Spaces:
Sleeping
Sleeping
File size: 14,211 Bytes
ccbdf6f |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 |
import openai
from openai import OpenAI
from scraper import *
from dotenv import load_dotenv
import streamlit as st
import math
USE_CLI = False
USE_HISTORY_LEN = 20
load_dotenv()
openai.api_key = os.getenv("OAI_KEY")
brave_key = os.getenv("BRAVE_KEY")
client = OpenAI()
chat_hist = []
def display_imgs(urls):
grid = st.columns(3)
col = 0
ctr = 0
for url in urls[:min(len(urls), 9)]:
ctr += 1
with grid[col]:
st.image(url)
col = ctr % 3
def update_query(raw_query, context):
context_str ="\n".join(context)
updated_query = f"""
{raw_query}
Context: {context_str}
"""
return updated_query
def openAI_api_call(mode, query, raw_query = None):
print("="*50)
print(f"Using mode {mode}")
print("="*50)
if mode == "router":
curr_msgs = [
{"role": "system", "content": """You are a helpful assistant with access to the chat history, user query and following tools and their descriptions:
TOOL NAME: get_relevant_context
TOOL_DESCRIPTION: Given user query, present relevant text information about the query.
TOOL NAME: get_relevant_images
TOOL_DESCRIPTION: Given user query, present relevant image URLs about the query.
The use of tools is optional. If based on user query and chat history, you feel that no tools are required, answer saying no_tools. Otherwise,
mention the tool name(s). Note, including relevant images whenever possible is highly encouraged to enhance user experience. The answer has to be one or more from [get_relevant_context, get_relevant_images, no_tools]. Think through this step by step and come up with the answer.
Here are some examples:
User: What is the name of a Tom Cruise Movie?
Assistant: get_relevant_context
User: Suggest some books by Enid Blyton
Assistant: get_relevant_context, get_relevant_images
User: Suggest some movies by Steven Speilberg
Assistant: get_relevant_context, get_relevant_images
User: Suggest a comedy movie
Assistant: get_relevant_context, get_relevant_images
User: Suggest a book for a seven year old
Assistant: get_relevant_context, get_relevant_images
User: Can you show me a poster of the movie Space Jam?
Assistant: get_relevant_images
User: Tell me a joke
Assistant: no_tools
User: Who are you?
Assistant: no_tools
User: Can you give me a summary of the third one?
Assistant: get_relevant_context
User: Can you give me a photo of this person?
Assistant: get_relevant_images
"""},
]
elif mode == "images":
curr_msgs = [
{"role": "system", "content": """
"Given a user query and chat history, use the chat history and user query to give key words such as title, names, etc. Consider incorporating terms, phrases, or topics discussed in the chat history that may provide additional context or refine the search. Ensure the query return keywords separated by commas. Avoid ambiguity or overly broad queries that may result in irrelevant images. If no relevant chat history is available, focus on refining the query based on the user's input alone. Think through this step by step and come up with the answer."
Example:
Chat History:
User: Suggest a book for a 5 year old
Assistant: A recommended book for a 5-year-old is "Where the Wild Things Are" by Maurice Sendak
User: Can you give a picture of the author?
Assistant: Maurice Sendak
Example:
Chat History:
User: Suggest a book for a 5 year old
Assistant: I recommend the book "The Very Hungry Caterpillar" by Eric Carle for a 5-year-old
User: Can you give a picture of the book?
Assistant: The Very Hungry Caterpillar
"""},
]
elif mode == "text":
curr_msgs = [{"role": "system", "content": """You are a knowledgeable assistant with access to user queries and chat history. Your task is to revise user queries using the user query and chat history for web search to retrieve relevant information. Below are examples of user queries and optimized responses:
Example 1:
User: "I'm in the mood for a thriller novel. Any recommendations?"
Assistant: "Best thriller novels of all time"
Example 2:
User: "Who directed the movie Inception?"
Assistant: "Director of Inception"
Example 3:
User: "Can you tell me about the cast of The Godfather?"
Assistant: "Cast of The Godfather"
Example 4:
User: "What genre does The Great Gatsby belong to?"
Assistant: "Genre of The Great Gatsby"
Example 5:
User: "Suggest a book for a 5 year old"
Assistant: "Recommended book for a 5 year old"
Please provide brief and concise responses by revising the user queries accordingly. Think through this step by step and come up with the answer."
"""
}]
elif mode == "direct":
curr_msgs = [{"role": "system", "content": """Your task is to provide a random fun fact about children's books or movies. Be concise with the response.
"""
}]
curr_msgs.append({"role": "user", "content": query})
response = client.chat.completions.create(
model="gpt-3.5-turbo-0125",
# response_format={ "type": "json_object" },
messages=curr_msgs
)
return response.choices[0].message.content
else:
curr_msgs = [{"role": "system", "content":"""You are a knowledgeable chat assistant specialized in answering questions related to books, movies, and related topics such as authors, genres, target age groups, summaries, titles, cast, directors, producers, and plot genres. Your responses should be based on the provided chat history and/or context.
