Compare commits

..

3 Commits

Author SHA1 Message Date
Niklas Müller f4b8029593 added tests for function call with location 2024-01-05 17:43:28 +01:00
Niklas Müller 376fafa171 added logging
added split_pdf and merge_pdf with location parameter
2024-01-05 17:43:01 +01:00
Niklas Müller 6cb54c31ba Chnage logging output 2024-01-05 17:40:16 +01:00
5 changed files with 219 additions and 18 deletions

2
app.py
View File

@ -14,7 +14,7 @@ logging.basicConfig(
# level=logging.ERROR,
# level=logging.INFO,
level=logging.DEBUG,
format="Start: " + str(dt.datetime.now()).replace(" ", "_") + " | %(asctime)s [%(levelname)s] %(message)s",
format=str(dt.datetime.now()).replace(" ", "_") + " | %(asctime)s [%(levelname)s] %(message)s",
handlers=[
logging.FileHandler("/var/log/" + str(dt.datetime.today().strftime('%Y-%m-%d')) + "_-_cron.log"),
logging.StreamHandler(sys.stdout)

View File

@ -1,9 +1,48 @@
from pypdf import PdfReader, PdfWriter
import uuid
import os
import shutil
import datetime as dt
import logging
import sys
from pdf_util.pdf_util import pdf_util
# Setup Logging
logging.basicConfig(
# level=logging.ERROR,
# level=logging.INFO,
level=logging.DEBUG,
format=str(dt.datetime.now()).replace(" ", "_") + " | %(asctime)s [%(levelname)s] %(message)s",
handlers=[
logging.FileHandler("/var/log/" + str(dt.datetime.today().strftime('%Y-%m-%d')) + "_-_pdf_project_manager.log"),
logging.StreamHandler(sys.stdout)
]
)
base_path = "/app/projects/"
class pdf_project_manager:
def __init__(self):
self.uuid = str(uuid.uuid4())
os.makedirs("/app/projects/" + self.uuid, exist_ok=True)
self.pdf_init = False
self.project_name = ""
os.makedirs(base_path + self.uuid, exist_ok=True)
"""
def add_pdf(self, pdf_path):
if self.pdf_init:
shutil.copyfile(pdf_path, base_path + self.uuid + "/complete.pdf")
else:
shutil.copyfile(pdf_path, base_path + self.uuid + "/tmp.pdf")
pdf_util(base_path + self.uuid + "/complete.pdf").merge_pdf_with_and_location(base_path + self.uuid + "/tmp.pdf", base_path + self.uuid + "/tmp_complete.pdf")
shutil.copyfile(base_path + self.uuid + "/tmp_complete.pdf", base_path + self.uuid + "/complete.pdf")
os.remove(base_path + self.uuid + "/tmp_complete.pdf")
os.remove(base_path + self.uuid + "/tmp.pdf")
# Splitt files in all single Pages in Subdirectory...
"""

