[2.x] utils module test suite (#82)

* ADD `test_data_path()` func

* CUT password prompting from `Info.__init__` for encrypted pdfs
 - fix `Info().decrypted()` method

* simplify `Info().encrypted()` method

* add todo

* ADD utils module test suite with test data

* MAKE test_utils_path.py for testing `add_suffix()` function

* MAKE test_utils_read.py for testing `pypdf3_reader()` function

* reformat todo

* reformat todo

* CUT unused test function

pdfconduit/utils/info.py

@@ -3,24 +3,14 @@
 
 
 class Info:
-    def __init__(self, path, password=None, prompt=True):
-        self.pdf = self._reader(path, password, prompt)
+    def __init__(self, path, password=None):
+        self.pdf = self._reader(path, password)
 
     @staticmethod
-    def _reader(path, password, prompt):
+    def _reader(path, password):
         """Read PDF and decrypt if encrypted."""
         pdf = PdfFileReader(path) if not isinstance(path, PdfFileReader) else path
-        # Check that PDF is encrypted
-        if pdf.isEncrypted:
-            # Check that password is none
-            if not password:
-                pdf.decrypt('')
-                # Try and decrypt PDF using no password, prompt for password
-                if pdf.isEncrypted and prompt:
-                    print('No password has been given for encrypted PDF ', path)
-                    password = input('Enter Password: ')
-                else:
-                    return False
+        if password:
             pdf.decrypt(password)
         return pdf
 
@@ -32,12 +22,12 @@ def _resolved_objects(pdf, xobject):
     @property
     def encrypted(self):
         """Check weather a PDF is encrypted"""
-        return True if self.pdf.isEncrypted else False
+        return self.pdf.isEncrypted
 
     @property
     def decrypted(self):
         """Check weather a PDF is encrypted"""
-        return True if self.pdf.isDecrypted else False
+        return not self.encrypted
 
     @property
     def pages(self):
@@ -51,6 +41,7 @@ def metadata(self):
 
     def resources(self):
         """Retrieve contents of each page of PDF"""
+        # todo: refactor to generator?
         return [self.pdf.getPage(i) for i in range(self.pdf.getNumPages())]
 
     @property
@@ -61,6 +52,8 @@ def security(self):
     @property
     def dimensions(self):
         """Get width and height of a PDF"""
+        # todo: add page parameter?
+        # todo: add height & width methods?
         size = self.pdf.getPage(0).mediaBox
         return {'w': float(size[2]), 'h': float(size[3])}
 
@@ -73,4 +66,7 @@ def size(self):
     @property
     def rotate(self):
         """Retrieve rotation info."""
+        # todo: add page param
+        # todo: refactor to `rotation()`
+        # todo: add is_rotated
         return self._resolved_objects(self.pdf, '/Rotate')

pdfconduit/utils/path.py

@@ -4,7 +4,6 @@
 from pathlib import Path
 
 if 'pathlib' in sys.modules:
-
     def _add_suffix(file_path, suffix, sep, ext):
         p = Path(file_path)
         _ext = p.suffix if ext is None else str('.' + ext.strip('.'))
@@ -24,6 +23,7 @@ def add_suffix(file_path, suffix='modified', sep='_', ext=None):
 
 
 def set_destination(source, suffix, filename=False, ext=None):
+    # todo: can be removed, not used
     """Create new pdf filename for temp files"""
     source_dirname = os.path.dirname(source)
 

pdfconduit/utils/receipt.py

@@ -3,6 +3,7 @@
 
 
 class Receipt:
+    # todo: refactor to write to logs
     def __init__(self, use=True, gui=False):
         self.dst = None
         self.use = use

pdfconduit/utils/write.py

@@ -3,6 +3,7 @@
 
 
 def overlay_pdfs(top_pdf, bottom_pdf, destination):
+    # todo: possibly remove?
     """
     Overlay PDF objects to files
     :param top_pdf: PDF object to be placed on top

tests/__init__.py

@@ -10,4 +10,9 @@
 pdf_path = os.path.join(test_data_dir, pdf_name)
 img_path = os.path.join(test_data_dir, img_name)
 
