Cleanup filter script CLI error handling (#102)

Fix exit code values to 1 and print error messages to stdout and update unit tests accordingly.

src/corppa/utils/filter.py

@@ -296,19 +296,19 @@ def main():
 
     if args.idfile:
         if not args.idfile.is_file():
-            print(f"Error: idfile {args.idfile} does not exist")
-            sys.exit(-1)
+            print(f"Error: idfile {args.idfile} does not exist", file=sys.stderr)
+            sys.exit(1)
         elif args.idfile.stat().st_size == 0:
-            print(f"Error: idfile {args.idfile} is zero size")
-            sys.exit(-1)
+            print(f"Error: idfile {args.idfile} is zero size", file=sys.stderr)
+            sys.exit(1)
 
     if args.pgfile:
         if not args.pgfile.is_file():
-            print(f"Error: pgfile {args.pgfile} does not exist")
+            print(f"Error: pgfile {args.pgfile} does not exist", file=sys.stderr)
             sys.exit(1)
         elif args.pgfile.stat().st_size == 0:
-            print(f"Error: pgfile {args.pgfile} is zero size")
-            sys.exit(-1)
+            print(f"Error: pgfile {args.pgfile} is zero size", file=sys.stderr)
+            sys.exit(1)
 
     # if requested output filename has no extension, add jsonl
     output_filepath = args.output
@@ -317,9 +317,10 @@ def main():
 
     if output_filepath.is_file():
         print(
-            f"Error: requested output file {args.output} already exists; not overwriting"
+            f"Error: requested output file {args.output} already exists; not overwriting",
+            file=sys.stderr,
         )
-        sys.exit(-1)
+        sys.exit(1)
 
     try:
         save_filtered_corpus(
@@ -334,12 +335,12 @@ def main():
     except (FileNotFoundError, JSONDecodeError) as err:
         # catch known possible errors and display briefly
         # with the type of error and the brief message
-        print(f"{err.__class__.__name__}: {err}")
-        sys.exit(-1)
+        print(f"{err.__class__.__name__}: {err}", file=sys.stderr)
+        sys.exit(1)
 
     # check if output file exists but is zero size (i.e., no pages selected)
     if output_filepath.is_file() and output_filepath.stat().st_size == 0:
-        # if claenup is disabled, remove and report
+        # if cleanup is disabled, remove and report
         if args.cleanup:
             output_filepath.unlink()
             print(

test/test_utils/test_filter.py

@@ -360,32 +360,35 @@ def test_main_idfile_nonexistent(mock_save_filtered_corpus, capsys):
     with patch(
         "sys.argv", ["f.py", "foo.jsonl", "out.jsonl", "--idfile", "/not/a/real/id.txt"]
     ):
-        with pytest.raises(SystemExit):
+        with pytest.raises(SystemExit) as execinfo:
             main()
+        assert execinfo.value.code == 1
     captured = capsys.readouterr()
-    assert "does not exist" in captured.out
+    assert "does not exist" in captured.err
 
 
 @patch("corppa.utils.filter.save_filtered_corpus")
 def test_main_idfile_empty(mock_save_filtered_corpus, capsys, tmp_path):
     idfile = tmp_path / "id.txt"
     idfile.touch()
     with patch("sys.argv", ["f.py", "foo.jsonl", "out.jsonl", "--idfile", str(idfile)]):
-        with pytest.raises(SystemExit):
+        with pytest.raises(SystemExit) as execinfo:
             main()
+        assert execinfo.value.code == 1
     captured = capsys.readouterr()
-    assert "is zero size" in captured.out
+    assert "is zero size" in captured.err
 
 
 @patch("corppa.utils.filter.save_filtered_corpus")
 def test_main_pgfile_empty(mock_save_filtered_corpus, capsys, tmp_path):
     pgfile = tmp_path / "pages.csv"
     pgfile.touch()
     with patch("sys.argv", ["f.py", "foo.jsonl", "out.jsonl", "--pgfile", str(pgfile)]):
-        with pytest.raises(SystemExit):
+        with pytest.raises(SystemExit) as execinfo:
             main()
+        assert execinfo.value.code == 1
     captured = capsys.readouterr()
-    assert "is zero size" in captured.out
+    assert "is zero size" in captured.err
 
 
 @patch("corppa.utils.filter.save_filtered_corpus")
@@ -397,7 +400,8 @@ def test_main_outfile_exists(mock_save_filtered_corpus, capsys, tmp_path):
     with patch(
         "sys.argv", ["f.py", "foo.jsonl", str(outfile), "--idfile", str(idfile)]
     ):
-        with pytest.raises(SystemExit):
+        with pytest.raises(SystemExit) as execinfo:
             main()
+        assert execinfo.value.code == 1
     captured = capsys.readouterr()
-    assert "already exists" in captured.out
+    assert "already exists" in captured.err