View File

@ -1,6 +1,23 @@
import os
from pypdf import PdfReader, PdfWriter
import datetime as dt
import logging
import sys
# Setup Logging
logging.basicConfig(
# level=logging.ERROR,
# level=logging.INFO,
level=logging.DEBUG,
format=str(dt.datetime.now()).replace(" ", "_") + " | %(asctime)s [%(levelname)s] %(message)s",
handlers=[
logging.FileHandler("/var/log/" + str(dt.datetime.today().strftime('%Y-%m-%d')) + "_-_pdf_util.log"),
logging.StreamHandler(sys.stdout)
]
)
class pdf_util:
def __init__(self, file_path):
self.file_path = file_path
@ -8,9 +25,9 @@ class pdf_util:
self.file_name_wo_extension = os.path.splitext(os.path.basename(file_path))[0]
def split_pdf(self):
def split_pdf_with_location(self, output_filepath, no_names=False):
out_filenames = []
os.makedirs(os.path.dirname(self.file_path) + "/split_pdf", exist_ok=True)
os.makedirs(os.path.dirname(output_filepath), exist_ok=True)
with open(self.file_path, 'rb') as pdf_file:
pdf_reader = PdfReader(pdf_file)
num_pages = len(pdf_reader.pages)
@ -19,16 +36,25 @@ class pdf_util:
writer = PdfWriter()
writer.add_page(pdf_reader.pages[page_num])
out_filename = os.path.dirname(self.file_path) + '/split_pdf/' + self.file_name_wo_extension + '_' + str(page_num + 1) + '.pdf'
if no_names:
out_filename = os.path.dirname(output_filepath) + '/' + str(page_num + 1) + '.pdf'
else:
out_filename = os.path.dirname(output_filepath) + '/' + self.file_name_wo_extension + '_' + str(page_num + 1) + '.pdf'
with open(out_filename, 'wb') as outfile:
writer.write(outfile)
out_filenames.append(out_filename)
return out_filenames
# Deprecate when pdf_project_manager takes effect
def split_pdf(self):
os.makedirs(os.path.dirname(self.file_path) + "/split_pdf", exist_ok=True)
return self.split_pdf_with_location(os.path.dirname(self.file_path) + "/split_pdf/", False)
def merge_pdf_with(self, merge_file_path, merged_name="merged"):
os.makedirs(os.path.dirname(self.file_path) + "/merge_pdf", exist_ok=True)
def merge_pdf_with_and_location(self, merge_file_path, output_filepath):
os.makedirs(os.path.dirname(output_filepath), exist_ok=True)
pdf_reader_1 = PdfReader(self.file_path)
pdf_reader_2 = PdfReader(merge_file_path)
writer = PdfWriter()
@ -39,8 +65,14 @@ class pdf_util:
for page_num in range(len(pdf_reader_2.pages)):
writer.add_page(pdf_reader_2.pages[page_num])
out_path = os.path.dirname(self.file_path) + "/merge_pdf" + '/merger.pdf'
with open(out_path, 'wb') as outfile:
with open(output_filepath, 'wb') as outfile:
writer.write(outfile)
return out_path
return output_filepath
# Deprecate when pdf_project_manager takes effect
def merge_pdf_with(self, merge_file_path, merged_name="merged"):
os.makedirs(os.path.dirname(self.file_path) + "/merge_pdf", exist_ok=True)
return self.merge_pdf_with_and_location(merge_file_path, os.path.dirname(self.file_path) + "/merge_pdf" + '/merger.pdf')

View File

@ -13,3 +13,14 @@ def test_folder_creation():
print(test_pdf_project_manager.uuid)
assert os.path.isdir('/app/projects/' + test_pdf_project_manager.uuid)
def test_Projecttest_1():
"""
test_project = pdf_project_manager()
print(test_pdf_project_manager.uuid)
assert os.path.isdir('/app/projects/' + test_pdf_project_manager.uuid)
"""
assert True

