|
| 1 | +import unittest |
| 2 | +from pathlib import Path |
| 3 | + |
| 4 | +import pandas as pd |
| 5 | + |
| 6 | +from pysus.ftp import File, Directory, Database, CACHE |
| 7 | +from pysus.data.local import ParquetSet |
| 8 | +from pysus.ftp.databases import ( |
| 9 | + ciha, |
| 10 | + cnes, |
| 11 | + pni, |
| 12 | + ibge_datasus, |
| 13 | + sinan, |
| 14 | + sih, |
| 15 | + sinasc, |
| 16 | + sia, |
| 17 | + sim, |
| 18 | +) |
| 19 | + |
| 20 | + |
| 21 | +def _test_file(testcase: unittest.TestCase, file: File): |
| 22 | + testcase.assertTrue(isinstance(file, File)) |
| 23 | + testcase.assertTrue(set(["size", "type", "modify"]) == set(file.info)) |
| 24 | + testcase.assertTrue(bool(file.basename)) |
| 25 | + testcase.assertTrue(bool(file.name)) |
| 26 | + testcase.assertTrue(bool(file.path)) |
| 27 | + testcase.assertTrue(str(Path(file.path).parent) == file.parent_path) |
| 28 | + |
| 29 | + |
| 30 | +def _test_database(testcase: unittest.TestCase, database: Database): |
| 31 | + testcase.assertTrue(isinstance(database, Database)) |
| 32 | + testcase.assertTrue(bool(database.content)) |
| 33 | + testcase.assertTrue( |
| 34 | + set(["description", "long_name", "source"]) == set(database.metadata) |
| 35 | + ) |
| 36 | + |
| 37 | + downloaded_file = ( |
| 38 | + database.download(database.files[0]) |
| 39 | + if database.files[0].extension != ".zip" |
| 40 | + else database.download(database.files[-1]) |
| 41 | + ) |
| 42 | + testcase.assertTrue(isinstance(downloaded_file, ParquetSet)) |
| 43 | + testcase.assertTrue(Path(downloaded_file.path).exists()) |
| 44 | + testcase.assertTrue( |
| 45 | + isinstance(downloaded_file.to_dataframe(), pd.DataFrame) |
| 46 | + ) |
| 47 | + testcase.assertTrue(not downloaded_file.to_dataframe().empty) |
| 48 | + |
| 49 | + |
| 50 | +class TestDirectoryAndFile(unittest.TestCase): |
| 51 | + def setUp(self): |
| 52 | + self.root = Directory("/").load() |
| 53 | + |
| 54 | + def test_root_load(self): |
| 55 | + self.assertTrue(self.root.loaded) |
| 56 | + self.assertTrue(Directory("/dissemin") in self.root.content) |
| 57 | + |
| 58 | + def test_root_reload(self): |
| 59 | + root = self.root.reload() |
| 60 | + self.assertTrue(root.content == self.root.content) |
| 61 | + |
| 62 | + def test_root_directory(self): |
| 63 | + self.assertTrue(self.root.name == "/") |
| 64 | + self.assertTrue(self.root.path == "/") |
| 65 | + self.assertTrue(self.root.parent == self.root) # outermost parent |
| 66 | + |
| 67 | + def test_directory_cache(self): |
| 68 | + self.assertTrue(CACHE["/"] == self.root) |
| 69 | + |
| 70 | + def test_sinan_file(self): |
| 71 | + file = Directory("/dissemin/publicos/SINAN/DADOS/FINAIS").content[0] |
| 72 | + _test_file(self, file) |
| 73 | + |
| 74 | + |
| 75 | +class TestDatabases(unittest.TestCase): |
| 76 | + def test_ciha(self): |
| 77 | + database = ciha.CIHA().load() |
| 78 | + _test_database(self, database) |
| 79 | + self.assertTrue(database.name == "CIHA") |
| 80 | + self.assertSetEqual( |
| 81 | + set(database.describe(database.files[0])), |
| 82 | + {'group', 'last_update', 'month', 'name', 'size', 'uf', 'year'} |
| 83 | + ) |
| 84 | + self.assertEqual(len(database.format(database.files[0])), 4) |
| 85 | + |
| 86 | + def test_cnes(self): |
| 87 | + database = cnes.CNES().load("DC") |
| 88 | + _test_database(self, database) |
