|
@@ -5,7 +5,8 @@ import unittest
|
|
# import pyautogui as ag
|
|
# import pyautogui as ag
|
|
import requests
|
|
import requests
|
|
import fitz
|
|
import fitz
|
|
-import PyPDF2
|
|
|
|
|
|
+import pypdf
|
|
|
|
+from lxml import etree
|
|
# import PyPDF225
|
|
# import PyPDF225
|
|
|
|
|
|
from receipt import Receipts
|
|
from receipt import Receipts
|
|
@@ -13,18 +14,25 @@ from receipt import Receipts
|
|
class Test(unittest.TestCase):
|
|
class Test(unittest.TestCase):
|
|
receipt = Receipts()
|
|
receipt = Receipts()
|
|
|
|
|
|
|
|
+ def test_xml(self):
|
|
|
|
+ xml = etree.parse('/home/deeejas/git/pyanaf/sample_files/D300_21480386_2021_02.xml')
|
|
|
|
+ root = xml.getroot()
|
|
|
|
+ print(root)
|
|
|
|
+ tree = xml.xpath('/xmlns:declaratie300/*', namespaces={'xmlns': 'mfp:anaf:dgti:d300:declaratie:v6'})
|
|
|
|
+ print(tree)
|
|
|
|
+
|
|
def test_pdfread(self):
|
|
def test_pdfread(self):
|
|
self.receipt.pdf_reader('sample_files/202550755_17259191_D394_1_2020.pdf')
|
|
self.receipt.pdf_reader('sample_files/202550755_17259191_D394_1_2020.pdf')
|
|
print(self.receipt.mesaj.strip())
|
|
print(self.receipt.mesaj.strip())
|
|
|
|
|
|
def test_pypdf_read(self):
|
|
def test_pypdf_read(self):
|
|
- pdf = PyPDF2.PdfFileReader('sample_files/202550755_17259191_D394_1_2020.pdf')
|
|
|
|
|
|
+ pdf = pypdf.PdfReader('sample_files/202550755_17259191_D394_1_2020.pdf')
|
|
pdf_page = pdf.getPage(0)
|
|
pdf_page = pdf.getPage(0)
|
|
text = pdf_page.extractText()
|
|
text = pdf_page.extractText()
|
|
print(text)
|
|
print(text)
|
|
|
|
|
|
def test_pypdf2_read(self):
|
|
def test_pypdf2_read(self):
|
|
- pdf = PyPDF2.PdfFileReader('sample_files/202550755_17259191_D394_1_2020.pdf')
|
|
|
|
|
|
+ pdf = pypdf.PdfReader('sample_files/202550755_17259191_D394_1_2020.pdf')
|
|
pdf_page = pdf.pages[0]
|
|
pdf_page = pdf.pages[0]
|
|
text = pdf_page.extract_text()
|
|
text = pdf_page.extract_text()
|
|
print(text)
|
|
print(text)
|