Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -7,60 +7,75 @@ import json
|
|
7 |
import mistune
|
8 |
import pytz
|
9 |
import math
|
|
|
|
|
10 |
from datetime import datetime
|
11 |
from openai import ChatCompletion
|
12 |
from xml.etree import ElementTree as ET
|
13 |
from bs4 import BeautifulSoup
|
14 |
from collections import deque
|
|
|
15 |
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
menu = ["txt", "htm", "md", "py"]
|
22 |
-
choice = st.sidebar.selectbox("Output file type:", menu)
|
23 |
-
choicePrefix = "Output file type is "
|
24 |
-
|
25 |
-
if choice == "txt":
|
26 |
-
st.sidebar.write(choicePrefix + "Text File.")
|
27 |
-
elif choice == "htm":
|
28 |
-
st.sidebar.write(choicePrefix + "HTML5.")
|
29 |
-
elif choice == "md":
|
30 |
-
st.sidebar.write(choicePrefix + "Markdown.")
|
31 |
-
elif choice == "py":
|
32 |
-
st.sidebar.write(choicePrefix + "Python Code.")
|
33 |
-
|
34 |
-
model_choice = st.sidebar.radio("Select Model:", ('gpt-3.5-turbo', 'gpt-3.5-turbo-0301'))
|
35 |
|
36 |
def chat_with_model(prompt, document_section):
|
37 |
model = model_choice
|
38 |
conversation = [{'role': 'system', 'content': 'You are a helpful assistant.'}]
|
39 |
conversation.append({'role': 'user', 'content': prompt})
|
40 |
-
|
|
|
41 |
response = openai.ChatCompletion.create(model=model, messages=conversation)
|
|
|
42 |
return response['choices'][0]['message']['content']
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
43 |
|
44 |
-
def
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
|
|
|
|
|
|
|
|
49 |
|
50 |
def create_file(filename, prompt, response):
|
51 |
if filename.endswith(".txt"):
|
52 |
with open(filename, 'w') as file:
|
53 |
-
file.write(f"
|
54 |
elif filename.endswith(".htm"):
|
55 |
with open(filename, 'w') as file:
|
56 |
-
file.write(f"
|
57 |
elif filename.endswith(".md"):
|
58 |
with open(filename, 'w') as file:
|
59 |
-
file.write(f"
|
60 |
-
|
61 |
def truncate_document(document, length):
|
62 |
return document[:length]
|
63 |
-
|
64 |
def divide_document(document, max_length):
|
65 |
return [document[i:i+max_length] for i in range(0, len(document), max_length)]
|
66 |
|
@@ -72,6 +87,12 @@ def get_table_download_link(file_path):
|
|
72 |
ext = os.path.splitext(file_name)[1] # get the file extension
|
73 |
if ext == '.txt':
|
74 |
mime_type = 'text/plain'
|
|
|
|
|
|
|
|
|
|
|
|
|
75 |
elif ext == '.htm':
|
76 |
mime_type = 'text/html'
|
77 |
elif ext == '.md':
|
@@ -109,15 +130,44 @@ def read_file_content(file,max_length):
|
|
109 |
else:
|
110 |
return ""
|
111 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
112 |
def main():
|
113 |
-
user_prompt = st.text_area("
|
114 |
|
115 |
collength, colupload = st.columns([2,3]) # adjust the ratio as needed
|
116 |
with collength:
|
117 |
#max_length = 12000 - optimal for gpt35 turbo. 2x=24000 for gpt4. 8x=96000 for gpt4-32k.