| 89 | + self.assertTrue(database.name == "CNES") |
| 90 | + self.assertSetEqual( |
| 91 | + set(database.describe(database.files[0])), |
| 92 | + {'group', 'last_update', 'month', 'name', 'size', 'uf', 'year'} |
| 93 | + ) |
| 94 | + self.assertEqual(len(database.format(database.files[0])), 4) |
| 95 | + |
| 96 | + def test_pni(self): |
| 97 | + database = pni.PNI().load() |
| 98 | + _test_database(self, database) |
| 99 | + self.assertTrue(database.name == "PNI") |
| 100 | + self.assertSetEqual( |
| 101 | + set(database.describe(database.files[0])), |
| 102 | + {'group', 'last_update', 'name', 'size', 'uf', 'year'} |
| 103 | + ) |
| 104 | + self.assertEqual(len(database.format(database.files[0])), 3) |
| 105 | + |
| 106 | + def test_ibge_datasus(self): |
| 107 | + database = ibge_datasus.IBGEDATASUS().load() |
| 108 | + _test_database(self, database) |
| 109 | + self.assertTrue(database.name == "IBGE-DataSUS") |
| 110 | + self.assertSetEqual( |
| 111 | + set(database.describe(database.files[0])), |
| 112 | + {'last_update', 'name', 'size', 'year'} |
| 113 | + ) |
| 114 | + self.assertEqual(len(database.format(database.files[0])), 1) |
| 115 | + |
| 116 | + def test_sinan(self): |
| 117 | + database = sinan.SINAN().load() |
| 118 | + _test_database(self, database) |
| 119 | + self.assertTrue(database.name == "SINAN") |
| 120 | + self.assertSetEqual( |
| 121 | + set(database.describe(database.files[0])), |
| 122 | + {'disease', 'last_update', 'name', 'size', 'year'} |
| 123 | + ) |
| 124 | + self.assertEqual(len(database.format(database.files[0])), 2) |
| 125 | + |
| 126 | + def test_sih(self): |
| 127 | + database = sih.SIH().load() |
| 128 | + _test_database(self, database) |
| 129 | + self.assertTrue(database.name == "SIH") |
| 130 | + self.assertSetEqual( |
| 131 | + set(database.describe(database.files[0])), |
| 132 | + {'group', 'last_update', 'month', 'name', 'size', 'uf', 'year'} |
| 133 | + ) |
| 134 | + self.assertEqual(len(database.format(database.files[0])), 4) |
| 135 | + |
| 136 | + def test_sinasc(self): |
| 137 | + database = sinasc.SINASC().load() |
| 138 | + _test_database(self, database) |
| 139 | + self.assertTrue(database.name == "SINASC") |
| 140 | + self.assertSetEqual( |
| 141 | + set(database.describe(database.files[0])), |
| 142 | + {'group', 'last_update', 'name', 'size', 'uf', 'year'} |
| 143 | + ) |
| 144 | + self.assertEqual(len(database.format(database.files[0])), 3) |
| 145 | + |
| 146 | + def test_sia(self): |
| 147 | + database = sia.SIA().load() |
| 148 | + _test_database(self, database) |
| 149 | + self.assertTrue(database.name == "SIA") |
| 150 | + self.assertSetEqual( |
| 151 | + set(database.describe(database.files[0])), |
| 152 | + {'group', 'last_update', 'month', 'name', 'size', 'uf', 'year'} |
| 153 | + ) |
| 154 | + self.assertEqual(len(database.format(database.files[0])), 4) |
| 155 | + |
| 156 | + def test_sim(self): |
| 157 | + database = sim.SIM().load() |
| 158 | + _test_database(self, database) |
| 159 | + self.assertTrue(database.name == "SIM") |
| 160 | + self.assertSetEqual( |
| 161 | + set(database.describe(database.files[0])), |
| 162 | + {'group', 'last_update', 'name', 'size', 'uf', 'year'} |
| 163 | + ) |
| 164 | + self.assertEqual(len(database.format(database.files[0])), 3) |
0 commit comments