On Mon, Oct 02, 2017 at 03:10:19AM -0700, Per Persson wrote:
> Probably it's best to ask here before I try to contact the Go developers.
>
> I had a file with space padded numbers (NumberOfPoints=" 266703") and the
> XML unmarshaller couldn't handle the spaces.
>
> Booleans are trimmed
> <https://github.com/golang/go/blob/master/src/encoding/xml/read.go#L647>,
> so why not numbers?
I'd take this question to be rather philosophical.
The question of whether the string " 266703" represents a valid
integer is hardly decidable in my opinion. For instance, would the
string "\v\v\v\n\n\n\t\t266703\n\n\n\v\v\v\v\t\x20\t" represent a valid
integer as well?
I don't know why the decoder from encoding/xml is more lax regarding
booleans; may be that's because the string representation of a boolean
value strconv.ParseBool() accepts as valid is pretty lax in itself,
or may be that's simply because when that code was written, that space
trimming was added semi-automatically by the programmer without giving
it much thought. ;-)
All in all, you could use the fact XML decoder from encoding/xml
checks whether the type of the variable it's going to unmarshal textual
data into implements the encoding.TextUnmarshaler interface [2],
and if it does, the UnmarshalText() method of that type is used to parse
that textual data.
So a way to go in your case is to define a special type to unmarshal
those "whitespace integers" from your XML data, and make that type
implement encoding.TextUnmarshaler:
----------------8<----------------
package main
import (
"bytes"
"encoding/xml"
"fmt"
"strconv"
)
type xmlInt int64
func (xi *xmlInt) UnmarshalText(b []byte) error {
v, err := strconv.ParseInt(string(bytes.TrimSpace(b)), 10, 64)
if err != nil {
return err
}
*xi = xmlInt(v)
return nil
}
type data struct {
NumberOfPoints xmlInt `xml:",attr"`
}
const s = `<data NumberOfPoints=" 266703"/>`
func main() {
var d data
err := xml.Unmarshal([]byte(s), &d)
if err != nil {
panic(err)
}
fmt.Println(d)
}
----------------8<----------------
Playground link: [1].
1. https://play.golang.org/p/Euy8Sag88P
2. https://golang.org/pkg/encoding/#TextUnmarshaler
--
You received this message because you are subscribed to the Google Groups
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to golang-nuts+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
