Update app.py
Browse files
app.py
CHANGED
|
@@ -3,6 +3,7 @@ import cv2
|
|
| 3 |
import Scraper
|
| 4 |
import Manager
|
| 5 |
import ImageProcessor
|
|
|
|
| 6 |
import os
|
| 7 |
import pickle
|
| 8 |
from bs4 import BeautifulSoup
|
|
@@ -32,6 +33,9 @@ if run:
|
|
| 32 |
shutil.rmtree('PDF Cases/Temp')
|
| 33 |
Scraper.download_loc(loc_link)
|
| 34 |
ImageProcessor.process_file('PDF Cases/Temp')
|
|
|
|
|
|
|
|
|
|
| 35 |
# Manager.run('Temp')
|
| 36 |
# file = open("PDF Cases/Temp/processed.pkl",'rb')
|
| 37 |
# C = pickle.load(file)
|
|
|
|
| 3 |
import Scraper
|
| 4 |
import Manager
|
| 5 |
import ImageProcessor
|
| 6 |
+
import Parser
|
| 7 |
import os
|
| 8 |
import pickle
|
| 9 |
from bs4 import BeautifulSoup
|
|
|
|
| 33 |
shutil.rmtree('PDF Cases/Temp')
|
| 34 |
Scraper.download_loc(loc_link)
|
| 35 |
ImageProcessor.process_file('PDF Cases/Temp')
|
| 36 |
+
header_texts, body_texts, footer_texts, paginated_dict = Parser.parse_doc('Temp')
|
| 37 |
+
all_body_text = " ".join(body_texts)
|
| 38 |
+
st.write(all_body_text)
|
| 39 |
# Manager.run('Temp')
|
| 40 |
# file = open("PDF Cases/Temp/processed.pkl",'rb')
|
| 41 |
# C = pickle.load(file)
|