line, parser.Delimiter)
if nextSpace != -1 {
parser.Fields = append(parser.Fields, parser.Line[:nextSpace])
parser.Line = parser.Line[nextSpace:]
} else {
parser.Fields = append(parser.Fields, parser.Line)
parser.Line = ""
break
}
}
}
return nil
}