Enhance and make parsing more lenient in no-todos

hooks/no_todos.py

@@ -27,8 +27,17 @@ def main(argv: Sequence[str] | None = None) -> int:
     result = 0
 
     for filename in args.filenames:
-        with open(filename, encoding='utf-8') as f:
-            lines = f.readlines()
+        try:
+            with open(filename, encoding='utf-8-sig') as f:
+                lines = f.readlines()
+        except (UnicodeDecodeError, UnicodeError):
+            print(f'{filename}: cannot be read as UTF-8, trying UTF-16')
+            try:
+                with open(filename, encoding='utf-16') as f:
+                    lines = f.readlines()
+            except (UnicodeDecodeError, UnicodeError):
+                print(f'{filename}: cannot be read as UTF-16, skipping it')
+                continue
 
         for tag in disallowed:
             for line in lines:

tests/test_cojira.py

@@ -1,11 +1,11 @@
 from __future__ import annotations
 
 import json
+import re
 import subprocess
-import unittest
 from os import environ
 from typing import Any
-from unittest.mock import call
+from unittest.mock import patch
 from urllib import request
 from urllib.error import HTTPError
 from urllib.error import URLError
@@ -461,7 +461,7 @@ def test_jira_pat_as_env_var(tmpdir):
             (
                 'pseudo_commit_msg.txt', '-e=none',
                 '-u=http://banana', '-p=$YYYJIRA_PAT',
-            )
+            ),
         ) == 0
 
 
@@ -470,55 +470,65 @@ def test_version_empty_via_env_var(tmpdir):
     environ['EMPTY_VERSION'] = ''
     with (
         tmpdir.as_cwd(),
-        unittest.mock.patch('builtins.print') as mocked_print,
+        patch('builtins.print') as mocked_print,
     ):
         exec_cmd('git', 'init')
         f = tmpdir.join('pseudo_commit_msg.txt')
         f.write_text('ABC-123: Banana', encoding='utf-8')
         assert cojira.main(
             (
                 'pseudo_commit_msg.txt',
-                f'-v=$EMPTY_VERSION',
+                '-v=$EMPTY_VERSION',
                 '-u=http://banana', '-p=pat_on_the_back',
             ),
         ) == 0
-        assert call('Ticket fix version not checked') \
-            in mocked_print.mock_calls
+        assert mocked_print.call_args_list[-2].args[0] \
+            .endswith('Ticket fix version not checked')
+        assert mocked_print.call_args_list[-1].args[0] \
+            .endswith('Ticket is OK according to COJIRA rules')
 
 
 def test_version_multiple_via_env_var(tmpdir):
     cojira.request.urlopen = mocked_response
     environ['EMPTY_VERSION'] = 'a,b,version'
     with (
         tmpdir.as_cwd(),
-        unittest.mock.patch('builtins.print') as mocked_print,
+        patch('builtins.print') as mocked_print,
     ):
         exec_cmd('git', 'init')
         f = tmpdir.join('pseudo_commit_msg.txt')
         f.write_text('ABC-123: Banana', encoding='utf-8')
         assert cojira.main(
             (
                 'pseudo_commit_msg.txt',
-                f'-v=$EMPTY_VERSION',
+                '-v=$EMPTY_VERSION',
                 '-u=http://banana', '-p=pat_on_the_back',
             ),
         ) == 0
-        assert call('Ticket fix version ("version") is allowed') \
-            in mocked_print.mock_calls
+        assert mocked_print.call_args_list[-3].args[0] \
+            .startswith('Ticket fix version ("version") is allowed')
         assert (
-            call('\t(allowed versions are: ({\'a\', \'b\', \'version\'}))') \
-                in mocked_print.mock_calls or
-            call('\t(allowed versions are: ({\'a\', \'version\', \'b\'}))') \
-                in mocked_print.mock_calls or
-            call('\t(allowed versions are: ({\'version\', \'a\', \'b\'}))') \
-                in mocked_print.mock_calls or
-            call('\t(allowed versions are: ({\'b\', \'a\', \'version\'}))') \
-                in mocked_print.mock_calls or
-            call('\t(allowed versions are: ({\'b\', \'version\', \'a\'}))') \
-                in mocked_print.mock_calls or
-            call('\t(allowed versions are: ({\'version\', \'b\', \'a\'}))') \
-                in mocked_print.mock_calls
+            mocked_print.call_args_list[-2].args[0].endswith(
+                '\t(allowed versions are: ({\'a\', \'b\', \'version\'}))',
+            ) or
+            mocked_print.call_args_list[-2].args[0].endswith(
+                '\t(allowed versions are: ({\'a\', \'version\', \'b\'}))',
+            ) or
+            mocked_print.call_args_list[-2].args[0].endswith(
+                '\t(allowed versions are: ({\'version\', \'a\', \'b\'}))',
+            ) or
+            mocked_print.call_args_list[-2].args[0].endswith(
+                '\t(allowed versions are: ({\'b\', \'a\', \'version\'}))',
+            ) or
+            mocked_print.call_args_list[-2].args[0].endswith(
+                '\t(allowed versions are: ({\'b\', \'version\', \'a\'}))',
+            ) or
+            mocked_print.call_args_list[-2].args[0].endswith(
+                '\t(allowed versions are: ({\'version\', \'b\', \'a\'}))',
+            )
         )
+        assert mocked_print.call_args_list[-1].args[0] \
+            .endswith('Ticket is OK according to COJIRA rules')
 
