add tests for derivation process

tests/test_annotations.py

@@ -1,17 +1,15 @@
 from ChildProject.projects import ChildProject
 from ChildProject.annotations import AnnotationManager
-from ChildProject.tables import IndexTable
 from ChildProject.converters import *
 from functools import partial
-import glob
+
 import pandas as pd
 import numpy as np
 import datetime
 import os
 import pytest
 import shutil
-import subprocess
-import sys
+from pathlib import Path
 import time
 
 
@@ -20,18 +18,19 @@ def standardize_dataframe(df, columns):
     return df.sort_index(axis=1).sort_values(list(columns)).reset_index(drop=True)
 
 
-DATA = os.path.join('tests', 'data')
-TRUTH = os.path.join('tests', 'truth')
+DATA = Path('tests', 'data')
+TRUTH = Path('tests', 'truth')
+PATH = Path('output', 'annotations')
 
 
 @pytest.fixture(scope="function")
 def project(request):
-    if os.path.exists("output/annotations"):
+    if os.path.exists(PATH):
         # shutil.copytree(src="examples/valid_raw_data", dst="output/annotations")
-        shutil.rmtree("output/annotations")
-    shutil.copytree(src="examples/valid_raw_data", dst="output/annotations")
+        shutil.rmtree(PATH)
+    shutil.copytree(src="examples/valid_raw_data", dst=PATH)
 
-    project = ChildProject("output/annotations")
+    project = ChildProject(PATH)
 
     yield project
 
@@ -286,13 +285,6 @@ def test_multiple_imports(project, am, input_file, ow, rimported, rerrors, excep
         assert len(errors) == 0 and len(warnings) == 0, "malformed annotation indexes detected"
 
 
-@pytest.mark.parametrize("input_set",
-                         [(1)]
-                         )
-def test_derive(project, am, input_set):
-    pass
-
-
 # function used as a derivation function, it should throw errors if not returning dataframe or without required columns
 def dv_func(a, b, x, type):
     if type == 'number':
@@ -303,17 +295,50 @@ def dv_func(a, b, x, type):
         return x
 
 
+@pytest.mark.parametrize("exists,ow",
+                         [(False, False),
+                          (True, False),
+                          (False, True),
+                          (True, True),
+                          ])
+def test_derive(project, am, exists, ow):
+    input_set = 'vtc_present'
+    output_set = 'output'
+    function = partial(dv_func, type='normal')
+    am.read()
+
+    # copy the input set to act as an existing output_set
+    if exists:
+        shutil.copytree(src=PATH / 'annotations' / input_set, dst=PATH / 'annotations' / output_set)
+        additions = am.annotations[am.annotations['set'] == input_set].copy()
+        additions['set'] = output_set
+        am.annotations = pd.concat([am.annotations, additions])
+
+    imported, errors = am.derive_annotations(input_set, output_set, function, overwrite_existing=ow)
+    assert imported.shape[0] == am.annotations[am.annotations['set'] == input_set].shape[0]
+    assert errors.shape[0] == 0
+
+    truth = am.annotations[am.annotations['set'] == input_set]
+    truth['merged_from'] = truth['set']
+    truth['set'] = output_set
+    truth['format'] = 'NA'
+    cols = ['imported_at', 'package_version']
+    pd.testing.assert_frame_equal(truth.drop(columns=cols).reset_index(drop=True),
+                                  imported.drop(columns=cols).reset_index(drop=True))
+
+
 # function used for derivation but does not hav correct signature
 def bad_func(a, b):
     return b
+
+
 @pytest.mark.parametrize("input_set,function,output_set,exists,ow,error",
                          [("missing", partial(dv_func, type='normal'), "output", False, False, AssertionError),
                           ("vtc_present", partial(dv_func, type='number'), "output", False, False, None),
                           ("vtc_present", partial(dv_func, type='columns'), "output", False, False, None),
                           ("vtc_present", bad_func, "output", False, False, None),
                           ("vtc_present", partial(dv_func, type='normal'), "vtc_present", False, False, AssertionError),
-                          ("vtc_present", partial(dv_func, type='normal'), "output", True, False, AssertionError),
-                          ("vtc_present", partial(dv_func, type='normal'), "output", True, True, AssertionError),
+                          ("vtc_present", 'not_a_function', "output", False, False, ValueError),
                           ])
 def test_derive_inputs(project, am, input_set, function, output_set, exists, ow, error):
     am.read()
@@ -326,7 +351,12 @@ def test_derive_inputs(project, am, input_set, function, output_set, exists, ow,
 
     if error:
         with pytest.raises(error):
-            am.rename_set(old, new)
+            am.derive_annotations(input_set, output_set, function, overwrite_existing=ow)
+    else:
+        imported, errors = am.derive_annotations(input_set, output_set, function, overwrite_existing=ow)
+        # check that 0 lines were imported because of bad input
+        assert imported.shape[0] == 0
+        assert errors.shape[0] == am.annotations[am.annotations['set'] == input_set].shape[0]
 
 def test_intersect(project, am):
     input_annotations = pd.read_csv("examples/valid_raw_data/annotations/intersect.csv")

