[Jason2866]

Description:

Refactor get_jpeg_size function to improve JPEG parsing and validation.

• Bug Fixes
  • Improved JPEG image handling so valid baseline images can be recognized without JFIF metadata, and their dimensions are extracted more reliably.
  • Invalid or incomplete JPEG data and progressive JPEG images are rejected.

Related (if applicable): #25219 @SHIFox8

Checklist:

• The pull request is done against the latest development branch
• Only relevant files were touched
• Only one feature/fix was added per PR and the code change compiles without warnings
• The code change is tested and works with Tasmota core ESP8266 V.2.7.8
• The code change is tested and works with Tasmota core ESP32 V.3.3.12-1 from Platform 2026.10.50
• I accept the CLA.

NOTE: The code change must pass CI tests. Your PR cannot be merged unless tests pass

[Jason2866]

@arendst Should be good to be merged. Tested with several jpeg files.

[s-hadinger]

I submitted an AI review, and it proposed some improvements:

• Outputs stay set when the function returns false, however the callers don't test for return value. Fix: store the SOF0 values in locals and assign them only on return true.
• The USE_BERRY_IMAGE guard adds code nothing calls. xdrv_52_3_berry_img.ino uses its own be_img_util::jpeg_size()
• Requiring EOI is not stricly required if we accept to decode partially truncated files. Returning true at SOF0 or
  at the first SOS would cut about 25 lines and leave truncation to the decoder.

Do you want me to update the files in place in this PR?

[Jason2866]

@s-hadinger i added the USE_BERRY_IMAGE since the jpeg size refactor is part of implementation of a new epd display.
The jpeg_size in xdrv_52_3_berry_img.ino is redundant code which can be removed and can call the function here.

Not complete jpeg is an error and should not be accepted.

[s-hadinger]

Ok, let me fix that

[s-hadinger]

Wait, let me clean up my mess

[s-hadinger]

Ok, I just limited to zeroing the height/width when the JPEG is rejected.
Overall it's not ideal to use unsigned short types, it would be more explicit to use uint16_t instead.