Your task is to provide accurate and relevant information to users' queries within the scope of books and movies.Remember to provide accurate and contextually relevant responses based on the user's queries and the information available from previous interactions. Think through this step by step and come up with the answer."""
}
]
n = 5
for msg in st.session_state.messages[-min(len(st.session_state.messages), USE_HISTORY_LEN):-1]:
curr_msgs.append(msg)
curr_msgs.append({"role": "user", "content": query})
print("~"*50)
print(curr_msgs)
print("~"*50)
response = client.chat.completions.create(
model="gpt-3.5-turbo-0125",
# response_format={ "type": "json_object" },
messages=curr_msgs
)
return response.choices[0].message.content
def make_router_call(query: str):
router_answer = openAI_api_call("router", query)
print("="*50)
print(f"Router answer is: {router_answer}")
print("="*50)
return router_answer
def make_context_call(query: str, is_chat=True):
print("="*50)
print(f"get_relevant_context")
print("="*50)
opt_query = openAI_api_call("text", query)
print("="*50)
print(f"opt_query {opt_query}")
print("="*50)
context = fetch_context(opt_query)
print("="*50)
print(f"context {context}")
print("="*50)
updated_query = update_query(opt_query, context)
print("="*50)
print(f"updated_query {updated_query}")
print("="*50)
answer = openAI_api_call("",updated_query)
if is_chat:
chat_hist.append({"role": "user", "content": query})
chat_hist.append({"role": "assistant", "content": answer})
print("@"*50)
print(f"Answer: {answer}")
print("@"*50)
skip = True
return answer
def make_img_search_call(query, answer):
print("="*50)
print(f"get_relevant_images")
print("="*50)
if answer:
opt_query = openAI_api_call("images", query + ", " + answer)
else:
opt_query = openAI_api_call("images", query)
st.session_state.messages.append({"role": "assistant", "content": ""})
print("="*50)
print(f"opt_query: {opt_query}")
print("="*50)
images_urls = fetch_images(opt_query)
print("@"*50)
print(f"Found images: {images_urls}")
print("@"*50)
skip = True
return images_urls
def make_default_call(query):
print("="*50)
print(f"Answering from past")
print("="*50)
opt_query = openAI_api_call("text", query)
print("="*50)
print(f"opt_query: {opt_query}")
print("="*50)
answer = openAI_api_call("",opt_query)
chat_hist.append({"role": "user", "content": query})
chat_hist.append({"role": "assistant", "content": answer})
print("@"*50)
print(f"Answer: {answer}")
print("@"*50)
return answer
if USE_CLI:
while True:
query = input("prompt: ")
router_answer = make_router_call(query)
skip = False
answer = None
if "get_relevant_context" in router_answer:
answer = make_context_call(query)
skip = True
if "get_relevant_images" in router_answer:
images_urls = make_img_search_call(query, answer)
skip = True
if (not skip):
answer = make_default_call(query)
print("!"*50)
print("ONE TURN FINISHED")
print("!"*50)
else:
if "facts" not in st.session_state:
st.session_state.facts = [openAI_api_call("direct", "Give one random fun fact about a childrens book or movie")]
st.set_page_config(page_title="Project BookWorm: Your own Librarian!", layout="centered", initial_sidebar_state="auto", menu_items=None)
st.title("Project BookWorm: Your own Librarian!")
st.markdown(f"""> ###### _{st.session_state.facts[0]}_""")
st.info("Use this app to get recommendations for books and movies")
# Initialize chat history
if "messages" not in st.session_state:
st.session_state.messages = []
# Display chat messages from history on app rerun
for message in st.session_state.messages:
with st.chat_message(message["role"]):
st.markdown(message["content"])
# Accept user input
if query := st.chat_input("What would you like to know today?"):
# Add user message to chat history
# Display user message in chat message container
with st.chat_message("user"):
st.markdown(query)
router_answer = make_router_call(query)
skip = False
answer = None
images_urls = None
if "get_relevant_context" in router_answer:
st.session_state.messages.append({"role": "user", "content": query})
answer = make_context_call(query)
skip = True
st.session_state.messages.append({"role": "assistant", "content": answer})
if "get_relevant_images" in router_answer:
st.session_state.messages.append({"role": "user", "content": query})
images_urls = make_img_search_call(query, answer)
skip = True
if (not skip):
st.session_state.messages.append({"role": "user", "content": query})
answer = make_default_call(query)
st.session_state.messages.append({"role": "assistant", "content": answer})
print("!"*50)
print("ONE TURN FINISHED")
print("!"*50)
# Display assistant response in chat message container
with st.chat_message("assistant"):
# response = st.write_stream(response_generator(answer))
if answer: st.markdown(answer)
if images_urls: display_imgs(images_urls)
# Add assistant response to chat history
|