aboutsummaryrefslogtreecommitdiff
path: root/backend/tests/enwiki/test_gen_desc_data.py
diff options
context:
space:
mode:
authorTerry Truong <terry06890@gmail.com>2022-10-01 21:07:59 +1000
committerTerry Truong <terry06890@gmail.com>2022-10-01 21:07:59 +1000
commita0b1e1a8a303504dd2cc743ab72937aee7f60f4d (patch)
tree8dfe88637c4c1f7830bb0f089ff630f8582310b2 /backend/tests/enwiki/test_gen_desc_data.py
parentde9d6642ad2a57830f559fce22e36e3d68c5c70f (diff)
Add unit tests for Wikipedia extraction
Diffstat (limited to 'backend/tests/enwiki/test_gen_desc_data.py')
-rw-r--r--backend/tests/enwiki/test_gen_desc_data.py37
1 files changed, 37 insertions, 0 deletions
diff --git a/backend/tests/enwiki/test_gen_desc_data.py b/backend/tests/enwiki/test_gen_desc_data.py
new file mode 100644
index 0000000..f6d4250
--- /dev/null
+++ b/backend/tests/enwiki/test_gen_desc_data.py
@@ -0,0 +1,37 @@
+import unittest
+import os, tempfile
+
+from tests.common import readTestDbTable
+from hist_data.enwiki.gen_desc_data import genData
+
+TEST_DUMP_FILE = os.path.join(os.path.dirname(__file__), 'sample_enwiki_pages_articles.xml.bz2')
+
+class TestGenData(unittest.TestCase):
+ def test_gen(self):
+ with tempfile.TemporaryDirectory() as tempDir:
+ # Run
+ dbFile = os.path.join(tempDir, 'descData.db')
+ genData(TEST_DUMP_FILE, dbFile)
+ # Check
+ self.assertEqual(
+ readTestDbTable(dbFile, 'SELECT id, title FROM pages'),
+ {
+ (10, 'AccessibleComputing'),
+ (13, 'AfghanistanHistory'),
+ (25, 'Autism'),
+ }
+ )
+ self.assertEqual(
+ readTestDbTable(dbFile, 'SELECT id, target FROM redirects'),
+ {
+ (10, 'Computer accessibility'),
+ (13, 'History of Afghanistan'),
+ }
+ )
+ descsRows = readTestDbTable(dbFile, 'SELECT id, desc FROM descs')
+ expectedDescPrefixes = {
+ 25: 'Kanner autism, or classic autism, is a neurodevelopmental disorder',
+ }
+ self.assertEqual({row[0] for row in descsRows}, set(expectedDescPrefixes.keys()))
+ for id, desc in descsRows:
+ self.assertTrue(id in expectedDescPrefixes and desc.startswith(expectedDescPrefixes[id]))