 
 @pytest.mark.parametrize(
@@ -531,7 +541,6 @@ def test_version_multiple_via_env_var(tmpdir):
             texts=[
                 'Lenient early exit, because no JIRA URI given',
             ],
-            alts=[],
         ),
         dict(
             e=4,
@@ -541,7 +550,6 @@ def test_version_multiple_via_env_var(tmpdir):
             texts=[
                 'Could not reify ticket from commit message',
             ],
-            alts=[],
         ),
         dict(
             e=3,
@@ -552,23 +560,21 @@ def test_version_multiple_via_env_var(tmpdir):
                 '-u=http://banana',
             ),
             texts=[
+                '\t\\(allowed versions are: \\({\'not-this-version\'}\\)\\)',
                 'Ticket has no fix version, but it is expected',
-                '\t(allowed versions are: ({\'not-this-version\'}))',
             ],
-            alts=[],
         ),
         dict(
             e=1,
             msg='ABC-123: Banana',
             u=lambda _: MockedResponse('<this><is not="a"/><json/></this>'),
             args=('pseudo_commit_msg.txt', '-u=http://banana'),
             texts=[
+                '\t disallowed categories are: \\({\'done\'}\\)\\)',
+                '\t\\(allowed categories are: \\(\\),',
+                r'Ticket status category \("None"\) is not allowed',
                 'Ticket fix version not checked',
-                'Ticket status category ("None") is not allowed',
-                '\t(allowed categories are: (),',
-                '\t disallowed categories are: ({\'done\'}))',
             ],
-            alts=[],
         ),
         dict(
             e=2,
@@ -579,13 +585,9 @@ def test_version_multiple_via_env_var(tmpdir):
                 '-u=http://banana',
             ),
             texts=[
-                'Fix version of ticket ("version") is not allowed',
-            ],
-            alts=[
-                [
-                    '\t(allowed versions are: ({\'version2\', \'version3\'}))',
-                    '\t(allowed versions are: ({\'version3\', \'version2\'}))',
-                ],
+                '\t\\(allowed versions are: \\({\'version[23]\','
+                ' \'version[23]\'}\\)\\)',
+                r'Fix version of ticket \("version"\) is not allowed',
             ],
         ),
         dict(
@@ -597,34 +599,28 @@ def test_version_multiple_via_env_var(tmpdir):
                 '-u=http://banana',
             ),
             texts=[
-                'Checking ticket "ABC-123"',
-                'Ticket fix version ("version") is allowed',
                 'Ticket is OK according to COJIRA rules',
-            ],
-            alts=[
-                [
-                    '\t(allowed versions are: ({\'version2\', \'version\'}))',
-                    '\t(allowed versions are: ({\'version\', \'version2\'}))',
-                ],
+                '\t\\(allowed versions are: \\({\'version2?\','
+                ' \'version2?\'}\\)\\)',
+                r'Ticket fix version \("version"\) is allowed',
+                'Checking ticket "ABC-123"',
             ],
         ),
     ],
 )
 def test_cojira_outputs(tmpdir, params):
     with (
         tmpdir.as_cwd(),
-        unittest.mock.patch('builtins.print') as mocked_print,
+        patch('builtins.print') as mocked_print,
     ):
         if params['u'] is not None:
             cojira.request.urlopen = params['u']
         exec_cmd('git', 'init')
         f = tmpdir.join('pseudo_commit_msg.txt')
         f.write_text(params['msg'], encoding='utf-8')
         assert cojira.main(params['args']) == params['e']
+        index = -1
         for t in params['texts']:
-            assert call(t) in mocked_print.mock_calls
-        for p in params['alts']:
-            assert (
-                call(p[0]) in mocked_print.mock_calls or
-                call(p[1]) in mocked_print.mock_calls
-            )
+            assert re.match(t, mocked_print.call_args_list[index].args[0]) \
+                is not None
+            index -= 1

