diff options
Diffstat (limited to 'tests')
| -rw-r--r-- | tests/test_arxiv.py | 62 | ||||
| -rw-r--r-- | tests/test_providers.py | 8 |
2 files changed, 68 insertions, 2 deletions
diff --git a/tests/test_arxiv.py b/tests/test_arxiv.py index 08ea2c4..2cb84cf 100644 --- a/tests/test_arxiv.py +++ b/tests/test_arxiv.py @@ -13,7 +13,11 @@ import shutil import tempfile import unittest -from paper2remarkable.providers.arxiv import DEARXIV_TEXT_REGEX, Arxiv +from paper2remarkable.providers.arxiv import ( + DEARXIV_TEXT_REGEX, + DEARXIV_URI_REGEX, + Arxiv, +) class TestArxiv(unittest.TestCase): @@ -39,6 +43,26 @@ class TestArxiv(unittest.TestCase): m = re.fullmatch(DEARXIV_TEXT_REGEX, key) self.assertIsNotNone(m) + def test_text_regex_3(self): + key = b"arXiv:physics/0605197v1 [physics.data-an] 23 May 2006" + m = re.fullmatch(DEARXIV_TEXT_REGEX, key) + self.assertIsNotNone(m) + + def test_text_regex_4(self): + key = b"arXiv:math/0309285v2 [math.NA] 9 Apr 2004" + m = re.fullmatch(DEARXIV_TEXT_REGEX, key) + self.assertIsNotNone(m) + + def test_uri_regex_1(self): + key = b"http://arxiv.org/abs/physics/0605197v1" + m = re.fullmatch(DEARXIV_URI_REGEX, key) + self.assertIsNotNone(m) + + def test_uri_regex_2(self): + key = b"https://arxiv.org/abs/1101.0028v3" + m = re.fullmatch(DEARXIV_URI_REGEX, key) + self.assertIsNotNone(m) + def test_stamp_removed_1(self): url = "https://arxiv.org/pdf/1703.06103.pdf" prov = Arxiv(upload=False) @@ -57,6 +81,42 @@ class TestArxiv(unittest.TestCase): data = fp.read() self.assertNotIn(b"arXiv:2003.06222v1 [stat.ML] 13 Mar 2020", data) + def test_stamp_removed_3(self): + url = "https://arxiv.org/abs/physics/0605197v1" + prov = Arxiv(upload=False) + filename = prov.run(url, filename="./target.pdf") + prov.uncompress_pdf(filename, "unc.pdf") + with open("unc.pdf", "rb") as fp: + data = fp.read() + self.assertNotIn( + b"arXiv:physics/0605197v1 [physics.data-an] 23 May 2006", data + ) + self.assertNotIn( + b"/URI (http://arxiv.org/abs/physics/0605197v1)", data + ) + + def test_stamp_removed_4(self): + url = "https://arxiv.org/abs/math/0309285v2" + prov = Arxiv(upload=False) + filename = prov.run(url, filename="./target.pdf") + prov.uncompress_pdf(filename, "unc.pdf") + with open("unc.pdf", "rb") as fp: + data = fp.read() + self.assertNotIn(b"arXiv:math/0309285v2 [math.NA] 9 Apr 2004", data) + self.assertNotIn(b"/URI (http://arXiv.org/abs/math/0309285v2)", data) + + def test_stamp_removed_5(self): + url = "https://arxiv.org/abs/astro-ph/9207001v1" + prov = Arxiv(upload=False) + filename = prov.run(url, filename="./target.pdf") + prov.uncompress_pdf(filename, "unc.pdf") + with open("unc.pdf", "rb") as fp: + data = fp.read() + self.assertNotIn( + b"/URI (http://arxiv.org/abs/astro-ph/9207001v1)", data + ) + self.assertNotIn(b"arXiv:astro-ph/9207001v1 13 Jul 1992", data) + if __name__ == "__main__": unittest.main() diff --git a/tests/test_providers.py b/tests/test_providers.py index a7f17ff..d2fdb0d 100644 --- a/tests/test_providers.py +++ b/tests/test_providers.py @@ -86,6 +86,13 @@ class TestProviders(unittest.TestCase): filename = prov.run(url) self.assertEqual(exp_filename, os.path.basename(filename)) + def test_arxiv_5(self): + prov = Arxiv(upload=False, verbose=VERBOSE, qpdf_path=None) + url = "https://arxiv.org/abs/2002.11523" + exp_filename = "Ponomarev_Oseledets_Cichocki_-_Using_Reinforcement_Learning_in_the_Algorithmic_Trading_Problem_2020.pdf" + filename = prov.run(url) + self.assertEqual(exp_filename, os.path.basename(filename)) + def test_pmc(self): prov = PubMed(upload=False, verbose=VERBOSE) url = "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3474301/" @@ -256,6 +263,5 @@ class TestProviders(unittest.TestCase): self.assertEqual(4, len(pdfplumber.open(filename).pages)) - if __name__ == "__main__": unittest.main() |