Problème de reconnaissance des champs ?
This commit is contained in:
parent
f9b8ec213a
commit
892e9a62ec
39
csvparser.go
39
csvparser.go
|
@ -6,13 +6,13 @@ import (
|
|||
"unicode/utf8"
|
||||
)
|
||||
|
||||
type Parser struct {
|
||||
type CsvParser struct {
|
||||
enclosers []string
|
||||
delimiter rune
|
||||
fields []string
|
||||
}
|
||||
|
||||
func (p *Parser) Initialize(delimiter string, enclosers []string, lineFormat string) error {
|
||||
func (p *CsvParser) Initialize(delimiter string, enclosers []string, lineFormat string) error {
|
||||
if utf8.RuneCountInString(delimiter) != 1 {
|
||||
return fmt.Errorf("delimiter shoud be one character")
|
||||
}
|
||||
|
@ -36,23 +36,40 @@ func (p *Parser) Initialize(delimiter string, enclosers []string, lineFormat str
|
|||
return nil
|
||||
}
|
||||
|
||||
func (p *Parser) Parse(line string) (map[string]string, error) {
|
||||
func (p *CsvParser) Parse(line string) (map[string]string, error) {
|
||||
inEnclosedField := false
|
||||
currentEncloserEnd := ' '
|
||||
escape := false
|
||||
currentFieldIndex := 0
|
||||
|
||||
ret := make(map[string]string)
|
||||
value := ""
|
||||
valueStart := 0
|
||||
valueEnd := 0
|
||||
for index, r := range line {
|
||||
if r == p.delimiter {
|
||||
currentFieldIndex++
|
||||
ret[p.fields[currentFieldIndex]] = value
|
||||
continue
|
||||
if r == '\\' {
|
||||
escape = !escape
|
||||
} else if inEnclosedField {
|
||||
if r == currentEncloserEnd && !escape {
|
||||
if currentFieldIndex < len(p.fields) {
|
||||
ret[p.fields[currentFieldIndex]] = line[valueStart : valueEnd-valueStart]
|
||||
}
|
||||
currentFieldIndex++
|
||||
inEnclosedField = false
|
||||
}
|
||||
} else if r == p.delimiter {
|
||||
valueEnd = index
|
||||
if currentFieldIndex < len(p.fields) {
|
||||
ret[p.fields[currentFieldIndex]] = line[valueStart : valueEnd-valueStart]
|
||||
}
|
||||
currentFieldIndex++
|
||||
} else {
|
||||
for _, encloser := range p.enclosers {
|
||||
runes := []rune(encloser)
|
||||
if r == runes[0] {
|
||||
// opening encloser
|
||||
tmpStr := line[index:]
|
||||
for tmpIndex, tmpR := range tmpStr {
|
||||
|
||||
inEnclosedField = true
|
||||
currentEncloserEnd = runes[1]
|
||||
break
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
|
@ -0,0 +1,39 @@
|
|||
package csvparser
|
||||
|
||||
import (
|
||||
"fmt"
|
||||
"testing"
|
||||
)
|
||||
|
||||
func TestCorrectLines(t *testing.T) {
|
||||
var csvParser CsvParser
|
||||
csvParser.Initialize(" ", []string{"\"\"", "[]"}, "firstname lastname complete_name")
|
||||
line := ""
|
||||
fmt.Println("parsing:", line)
|
||||
event, err := csvParser.Parse(line)
|
||||
if err != nil {
|
||||
t.Fatalf("Parsing of empty line failed %v %v", err, event)
|
||||
}
|
||||
fmt.Println("event is", event)
|
||||
line = "John Doe \"John Doe\""
|
||||
fmt.Println("parsing:", line)
|
||||
event, err = csvParser.Parse(line)
|
||||
if err != nil {
|
||||
t.Fatalf("Parsing of empty line failed %v %v", err, event)
|
||||
}
|
||||
fmt.Println("event is", event)
|
||||
line = "John Doe"
|
||||
fmt.Println("parsing:", line)
|
||||
event, err = csvParser.Parse(line)
|
||||
if err != nil {
|
||||
t.Fatalf("Parsing of empty line failed %v %v", err, event)
|
||||
}
|
||||
fmt.Println("event is", event)
|
||||
line = "John Doe I don't know him"
|
||||
fmt.Println("parsing:", line)
|
||||
event, err = csvParser.Parse(line)
|
||||
if err != nil {
|
||||
t.Fatalf("Parsing of empty line failed %v %v", err, event)
|
||||
}
|
||||
fmt.Println("event is", event)
|
||||
}
|
Loading…
Reference in New Issue