Update app.py
Browse files
app.py
CHANGED
@@ -188,6 +188,9 @@ def get_message_history():
|
|
188 |
docs = load_data()
|
189 |
qa_chain = load_model(docs)
|
190 |
|
|
|
|
|
|
|
191 |
print('2')
|
192 |
print(time.time())
|
193 |
if prompt := st.chat_input("How can I help you today?"):
|
@@ -206,9 +209,7 @@ if prompt := st.chat_input("How can I help you today?"):
|
|
206 |
result = qa_chain(prompt)
|
207 |
print('4')
|
208 |
print(time.time())
|
209 |
-
|
210 |
-
tts = gTTS(result['result'], lang='en')
|
211 |
-
tts.write_to_fp(sound_file)
|
212 |
output = [result['result']]
|
213 |
|
214 |
for item in output:
|
@@ -220,116 +221,60 @@ if prompt := st.chat_input("How can I help you today?"):
|
|
220 |
|
221 |
print('5')
|
222 |
print(time.time())
|
223 |
-
### READ IN PDF
|
224 |
-
page_number = int(result['source_documents'][0].metadata['page'])
|
225 |
-
doc = fitz.open(str(result['source_documents'][0].metadata['source']))
|
226 |
-
|
227 |
-
text = str(result['source_documents'][0].page_content)
|
228 |
-
if text != '':
|
229 |
-
for page in doc:
|
230 |
-
### SEARCH
|
231 |
-
text_instances = page.search_for(text)
|
232 |
-
|
233 |
-
### HIGHLIGHT
|
234 |
-
for inst in text_instances:
|
235 |
-
highlight = page.add_highlight_annot(inst)
|
236 |
-
highlight.update()
|
237 |
-
|
238 |
-
### OUTPUT
|
239 |
-
doc.save("/home/user/app/pdf2image/output.pdf", garbage=4, deflate=True, clean=True)
|
240 |
-
|
241 |
-
# pdf_to_open = repr(result['source_documents'][0].metadata['source'])
|
242 |
-
|
243 |
-
def pdf_page_to_image(pdf_file, page_number, output_image):
|
244 |
-
# Open the PDF file
|
245 |
-
pdf_document = fitz.open(pdf_file)
|
246 |
-
|
247 |
-
# Get the specific page
|
248 |
-
page = pdf_document[page_number]
|
249 |
-
|
250 |
-
# Define the image DPI (dots per inch)
|
251 |
-
dpi = 300 # You can adjust this as needed
|
252 |
-
|
253 |
-
# Convert the page to an image
|
254 |
-
pix = page.get_pixmap(matrix=fitz.Matrix(dpi / 100, dpi / 100))
|
255 |
-
|
256 |
-
# Save the image as a PNG file
|
257 |
-
pix.save(output_image, "png")
|
258 |
-
|
259 |
-
# Close the PDF file
|
260 |
-
pdf_document.close()
|
261 |
-
|
262 |
-
|
263 |
-
pdf_page_to_image('/home/user/app/pdf2image/output.pdf', page_number, '/home/user/app/pdf2image/output.png')
|
264 |
|
265 |
-
|
266 |
-
|
267 |
-
|
268 |
-
|
269 |
-
# if 'clickedR' not in st.session_state:
|
270 |
-
# st.session_state.clickedR = False
|
271 |
-
|
272 |
-
# def click_buttonR():
|
273 |
-
# st.session_state.clickedR = True
|
274 |
-
# if st.session_state.clickedR:
|
275 |
-
# message_placeholder.markdown(full_response+repr(result['source_documents'][0]))
|
276 |
-
|
277 |
-
# ref = st.button('References', on_click = click_buttonR)
|
278 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
279 |
|
280 |
-
|
281 |
-
|
282 |
-
|
283 |
-
|
284 |
-
|
285 |
-
|
286 |
-
|
287 |
-
|
288 |
-
|
289 |
-
|
290 |
-
|
291 |
-
|
292 |
-
|
293 |
-
|
294 |
-
|
295 |
-
|
296 |
-
|
297 |
-
|
298 |
-
#st.session_state.clickedR = False
|
299 |
-
|
300 |
-
# #if ref:
|
301 |
-
# message_placeholder.markdown(full_response+repr(result['source_documents'][0]))
|
302 |
-
# #if sound:
|
303 |
-
# sound_file = BytesIO()
|
304 |
-
# tts = gTTS(result['result'], lang='en')
|
305 |
-
# tts.write_to_fp(sound_file)
|
306 |
-
# html_string = """
|
307 |
-
# <audio controls autoplay>
|
308 |
-
# <source src="/content/sound_file" type="audio/wav">
|
309 |
-
# </audio>
|
310 |
-
# """
|
311 |
-
# message_placeholder.markdown(html_string, unsafe_allow_html=True) # will display a st.audio with the sound you specified in the "src" of the html_string and autoplay it
|
312 |
-
# #time.sleep(5) # wait for 2 seconds to finish the playing of the audio
|
313 |
-
