Recording & pasting responses¶

← docs home

OkLine captures every request/response by default, so you can inspect or paste the full exchange for any endpoint. Implementation: okline/recorder.py.

The basics¶

from okline import OkLine

api = OkLine(access_token="...")  # record=True by default
api.get_profile()
api.send_text("u...", "hi")

api.last  # the most recent Exchange (or None)
api.history  # list[Exchange], oldest → newest
print(api.last.pretty())  # one HTTP transcript, secrets redacted
print(api.dump())  # every call this session, as one string

A transcript looks like:

#1 [OK] POST /api/talk/thrift/Talk/TalkService/getProfile   (Talk.TalkService.getProfile)
======================================================================
  -> POST https://line-chrome-gw.line-apps.com/api/talk/thrift/Talk/TalkService/getProfile
  >  X-Line-Chrome-Version: 3.7.2
  >  X-Line-Access: <redacted>
  >  X-Hmac: <redacted>
  >  body: [ 2 ]
  <- HTTP 200   109 ms
  <  resp: { "mid": "u...", "displayName": "...", "regionCode": "TH" }

The `Exchange` object¶

Attribute	Meaning
`seq`	call number this session
`method`, `url`, `path`, `endpoint`	the request target (`endpoint` = `Namespace.Service.method`)
`request_headers`, `request_body`	what was sent
`status`, `response_headers`, `response_body`, `response_text`	what came back (`response_body` is the unwrapped result; `response_text` is the raw body)
`duration_ms`, `ok`, `error`, `started_at`	timing / outcome

Methods: .pretty(redact=True), .to_dict(redact=True), .to_har_entry(redact=True).

Filtering¶

api.recorder.find("Talk.TalkService.sendMessage")  # all calls to one endpoint
api.history[-5:]  # last 5 calls

Exporting¶

api.save_log("session.txt")  # plain transcript (fmt="text", the default)
api.save_log("session.json", fmt="json")  # structured JSON (a list of exchanges)
api.save_log("session.har", fmt="har")  # open in browser DevTools → Network

The accepted fmt values are "text", "json" and "har". HAR files import straight into Chrome/Firefox DevTools for a familiar Network-tab view.

You can also print the last exchange directly (UTF-8 safe, even on a Windows console):

api.print_last()

Redaction (important)¶

By default OkLine masks secrets so you can safely paste output:

request headers: X-Line-Access, X-Hmac, Authorization, X-Line-ChannelToken, Cookie, Set-Cookie.
request/response body keys: password, accessToken, refreshToken, secret, encryptedKeyChain, verifier, certificate.

Reveal them only when you really need to:

print(api.last.pretty(redact=False))  # one-off
api = OkLine(access_token="...", redact=False)  # whole session
api.save_log("raw.json", fmt="json", redact=False)

🔒 Even with redaction on, don't paste production logs publicly without a second look — response bodies can contain personal data (mids, names, …), and a saved session file (and any redact=False export) contains live credentials, including the E2EE keychain.

Live hooks¶

React to each call as it completes:

@api.on_exchange
def _(ex):
    print(ex.seq, ex.endpoint, ex.status, f"{ex.duration_ms:.0f}ms")

Or pass OkLine(on_exchange=callback). Hooks never break a call — exceptions in a hook are swallowed.

Controls¶

OkLine(record=False)  # disable recording entirely (api.last is None)
OkLine(record_capacity=2000)  # keep more history (default 500, ring buffer)
api.clear_log()  # drop captured exchanges

The recorder is a fixed-size ring buffer, so the oldest exchange is dropped once you exceed record_capacity.

Set LINE_DEBUG=1 in the environment to also dump raw response bodies to stderr as they arrive:

LINE_DEBUG=1 python your_script.py

Next: troubleshooting · architecture