View File

@ -1,19 +1,18 @@
import pytest
import os
import shutil
from pdf_util.pdf_util import pdf_util
def test_split_pages():
def test_split_pdf():
# Single Pages
test_file = pdf_util("/app/tests/sample_pdfs/sample_1_page.pdf").split_pdf()
print(test_file)
print(os.stat("/app/tests/sample_pdfs/split_pdf/sample_1_page_1.pdf").st_size)
assert os.stat("/app/tests/sample_pdfs/split_pdf/sample_1_page_1.pdf").st_size == 69339
# Two Pages
test_file = pdf_util("/app/tests/sample_pdfs/sample_2_page.pdf").split_pdf()
print(test_file)
print(os.stat("/app/tests/sample_pdfs/split_pdf/sample_2_page_1.pdf").st_size)
assert os.stat("/app/tests/sample_pdfs/split_pdf/sample_2_page_1.pdf").st_size == 1804
print(os.stat("/app/tests/sample_pdfs/split_pdf/sample_2_page_2.pdf").st_size)
@ -22,7 +21,6 @@ def test_split_pages():
# Ten Pages
test_file = pdf_util("/app/tests/sample_pdfs/sample_10_page.pdf").split_pdf()
print(test_file)
print(os.stat("/app/tests/sample_pdfs/split_pdf/sample_10_page_1.pdf").st_size)
assert os.stat("/app/tests/sample_pdfs/split_pdf/sample_10_page_1.pdf").st_size == 3167
print(os.stat("/app/tests/sample_pdfs/split_pdf/sample_10_page_2.pdf").st_size)
@ -44,10 +42,131 @@ def test_split_pages():
print(os.stat("/app/tests/sample_pdfs/split_pdf/sample_10_page_10.pdf").st_size)
assert os.stat("/app/tests/sample_pdfs/split_pdf/sample_10_page_10.pdf").st_size == 3789
shutil.rmtree("/app/tests/sample_pdfs/split_pdf/")
def test_merge_with():
# Write test code to verify the behavior of the merge_with method
pass
def test_split_pdf_and_location():
# Single Pages
test_file = pdf_util("/app/tests/sample_pdfs/sample_1_page.pdf").split_pdf_with_location("/tmp/test_directory/", False)
print(test_file)
print(os.stat("/tmp/test_directory/sample_1_page_1.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_1_page_1.pdf").st_size == 69339
test_file = pdf_util("/app/tests/sample_pdfs/sample_1_page.pdf").split_pdf_with_location("/tmp/test_directory/", True)
print(test_file)
print(os.stat("/tmp/test_directory/1.pdf").st_size)
assert os.stat("/tmp/test_directory/1.pdf").st_size == 69339
shutil.rmtree("/tmp/test_directory/")
# Two Pages
test_file = pdf_util("/app/tests/sample_pdfs/sample_2_page.pdf").split_pdf_with_location("/tmp/test_directory/", False)
print(test_file)
print(os.stat("/tmp/test_directory/sample_2_page_1.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_2_page_1.pdf").st_size == 1804
print(os.stat("/tmp/test_directory/sample_2_page_2.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_2_page_2.pdf").st_size == 1405
# Two Pages
test_file = pdf_util("/app/tests/sample_pdfs/sample_2_page.pdf").split_pdf_with_location("/tmp/test_directory/", True)
print(test_file)
print(os.stat("/tmp/test_directory/1.pdf").st_size)
assert os.stat("/tmp/test_directory/1.pdf").st_size == 1804
print(os.stat("/tmp/test_directory/2.pdf").st_size)
assert os.stat("/tmp/test_directory/2.pdf").st_size == 1405
shutil.rmtree("/tmp/test_directory/")
# Ten Pages
test_file = pdf_util("/app/tests/sample_pdfs/sample_10_page.pdf").split_pdf_with_location("/tmp/test_directory/", False)
print(test_file)
print(os.stat("/tmp/test_directory/sample_10_page_1.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_10_page_1.pdf").st_size == 3167
print(os.stat("/tmp/test_directory/sample_10_page_2.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_10_page_2.pdf").st_size == 2888
print(os.stat("/tmp/test_directory/sample_10_page_3.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_10_page_3.pdf").st_size == 6670
print(os.stat("/tmp/test_directory/sample_10_page_4.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_10_page_4.pdf").st_size == 3043
print(os.stat("/tmp/test_directory/sample_10_page_5.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_10_page_5.pdf").st_size == 9968
print(os.stat("/tmp/test_directory/sample_10_page_6.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_10_page_6.pdf").st_size == 5367
