Create file_processing.py
Browse files- file_processing.py +17 -0
file_processing.py
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import PyPDF2
|
| 2 |
+
|
| 3 |
+
def extract_text_from_pdf(pdf_file):
|
| 4 |
+
pdf_reader = PyPDF2.PdfFileReader(pdf_file)
|
| 5 |
+
num_pages = pdf_reader.numPages
|
| 6 |
+
|
| 7 |
+
text = ""
|
| 8 |
+
for page_num in range(num_pages):
|
| 9 |
+
page = pdf_reader.getPage(page_num)
|
| 10 |
+
text += page.extract_text()
|
| 11 |
+
|
| 12 |
+
return text
|
| 13 |
+
|
| 14 |
+
def process_text_file(text_file):
|
| 15 |
+
text_content = text_file.read().decode('utf-8')
|
| 16 |
+
# Führe hier die Verarbeitung des Textdokuments durch und gib das Ergebnis zurück
|
| 17 |
+
return text_content
|