-__all__ = ['pdf_path', 'img_path', 'test_data_dir']
+
+def test_data_path(filename):
+    return os.path.join(test_data_dir, filename)
+
+
+__all__ = ['pdf_path', 'img_path', 'test_data_dir', 'test_data_path']

tests/test_utils_info.py

@@ -0,0 +1,143 @@
+import unittest
+
+from looptools import Timer
+
+from pdfconduit import Info
+from tests import *
+
+
+class TestInfo(unittest.TestCase):
+    @Timer.decorator
+    def test_pypdf3_is_encrypted(self):
+        info = self._get_info('encrypted.pdf')
+
+        self.assertTrue(info.encrypted)
+
+    @Timer.decorator
+    def test_pypdf3_is_decrypted(self):
+        info = self._get_info('article.pdf')
+
+        self.assertTrue(info.decrypted)
+
+    @Timer.decorator
+    def test_pypdf3_is_not_encrypted(self):
+        info = self._get_info('article.pdf')
+
+        self.assertFalse(info.encrypted)
+
+    @Timer.decorator
+    def test_pypdf3_is_not_decrypted(self):
+        info = self._get_info('encrypted.pdf')
+
+        self.assertFalse(info.decrypted)
+
+    @Timer.decorator
+    def test_pypdf3_pages(self):
+        info = self._get_info('article.pdf')
+
+        self.assertIsInstance(info.pages, int)
+        self.assertEqual(info.pages, 1)
+        self.assertEqual(Info(test_data_path('document.pdf')).pages, 11)
+
+    @Timer.decorator
+    def test_pypdf3_metadata(self):
+        info = self._get_info('article.pdf')
+
+        self.assertIsInstance(info.metadata, dict)
+        self.assertEqual(info.metadata['/Creator'], 'This PDF is created by PDF4U Pro 2.0')
+        self.assertEqual(info.metadata['/CreationDate'], 'D:20040120105826',)
+        self.assertEqual(info.metadata['/Producer'],  'PDF4U Adobe PDF Creator 2.0',)
+
+    @Timer.decorator
+    def test_pypdf3_resources(self):
+        info = self._get_info('article.pdf')
+
+        self.assertEqual(info.pages, len(info.resources()))
+
+        resources = info.resources()[0]
+        self.assertIsInstance(resources, dict)
+        self.assertEqual(resources['/Type'], '/Page')
+
+        mediabox = list(resources['/MediaBox'])
+        self.assertEqual(mediabox[0], 0)
+        self.assertEqual(mediabox[1], 0)
+        self.assertEqual(float(mediabox[2]), 595.276)
+        self.assertEqual(float(mediabox[3]), 841.89)
+
+    @Timer.decorator
+    def test_pypdf3_security_encrypted_pdf(self):
+        info = self._get_info('encrypted.pdf')
+
+        self.assertIsInstance(info.security, dict)
+        self.assertEqual(info.security, {})
+
+    @Timer.decorator
+    def test_pypdf3_security_decrypted_pdf(self):
+        info = self._get_info('encrypted.pdf', 'foo')
+
+        self.assertIsInstance(info.security, dict)
+        self.assertTrue('/V' in info.security)
+        self.assertTrue('/R' in info.security)
+        self.assertTrue('/Length' in info.security)
+        self.assertTrue('/P' in info.security)
+        self.assertTrue('/Filter' in info.security)
+        self.assertTrue('/O' in info.security)
+        self.assertTrue('/U' in info.security)
+
+    @Timer.decorator
+    def test_pypdf3_security_passwordless_pdf(self):
+        info = self._get_info('article.pdf')
+
+        self.assertIsInstance(info.security, dict)
+        self.assertEqual(info.security, {})
+
+    @Timer.decorator
+    def test_pypdf3_dimensions(self):
+        info = self._get_info('article.pdf')
+
+        self.assertIsInstance(info.dimensions, dict)
+        self.assertTrue('w' in info.dimensions)
+        self.assertTrue('h' in info.dimensions)
+        self.assertIsInstance(info.dimensions['w'], float)
+        self.assertIsInstance(info.dimensions['h'], float)
+        self.assertEqual(info.dimensions['w'], 595.276)
+        self.assertEqual(info.dimensions['h'], 841.89)
+
+    @Timer.decorator
+    def test_pypdf3_size(self):
+        info = self._get_info('article.pdf')
+
+        self.assertIsInstance(info.size, tuple)
+        self.assertEqual(len(info.size), 2)
+        self.assertIsInstance(info.size[0], float)
+        self.assertIsInstance(info.size[1], float)
+        self.assertEqual(info.size[0], 595.276)
+        self.assertEqual(info.size[1], 841.89)
+
+    @Timer.decorator
+    def test_pypdf3_size_and_dimensions_are_equal(self):
+        info = self._get_info('article.pdf')
+
+        self.assertEqual(info.size[0], info.dimensions['w'])
+        self.assertEqual(info.size[1], info.dimensions['h'])
+
+    @Timer.decorator
+    def test_pypdf3_rotate_no_rotation(self):
+        info = self._get_info('article.pdf')
+
+        self.assertEqual(info.rotate, None)
+
+    @Timer.decorator
+    def test_pypdf3_rotate_rotated(self):
+        info = self._get_info('rotated.pdf')
+
+        self.assertIsInstance(info.rotate, int)
+        self.assertEqual(info.rotate, 90)
+
+    @staticmethod
+    def _get_info(filename, password=None):
+        return Info(test_data_path(filename), password=password)
+
+
+if __name__ == '__main__':
+    unittest.main()

