ahserver_g/ahserver/llm_client.py

import re
import json
from aiohttp import web
from appPublic.dictObject import DictObject
from appPublic.httpclient import HttpClient, RESPONSE_TEXT, RESPONSE_JSON, RESPONSE_BIN,RESPONSE_FILE, RESPONSE_STREAM
from appPublic.argsConvert import ArgsConvert

class LlmProxy:
	def __init__(self, processor, desc):
		assert desc.name
		self.name = desc.name
		self.processor = processor
		self.auth_api = desc.auth
		self.desc = desc
		self.api_name = desc.name
		self.data = DictObject()
		self.ac = ArgsConvert('${', '}')
	
	def line_chunk_match(self, l):
		if self.chunk_match:
			match = re.search(self.chunk_match, l)
			if match:
				return match.group(1)
		return l

	async def write_chunk(self, l):
		l = self.line_chunk_match(l)
		d = DictObject(** json.loads(l))
		content = d.get_data_by_keys(self.api.content_keys)
		finished = d.get_data_by_keys(self.api.finished_keys)
		tokens = {}
		if finished:
			tokens = d.get_data_by_keys(self.api.tokens_keys)

		j = {
			'content':content,
			'finish':False,
			'token':tokens
		}
		jstr = json.dumps(j) + '\n'
		bin = jstr.encode('utf-8')
		await self.resp.write(bin)

	async def stream_handle(self, chunk):
		chunk = chunk.decode('utf-8')
		print(f'{chunk=}')
		chunk = self.remain_str + chunk
		lines = chunk.split('\n')
		self.remain_str = lines[-1]
		ls = lines[:-1]
		for l in ls:	
			if l == '':
				continue
			await self.write_chunk(l)

	async def get_apikey(self, apiname, user):
		f = self.processor.run_ns.get_llm_user_apikey
		if f:
			# return a DictObject instance
			return await f(apiname, user)
		raise Exception('get_llm_user_apikey() function not found in ServerEnv')

	async def do_auth(self, request):
		d = self.desc.auth
		user = await self.processor.run_ns.get_user()
		self.data = self.get_data(self.name, user)
		if self.data.authed:
			return
		self.data = await self.get_apikey(self.name, user)
		if self.data is None:
			raise Exception(f'user({user}) do not has a apikey for {self.name}')
		method = d.get('method', 'POST')
		headers = {}
		for h in d.get('headers',{}):
			headers[h.get('name')] = h.get('value')
		mydata = {}
		for p in d.data or []:
			mydata[p.get('name')] = p.get('value')
		myparams = {}
		for p in d.params or []:
			myparams[p.get('name')] = p.get('value')
		url = d.get('url')
		params = {}
		_params = self.datalize(myparams, params)
		_headers = self.datalize(headers, params)
		_data = self.datalize(mydata, params)
		hc = HttpClient()
		resp_data = await hc.request(url, method, response_type=RESPONSE_JSON,
								params=_params,
								data=None if _data == {} else json.dumps(_data),
								headers=_headers)
		resp_data = DictObject(**resp_data)
		for sd in d.set_data:
			self.data[sd.name] = resp_data.get_data_by_keys(sd.field)
		self.data.authed = True
		self.set_data(self.name, user, self.data)

	def data_key(self, apiname, user):
		if user is None:
			user = 'anoumous'
		return apiname + '_a_' + user

	def set_data(self, apiname, user, data):
		request = self.processor.run_ns.request
		app = request.app
		app.set_data(self.data_key(apiname, user), data)
	
	def get_data(self, apiname, user):
		request = self.processor.run_ns.request
		app = request.app
		return app.get_data(self.data_key(apiname, user))
		
	async def __call__(self, request, params):
		mapi = params.mapi
		stream = params.stream
		self.resp = web.StreamResponse()
		await self.resp.prepare(request)
		if stream is None:
			stream = True
		self.remain_str = ''
		if not self.desc[mapi]:
			raise Exception(f'{mapi} not defined')
		d = self.desc[mapi]
		self.api = d
		self.chunk_match = d.chunk_match
		if self.api.need_auth and self.auth_api:
			await self.do_auth(request)
		assert d.get('url')
		method = d.get('method', 'POST')
		headers = {}
		for h in d.get('headers',{}):
			headers[h.get('name')] = h.get('value')
		mydata = {}
		for p in d.get('data', {}):
			mydata[p.get('name')] = p.get('value')
		myparams = {}
		for p in d.get('params', {}):
			myparams[p.get('name')] = p.get('value')
		url = d.get('url')
		_params = self.datalize(myparams, params)
		_headers = self.datalize(headers, params)
		_data = self.datalize(mydata, params)
		response_type =  RESPONSE_STREAM
		hc = HttpClient()
		resp_data = await hc.request(url, method, response_type=response_type,
								params=_params,
								data=None if _data == {} else json.dumps(_data),
								stream_func=self.stream_handle,
								headers=_headers)
		if self.remain_str != '':
			await self.write_chunk(self.remain_str)
		return self.resp
		
	def datalize(self, dic, data={}):
		mydata = self.data.copy()
		mydata.update(data)
		s1 = self.ac.convert(dic, mydata)
		return s1