Spaces:
Runtime error
Runtime error
File size: 8,524 Bytes
e28221f 3a09006 e28221f 82cccd9 1375c75 82cccd9 d37853e 793eb2a e28221f 1e78474 341f0bd 0aebbbc b240a3f 3a09006 1e78474 793eb2a d37853e 1c4beff 3a09006 82cccd9 9131fdd 0aeab64 3a09006 82cccd9 0aebbbc 3a09006 0aebbbc 3a09006 73a00d6 0aebbbc 3a09006 4fbe5ad 3a09006 82cccd9 1e78474 f22c406 63cd388 1e78474 63cd388 4fbe5ad 63cd388 4fbe5ad 0aebbbc 4fbe5ad 73a00d6 4fbe5ad 0aebbbc 3a09006 82cccd9 50955d9 82cccd9 50955d9 82cccd9 4fbe5ad 5b03380 4fbe5ad d37853e 4fbe5ad 5b03380 4fbe5ad d37853e 13533c1 9062b52 b240a3f 9062b52 c36683d b240a3f c36683d 13533c1 9062b52 13533c1 d37853e 3a09006 82cccd9 0aebbbc 82cccd9 3a09006 0aebbbc 82cccd9 3a09006 82cccd9 4fbe5ad d37853e 3a09006 e28221f 3a09006 1c4beff 3a09006 e28221f |
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 |
import argparse
import uvicorn
import sys
import os
import io
from transformers import M2M100Tokenizer, M2M100ForConditionalGeneration
import time
import json
from typing import List
import torch
import logging
import string
import random
import base64
from fastapi import FastAPI, Response
from fastapi.encoders import jsonable_encoder
from fastapi.responses import JSONResponse, StreamingResponse
from pydantic import BaseModel, Field
from sse_starlette.sse import EventSourceResponse
from utils.logger import logger
from networks.message_streamer import MessageStreamer
from messagers.message_composer import MessageComposer
from googletrans import Translator
from io import BytesIO
from gtts import gTTS
from fastapi.middleware.cors import CORSMiddleware
class ChatAPIApp:
def __init__(self):
self.app = FastAPI(
docs_url="/",
title="HuggingFace LLM API",
swagger_ui_parameters={"defaultModelsExpandDepth": -1},
version="1.0",
)
self.setup_routes()
def get_available_langs(self):
f = open('apis/lang_name.json', "r")
self.available_models = json.loads(f.read())
return self.available_models
class TranslateCompletionsPostItem(BaseModel):
from_language: str = Field(
default="auto",
description="(str) `Detect`",
)
to_language: str = Field(
default="en",
description="(str) `en`",
)
input_text: str = Field(
default="Hello",
description="(str) `Text for translate`",
)
def translate_completions(self, item: TranslateCompletionsPostItem):
translator = Translator()
f = open('apis/lang_name.json', "r")
available_langs = json.loads(f.read())
from_lang = 'en'
to_lang = 'en'
for lang_item in available_langs:
if item.to_language == lang_item['code']:
to_lang = item.to_language
break
translated = translator.translate(item.input_text, dest=to_lang)
item_response = {
"from_language": translated.src,
"to_language": translated.dest,
"text": item.input_text,
"translate": translated.text
}
json_compatible_item_data = jsonable_encoder(item_response)
return JSONResponse(content=json_compatible_item_data)
def translate_ai_completions(self, item: TranslateCompletionsPostItem):
translator = Translator()
f = open('apis/lang_name.json', "r")
available_langs = json.loads(f.read())
from_lang = 'en'
to_lang = 'en'
for lang_item in available_langs:
if item.to_language == lang_item['code']:
to_lang = item.to_language
if item.from_language == lang_item['code']:
from_lang = item.from_language
if to_lang == 'auto':
to_lang = 'en'
if from_lang == 'auto':
from_lang = translator.detect(item.input_text).lang
if torch.cuda.is_available():
device = torch.device("cuda:0")
else:
device = torch.device("cpu")
logging.warning("GPU not found, using CPU, translation will be very slow.")