tests/test_derivations.py

@@ -0,0 +1,49 @@
+import pandas as pd
+from pathlib import Path
+
+from ChildProject.projects import ChildProject
+import ChildProject.pipelines.derivations as deriv
+
+CP_PATH = Path('examples','valid_raw_data')
+CSV_DF = pd.read_csv(Path('examples', 'valid_raw_data', 'annotations', 'vtc_present', 'converted',
+                          'sound2_0_4000.csv'))
+TRUTH = Path('tests', 'truth', 'derivations')
+
+
+def test_conversations():
+    df = CSV_DF.copy()
+    meta = {}
+    project = None
+
+    res = deriv.conversations(project, meta, df)
+    # res.to_csv(TRUTH / 'conversations.csv', index=False)
+    truth = pd.read_csv(TRUTH / 'conversations.csv')
+
+    pd.testing.assert_frame_equal(res, truth, check_dtype=False)
+
+def test_acoustics():
+    df = CSV_DF.copy()
+    project = ChildProject(CP_PATH)
+    project.read()
+    meta = {'recording_filename': 'sound.wav'}
+
+    res = deriv.acoustics(project, meta, df, profile=None, target_sr=4096)
+    # res.to_csv(TRUTH / 'acoustics.csv', index=False)
+    truth = pd.read_csv(TRUTH / 'acoustics.csv')
+
+    print(truth.to_string())
+    print(res.to_string())
+
+    pd.testing.assert_frame_equal(res, truth)
+
+
+def test_remove_overlaps():
+    df = CSV_DF.copy()
+    meta = {}
+    project = None
+
+    res = deriv.remove_overlaps(project, meta, df)
+    # res.to_csv(TRUTH / 'remove-overlaps.csv', index=False)
+    truth = pd.read_csv(TRUTH / 'remove-overlaps.csv')
+
+    pd.testing.assert_frame_equal(res, truth, check_dtype=False)

tests/truth/derivations/acoustics.csv

@@ -0,0 +1,17 @@
+segment_onset,segment_offset,speaker_type,raw_filename,mean_pitch_semitone,median_pitch_semitone,p5_pitch_semitone,p95_pitch_semitone,pitch_range_semitone
+0,342,CHI,example.rttm,30.556933071954266,40.273725722703276,14.038385565680954,40.273725722703276,26.23534015702232
+113,486,FEM,example.rttm,39.84245110739157,40.273725722703276,39.10928426136167,40.273725722703276,1.1644414613416032
+200,901,OCH,example.rttm,34.18339779437408,40.273725722703276,12.867250045221894,40.273725722703276,27.40647567748138
+782,1421,MAL,example.rttm,38.294152588657674,40.273725722703276,31.36564661949806,40.273725722703276,8.908079103205214
+1401,1753,,example.rttm,40.273725722703276,40.273725722703276,40.273725722703276,40.273725722703276,0.0
+203,1000,CHI,example.rttm,35.17511859935001,40.273725722703276,15.29055081827227,40.273725722703276,24.983174904431007
+1200,1656,OCH,example.rttm,34.6622956535196,40.273725722703276,21.19486348747878,40.273725722703276,19.078862235224495
+1350,2111,MAL,example.rttm,39.05753217752032,40.273725722703276,34.3143773513068,40.273725722703276,5.959348371396473
+1821,2324,FEM,example.rttm,38.991258223371645,40.273725722703276,35.14385572537674,40.273725722703276,5.129869997326537
+2301,2845,,example.rttm,37.46240668323427,40.273725722703276,29.02844956482724,40.273725722703276,11.245276157876035
+2556,2890,CHI,example.rttm,33.547635907323105,40.273725722703276,22.11328322117681,40.273725722703276,18.160442501526465
+2901,3205,FEM,example.rttm,36.16407064848453,40.273725722703276,29.177657022312662,40.273725722703276,11.096068700390614
+3056,3457,OCH,example.rttm,32.7496604661,40.273725722703276,14.691903850252153,40.273725722703276,25.581821872451123
+3301,3654,MAL,example.rttm,30.550152227679973,40.273725722703276,14.020077286140356,40.273725722703276,26.25364843656292
+3456,3705,,example.rttm,36.88339947865374,36.88339947865374,33.83210585900916,39.934693098298325,6.102587239289164
+3900,3987,CHI,example.rttm,34.47045449069267,34.47045449069267,34.47045449069267,34.47045449069267,0.0

tests/truth/derivations/conversations.csv

@@ -0,0 +1,14 @@
+segment_onset,segment_offset,speaker_type,raw_filename,iti,prev_speaker_type,delay,is_CT,conv_count
+0,342,CHI,example.rttm,,,,False,1
+113,486,FEM,example.rttm,-229.0,CHI,113.0,True,1
+200,901,OCH,example.rttm,-286.0,FEM,87.0,True,1
+782,1421,MAL,example.rttm,-119.0,OCH,582.0,True,1
+203,1000,CHI,example.rttm,-1218.0,MAL,-579.0,False,2
+1200,1656,OCH,example.rttm,200.0,CHI,997.0,True,2
+1350,2111,MAL,example.rttm,-306.0,OCH,150.0,True,2
+1821,2324,FEM,example.rttm,-290.0,MAL,471.0,True,2
+2556,2890,CHI,example.rttm,232.0,FEM,735.0,True,2
+2901,3205,FEM,example.rttm,11.0,CHI,345.0,True,2
+3056,3457,OCH,example.rttm,-149.0,FEM,155.0,True,2
+3301,3654,MAL,example.rttm,-156.0,OCH,245.0,True,2
+3900,3987,CHI,example.rttm,246.0,MAL,599.0,True,2

tests/truth/derivations/remove-overlaps.csv

@@ -0,0 +1,10 @@
+segment_onset,segment_offset,speaker_type,raw_filename
+0,113,CHI,example.rttm
+1000,1200,MAL,example.rttm
+1656,1821,MAL,example.rttm
+2111,2324,FEM,example.rttm
+2556,2890,CHI,example.rttm
+2901,3056,FEM,example.rttm
+3205,3301,OCH,example.rttm
+3457,3654,MAL,example.rttm
+3900,3987,CHI,example.rttm