Forbidden 403 "Failed to parse Connect Session Auth Token" when using Confluence API Token, workaround with header "Authorization:Basic <email:token | base64>"

[aerickson-clt]

https://github.com/atlassian-api/atlassian-python-api/blob/5d9cd15a91eea0384333d408ae0e68148c7dc302/atlassian/rest_client.py#L102

A couple of months ago this worked as is shown here. I generated an API token with my Confluence "manage account" page and simply used it here. Around the end of june this stopped working and I just figured out a fix.

Instead of simply "Bearer " I needed to use "Basic <email:token | base64>", where the string "email:token" is encoded in base 64. This is following the instructions here https://developer.atlassian.com/cloud/confluence/basic-auth-for-rest-apis/#supplying-basic-auth-headers

Note the word "Basic" instead of "Bearer".

For completeness, here is the text from that URL

---

Supplying basic auth headers You can construct and send basic auth headers yourself, including a base64-encoded string that contains your Atlassian account email and API token.

To use basic auth headers, perform the following steps:

Generate an API Token for your Atlassian Account: https://id.atlassian.com/manage/api-tokens Build a string of the form [email protected]:your_user_api_token. You'll need to encode your authorization credentials to Base64 encoded. You can do this locally: Linux/Unix/MacOS:

echo -n [email protected]:your_user_api_token | base64

Windows 7 and later, using Microsoft Powershell:

$Text = ‘[email protected]:your_user_api_token’
$Bytes = [System.Text.Encoding]::UTF8.GetBytes($Text)
$EncodedText = [Convert]::ToBase64String($Bytes)
$EncodedText

Supply an Authorization header with content Basic followed by the encoded string. Example: Authorization: Basic eW91cl9lbWFpbEBkb21haW4uY29tOnlvdXJfdXNlcl9hcGlfdG9rZW4=

curl -D- \
   -X GET \
   -H "Authorization: Basic <your_encoded_string>" \
   -H "Content-Type: application/json" \
   "https://<your-domain.atlassian.net>/wiki/rest/api/space"

---

I don't see any changes in the the atlassian-api code that would have introduced the errors that I saw, so maybe this is a change with Atlassian, or with my own organization. In any case, the rest_client does not work for me.

The non-working code I used to call this is below, where CONFLUENCE_API_TOKEN is just my plain unencoded API token.

from atlassian.confluence import Confluence
#%%
my_confluence = Confluence(url=base_url, token=CONFLUENCE_API_TOKEN, cloud=True)
#%%
my_confluence.get_page_by_id(<page_id>)

I also tried in Postman and found the actual 403 message is "Failed to parse Connect Session Auth Token". Below I show how I finally got Authentication working using the method described above.

[gonchik]

@aerickson-clt do you personal access token ?

[jpinxten]

You can work around it like this:

token = "your-api-token"
email = "[email protected]"

base64_token = base64.b64encode(f"{email}:{token}".encode()).decode()
confluence_cloud = Confluence("https://<your-domain>.atlassian.net/wiki", token=token, cloud=True)
confluence_cloud._update_header("Authorization", "Basic {token}".format(token=base64_token))

[johanatzoopla]

These docs still say to use a "Bearer" prefix for PAT: https://confluence.atlassian.com/enterprise/using-personal-access-tokens-1026032365.html

[gzurl]

I just observed the same behavior described by @aerickson-clt above. And I only managed to make it work by changing Bearer to Basic.

After some more investigations, this seems target-dependent:

• Using an older self-hosted Confluence instance for the enterprise, the way to authorize using a PAT is with Bearer.
• With the latest Cloud Confluence (you can get a free tier to reproduce it), you must use Basic with the base64 encoding.