Merge pull request #26 from eecs485staff/fix-tests

test improvements

Author: jaredzh

tests/test_api.py

@@ -1,25 +1,19 @@
 """System tests for the API interface."""
-import pathlib
-import filecmp
 import madoop
-
-
-# Directory containing unit test input files, etc.
-TESTDATA_DIR = pathlib.Path(__file__).parent/"testdata"
+from . import utils
+from .utils import TESTDATA_DIR
 
 
 def test_simple(tmpdir):
     """Run a simple MapReduce job and verify the output."""
     with tmpdir.as_cwd():
         madoop.mapreduce(
-            input_dir=str(TESTDATA_DIR/"word_count/input"),
+            input_dir=TESTDATA_DIR/"word_count/input",
             output_dir="output",
-            map_exe=str(TESTDATA_DIR/"word_count/map.py"),
-            reduce_exe=str(TESTDATA_DIR/"word_count/reduce.py"),
-        )
-    for path in (TESTDATA_DIR/"word_count/correct").glob("part-*"):
-        assert filecmp.cmp(
-            path,
-            TESTDATA_DIR/"word_count/correct"/path,
-            shallow=False,
+            map_exe=TESTDATA_DIR/"word_count/map.py",
+            reduce_exe=TESTDATA_DIR/"word_count/reduce.py",
         )
+    utils.assert_dirs_eq(
+        TESTDATA_DIR/"word_count/correct/output",
+        tmpdir/"output",
+    )

tests/test_cli.py

@@ -1,11 +1,8 @@
 """System tests for the command line interface."""
-import pathlib
 import subprocess
-import filecmp
 import pkg_resources
-
-# Directory containing unit test input files, etc.
-TESTDATA_DIR = pathlib.Path(__file__).parent/"testdata"
+from . import utils
+from .utils import TESTDATA_DIR
 
 
 def test_version():
@@ -45,10 +42,10 @@ def test_simple(tmpdir):
             stdout=subprocess.PIPE,
             check=True,
         )
-    correct_list = sorted((TESTDATA_DIR/"word_count/correct").glob("part-*"))
-    actual_list = sorted(pathlib.Path(tmpdir/"output").glob("part-*"))
-    for correct, actual in zip(correct_list, actual_list):
-        assert filecmp.cmp(correct, actual, shallow=False)
+    utils.assert_dirs_eq(
+        TESTDATA_DIR/"word_count/correct/output",
+        tmpdir/"output",
+    )
 
 
 def test_hadoop_arguments(tmpdir):

tests/test_stages.py

@@ -1,11 +1,7 @@
 """System tests for the map stage of Michigan Hadoop."""
-import pathlib
-import filecmp
 from madoop.mapreduce import map_stage, group_stage, reduce_stage
-
-
-# Directory containing unit test input files, etc.
-TESTDATA_DIR = pathlib.Path(__file__).parent/"testdata"
+from . import utils
+from .utils import TESTDATA_DIR
 
 
 def test_map_stage(tmpdir):
@@ -16,11 +12,10 @@ def test_map_stage(tmpdir):
         output_dir=tmpdir,
         num_map=2,
     )
-    correct_dir = TESTDATA_DIR/"word_count/correct/mapper-output"
-    correct_list = sorted(correct_dir.glob("part-*"))
-    actual_list = sorted(pathlib.Path(tmpdir/"output").glob("part-*"))
-    for correct, actual in zip(correct_list, actual_list):
-        assert filecmp.cmp(correct, actual, shallow=False)
+    utils.assert_dirs_eq(
+        TESTDATA_DIR/"word_count/correct/mapper-output",
+        tmpdir,
+    )
 
 
 def test_group_stage(tmpdir):
@@ -29,11 +24,10 @@ def test_group_stage(tmpdir):
         input_dir=TESTDATA_DIR/"word_count/correct/mapper-output",
         output_dir=tmpdir,
     )
-    correct_dir = TESTDATA_DIR/"word_count/correct/grouper-output"
-    correct_list = sorted(correct_dir.glob("part-*"))
-    actual_list = sorted(pathlib.Path(tmpdir).glob("part-*"))
-    for correct, actual in zip(correct_list, actual_list):
-        assert filecmp.cmp(correct, actual, shallow=False)
+    utils.assert_dirs_eq(
+        TESTDATA_DIR/"word_count/correct/grouper-output",
+        tmpdir,
+    )
 
 
 def test_reduce_stage(tmpdir):
@@ -42,10 +36,9 @@ def test_reduce_stage(tmpdir):
         exe=TESTDATA_DIR/"word_count/reduce.py",
         input_dir=TESTDATA_DIR/"word_count/correct/grouper-output",
         output_dir=tmpdir,
-        num_reduce=2,
+        num_reduce=4,
+    )
+    utils.assert_dirs_eq(
+        TESTDATA_DIR/"word_count/correct/reducer-output",
+        tmpdir,
     )
-    correct_dir = TESTDATA_DIR/"word_count/correct/reducer-output"
-    correct_list = sorted(correct_dir.glob("part-*"))
-    actual_list = sorted(pathlib.Path(tmpdir).glob("part-*"))
-    for correct, actual in zip(correct_list, actual_list):
-        assert filecmp.cmp(correct, actual, shallow=False)

tests/utils.py

@@ -0,0 +1,39 @@
+"""Unit test utilities."""
+import filecmp
+import pathlib
+
+
+# Directory containing unit test input files, etc.
+TESTDATA_DIR = pathlib.Path(__file__).parent/"testdata"
+
+
+def assert_dirs_eq(dir1, dir2):
+    """Compare two directories of files."""
+    assert dir1 != dir2, (
+        "Refusing to compare a directory to itself:\n"
+        f"dir1 = {dir1}\n"
+        f"dir2 = {dir2}\n"
+    )
+
+    # Get a list of files in each directory
+    dir1 = pathlib.Path(dir1)
+    dir2 = pathlib.Path(dir2)
+    paths1 = list(dir1.iterdir())
+    paths2 = list(dir2.iterdir())
+
+    # Sanity checks
+    assert paths1, f"Empty directory: {dir1}"
+    assert paths2, f"Empty directory: {dir2}"
+    assert all(p.is_file() for p in paths1)
+    assert all(p.is_file() for p in paths2)
+    assert len(paths1) == len(paths2), (
+        "Number of output files does not match\n"
+        f"dir1 = {dir1}\n"
+        f"dir2 = {dir2}\n"
+        f"number of files in dir1 = {len(paths1)}\n"
+        f"number of files in dir2 = {len(paths2)}\n"
+    )
+
+    # Compare files pairwise
+    for path1, path2 in zip(sorted(paths1), sorted(paths2)):
+        assert filecmp.cmp(path1, path2, shallow=False)