Spaces:
Running
Running
File size: 10,047 Bytes
5c263d5 fef773e 5c263d5 fef773e 5c263d5 fef773e 5c263d5 fef773e 5c263d5 fef773e 5c263d5 fef773e 5c263d5 fef773e 5c263d5 fef773e 5c263d5 fef773e 5c263d5 fef773e 5c263d5 fef773e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 |
# -------------------------------------------------------------------
# This source file is available under the terms of the
# Pimcore Open Core License (POCL)
# Full copyright and license information is available in
# LICENSE.md which is distributed with this source code.
#
# @copyright Copyright (c) Pimcore GmbH (https://www.pimcore.com)
# @license Pimcore Open Core License (POCL)
# -------------------------------------------------------------------
import os
import torch
from fastapi import FastAPI, Path, Depends, HTTPException, UploadFile, Form, File, status, Request
from fastapi.security import HTTPBearer, HTTPAuthorizationCredentials
from pydantic import BaseModel
from typing import Annotated
import json
import logging
import sys
import base64
from transformers import pipeline
app = FastAPI(
title="Pimcore Local Inference Service",
description="This services allows HF inference provider compatible inference to models which are not available at HF inference providers.",
version="1.0.0"
)
logging.basicConfig(format='%(asctime)s %(levelname)-8s %(message)s')
logger = logging.getLogger(__name__)
logger.setLevel(logging.DEBUG)
class StreamToLogger(object):
def __init__(self, logger, log_level):
self.logger = logger
self.log_level = log_level
self.linebuf = ''
def write(self, buf):
for line in buf.rstrip().splitlines():
self.logger.log(self.log_level, line.rstrip())
def flush(self):
pass
sys.stdout = StreamToLogger(logger, logging.INFO)
sys.stderr = StreamToLogger(logger, logging.ERROR)
class ResponseModel(BaseModel):
""" Default response model for endpoints. """
message: str
success: bool = True
@app.get("/gpu_check")
async def gpu_check():
""" Check if a GPU is available """
gpu = 'GPU not available'
if torch.cuda.is_available():
gpu = 'GPU is available'
print("GPU is available")
else:
print("GPU is not available")
return {'success': True, 'gpu': gpu}
from typing import Optional
# =========================
# Translation Task
# =========================
class TranslationRequest(BaseModel):
inputs: str
parameters: Optional[dict] = None
options: Optional[dict] = None
async def get_translation_request(
request: Request
) -> TranslationRequest:
content_type = request.headers.get("content-type", "")
if content_type.startswith("application/json"):
data = await request.json()
return TranslationRequest(**data)
if content_type.startswith("application/x-www-form-urlencoded"):
raw = await request.body()
try:
data = json.loads(raw)
return TranslationRequest(**data)
except Exception:
try:
data = json.loads(raw.decode("utf-8"))
return TranslationRequest(**data)
except Exception:
raise HTTPException(status_code=400, detail="Invalid request body")
raise HTTPException(status_code=400, detail="Unsupported content type")
@app.post(
"/translation/{model_name:path}/",
openapi_extra={
"requestBody": {
"content": {
"application/json": {
"example": {
"inputs": "Hello, world! foo bar",
"parameters": {"repetition_penalty": 1.6}
}
}
}
}
}
)
async def translate(
request: Request,
model_name: str = Path(
...,
description="The name of the translation model (e.g. Helsinki-NLP/opus-mt-en-de)",
example="Helsinki-NLP/opus-mt-en-de"
)
):
"""
Execute translation tasks.
Returns:
list: The translation result(s) as returned by the pipeline.