print(os.stat("/tmp/test_directory/sample_10_page_7.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_10_page_7.pdf").st_size == 10093
print(os.stat("/tmp/test_directory/sample_10_page_8.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_10_page_8.pdf").st_size == 8578
print(os.stat("/tmp/test_directory/sample_10_page_9.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_10_page_9.pdf").st_size == 30188
print(os.stat("/tmp/test_directory/sample_10_page_10.pdf").st_size)
assert os.stat("/tmp/test_directory/sample_10_page_10.pdf").st_size == 3789
test_file = pdf_util("/app/tests/sample_pdfs/sample_10_page.pdf").split_pdf_with_location("/tmp/test_directory/", True)
print(test_file)
print(os.stat("/tmp/test_directory/1.pdf").st_size)
assert os.stat("/tmp/test_directory/1.pdf").st_size == 3167
print(os.stat("/tmp/test_directory/2.pdf").st_size)
assert os.stat("/tmp/test_directory/2.pdf").st_size == 2888
print(os.stat("/tmp/test_directory/3.pdf").st_size)
assert os.stat("/tmp/test_directory/3.pdf").st_size == 6670
print(os.stat("/tmp/test_directory/4.pdf").st_size)
assert os.stat("/tmp/test_directory/4.pdf").st_size == 3043
print(os.stat("/tmp/test_directory/5.pdf").st_size)
assert os.stat("/tmp/test_directory/5.pdf").st_size == 9968
print(os.stat("/tmp/test_directory/6.pdf").st_size)
assert os.stat("/tmp/test_directory/6.pdf").st_size == 5367
print(os.stat("/tmp/test_directory/7.pdf").st_size)
assert os.stat("/tmp/test_directory/7.pdf").st_size == 10093
print(os.stat("/tmp/test_directory/8.pdf").st_size)
assert os.stat("/tmp/test_directory/8.pdf").st_size == 8578
print(os.stat("/tmp/test_directory/9.pdf").st_size)
assert os.stat("/tmp/test_directory/9.pdf").st_size == 30188
print(os.stat("/tmp/test_directory/10.pdf").st_size)
assert os.stat("/tmp/test_directory/10.pdf").st_size == 3789
shutil.rmtree("/tmp/test_directory/")
def test_merge_pdf_with():
test_file = pdf_util("/app/tests/sample_pdfs/sample_1_page.pdf").merge_pdf_with("/app/tests/sample_pdfs/sample_2_page.pdf")
print(test_file)
print(os.stat("/app/tests/sample_pdfs/merge_pdf/merger.pdf").st_size)
assert os.stat("/app/tests/sample_pdfs/merge_pdf/merger.pdf").st_size == 71761
test_file = pdf_util("/app/tests/sample_pdfs/sample_1_page.pdf").merge_pdf_with("/app/tests/sample_pdfs/sample_10_page.pdf")
print(test_file)
print(os.stat("/app/tests/sample_pdfs/merge_pdf/merger.pdf").st_size)
assert os.stat("/app/tests/sample_pdfs/merge_pdf/merger.pdf").st_size == 115044
test_file = pdf_util("/app/tests/sample_pdfs/sample_2_page.pdf").merge_pdf_with("/app/tests/sample_pdfs/sample_10_page.pdf")
print(test_file)
print(os.stat("/app/tests/sample_pdfs/merge_pdf/merger.pdf").st_size)
assert os.stat("/app/tests/sample_pdfs/merge_pdf/merger.pdf").st_size == 48427
shutil.rmtree("/app/tests/sample_pdfs/merge_pdf/")
def test_merge_pdf_with_and_location():
test_file = pdf_util("/app/tests/sample_pdfs/sample_1_page.pdf").merge_pdf_with_and_location("/app/tests/sample_pdfs/sample_2_page.pdf", "/tmp/test_directory/merged_file.pdf")
print(test_file)
print(os.stat("/tmp/test_directory/merged_file.pdf").st_size)
assert os.stat("/tmp/test_directory/merged_file.pdf").st_size == 71761
shutil.rmtree("/tmp/test_directory/")
test_file = pdf_util("/app/tests/sample_pdfs/sample_1_page.pdf").merge_pdf_with_and_location("/app/tests/sample_pdfs/sample_10_page.pdf", "/tmp/test_directory/merge_pdf/merger.pdf")
print(test_file)
print(os.stat("/tmp/test_directory/merge_pdf/merger.pdf").st_size)
assert os.stat("/tmp/test_directory/merge_pdf/merger.pdf").st_size == 115044
shutil.rmtree("/tmp/test_directory/")
test_file = pdf_util("/app/tests/sample_pdfs/sample_2_page.pdf").merge_pdf_with_and_location("/app/tests/sample_pdfs/sample_10_page.pdf", "/tmp/test_directory/merge_pdf/bigfile.pdf")
print(test_file)
print(os.stat("/tmp/test_directory/merge_pdf/bigfile.pdf").st_size)
assert os.stat("/tmp/test_directory/merge_pdf/bigfile.pdf").st_size == 48427
shutil.rmtree("/tmp/test_directory/")
def test_rotate_pages():
# Write test code to verify the behavior of the rotate_pages method