Handle chunked / partial data from server

[cellerich]

Updated two files to correctly process incoming data that may arrive in multiple fragments instead of a single complete payload.

[lanrat]

Thanks for looking into this! Several of your improvements are great — checking http.begin() return value, adding waitForWiFi(), better error logging, and explicit timeouts. I'd love to see those in a follow-up.

However, after some research, I think the chunk-parsing approach needs to be reworked.

Use the library's built-in chunked handling

The ESP32 HTTPClient already handles chunked transfer encoding transparently — see the source (search for HTTPC_TE_CHUNKED). The reason we see raw chunk data is that httpGet() reads from http.getStreamPtr() (raw WiFiClient*), bypassing the library's decoding. The fix should be using http.writeToStream() in httpGet() instead.

Concerns with manual chunk parsing in trmnl.cpp

• The detection heuristic (buff[0] >= '0' && buff[0] <= 'f') can false-positive on valid JSON starting with a hex character, and the trailing removal (jsonLen -= 4) could corrupt valid data.
• Parsing HTTP transport encoding in each consumer is fragile — if we fix it in the network layer, no caller needs to worry about it.

Suggested approach

1. Replace the manual stream->readBytes() loop in httpGet() with http.writeToStream() to handle chunked encoding transparently.
2. Update *defaultLen with actual bytes received.
3. Keep your good additions: http.begin() check, waitForWiFi(), timeouts, error logging.
4. Drop the chunk-parsing code from trmnl.cpp.

Would you be open to reworking along these lines?

Thanks again!

[cellerich]

I have to admit that I "vibe coded" the manual chunk parsing just to get the thing working.
Thank you for the deeper investigation. I will try to implement your suggested approach.

Will close this PR now and we'll open a new one once I'm able to get the new code.
I cannot commit on a timeline, but I'm sure I should be able to do it within the next seven days.

[lanrat]

No worries. I also used an LLM to review the submission and investigate if the underlying library already had support.

I'd be happy to take a look at another PR whenever you are ready.