|
118 |
-
max_length = st.slider("
|
119 |
with colupload:
|
120 |
-
uploaded_file = st.file_uploader("
|
121 |
|
122 |
document_sections = deque()
|
123 |
document_responses = {}
|
@@ -139,8 +189,8 @@ def main():
|
|
139 |
st.markdown(f"**Section {i+1}**\n{document_responses[i]}")
|
140 |
else:
|
141 |
if st.button(f"Chat about Section {i+1}"):
|
142 |
-
st.write('
|
143 |
-
response = chat_with_model(user_prompt, section)
|
144 |
st.write('Response:')
|
145 |
st.write(response)
|
146 |
document_responses[i] = response
|
@@ -149,26 +199,60 @@ def main():
|
|
149 |
st.sidebar.markdown(get_table_download_link(filename), unsafe_allow_html=True)
|
150 |
|
151 |
if st.button('π¬ Chat'):
|
152 |
-
st.write('
|
153 |
-
response = chat_with_model(user_prompt, ''.join(list(document_sections)))
|
154 |
st.write('Response:')
|
155 |
st.write(response)
|
156 |
|
157 |
filename = generate_filename(user_prompt, choice)
|
158 |
create_file(filename, user_prompt, response)
|
159 |
st.sidebar.markdown(get_table_download_link(filename), unsafe_allow_html=True)
|
160 |
-
|
161 |
all_files = glob.glob("*.*")
|
162 |
all_files = [file for file in all_files if len(os.path.splitext(file)[0]) >= 20] # exclude files with short names
|
163 |
-
all_files
|
|
|
|
|
|
|
|
|
164 |
for file in all_files:
|
165 |
-
col1, col3 = st.sidebar.columns([
|
166 |
with col1:
|
|
|
|
|
|
|
|
|
|
|
167 |
st.markdown(get_table_download_link(file), unsafe_allow_html=True)
|
168 |
with col3:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
169 |
if st.button("π", key="delete_"+file):
|
170 |
os.remove(file)
|
171 |
st.experimental_rerun()
|
172 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
173 |
if __name__ == "__main__":
|
174 |
-
main()
|
|
|
7 |
import mistune
|
8 |
import pytz
|
9 |
import math
|
10 |
+
import requests
|
11 |
+
|
12 |
from datetime import datetime
|
13 |
from openai import ChatCompletion
|
14 |
from xml.etree import ElementTree as ET
|
15 |
from bs4 import BeautifulSoup
|
16 |
from collections import deque
|
17 |
+
from audio_recorder_streamlit import audio_recorder
|
18 |
|
19 |
+
def generate_filename(prompt, file_type):
|
20 |
+
central = pytz.timezone('US/Central')
|
21 |
+
safe_date_time = datetime.now(central).strftime("%m%d_%I%M")
|
22 |
+
safe_prompt = "".join(x for x in prompt if x.isalnum())[:45]
|
23 |
+
return f"{safe_date_time}_{safe_prompt}.{file_type}"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
|
25 |
def chat_with_model(prompt, document_section):
|
26 |
model = model_choice
|
27 |
conversation = [{'role': 'system', 'content': 'You are a helpful assistant.'}]
|
28 |
conversation.append({'role': 'user', 'content': prompt})
|
29 |
+
if len(document_section)>0:
|
30 |
+
conversation.append({'role': 'assistant', 'content': document_section})
|
31 |
response = openai.ChatCompletion.create(model=model, messages=conversation)
|
32 |
+
#return response
|
33 |
return response['choices'][0]['message']['content']
|
34 |
+
|
35 |
+
def transcribe_audio(openai_key, file_path, model):
|
36 |
+
OPENAI_API_URL = "https://api.openai.com/v1/audio/transcriptions"
|
37 |
+
headers = {
|
38 |
+
"Authorization": f"Bearer {openai_key}",
|
39 |
+
}
|
40 |
+
with open(file_path, 'rb') as f:
|
41 |
+
data = {'file': f}
|
42 |
+
response = requests.post(OPENAI_API_URL, headers=headers, files=data, data={'model': model})
|
43 |
+
if response.status_code == 200:
|
44 |
+
st.write(response.json())
|
45 |
+
|
46 |
+
response2 = chat_with_model(response.json().get('text'), '') # *************************************
|
47 |
+
st.write('Responses:')
|
48 |
+
#st.write(response)
|
49 |
+
st.write(response2)
|
50 |
+
return response.json().get('text')
|
51 |
+
else:
|
52 |
+
st.write(response.json())
|
53 |
+
st.error("Error in API call.")