time_start = time.time()
#TRANSFORMERS_CACHE
pretrained_model = "facebook/m2m100_1.2B"
cache_dir = "models/"
tokenizer = M2M100Tokenizer.from_pretrained(pretrained_model, cache_dir=cache_dir)
model = M2M100ForConditionalGeneration.from_pretrained(
pretrained_model, cache_dir=cache_dir
).to(device)
model.eval()
tokenizer.src_lang = from_lang
with torch.no_grad():
encoded_input = tokenizer(item.input_text, return_tensors="pt").to(device)
generated_tokens = model.generate(
**encoded_input, forced_bos_token_id=tokenizer.get_lang_id(to_lang)
)
translated_text = tokenizer.batch_decode(
generated_tokens, skip_special_tokens=True
)[0]
time_end = time.time()
translated = translated_text
item_response = {
"from_language": from_lang,
"to_language": to_lang,
"text": item.input_text,
"translate": translated,
"start": str(time_start),
"end": str(time_end)
}
json_compatible_item_data = jsonable_encoder(item_response)
return JSONResponse(content=json_compatible_item_data)
class DetectLanguagePostItem(BaseModel):
input_text: str = Field(
default="Hello",
description="(str) `Text for detection`",
)
def detect_language(self, item: DetectLanguagePostItem):
translator = Translator()
detected = translator.detect(item.input_text)
item_response = {
"lang": detected.lang,
"confidence": detected.confidence,
}
json_compatible_item_data = jsonable_encoder(item_response)
return JSONResponse(content=json_compatible_item_data)
class TTSPostItem(BaseModel):
input_text: str = Field(
default="Hello",
description="(str) `Text for TTS`",
)
from_language: str = Field(
default="en",
description="(str) `TTS language`",
)
def text_to_speech(self, item: TTSPostItem):
try:
audioobj = gTTS(text = item.input_text, lang = item.from_language, slow = False)
fileName = ''.join(random.SystemRandom().choice(string.ascii_uppercase + string.digits) for _ in range(10));
fileName = fileName + ".mp3";
mp3_fp = BytesIO()
#audioobj.save(fileName)
#audioobj.write_to_fp(mp3_fp)
#buffer = bytearray(mp3_fp.read())
#base64EncodedStr = base64.encodebytes(buffer)
#mp3_fp.read()
#return Response(content=mp3_fp.tell(), media_type="audio/mpeg")
return StreamingResponse(audioobj.stream())
except:
item_response = {
"status": 400
}
json_compatible_item_data = jsonable_encoder(item_response)
return JSONResponse(content=json_compatible_item_data)
def setup_routes(self):
for prefix in ["", "/v1"]:
self.app.get(
prefix + "/langs",
summary="Get available languages",
)(self.get_available_langs)
self.app.post(
prefix + "/translate",
summary="translate text",
)(self.translate_completions)
self.app.post(
prefix + "/translate/ai",
summary="translate text with ai",
)(self.translate_ai_completions)
self.app.post(
prefix + "/detect",
summary="detect language",
)(self.detect_language)
self.app.post(
prefix + "/tts",
summary="text to speech",
)(self.text_to_speech)
class ArgParser(argparse.ArgumentParser):
def __init__(self, *args, **kwargs):
super(ArgParser, self).__init__(*args, **kwargs)
self.add_argument(
"-s",
"--server",
type=str,
default="0.0.0.0",
help="Server IP for HF LLM Chat API",
)
self.add_argument(
"-p",
"--port",
type=int,
default=23333,
help="Server Port for HF LLM Chat API",
)
self.add_argument(
"-d",
"--dev",
default=False,
action="store_true",
help="Run in dev mode",
)
self.args = self.parse_args(sys.argv[1:])
app = ChatAPIApp().app
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
if __name__ == "__main__":
args = ArgParser().args
if args.dev:
uvicorn.run("__main__:app", host=args.server, port=args.port, reload=True)
else:
uvicorn.run("__main__:app", host=args.server, port=args.port, reload=False)
# python -m apis.chat_api # [Docker] on product mode
# python -m apis.chat_api -d # [Dev] on develop mode
|