Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -16,13 +16,6 @@ from bs4 import BeautifulSoup
|
|
16 |
from collections import deque
|
17 |
from audio_recorder_streamlit import audio_recorder
|
18 |
|
19 |
-
openai.api_key = os.getenv('OPENAI_KEY')
|
20 |
-
st.set_page_config(page_title="GPT Streamlit Document Reasoner",layout="wide")
|
21 |
-
|
22 |
-
menu = ["txt", "htm", "md", "py"]
|
23 |
-
choice = st.sidebar.selectbox("Output File Type:", menu)
|
24 |
-
model_choice = st.sidebar.radio("Select Model:", ('gpt-3.5-turbo', 'gpt-3.5-turbo-0301'))
|
25 |
-
|
26 |
def generate_filename(prompt, file_type):
|
27 |
central = pytz.timezone('US/Central')
|
28 |
safe_date_time = datetime.now(central).strftime("%m%d_%I%M")
|
@@ -33,10 +26,12 @@ def chat_with_model(prompt, document_section):
|
|
33 |
model = model_choice
|
34 |
conversation = [{'role': 'system', 'content': 'You are a helpful assistant.'}]
|
35 |
conversation.append({'role': 'user', 'content': prompt})
|
36 |
-
|
|
|
37 |
response = openai.ChatCompletion.create(model=model, messages=conversation)
|
|
|
38 |
return response['choices'][0]['message']['content']
|
39 |
-
|
40 |
def transcribe_audio(openai_key, file_path, model):
|
41 |
OPENAI_API_URL = "https://api.openai.com/v1/audio/transcriptions"
|
42 |
headers = {
|
@@ -47,7 +42,8 @@ def transcribe_audio(openai_key, file_path, model):
|
|
47 |
response = requests.post(OPENAI_API_URL, headers=headers, files=data, data={'model': model})
|
48 |
if response.status_code == 200:
|
49 |
st.write(response.json())
|
50 |
-
|
|
|
51 |
st.write('Responses:')
|
52 |
#st.write(response)
|
53 |
st.write(response2)
|
@@ -67,27 +63,19 @@ def save_and_play_audio(audio_recorder):
|
|
67 |
return filename
|
68 |
return None
|
69 |
|
70 |
-
filename = save_and_play_audio(audio_recorder)
|
71 |
-
if filename is not None:
|
72 |
-
if st.button("Transcribe"):
|
73 |
-
transcription = transcribe_audio(openai.api_key, filename, "whisper-1")
|
74 |
-
st.write(transcription)
|
75 |
-
chat_with_model(transcription, '') # push transcript through as prompt
|
76 |
-
|
77 |
def create_file(filename, prompt, response):
|
78 |
if filename.endswith(".txt"):
|
79 |
with open(filename, 'w') as file:
|
80 |
-
file.write(f"
|
81 |
elif filename.endswith(".htm"):
|
82 |
with open(filename, 'w') as file:
|
83 |
-
file.write(f"
|
84 |
elif filename.endswith(".md"):
|
85 |
with open(filename, 'w') as file:
|
86 |
-
file.write(f"
|
87 |
-
|
88 |
def truncate_document(document, length):
|
89 |
return document[:length]
|
90 |
-
|
91 |
def divide_document(document, max_length):
|
92 |
return [document[i:i+max_length] for i in range(0, len(document), max_length)]
|
93 |
|
@@ -99,6 +87,12 @@ def get_table_download_link(file_path):
|
|
99 |
ext = os.path.splitext(file_name)[1] # get the file extension
|
100 |
if ext == '.txt':
|
101 |
mime_type = 'text/plain'
|
|
|
|
|
|
|
|
|
|
|
|
|
102 |
elif ext == '.htm':
|
103 |
mime_type = 'text/html'
|
104 |
elif ext == '.md':
|
@@ -136,6 +130,35 @@ def read_file_content(file,max_length):
|
|
136 |
else:
|
137 |
return ""
|
138 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
139 |
def main():
|
140 |
user_prompt = st.text_area("Enter prompts, instructions & questions:", '', height=100)
|
141 |
|
@@ -144,7 +167,7 @@ def main():
|
|
144 |
#max_length = 12000 - optimal for gpt35 turbo. 2x=24000 for gpt4. 8x=96000 for gpt4-32k.
|
145 |
max_length = st.slider("File section length for large files", min_value=1000, max_value=128000, value=12000, step=1000)
|
146 |
with colupload:
|
147 |
-
uploaded_file = st.file_uploader("Add a file for context:", type=["xml", "json", "html", "htm", "md", "txt"])
|
148 |
|
149 |
document_sections = deque()
|
150 |
document_responses = {}
|
@@ -167,7 +190,7 @@ def main():
|
|
167 |
else:
|
168 |
if st.button(f"Chat about Section {i+1}"):
|
169 |
st.write('Reasoning with your inputs...')
|
170 |
-
response = chat_with_model(user_prompt, section)
|
171 |
st.write('Response:')
|
172 |
st.write(response)
|
173 |
document_responses[i] = response
|
@@ -177,7 +200,7 @@ def main():
|
|
177 |
|
178 |
if st.button('π¬ Chat'):
|
179 |
st.write('Reasoning with your inputs...')
|
180 |
-
response = chat_with_model(user_prompt, ''.join(list(document_sections)))
|
181 |
st.write('Response:')
|
182 |
st.write(response)
|
183 |
|
@@ -188,16 +211,48 @@ def main():
|
|
188 |
all_files = glob.glob("*.*")
|
189 |
all_files = [file for file in all_files if len(os.path.splitext(file)[0]) >= 20] # exclude files with short names
|
190 |
all_files.sort(key=lambda x: (os.path.splitext(x)[1], x), reverse=True) # sort by file type and file name in descending order
|
191 |
-
|
|
|
|
|
|
|
192 |
for file in all_files:
|
193 |
-
col1, col3 = st.sidebar.columns([
|
194 |
with col1:
|
|
|
|
|
|
|
|
|
|
|
195 |
st.markdown(get_table_download_link(file), unsafe_allow_html=True)
|
196 |
with col3:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
197 |
if st.button("π", key="delete_"+file):
|
198 |
os.remove(file)
|
199 |
st.experimental_rerun()
|
200 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
201 |
if __name__ == "__main__":
|
202 |
-
main()
|
203 |
-
|
|
|
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")
|
|
|
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 = {
|
|
|
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)
|
|
|
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 |
|
|
|
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 = {}
|
|
|
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
|
|
|
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 |
|
|
|
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(user_prompt, 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()
|
|