csv-parser/csvparser.go

96 lines
2.2 KiB
Go

package csvparser
import (
"fmt"
"strings"
"unicode/utf8"
)
type CsvParser struct {
enclosers []string
delimiter rune
fields []string
}
func (p *CsvParser) Initialize(delimiter string, enclosers []string, lineFormat string) error {
if utf8.RuneCountInString(delimiter) != 1 {
return fmt.Errorf("delimiter shoud be one character")
}
p.enclosers = make([]string, 0)
for _, encloser := range enclosers {
if utf8.RuneCountInString(encloser) != 2 {
return fmt.Errorf("encolser should have to characters")
}
}
p.enclosers = enclosers
p.delimiter = []rune(delimiter)[0]
for _, pair := range enclosers {
if utf8.RuneCountInString(pair) != 2 {
return fmt.Errorf("encoloser should contain two characters: %s", pair)
}
}
// line format is in the form of: field1 field2 ignore ...
// if field name is ignore, it is parsed but not retained
p.fields = strings.Split(lineFormat, " ")
return nil
}
func (p *CsvParser) Parse(line string) (map[string]string, error) {
currentFieldIndex := 0
valueStart := 0
escape := false
enclosed := false
ret := make(map[string]string)
indexMax := len(line) - 1
maxFieldIndex := len(p.fields) - 1
delimiter := p.delimiter
for index, r := range line {
if r == '\\' {
// Check if EOL before continue
escape=true
continue
}
if escape {
escape=false
continue
}
if r == delimiter {
if p.fields[currentFieldIndex] != "ignore" {
ret[p.fields[currentFieldIndex]] = line[valueStart : index]
}
currentFieldIndex++
if currentFieldIndex > maxFieldIndex {
break
}
valueStart = index + 1
if enclosed {
enclosed=false
delimiter = p.delimiter
// Omit next delimiter
escape=true
}
continue
}
if index >= indexMax {
if p.fields[currentFieldIndex] != "ignore" {
ret[p.fields[currentFieldIndex]] = line[valueStart:]
}
continue
}
for _, encloser := range p.enclosers {
runes := []rune(encloser)
if r == runes[0] {
// opening encloser
enclosed = true
delimiter = runes[1]
valueStart++
break
}
}
}
return ret, nil
}