Spaces:
Sleeping
Sleeping
File size: 5,431 Bytes
98dceee |
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 |
import asyncio
import base64
import os
import tempfile
import numpy as np
from project.config import settings
import pandas as pd
class ChatBot:
chat_history = []
def __init__(self, memory=None):
self.chat_history.append({
"role": 'assistant',
'content': "Hi! What would you like to order from the food?"
})
@staticmethod
def _transform_bytes_to_file(data_bytes) -> str:
audio_bytes = base64.b64decode(data_bytes)
temp_file = tempfile.NamedTemporaryFile(delete=False, suffix='.mp3')
try:
temp_file.write(audio_bytes)
filepath = temp_file.name
finally:
temp_file.close()
return filepath
@staticmethod
async def _transcript_audio(temp_filepath: str) -> str:
with open(temp_filepath, 'rb') as file:
transcript = await settings.OPENAI_CLIENT.audio.transcriptions.create(
model='whisper-1',
file=file,
# language='nl'
)
text = transcript.text
return text
@staticmethod
async def _convert_to_embeddings(query: str):
response = await settings.OPENAI_CLIENT.embeddings.create(
input=query,
model='text-embedding-3-large'
)
embeddings = response.data[0].embedding
return embeddings
@staticmethod
async def _convert_response_to_voice(ai_response: str) -> str:
audio = await settings.OPENAI_CLIENT.audio.speech.create(
model="tts-1",
voice="alloy",
input=ai_response
)
encoded_audio = base64.b64encode(audio.content).decode('utf-8')
return encoded_audio
@staticmethod
async def _get_context_data(query: list[float]) -> str:
query = np.array([query]).astype('float32')
_, distances, indices = settings.FAISS_INDEX.range_search(query.astype('float32'), settings.SEARCH_RADIUS)
indices_distances_df = pd.DataFrame({'index': indices, 'distance': distances})
filtered_data_df = settings.products_dataset.iloc[indices]
filtered_data_df['distance'] = indices_distances_df['distance'].values
sorted_data_df: pd.DataFrame = filtered_data_df.sort_values(by='distance').reset_index(drop=True)
sorted_data_df = sorted_data_df.drop('distance', axis=1)
data = sorted_data_df.head(1).to_dict(orient='records')
context_str = ''
for row in data:
context_str += f'{row["Search"]}\n\n'
return context_str
async def _rag(self, query: str, query_type: str, context: str = None):
if context:
self.chat_history.append({'role': 'assistant', 'content': context})
prompt = settings.PRODUCT_PROMPT
else:
if 'search' in query_type.lower():
prompt = settings.EMPTY_PRODUCT_PROMPT
elif 'purchase' in query_type.lower():
prompt = settings.ADD_TO_CART_PROMPT
elif 'product_list' in query_type.lower():
prompt = settings.PRODUCT_LIST_PROMPT
else:
prompt = settings.EMPTY_PRODUCT_PROMPT
self.chat_history.append({
'role': 'user',
'content': query
})
messages = [
{
'role': 'system',
'content': f"{prompt}"
},
]
messages += self.chat_history
completion = await settings.OPENAI_CLIENT.chat.completions.create(
messages=messages,
temperature=0,
n=1,
model="gpt-3.5-turbo",
)
response = completion.choices[0].message.content
self.chat_history.append({'role': 'assistant', 'content': response})
return response
async def _get_query_type(self, query: str) -> str:
assistant_message = self.chat_history[-1]['content']
messages = [
{
"role": 'system',
'content': settings.ANALYZER_PROMPT
},
{
"role": 'user',
"content": f"Assistant message: {assistant_message}\n"
f"User response: {query}"
}
]
completion = await settings.OPENAI_CLIENT.chat.completions.create(
messages=messages,
temperature=0,
n=1,
model="gpt-3.5-turbo",
)
response = completion.choices[0].message.content
return response
async def ask(self, data: dict):
audio = data['audio']
temp_filepath = self._transform_bytes_to_file(audio)
transcript = await self._transcript_audio(temp_filepath)
query_type = await self._get_query_type(transcript)
context = None
if query_type == 'search':
transformed_query = await self._convert_to_embeddings(transcript)
context = await self._get_context_data(transformed_query)
ai_response = await self._rag(transcript, query_type, context)
voice_ai_response = await self._convert_response_to_voice(ai_response)
data = {
'user_query': transcript,
'ai_response': ai_response,
'voice_response': voice_ai_response
}
try:
os.remove(temp_filepath)
except FileNotFoundError:
pass
return data
|