tests/test_utils_path.py

@@ -0,0 +1,45 @@
+import os.path
+import unittest
+
+from looptools import Timer
+
+from pdfconduit.utils import add_suffix, set_destination
+from tests import *
+
+
+class TestPath(unittest.TestCase):
+    @Timer.decorator
+    def test_add_suffix(self):
+        file_path = test_data_path('article.pdf')
+        with_suffix = add_suffix(file_path)
+
+        self.assertIsInstance(with_suffix, str)
+        self.assertEqual(with_suffix, os.path.join(os.path.dirname(file_path), 'article_modified.pdf'))
+
+    @Timer.decorator
+    def test_add_suffix_suffix(self):
+        file_path = test_data_path('article.pdf')
+        with_suffix = add_suffix(file_path, 'new')
+
+        self.assertIsInstance(with_suffix, str)
+        self.assertEqual(with_suffix, os.path.join(os.path.dirname(file_path), 'article_new.pdf'))
+
+    @Timer.decorator
+    def test_add_suffix_suffix_sep(self):
+        file_path = test_data_path('article.pdf')
+        with_suffix = add_suffix(file_path, 'old', '-')
+
+        self.assertIsInstance(with_suffix, str)
+        self.assertEqual(with_suffix, os.path.join(os.path.dirname(file_path), 'article-old.pdf'))
+
+    @Timer.decorator
+    def test_add_suffix_suffix_sep_ext(self):
+        file_path = test_data_path('article.pdf')
+        with_suffix = add_suffix(file_path, 'backup', '-', 'zip')
+
+        self.assertIsInstance(with_suffix, str)
+        self.assertEqual(with_suffix, os.path.join(os.path.dirname(file_path), 'article-backup.zip'))
+
+
+if __name__ == '__main__':
+    unittest.main()

tests/test_utils_read.py

@@ -0,0 +1,42 @@
+import unittest
+
+from PyPDF3 import PdfFileReader
+from PyPDF3.utils import PdfReadError
+from looptools import Timer
+
+from pdfconduit.utils import pypdf3_reader
+from tests import *
+
+
+class TestRead(unittest.TestCase):
+    @Timer.decorator
+    def test_pypdf3_reader_can_read_unencrypted(self):
+        file_path = test_data_path('document.pdf')
+        reader = pypdf3_reader(file_path)
+
+        self.assertIsInstance(reader, PdfFileReader)
+        self.assertEqual(reader.getNumPages(), 11)
+
+    @Timer.decorator
+    def test_pypdf3_reader_cant_read_encrypted(self):
+        file_path = test_data_path('encrypted.pdf')
+        reader = pypdf3_reader(file_path)
+
+        with self.assertRaises(PdfReadError) as context:
+            reader.getNumPages()
+
+        self.assertIsInstance(context.exception, PdfReadError)
+        self.assertEqual('File has not been decrypted', context.exception.__str__())
+
+    @Timer.decorator
+    def test_pypdf3_reader_can_read_encrypted_with_password(self):
+        file_path = test_data_path('encrypted.pdf')
+        reader = pypdf3_reader(file_path, 'foo')
+
+        self.assertIsInstance(reader, PdfFileReader)
+        self.assertTrue(reader.isEncrypted)
+        self.assertEqual(reader.getNumPages(), 11)
+
+
+if __name__ == '__main__':
+    unittest.main()