#response_sentiment = st.radio(
|
314 |
-
# "How was the Assistant's response?",
|
315 |
-
# ["😁", "😕", "😢"],
|
316 |
-
# key="response_sentiment",
|
317 |
-
# disabled=st.session_state.disabled,
|
318 |
-
# horizontal=True,
|
319 |
-
# index=1,
|
320 |
-
# help="This helps us improve the model.",
|
321 |
-
# # hide the radio button on click
|
322 |
-
# on_change=on_select(),
|
323 |
-
#)
|
324 |
-
#logger.info(f"{user_session_id} | {full_response} | {response_sentiment}")
|
325 |
-
|
326 |
-
# # Logging to FastAPI Endpoint
|
327 |
-
# headers = {"Authorization": f"Bearer {secret_token}"}
|
328 |
-
# log_data = {"log": f"{user_session_id} | {full_response} | {response_sentiment}"}
|
329 |
-
# response = requests.post(fastapi_endpoint, json=log_data, headers=headers, timeout=10)
|
330 |
-
# if response.status_code == 200:
|
331 |
-
# logger.info("Query logged successfully")
|
332 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
333 |
st.session_state.messages.append({"role": "assistant", "content": full_response})
|
334 |
|
335 |
|
|
|
188 |
docs = load_data()
|
189 |
qa_chain = load_model(docs)
|
190 |
|
191 |
+
|
192 |
+
|
193 |
+
|
194 |
print('2')
|
195 |
print(time.time())
|
196 |
if prompt := st.chat_input("How can I help you today?"):
|
|
|
209 |
result = qa_chain(prompt)
|
210 |
print('4')
|
211 |
print(time.time())
|
212 |
+
|
|
|
|
|
213 |
output = [result['result']]
|
214 |
|
215 |
for item in output:
|
|
|
221 |
|
222 |
print('5')
|
223 |
print(time.time())
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
224 |
|
225 |
+
def generate_pdf():
|
226 |
+
page_number = int(result['source_documents'][0].metadata['page'])
|
227 |
+
doc = fitz.open(str(result['source_documents'][0].metadata['source']))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
228 |
|
229 |
+
text = str(result['source_documents'][0].page_content)
|
230 |
+
if text != '':
|
231 |
+
for page in doc:
|
232 |
+
### SEARCH
|
233 |
+
text_instances = page.search_for(text)
|
234 |
+
|
235 |
+
### HIGHLIGHT
|
236 |
+
for inst in text_instances:
|
237 |
+
highlight = page.add_highlight_annot(inst)
|
238 |
+
highlight.update()
|
239 |
+
|
240 |
+
### OUTPUT
|
241 |
+
doc.save("/home/user/app/pdf2image/output.pdf", garbage=4, deflate=True, clean=True)
|
242 |
+
|
243 |
+
# pdf_to_open = repr(result['source_documents'][0].metadata['source'])
|
244 |
+
|
245 |
+
def pdf_page_to_image(pdf_file, page_number, output_image):
|
246 |
+
# Open the PDF file
|
247 |
+
pdf_document = fitz.open(pdf_file)
|
248 |
+
|
249 |
+
# Get the specific page
|
250 |
+
page = pdf_document[page_number]
|
251 |
+
|
252 |
+
# Define the image DPI (dots per inch)
|
253 |
+
dpi = 300 # You can adjust this as needed
|
254 |
|
255 |
+
# Convert the page to an image
|
256 |
+
pix = page.get_pixmap(matrix=fitz.Matrix(dpi / 100, dpi / 100))
|
257 |
+
|
258 |
+
# Save the image as a PNG file
|
259 |
+
pix.save(output_image, "png")
|
260 |
+
|
261 |
+
# Close the PDF file
|
262 |
+
pdf_document.close()
|
263 |
+
|
264 |
+
|
265 |
+
pdf_page_to_image('/home/user/app/pdf2image/output.pdf', page_number, '/home/user/app/pdf2image/output.png')
|
266 |
+
|
267 |
+
image = Image.open('/home/user/app/pdf2image/output.png')
|
268 |
+
st.image(image)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
269 |
|
270 |
+
def generate_audio():
|
271 |
+
sound_file = BytesIO()
|
272 |
+
tts = gTTS(result['result'], lang='en')
|
273 |
+
tts.write_to_fp(sound_file)
|
274 |
+
st.audio(sound_file)
|
275 |
+
|
276 |
+
st.button(':speaker:', type='primary',on_click=generate_audio)
|
277 |
+
st.button('Reference',type='primary',on_click=generate_pdf)
|
278 |
st.session_state.messages.append({"role": "assistant", "content": full_response})
|
279 |
|
280 |
|