test_receipt.py 1.4 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152
  1. # coding: UTF-8
  2. import urllib.request
  3. import unittest
  4. # import pyautogui as ag
  5. import requests
  6. import fitz
  7. import pypdf
  8. from lxml import etree
  9. # import PyPDF225
  10. from receipt import Receipts
  11. class Test(unittest.TestCase):
  12. receipt = Receipts()
  13. def test_xml(self):
  14. xml = etree.parse('/home/deeejas/git/pyanaf/sample_files/D300_21480386_2021_02.xml')
  15. root = xml.getroot()
  16. print(root)
  17. tree = xml.xpath('/xmlns:declaratie300/*', namespaces={'xmlns': 'mfp:anaf:dgti:d300:declaratie:v6'})
  18. print(tree)
  19. def test_pdfread(self):
  20. self.receipt.pdf_reader('sample_files/202550755_17259191_D394_1_2020.pdf')
  21. print(self.receipt.mesaj.strip())
  22. def test_pypdf_read(self):
  23. pdf = pypdf.PdfReader('sample_files/202550755_17259191_D394_1_2020.pdf')
  24. pdf_page = pdf.getPage(0)
  25. text = pdf_page.extractText()
  26. print(text)
  27. def test_pypdf2_read(self):
  28. pdf = pypdf.PdfReader('sample_files/202550755_17259191_D394_1_2020.pdf')
  29. pdf_page = pdf.pages[0]
  30. text = pdf_page.extract_text()
  31. print(text)
  32. def test_pymupdf_read(self):
  33. pdf = fitz.Document('sample_files/202550755_17259191_D394_1_2020.pdf')
  34. pdf_page = pdf.load_page(0)
  35. text = pdf_page.get_text('text')
  36. print(text)
  37. def test_dir(self):
  38. receipt = Receipts()
  39. print(receipt._dir_fdate())
  40. if __name__ == '__main__':
  41. unittest.main()