tests/test_commiticketing.py

@@ -2,10 +2,9 @@
 
 import random
 import subprocess
-import unittest
 from pathlib import Path
 from typing import Any
-from unittest.mock import call
+from unittest.mock import patch
 
 import pytest
 from typing_extensions import Buffer
@@ -166,13 +165,13 @@ def test_get_prefix(tmpdir, params):
 
 
 def test_not_reified(tmpdir):
-    with unittest.mock.patch('builtins.print') as mocked_print:
+    with patch('builtins.print') as mocked_print:
         with tmpdir.as_cwd():
             f = tmpdir.join('pseudo_commit_msg.txt')
             f.write_text('Abracadabra', encoding='utf-8')
             assert commiticketing.main((str(f),)) == 3
-        assert call('Could not reify branch name.') \
-               in mocked_print.mock_calls
+        assert mocked_print.call_args_list[-1].args[0] \
+            .endswith('Could not reify branch name.')
 
 
 @pytest.mark.parametrize('branch_name', test_set_2)
@@ -193,7 +192,7 @@ def test_out_of_scope_branch_default(tmpdir, branch_name):
     ],
 )
 def test_out_of_scope_branch_non_default(tmpdir, params):
-    with unittest.mock.patch('builtins.print') as mocked_print:
+    with patch('builtins.print') as mocked_print:
         with tmpdir.as_cwd():
             exec_cmd('git', 'init')
             exec_cmd('git', 'checkout', '-b', params['branch'])
@@ -202,10 +201,11 @@ def test_out_of_scope_branch_non_default(tmpdir, params):
             assert commiticketing.main(
                 ('-b', params['proc'], '-t', params['leveled'], str(f)),
             ) == 1
-        assert call(
+        assert mocked_print.call_args_list[-1].args[0] \
+            .endswith(
             f"You wanted to commit to a branch [{params['branch']}], "
             'which does not correspond to the commiticketing setup.',
-        ) in mocked_print.mock_calls
+            )
 
 
 @pytest.mark.parametrize(
@@ -216,17 +216,18 @@ def test_out_of_scope_branch_non_default(tmpdir, params):
     ),
 )
 def test_in_scope_mismatched_default(tmpdir, branch_name):
-    with unittest.mock.patch('builtins.print') as mocked_print:
+    with patch('builtins.print') as mocked_print:
         with tmpdir.as_cwd():
             exec_cmd('git', 'init')
             exec_cmd('git', 'checkout', '-b', branch_name)
             f = tmpdir.join('pseudo_commit_msg.txt')
             f.write_text('Abracadabra', encoding='utf-8')
             assert commiticketing.main((str(f),)) == 2
-        assert call(
-            f'[{branch_name}] does not correspond to branch naming '
-            'rules, consult guidelines.',
-        ) in mocked_print.mock_calls
+        assert mocked_print.call_args_list[-1].args[0] \
+            .endswith(
+                f'[{branch_name}] does not correspond to branch naming '
+                'rules, consult guidelines.',
+            )
 
