Implement LoRA tag extraction in /sdapi/v1/txt2img and /sdapi/v1/img2img API endpoints.

[JackBinary]

Problem

The sd-server ignores <lora:name:weight> tags in prompts sent via the /sdapi/v1/txt2img and /sdapi/v1/img2img API endpoints.

LoRA tags work correctly in the CLI (sd-cli) because it calls extract_and_remove_lora() on every generation. However, the server only calls this function once at startup when parsing command-line arguments. When handling API requests, the server copies default_gen_params, updates the prompt, but never re-parses the new prompt for LoRA tags.

Solution

This patch adds extract_and_remove_lora() to the sdapi_any2img handler function, which services both txt2img and img2img endpoints. The fix clears any pre-existing LoRA data and extracts LoRA tags from the incoming prompt, matching the behavior of the CLI.

[wbruna]

This is unsafe to enable as-is, since it'd allow loading arbitrary server files (see #1156 and #830 ).

[leejet]

Thank you for your contribution. However, personally, I’m not inclined toward this approach. I’m planning to extend the SD API to provide an interface for retrieving supported LoRAs, and then allow specifying LoRAs via additional fields in txt2img / img2img.

[JackBinary]

Understood, I'll continue maintaining my fork for local sd-server use then and await official implementation.