patch(js): utf-8 encode multipart parts #1487
Conversation
js/src/client.ts
Outdated
| @@ -1144,10 +1145,11 @@ export class Client implements LangSmithTracingClientInterface { | |||
| const fields = { inputs, outputs, events }; | |||
| // encode the main run payload | |||
| const stringifiedPayload = stringifyForTracing(payload); | |||
| const valueBytes = encoder.encode(stringifiedPayload).length; | |||
There was a problem hiding this comment.
Is this necessary? Shouldn't/can't stringifyForTracing already return a properly encoded string?
There was a problem hiding this comment.
it is necessary. The stringify function itself doesn't actually encode anything. It converts the js object into a JSON string.
JSON.stringify() produces a JavaScript string, which is always UTF-16.
There was a problem hiding this comment.
Right, but could/should our stringifyForTracing wrapper serialize it as a utf-8 string?
Then we could pass it directly into the Blob constructor?
I think that may save us some bugs in the future
There was a problem hiding this comment.
If we don’t need to parse a length=... parameter (just application/json), then JSON.stringify(...) should be sufficient. The browser will treat it as a UTF-8 text blob, and the server sees it as JSON with no mismatch in length.
There was a problem hiding this comment.
But I can add an explicit encoding
No description provided.