Spaces:
Running
Running
adibak
commited on
Commit
·
b4b5aad
1
Parent(s):
8982261
added page range slider by modifying get_pdf_content and app.py
Browse files- app.py +10 -1
- helpers/file_manager.py +11 -4
app.py
CHANGED
|
@@ -222,6 +222,12 @@ with st.sidebar:
|
|
| 222 |
value='2024-05-01-preview',
|
| 223 |
)
|
| 224 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 225 |
|
| 226 |
def build_ui():
|
| 227 |
"""
|
|
@@ -284,7 +290,10 @@ def set_up_chat_ui():
|
|
| 284 |
if prompt['files']:
|
| 285 |
# Apparently, Streamlit stores uploaded files in memory and clears on browser close
|
| 286 |
# https://docs.streamlit.io/knowledge-base/using-streamlit/where-file-uploader-store-when-deleted
|
| 287 |
-
|
|
|
|
|
|
|
|
|
|
| 288 |
print(f'{prompt["files"]=}')
|
| 289 |
|
| 290 |
provider, llm_name = llm_helper.get_provider_model(
|
|
|
|
| 222 |
value='2024-05-01-preview',
|
| 223 |
)
|
| 224 |
|
| 225 |
+
page_range_slider = st.slider(label=('4: Specify a page range to examine:\n\n'
|
| 226 |
+
'(min=1, max=50)'),
|
| 227 |
+
min_value=1, max_value=50,
|
| 228 |
+
value=(1, 50))
|
| 229 |
+
st.session_state['page_range'] = page_range_slider
|
| 230 |
+
|
| 231 |
|
| 232 |
def build_ui():
|
| 233 |
"""
|
|
|
|
| 290 |
if prompt['files']:
|
| 291 |
# Apparently, Streamlit stores uploaded files in memory and clears on browser close
|
| 292 |
# https://docs.streamlit.io/knowledge-base/using-streamlit/where-file-uploader-store-when-deleted
|
| 293 |
+
page_range = st.session_state.get('page_range', (1, 50)) # fallback default
|
| 294 |
+
st.session_state[ADDITIONAL_INFO] = filem.get_pdf_contents(
|
| 295 |
+
prompt['files'][0], page_range
|
| 296 |
+
)
|
| 297 |
print(f'{prompt["files"]=}')
|
| 298 |
|
| 299 |
provider, llm_name = llm_helper.get_provider_model(
|
helpers/file_manager.py
CHANGED
|
@@ -19,6 +19,7 @@ logger = logging.getLogger(__name__)
|
|
| 19 |
|
| 20 |
def get_pdf_contents(
|
| 21 |
pdf_file: st.runtime.uploaded_file_manager.UploadedFile,
|
|
|
|
| 22 |
max_pages: int = GlobalConfig.MAX_PAGE_COUNT
|
| 23 |
) -> str:
|
| 24 |
"""
|
|
@@ -30,11 +31,17 @@ def get_pdf_contents(
|
|
| 30 |
"""
|
| 31 |
|
| 32 |
reader = PdfReader(pdf_file)
|
| 33 |
-
n_pages = min(max_pages, len(reader.pages))
|
| 34 |
-
text = ''
|
| 35 |
|
| 36 |
-
|
| 37 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 38 |
text += page.extract_text()
|
| 39 |
|
| 40 |
return text
|
|
|
|
| 19 |
|
| 20 |
def get_pdf_contents(
|
| 21 |
pdf_file: st.runtime.uploaded_file_manager.UploadedFile,
|
| 22 |
+
page_range: tuple[int, int],
|
| 23 |
max_pages: int = GlobalConfig.MAX_PAGE_COUNT
|
| 24 |
) -> str:
|
| 25 |
"""
|
|
|
|
| 31 |
"""
|
| 32 |
|
| 33 |
reader = PdfReader(pdf_file)
|
|
|
|
|
|
|
| 34 |
|
| 35 |
+
total_pages = len(reader.pages)
|
| 36 |
+
n_pages = min(max_pages, total_pages)
|
| 37 |
+
|
| 38 |
+
start, end = page_range
|
| 39 |
+
start = max(1, start)
|
| 40 |
+
end = min(n_pages, end)
|
| 41 |
+
|
| 42 |
+
text = ''
|
| 43 |
+
for page_num in range(start - 1, end):
|
| 44 |
+
page = reader.pages[page_num]
|
| 45 |
text += page.extract_text()
|
| 46 |
|
| 47 |
return text
|