Mdean77 commited on
Commit
004e22c
·
1 Parent(s): f08bee6

Updated ignore file and corrected error on start and stop pages in app.py

Browse files
Files changed (2) hide show
  1. .gitignore +4 -4
  2. app.py +6 -0
.gitignore CHANGED
@@ -4,7 +4,7 @@
4
  appOLD.py
5
  Clinical_Trial_Project.pages
6
  testing.ipynb
7
- .chainlit/
8
- .files/
9
- __pycache__/
10
- documents/
 
4
  appOLD.py
5
  Clinical_Trial_Project.pages
6
  testing.ipynb
7
+ .chainlit
8
+ .files
9
+ __pycache__
10
+ documents
app.py CHANGED
@@ -30,17 +30,23 @@ async def on_chat_start():
30
  doc = pymupdf.Document(file.path)
31
  toc = doc.get_toc()
32
  # Want to find the List Of Figures page because that is the last page I want to skip
 
 
33
  for _, title, page in toc:
34
  if title == "List of Figures":
35
  print(f"{title} on page {page}")
36
  start_page = page + 1
37
 
 
38
  # get the last page I want included
 
 
39
  for _, title, page in toc:
40
  if ("References" in title) or ("Bibliography" in title):
41
  print(f"{title} on page {page}")
42
  end_page = page
43
 
 
44
  print(f"Extraction should start on page {start_page} and end on page {end_page}")
45
 
46
 
 
30
  doc = pymupdf.Document(file.path)
31
  toc = doc.get_toc()
32
  # Want to find the List Of Figures page because that is the last page I want to skip
33
+ # Default is 1 if I do not find better start location
34
+ start_page = 1
35
  for _, title, page in toc:
36
  if title == "List of Figures":
37
  print(f"{title} on page {page}")
38
  start_page = page + 1
39
 
40
+
41
  # get the last page I want included
42
+ # default is last page of document
43
+ end_page = len(doc)
44
  for _, title, page in toc:
45
  if ("References" in title) or ("Bibliography" in title):
46
  print(f"{title} on page {page}")
47
  end_page = page
48
 
49
+
50
  print(f"Extraction should start on page {start_page} and end on page {end_page}")
51
 
52