|
|
from transformers import AutoTokenizer, PreTrainedTokenizerFast |
|
|
from http.server import HTTPServer, BaseHTTPRequestHandler |
|
|
import json |
|
|
import argparse |
|
|
from tokenizer.tokenizer import get_qwen_tokenizer |
|
|
|
|
|
class Tokenizer_Http(): |
|
|
|
|
|
def __init__(self): |
|
|
|
|
|
self.tokenizer = get_qwen_tokenizer("CosyVoice-BlankEN", True) |
|
|
|
|
|
def encode(self, prompt): |
|
|
|
|
|
token_ids = self.tokenizer.encode(prompt, allowed_special="all") |
|
|
return token_ids |
|
|
|
|
|
def decode(self, token_ids): |
|
|
return self.tokenizer.decode(token_ids) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@property |
|
|
def eos_id(self): |
|
|
return 1773 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@property |
|
|
def eos_token(self): |
|
|
return "<|eot_id|>" |
|
|
|
|
|
|
|
|
tokenizer = Tokenizer_Http() |
|
|
|
|
|
|
|
|
print(tokenizer.encode("hello world")) |
|
|
|
|
|
|
|
|
class Request(BaseHTTPRequestHandler): |
|
|
|
|
|
timeout = 5 |
|
|
server_version = 'Apache' |
|
|
|
|
|
def do_GET(self): |
|
|
print(self.path) |
|
|
|
|
|
self.send_response(200) |
|
|
self.send_header("type", "get") |
|
|
self.end_headers() |
|
|
|
|
|
if self.path == '/bos_id': |
|
|
bos_id = tokenizer.bos_id |
|
|
|
|
|
|
|
|
if bos_id is None: |
|
|
msg = json.dumps({'bos_id': -1}) |
|
|
else: |
|
|
msg = json.dumps({'bos_id': bos_id}) |
|
|
elif self.path == '/eos_id': |
|
|
eos_id = tokenizer.eos_id |
|
|
if eos_id is None: |
|
|
msg = json.dumps({'eos_id': -1}) |
|
|
else: |
|
|
msg = json.dumps({'eos_id': eos_id}) |
|
|
else: |
|
|
msg = 'error' |
|
|
|
|
|
print(msg) |
|
|
msg = str(msg).encode() |
|
|
|
|
|
self.wfile.write(msg) |
|
|
|
|
|
def do_POST(self): |
|
|
|
|
|
data = self.rfile.read(int( |
|
|
self.headers['content-length'])) |
|
|
data = data.decode() |
|
|
|
|
|
self.send_response(200) |
|
|
self.send_header("type", "post") |
|
|
self.end_headers() |
|
|
|
|
|
if self.path == '/encode': |
|
|
req = json.loads(data) |
|
|
prompt = req['text'] |
|
|
|
|
|
token_ids = tokenizer.encode(prompt) |
|
|
if token_ids is None: |
|
|
msg = json.dumps({'token_ids': -1}) |
|
|
else: |
|
|
msg = json.dumps({'token_ids': token_ids}) |
|
|
|
|
|
elif self.path == '/decode': |
|
|
req = json.loads(data) |
|
|
token_ids = req['token_ids'] |
|
|
text = tokenizer.decode(token_ids) |
|
|
if text is None: |
|
|
msg = json.dumps({'text': ""}) |
|
|
else: |
|
|
msg = json.dumps({'text': text}) |
|
|
else: |
|
|
msg = 'error' |
|
|
print(msg) |
|
|
msg = str(msg).encode() |
|
|
|
|
|
self.wfile.write(msg) |
|
|
|
|
|
|
|
|
if __name__ == "__main__": |
|
|
|
|
|
args = argparse.ArgumentParser() |
|
|
args.add_argument('--host', type=str, default='localhost') |
|
|
args.add_argument('--port', type=int, default=12345) |
|
|
args = args.parse_args() |
|
|
|
|
|
host = (args.host, args.port) |
|
|
print('http://%s:%s' % host) |
|
|
server = HTTPServer(host, Request) |
|
|
server.serve_forever() |
|
|
|