csv-parser/csv-parser.go

101 lines
2.1 KiB
Go

package csvparser
import (
"strings"
)
type EncloserId int
const (
DoubleQuotes EncloserId = iota
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
Delimiter rune
DelimiterString string
}
func (parser *CsvParser) Init() {
parser.Enclosers = []EncloserId{DoubleQuotes, SquareBrackets}
parser.Delimiter = ' '
parser.DelimiterString = string(parser.Delimiter)
}
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
}