fix(api): enable prompt alternatives for SDXL

api/onnx_web/chain/blend_img2img.py

@@ -84,9 +84,7 @@ class BlendImg2ImgStage(BaseStage):
                 prompt_embeds = encode_prompt(
                     pipe, prompt_pairs, params.batch, params.do_cfg()
                 )
-
-                if not params.is_xl():
-                    pipe.unet.set_prompts(prompt_embeds)
+                pipe.unet.set_prompts(prompt_embeds)
 
                 rng = np.random.RandomState(params.seed)
                 result = pipe(

api/onnx_web/chain/source_txt2img.py

@@ -133,9 +133,7 @@ class SourceTxt2ImgStage(BaseStage):
             prompt_embeds = encode_prompt(
                 pipe, prompt_pairs, params.batch, params.do_cfg()
             )
-
-            if not params.is_xl():
-                pipe.unet.set_prompts(prompt_embeds)
+            pipe.unet.set_prompts(prompt_embeds)
 
             rng = np.random.RandomState(params.seed)
             result = pipe(

api/onnx_web/diffusers/utils.py

@@ -17,14 +17,15 @@ LATENT_CHANNELS = 4
 LATENT_FACTOR = 8
 MAX_TOKENS_PER_GROUP = 77
 
+ANY_TOKEN = compile(r"\<([^\>]*)\>")
 CLIP_TOKEN = compile(r"\<clip:([-\w]+):(\d+)\>")
 INVERSION_TOKEN = compile(r"\<inversion:([^:\>]+):(-?[\.|\d]+)\>")
 LORA_TOKEN = compile(r"\<lora:([^:\>]+):(-?[\.|\d]+)\>")
-WILDCARD_TOKEN = compile(r"__([-/\\\w]+)__")
 REGION_TOKEN = compile(
     r"\<region:(\d+):(\d+):(\d+):(\d+):(-?[\.|\d]+):(-?[\.|\d]+_?[TLBR]*):([^\>]+)\>"
 )
 RESEED_TOKEN = compile(r"\<reseed:(\d+):(\d+):(\d+):(\d+):(-?\d+)\>")
+WILDCARD_TOKEN = compile(r"__([-/\\\w]+)__")
 
 INTERVAL_RANGE = compile(r"(\w+)-{(\d+),(\d+)(?:,(\d+))?}")
 ALTERNATIVE_RANGE = compile(r"\(([^\)]+)\)")
@@ -380,10 +381,10 @@ def encode_prompt(
 ) -> List[np.ndarray]:
     return [
         pipe._encode_prompt(
-            prompt,
+            remove_tokens(prompt),
             num_images_per_prompt=num_images_per_prompt,
             do_classifier_free_guidance=do_classifier_free_guidance,
-            negative_prompt=neg_prompt,
+            negative_prompt=remove_tokens(neg_prompt),
         )
         for prompt, neg_prompt in prompt_pairs
     ]
@@ -504,3 +505,8 @@ def parse_reseed_group(group) -> Region:
 
 def parse_reseed(prompt: str) -> Tuple[str, List[Reseed]]:
     return get_tokens_from_prompt(prompt, RESEED_TOKEN, parser=parse_reseed_group)
+
+
+def remove_tokens(prompt: str) -> str:
+    remainder, tokens = get_tokens_from_prompt(prompt, ANY_TOKEN)
+    return remainder