When Base64-decoding a JWT, I found that the JSON data was incomplete. This article mainly introduces the relevant knowledge and solves this problem.
1. Introduction to JWT
JWT passes authentication by setting Authorization: Bearer <token> in the Header.
A JWT Token is a Base64-encoded string joined by dots, something like Header.Payload.Signature, and consists of three parts:
- Header, which defines the Token type and the encryption algorithm
1
2
3
4
| {
"alg": "HS256",
"typ": "JWT"
}
|
- Payload, the payload information, usually iss (issuer), exp (expiration time), sub (subject), aud (audience), iat (issued at), etc.
1
2
3
4
5
| {
"sub": "1234567890",
"name": "John Doe",
"admin": true
}
|
- Signature, which signs the Base64-encoded Header and Playload to prevent the information from being tampered with.
1
2
3
4
5
| HMACSHA256(
base64UrlEncode(header) + "." +
base64UrlEncode(payload),
your-256-bit-secret
)
|
jwt.io provides an online tool for parsing Tokens.
2. Base64 Decoding
“encoding/base64” provides four encoding and decoding methods:
- StdEncoding, standard encoding; when the length is not a multiple of 3, pad with =
- URLEncoding, URL-safe encoding; replaces the special characters +/ in the string with -_
- RawStdEncoding, standard encoding; no = padding at the end
- RawURLEncoding, URL-safe encoding; no = padding at the end
Below, let’s look at the differences between them through a concrete example.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
| package main
import (
"encoding/base64"
"fmt"
)
func coding(msg []byte){
fmt.Println("Input :", string(msg))
encoded := base64.StdEncoding.EncodeToString(msg)
fmt.Println("StdEncoding :", encoded)
decoded, _ := base64.StdEncoding.DecodeString(encoded)
fmt.Println("StdEncoding :", string(decoded))
encoded = base64.URLEncoding.EncodeToString(msg)
fmt.Println("URLEncoding :", encoded)
decoded, _ = base64.URLEncoding.DecodeString(encoded)
fmt.Println("URLEncoding :", string(decoded))
encoded = base64.RawStdEncoding.EncodeToString(msg)
fmt.Println("RawStdEncoding :", encoded)
decoded, _ = base64.RawStdEncoding.DecodeString(encoded)
fmt.Println("RawStdEncoding :", string(decoded))
encoded = base64.RawURLEncoding.EncodeToString(msg)
fmt.Println("RawURLEncoding :", encoded)
decoded, _ = base64.RawURLEncoding.DecodeString(encoded)
fmt.Println("RawURLEncoding :", string(decoded))
}
func main() {
// 补齐
coding([]byte("https://www.chenshaowen.com/"))
// URL Safe 编码
coding([]byte("abc123!?$*&()'-=@~"))
}
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
| Input : https://www.chenshaowen.com/
StdEncoding : aHR0cHM6Ly93d3cuY2hlbnNoYW93ZW4uY29tLw==
StdEncoding : https://www.chenshaowen.com/
URLEncoding : aHR0cHM6Ly93d3cuY2hlbnNoYW93ZW4uY29tLw==
URLEncoding : https://www.chenshaowen.com/
RawStdEncoding : aHR0cHM6Ly93d3cuY2hlbnNoYW93ZW4uY29tLw
RawStdEncoding : https://www.chenshaowen.com/
RawURLEncoding : aHR0cHM6Ly93d3cuY2hlbnNoYW93ZW4uY29tLw
RawURLEncoding : https://www.chenshaowen.com/
Input : abc123!?$*&()'-=@~
StdEncoding : YWJjMTIzIT8kKiYoKSctPUB+
StdEncoding : abc123!?$*&()'-=@~
URLEncoding : YWJjMTIzIT8kKiYoKSctPUB-
URLEncoding : abc123!?$*&()'-=@~
RawStdEncoding : YWJjMTIzIT8kKiYoKSctPUB+
RawStdEncoding : abc123!?$*&()'-=@~
RawURLEncoding : YWJjMTIzIT8kKiYoKSctPUB-
RawURLEncoding : abc123!?$*&()'-=@~
|
Judging from the output:
- Stdxxx variants apply padding to the Base64 encoding
- URLxxx variants transcode the Base64 encoding
Base64 is a public, standard encoding rule, but different library implementations expose different interfaces; only by using the right interface can you get the expected result.
3. The JWT Playload Is Missing a Part
The following code extracts the Playload part and decodes it:
1
2
3
4
5
6
7
8
9
10
11
12
| package main
import (
"encoding/base64"
"fmt"
)
func main() {
decoded := "eyJ1c2VybmFtZSI6ImFkbWluIiwidWlkIjoiYjhiZTZlZGQtMmM5Mi00NTM1LTliMmEtZGY2MzI2NDc0NDU4IiwiaWF0IjoxNTkxMzU0MDEwLCJpc3MiOiJrdWJlc3BoZXJlIiwibmJmIjoxNTkxMzU0MDEwfQ"
encoded, _ := base64.StdEncoding.DecodeString(decoded)
fmt.Println(string(encoded))
}
|
The result is:
1
| {"username":"admin","uid":"b8be6edd-2c92-4535-9b2a-df6326474458","iat":1591354010,"iss":"","nbf":1591354010
|
Notice that this is not a complete Json object. In the dgrijalva/jwt-go library, you can see the implementation of the EncodeSegment function:
1
2
3
4
| // Encode JWT specific base64url encoding with padding stripped
func EncodeSegment(seg []byte) string {
return strings.TrimRight(base64.URLEncoding.EncodeToString(seg), "=")
}
|
Clearly, dgrijalva/jwt-go uses RawURLEncoding for encoding.
After adjusting, run the following code:
1
2
3
4
5
6
7
8
9
10
11
12
| package main
import (
"encoding/base64"
"fmt"
)
func main() {
decoded := "eyJ1c2VybmFtZSI6ImFkbWluIiwidWlkIjoiYjhiZTZlZGQtMmM5Mi00NTM1LTliMmEtZGY2MzI2NDc0NDU4IiwiaWF0IjoxNTkxMzU0MDEwLCJpc3MiOiJrdWJlc3BoZXJlIiwibmJmIjoxNTkxMzU0MDEwfQ"
encoded, _ := base64.RawURLEncoding.DecodeString(decoded)
fmt.Println(string(encoded))
}
|
The correct result is:
1
| {"username":"admin","uid":"b8be6edd-2c92-4535-9b2a-df6326474458","iat":1591354010,"iss":"","nbf":1591354010}
|
Another approach is to use the built-in parser in dgrijalva/jwt-go and provide the complete JWT Token for parsing. Take a look at the code below:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
| package main
import (
"github.com/dgrijalva/jwt-go"
"fmt"
)
func main() {
decoded := "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJ1c2VybmFtZSI6ImFkbWluIiwidWlkIjoiYjhiZTZlZGQtMmM5Mi00NTM1LTliMmEtZGY2MzI2NDc0NDU4IiwiaWF0IjoxNTkxMzU0MDEwLCJpc3MiOiJrdWJlc3BoZXJlIiwibmJmIjoxNTkxMzU0MDEwfQ.psKkj8vYWm9Crf9jnbB_PNestLNksaS9vuMvQI3C-dU"
type Claims struct {
Username string `json:"username"`
UID string `json:"uid"`
jwt.StandardClaims
}
claim := Claims{}
parser := jwt.Parser{}
parser.ParseUnverified(decoded, &claim)
fmt.Println(claim.Username)
}
|
The expected result is:
admin
4. References