TEST: Test description file generation

ukbparse/tests/test_main.py

@@ -6,6 +6,7 @@
 #
 
 import textwrap  as tw
+import itertools as it
 import os.path   as op
 import              os
 import              shlex
@@ -1609,3 +1610,82 @@ def test_main_argparse_bug_workaround():
 
         assert np.all(got['1-0.0'] == range( 11,  16))
         assert np.all(got['2-0.0'] == range(-11, -16, -1))
+
+
+@patch_logging
+def test_main_description_file():
+
+    vartable = tw.dedent("""
+    ID\tType\tDescription\tDataCoding\tNAValues\tRawLevels\tNewLevels\tParentValues\tChildValues\tClean
+    1\t\tvar one
+    2\t\tvar two
+    3\t\tvar three
+    4
+    """).strip()
+
+    proctable = tw.dedent("""
+    Variable\tProcess
+    1\tmyprocess
+    2\tmyprocess(metaproc='mymetaproc')
+    """)
+
+    @custom.processor()
+    def myprocess(dtable, vids):
+        columns = it.chain(*[dtable.columns(v) for v in vids])
+
+        add     = []
+        addvid  = []
+        addmeta = []
+
+        for col in columns:
+            series = dtable[:, col.name]
+            col.metadata = col.vid + 20
+
+            newseries = pd.Series(series + 5, name=col.name + '_a')
+            dtable[:, col.name] = dtable[:, col.name] + 3
+
+            add    .append(newseries)
+            addvid .append(col.vid)
+            addmeta.append(col.vid + 40)
+
+        return [], add, addvid, addmeta
+
+    @custom.metaproc()
+    def mymetaproc(val):
+        return str(val) + ' metaprocced'
+
+    with tempdir():
+        with open('vartable.tsv',  'wt') as f: f.write(vartable)
+        with open('proctable.tsv', 'wt') as f: f.write(proctable)
+
+        gen_test_data(4, 10, 'data.tsv')
+
+        main.main('-nb -vf vartable.tsv -pf proctable.tsv '
+                  '-def descriptions.tsv '
+                  'out.tsv data.tsv'.split())
+
+        inp   = pd.read_csv('data.tsv',         delimiter='\t', index_col=0)
+        got   = pd.read_csv('out.tsv',          delimiter='\t', index_col=0)
+        descs = pd.read_csv('descriptions.tsv',
+                            delimiter='\t',
+                            index_col=0,
+                            header=None,
+                            names=('column', 'description'))
+
+        assert all(got.columns == ['1-0.0', '1-0.0_a',
+                                   '2-0.0', '2-0.0_a',
+                                   '3-0.0', '4-0.0'])
+
+        assert (got['1-0.0']   == (inp['1-0.0'] + 3)).all()
+        assert (got['1-0.0_a'] == (inp['1-0.0'] + 5)).all()
+        assert (got['2-0.0']   == (inp['2-0.0'] + 3)).all()
+        assert (got['2-0.0_a'] == (inp['2-0.0'] + 5)).all()
+        assert (got['3-0.0']   == (inp['3-0.0']))    .all()
+        assert (got['4-0.0']   == (inp['4-0.0']))    .all()
+
+        assert descs.loc['1-0.0',   'description'] == 'var one (21)'
+        assert descs.loc['1-0.0_a', 'description'] == 'var one (41)'
+        assert descs.loc['2-0.0',   'description'] == 'var two (22)'
+        assert descs.loc['2-0.0_a', 'description'] == 'var two (42 metaprocced)'
+        assert descs.loc['3-0.0',   'description'] == 'var three (0.0)'
+        assert descs.loc['4-0.0',   'description'] == 'n/a (0.0)'