IssueTracker

Description

If the /batch/gmail/v1 path is used to batch multiple GET requests, the JSON objects in the response show gibberish characters, e.g. ü, in the subject and snippet where it is expected to see an accented letter such as German ü.

Meanwhile, when sending an individual request, e.g. GET /v1/users/me/messages/{message id}?format=full, the returned JSON shows the correct snippets and subject, with international alphabet showing correctly!

Batching Requests

Code

Prepare multiple content parts, each with Content-Type "application/http", encoding UTF-8 and body comprising of the verb, endpoint and content (if applicable). Send the multipart content to the batch endpoint.

POST /batch/gmail/v1 HTTP/1.1
Content-Type: multipart/mixed; boundary=a8ad44e6-221d-464f-a39a-feb66b553b0c

--a8ad44e6-221d-464f-a39a-feb66b553b0c
Content-Type: application/http; charset=UTF-8

GET /gmail/v1/users/me/messages/19193ba8f39fc6c5?format=metadata

--a8ad44e6-221d-464f-a39a-feb66b553b0c

Output

Among multiple parts, here is a substring. Notice Ich heiÃe... and Frage bezüglich des verfügbaren Zimmers and their expected correct output in the next section.

--batch_EmBtRoxoO2Ww8t4Redd9-G97k4feX9pD
Content-Type: application/http

HTTP/1.1 200 OK
Content-Type: application/json; charset=UTF-8

{
  "snippet": "Sehr geehrter Herr Barghash, Ich heiÃe NAME und habe von",
  "payload": { "headers": [ { "name": "Subject", "value": "Frage bezüglich des verfügbaren Zimmers" } ] }
}

Individual Requests

Code

Prepare and send a normal GET request to the resource endpoint.

GET /gmail/v1/users/me/messages/19192d4cc9b0281e?format=metadata

Output

Here is a part of the response content as string, displaying the German letters correctly without having to make any modification to the string yet.

{
  "snippet": "Sehr geehrte Herr Barghash, Ich heiße NAME und habe von",
  "payload": { "headers": [ { "name": "Subject", "value": "Frage bezüglich des verfügbaren Zimmers" } ] }
}

In that case, the user does not worry about the transfer encoding or such characters because the JSON response is UTF-8 and the backend code has already parsed the =?utf-8?q?....?= and all that stuff from the raw content. However, optimizing the network usage was going to be a great advantage of this batched requests service handler (and also recommended by the Gmail API Guide as a best practice) should the text be encoded in the multipart as it is in the individual requests.

Remarks

• This issue exists in messages as well as threads for all ?format={x} values except 'raw' (e.g. full, metadata, minimal) as it is reflected on the snippet field and the Subject header value. That being said, it could also occur for a sender's name if it uses non-ASCII characters, such as 'Hans Müller'.