|
54 |
+
return None
|
55 |
|
56 |
+
def save_and_play_audio(audio_recorder):
|
57 |
+
audio_bytes = audio_recorder()
|
58 |
+
if audio_bytes:
|
59 |
+
filename = generate_filename("Recording", "wav")
|
60 |
+
with open(filename, 'wb') as f:
|
61 |
+
f.write(audio_bytes)
|
62 |
+
st.audio(audio_bytes, format="audio/wav")
|
63 |
+
return filename
|
64 |
+
return None
|
65 |
|
66 |
def create_file(filename, prompt, response):
|
67 |
if filename.endswith(".txt"):
|
68 |
with open(filename, 'w') as file:
|
69 |
+
file.write(f"{prompt}\n{response}")
|
70 |
elif filename.endswith(".htm"):
|
71 |
with open(filename, 'w') as file:
|
72 |
+
file.write(f"{prompt} {response}")
|
73 |
elif filename.endswith(".md"):
|
74 |
with open(filename, 'w') as file:
|
75 |
+
file.write(f"{prompt}\n\n{response}")
|
76 |
+
|
77 |
def truncate_document(document, length):
|
78 |
return document[:length]
|
|
|
79 |
def divide_document(document, max_length):
|
80 |
return [document[i:i+max_length] for i in range(0, len(document), max_length)]
|
81 |
|
|
|
87 |
ext = os.path.splitext(file_name)[1] # get the file extension
|
88 |
if ext == '.txt':
|
89 |
mime_type = 'text/plain'
|
90 |
+
elif ext == '.py':
|
91 |
+
mime_type = 'text/plain'
|
92 |
+
elif ext == '.xlsx':
|
93 |
+
mime_type = 'text/plain'
|
94 |
+
elif ext == '.csv':
|
95 |
+
mime_type = 'text/plain'
|
96 |
elif ext == '.htm':
|
97 |
mime_type = 'text/html'
|
98 |
elif ext == '.md':
|
|
|
130 |
else:
|
131 |
return ""
|
132 |
|
133 |
+
|
134 |
+
|
135 |
+
def chat_with_file_contents(prompt, file_content):
|
136 |
+
conversation = [{'role': 'system', 'content': 'You are a helpful assistant.'}]
|
137 |
+
conversation.append({'role': 'user', 'content': prompt})
|
138 |
+
if len(file_content)>0:
|
139 |
+
conversation.append({'role': 'assistant', 'content': file_content})
|
140 |
+
response = openai.ChatCompletion.create(model=model_choice, messages=conversation)
|
141 |
+
return response['choices'][0]['message']['content']
|
142 |
+
|
143 |
+
|
144 |
+
# Sidebar and global
|
145 |
+
openai.api_key = os.getenv('OPENAI_KEY')
|
146 |
+
st.set_page_config(page_title="GPT Streamlit Document Reasoner",layout="wide")
|
147 |
+
menu = ["htm", "txt", "xlsx", "csv", "md", "py"] #619
|
148 |
+
choice = st.sidebar.selectbox("Output File Type:", menu)
|
149 |
+
model_choice = st.sidebar.radio("Select Model:", ('gpt-3.5-turbo', 'gpt-3.5-turbo-0301'))
|
150 |
+
|
151 |
+
# Audio, transcribe, GPT:
|
152 |
+
filename = save_and_play_audio(audio_recorder)
|
153 |
+
if filename is not None:
|
154 |
+
transcription = transcribe_audio(openai.api_key, filename, "whisper-1")
|
155 |
+
st.write(transcription)
|
156 |
+
gptOutput = chat_with_model(transcription, '') # *************************************
|
157 |
+
filename = generate_filename(transcription, choice)
|
158 |
+
create_file(filename, transcription, gptOutput)
|
159 |
+
st.sidebar.markdown(get_table_download_link(filename), unsafe_allow_html=True)
|
160 |
+
|
161 |
+
|
162 |
def main():
|
163 |
+
user_prompt = st.text_area("Enter prompts, instructions & questions:", '', height=100)
|
164 |
|
165 |
collength, colupload = st.columns([2,3]) # adjust the ratio as needed
|
166 |
with collength:
|
167 |
#max_length = 12000 - optimal for gpt35 turbo. 2x=24000 for gpt4. 8x=96000 for gpt4-32k.
