test: added tests based on KTH workload

test/test_integration.py

-import os
-from main import main
+import os, sys
+import subprocess
+import pytest
 
-KTH_WL = "workloads/KTH-SP2-1996-2.1-cln.swf"
+sys.path.append('../user_session_decompo')
+from swf2sessions import swf2sessions
+
+class KTH_WL:
+    swf_file = "workloads/KTH-SP2-1996-2.1-cln.swf"
+    nb_users = 214
+    nb_jobs = 28476
+
+thresholds = [0, 30, 1440]
 
 def create_dir_rec_if_needed(dirname):
     if not os.path.exists(dirname):
-            os.makedirs(dirname)
+        os.makedirs(dirname)
+
 
-def run_test(delim, threshold, input_swf=KTH_WL, dyn_red=True, graph=False):
+def run_test(delim, threshold, input_swf=KTH_WL.swf_file, dyn_red=True, graph=False):
     test_name = f"KTH_{delim}_t{threshold}"
-    out_folder = "test-out/" + test_name
-    create_dir_rec_if_needed(out_folder)
+    out_dir = os.path.abspath(f'test-out/{test_name}')
+    create_dir_rec_if_needed(out_dir)
+
+    cp = subprocess.run(['python3', 'swf2sessions.py', '-at', '30', input_swf, out_dir], check=True)
+
+    swf2sessions(input_swf, out_dir, delim, threshold, dyn_red, graph, False)
+
+    return out_dir, cp
+
+def test_bad_input():
+    with pytest.raises(subprocess.CalledProcessError):
+        run_test('arrival', 30, input_swf="grumpf")
+
+def test_bad_output():
+    bad_output_name = "/grumpf"
+    with pytest.raises(subprocess.CalledProcessError):
+        subprocess.run(['python3', 'swf2sessions.py', '-at', '30', KTH_WL.swf_file, bad_output_name], check=True)
+
+
+def test_kth_delim_last():
+    """Launch swf2sessions with 'last' delimitation approach and several threshold values"""
+
+    for threshold in thresholds:
+        out_dir, _ = run_test('last', threshold)
+
+        assert os.path.exists(out_dir)
+        assert len(os.listdir(out_dir)) > 0
+
+        kth_sanity_check(out_dir)
+
+
+def test_kth_delim_arrival():
+    """Launch swf2sessions with 'arrival' delimitation approach and several threshold values"""
+
+    for threshold in thresholds:
+        out_dir, _ = run_test('arrival', threshold)
+
+        assert os.path.exists(out_dir)
+        assert len(os.listdir(out_dir)) > 0
+
+        kth_sanity_check(out_dir)
+
+
+def test_kth_delim_max():
+    """Launch swf2sessions with 'max' delimitation approach and several threshold values"""
+
+    for threshold in thresholds:
+        out_dir, _ = run_test('max', threshold)
+
+        assert os.path.exists(out_dir)
+        assert len(os.listdir(out_dir)) > 0
+
+        kth_sanity_check(out_dir)
 
-    main(input_swf, out_folder, delim, threshold, dyn_red, graph, True)
 
+def nb_users(out_dir):
+    """Return the number of users for which a SABjson file has been created"""
+    assert os.path.exists(out_dir)
+    
+    SABjsons = [file for file in os.listdir(out_dir) if file[-8:]==".SABjson"]
+    return len(SABjsons)
 
-def test_delim_approaches():
-    """Launch TODO(name of the parser) with different delimitation approaches"""
 
-    delims = ['last', 'max', 'arrival']
-    thresholds = [0, 30, 1440]
+def kth_sanity_check(out_dir):
+    """Perform some sanity checks on the output files"""
+    
+    assert nb_users(out_dir) == KTH_WL.nb_users
 
-    for delim in delims:
-        for threshold in thresholds:
-            run_test(delim, threshold)