 
 @pytest.mark.parametrize(
@@ -264,22 +265,25 @@ def test_in_scope_mismatched_non_default(tmpdir, params):
 )
 def test_prefix_if_not_there(tmpdir, branch_name):
     prefix = commiticketing.get_prefix(branch_name, ['user', 'backup'])
-    with tmpdir.as_cwd():
-        with unittest.mock.patch('builtins.print') as mocked_print:
-            exec_cmd('git', 'init')
-            exec_cmd('git', 'checkout', '-b', branch_name)
-            f = tmpdir.join('pseudo_commit_msg.txt')
-            f.write_text('abracadabra', encoding='utf-8')
-            assert commiticketing.main((str(f),)) == 0
-            assert call('Commiticketing did not change your subject line.') \
-                   not in mocked_print.mock_calls
-            assert call(
+    with (
+        tmpdir.as_cwd(),
+        patch('builtins.print') as mocked_print,
+    ):
+        exec_cmd('git', 'init')
+        exec_cmd('git', 'checkout', '-b', branch_name)
+        f = tmpdir.join('pseudo_commit_msg.txt')
+        f.write_text('abracadabra', encoding='utf-8')
+        assert commiticketing.main((str(f),)) == 0
+        assert not mocked_print.call_args_list[-1].args[0] \
+            .endswith('Commiticketing did not change your subject line.')
+        assert mocked_print.call_args_list[-1].args[0] \
+            .endswith(
                 'Commiticketing prefixed your subject line '
                 f'with [{prefix}: ] and made it sentence case'
                 ' after.',
-            ) in mocked_print.mock_calls
-            with open(f) as msg:
-                assert msg.readline() == f'{prefix}: Abracadabra'
+            )
+        with open(f) as msg:
+            assert msg.readline() == f'{prefix}: Abracadabra'
 
 
 @pytest.mark.parametrize(
@@ -299,24 +303,26 @@ def test_prefix_if_not_there(tmpdir, branch_name):
 )
 def test_prefix_not_doubling(tmpdir, params):
     prefix = commiticketing.get_prefix(params['branch'], ['user', 'backup'])
-    with tmpdir.as_cwd():
-        with unittest.mock.patch('builtins.print') as mocked_print:
-            exec_cmd('git', 'init')
-            exec_cmd('git', 'checkout', '-b', params['branch'])
-            f = tmpdir.join('pseudo_commit_msg.txt')
-            f.write_text(
-                f"{prefix}: {params['msg']}",
-                encoding='utf-8',
-            )
-            assert commiticketing.main((str(f),)) == 0
-            with open(f) as msg:
-                assert msg.readline() == f'{prefix}: Abracadabra'
+    with (
+        tmpdir.as_cwd(),
+        patch('builtins.print') as mocked_print,
+    ):
+        exec_cmd('git', 'init')
+        exec_cmd('git', 'checkout', '-b', params['branch'])
+        f = tmpdir.join('pseudo_commit_msg.txt')
+        f.write_text(
+            f"{prefix}: {params['msg']}",
+            encoding='utf-8',
+        )
+        assert commiticketing.main((str(f),)) == 0
+        with open(f) as msg:
+            assert msg.readline() == f'{prefix}: Abracadabra'
     if params['msg'] == 'Abracadabra':
-        assert call('Commiticketing did not change your subject line.') \
-               in mocked_print.mock_calls
+        assert mocked_print.call_args_list[-1].args[0] \
+            .endswith('Commiticketing did not change your subject line.')
     elif params['msg'] == 'abracadabra':
-        assert call('Commiticketing did not change your subject line.') \
-               not in mocked_print.mock_calls
+        assert not mocked_print.call_args_list[-1].args[0] \
+            .endswith('Commiticketing did not change your subject line.')
 
 
 @pytest.mark.parametrize(