|
168 |
+
max_length = st.slider("File section length for large files", min_value=1000, max_value=128000, value=12000, step=1000)
|
169 |
with colupload:
|
170 |
+
uploaded_file = st.file_uploader("Add a file for context:", type=["xml", "json", "xlsx","csv","html", "htm", "md", "txt"])
|
171 |
|
172 |
document_sections = deque()
|
173 |
document_responses = {}
|
|
|
189 |
st.markdown(f"**Section {i+1}**\n{document_responses[i]}")
|
190 |
else:
|
191 |
if st.button(f"Chat about Section {i+1}"):
|
192 |
+
st.write('Reasoning with your inputs...')
|
193 |
+
response = chat_with_model(user_prompt, section) # *************************************
|
194 |
st.write('Response:')
|
195 |
st.write(response)
|
196 |
document_responses[i] = response
|
|
|
199 |
st.sidebar.markdown(get_table_download_link(filename), unsafe_allow_html=True)
|
200 |
|
201 |
if st.button('π¬ Chat'):
|
202 |
+
st.write('Reasoning with your inputs...')
|
203 |
+
response = chat_with_model(user_prompt, ''.join(list(document_sections))) # *************************************
|
204 |
st.write('Response:')
|
205 |
st.write(response)
|
206 |
|
207 |
filename = generate_filename(user_prompt, choice)
|
208 |
create_file(filename, user_prompt, response)
|
209 |
st.sidebar.markdown(get_table_download_link(filename), unsafe_allow_html=True)
|
210 |
+
|
211 |
all_files = glob.glob("*.*")
|
212 |
all_files = [file for file in all_files if len(os.path.splitext(file)[0]) >= 20] # exclude files with short names
|
213 |
+
all_files.sort(key=lambda x: (os.path.splitext(x)[1], x), reverse=True) # sort by file type and file name in descending order
|
214 |
+
|
215 |
+
# sidebar of files
|
216 |
+
file_contents=''
|
217 |
+
next_action=''
|
218 |
for file in all_files:
|
219 |
+
col1, col2, col3, col4, col5 = st.sidebar.columns([1,6,1,1,1]) # adjust the ratio as needed
|
220 |
with col1:
|
221 |
+
if st.button("π", key="md_"+file): # md emoji button
|
222 |
+
with open(file, 'r') as f:
|
223 |
+
file_contents = f.read()
|
224 |
+
next_action='md'
|
225 |
+
with col2:
|
226 |
st.markdown(get_table_download_link(file), unsafe_allow_html=True)
|
227 |
with col3:
|
228 |
+
if st.button("π", key="open_"+file): # open emoji button
|
229 |
+
with open(file, 'r') as f:
|
230 |
+
file_contents = f.read()
|
231 |
+
next_action='open'
|
232 |
+
with col4:
|
233 |
+
if st.button("π", key="read_"+file): # search emoji button
|
234 |
+
with open(file, 'r') as f:
|
235 |
+
file_contents = f.read()
|
236 |
+
next_action='search'
|
237 |
+
with col5:
|
238 |
if st.button("π", key="delete_"+file):
|
239 |
os.remove(file)
|
240 |
st.experimental_rerun()
|
241 |
+
|
242 |
+
if len(file_contents) > 0:
|
243 |
+
if next_action=='open':
|
244 |
+
file_content_area = st.text_area("File Contents:", file_contents, height=500)
|
245 |
+
if next_action=='md':
|
246 |
+
st.markdown(file_contents)
|
247 |
+
if next_action=='search':
|
248 |
+
file_content_area = st.text_area("File Contents:", file_contents, height=500)
|
249 |
+
st.write('Reasoning with your inputs...')
|
250 |
+
response = chat_with_file_contents(file_contents)
|
251 |
+
st.write('Response:')
|
252 |
+
st.write(response)
|
253 |
+
filename = generate_filename(file_content_area, choice)
|
254 |
+
create_file(filename, file_content_area, response)
|
255 |
+
st.sidebar.markdown(get_table_download_link(filename), unsafe_allow_html=True)
|
256 |
+
|
257 |
if __name__ == "__main__":
|
258 |
+
main()
|