"""
translationRequest: TranslationRequest = await get_translation_request(request)
try:
pipe = pipeline("translation", model=model_name)
except Exception as e:
logger.error(f"Failed to load model '{model_name}': {str(e)}")
raise HTTPException(
status_code=404,
detail=f"Model '{model_name}' could not be loaded: {str(e)}"
)
try:
result = pipe(translationRequest.inputs, **(translationRequest.parameters or {}))
except Exception as e:
logger.error(f"Inference failed for model '{model_name}': {str(e)}")
raise HTTPException(
status_code=500,
detail=f"Inference failed: {str(e)}"
)
return result
# =========================
# Zero-Shot Image Classification Task
# =========================
class ZeroShotImageClassificationRequest(BaseModel):
inputs: str
parameters: Optional[dict] = None
async def get_zero_shot_image_classification_request(
request: Request
) -> ZeroShotImageClassificationRequest:
content_type = request.headers.get("content-type", "")
if content_type.startswith("application/json"):
data = await request.json()
return ZeroShotImageClassificationRequest(**data)
if content_type.startswith("application/x-www-form-urlencoded"):
raw = await request.body()
try:
data = json.loads(raw)
return ZeroShotImageClassificationRequest(**data)
except Exception:
try:
data = json.loads(raw.decode("utf-8"))
return ZeroShotImageClassificationRequest(**data)
except Exception:
raise HTTPException(status_code=400, detail="Invalid request body")
raise HTTPException(status_code=400, detail="Unsupported content type")
@app.post(
"/zero-shot-image-classification/{model_name:path}/",
openapi_extra={
"requestBody": {
"content": {
"application/json": {
"example": {
"inputs": "base64_encoded_image_string",
"parameters": {"candidate_labels": "green, yellow, blue, white, silver"}
}
}
}
}
}
)
async def zero_shot_image_classification(
request: Request,
model_name: str = Path(
...,
description="The name of the zero-shot classification model (e.g., openai/clip-vit-large-patch14-336)",
example="openai/clip-vit-large-patch14-336"
)
):
"""
Execute zero-shot image classification tasks.
Returns:
list: The classification result(s) as returned by the pipeline.
"""
zeroShotRequest: ZeroShotImageClassificationRequest = await get_zero_shot_image_classification_request(request)
try:
pipe = pipeline("zero-shot-image-classification", model=model_name)
except Exception as e:
logger.error(f"Failed to load model '{model_name}': {str(e)}")
raise HTTPException(
status_code=404,
detail=f"Model '{model_name}' could not be loaded: {str(e)}"
)
try:
candidate_labels = []
if zeroShotRequest.parameters:
candidate_labels = zeroShotRequest.parameters.get('candidate_labels', [])
if isinstance(candidate_labels, str):
candidate_labels = [label.strip() for label in candidate_labels.split(',')]
result = pipe(zeroShotRequest.inputs, candidate_labels=candidate_labels)
except Exception as e:
logger.error(f"Inference failed for model '{model_name}': {str(e)}")
raise HTTPException(
status_code=500,
detail=f"Inference failed: {str(e)}"
)
return result
# =========================
# Image to Text Task
# =========================
async def get_encoded_image(
request: Request
) -> str:
content_type = request.headers.get("content-type", "")
if content_type.startswith("multipart/form-data"):
form = await request.form()
image = form.get("image")
if image:
image_bytes = await image.read()
return base64.b64encode(image_bytes).decode("utf-8")
if content_type.startswith("image/"):
image_bytes = await request.body()
return base64.b64encode(image_bytes).decode("utf-8")
raise HTTPException(status_code=400, detail="Unsupported content type")
@app.post(
"/image-to-text/{model_name:path}/",
openapi_extra={
"requestBody": {
"content": {
"multipart/form-data": {
"schema": {
"type": "object",
"properties": {
"image": {
"type": "string",
"format": "binary",
"description": "Image file to upload"
}
},
"required": ["image"]
}
}
}
}
}
)
async def image_to_text(
request: Request,
model_name: str = Path(
...,
description="The name of the image-to-text (e.g., Salesforce/blip-image-captioning-base)",
example="Salesforce/blip-image-captioning-base"
)
):
"""
Execute image-to-text tasks.
Returns:
list: The generated text as returned by the pipeline.
"""
encoded_image = await get_encoded_image(request)
try:
pipe = pipeline("image-to-text", model=model_name, use_fast=True)
except Exception as e:
logger.error(f"Failed to load model '{model_name}': {str(e)}")
raise HTTPException(
status_code=404,
detail=f"Model '{model_name}' could not be loaded: {str(e)}"
)
try:
result = pipe(encoded_image)
except Exception as e:
logger.error(f"Inference failed for model '{model_name}': {str(e)}")
raise HTTPException(
status_code=500,
detail=f"Inference failed: {str(e)}"
)
return result |