Normalize whitespace to support cov_to_lint.py

Indentation simplification would prevent matching `cov_to_lint` output
from different commits.

src/darker/linting.py

@@ -21,6 +21,7 @@
 
 import logging
 import os
+import re
 import shlex
 import sys
 from collections import defaultdict
@@ -29,7 +30,18 @@
 from pathlib import Path
 from subprocess import PIPE, Popen  # nosec
 from tempfile import TemporaryDirectory
-from typing import IO, Collection, Dict, Generator, Iterable, List, Set, Tuple, Union
+from typing import (
+    IO,
+    Callable,
+    Collection,
+    Dict,
+    Generator,
+    Iterable,
+    List,
+    Set,
+    Tuple,
+    Union,
+)
 
 from darker.diff import map_unmodified_lines
 from darker.git import (
@@ -110,6 +122,28 @@ def get(self, new_location: MessageLocation) -> MessageLocation:
         return NO_MESSAGE_LOCATION
 
 
+def normalize_whitespace(message: LinterMessage) -> LinterMessage:
+    """Given a line of linter output, shortens runs of whitespace to a single space
+
+    Also removes any leading or trailing whitespace.
+
+    This is done to support comparison of different ``cov_to_lint.py`` runs. To make the
+    output more readable and compact, the tool adjusts whitespace. This is done to both
+    align runs of lines and to remove blocks of extra indentation. For differing sets of
+    coverage messages from ``pytest-cov`` runs of different versions of the code, these
+    whitespace adjustments can differ, so we need to normalize them to compare and match
+    them.
+
+    :param message: The linter message to normalize
+    :return: The normalized linter message with leading and trailing whitespace stripped
+             and runs of whitespace characters collapsed into single spaces
+
+    """
+    return LinterMessage(
+        message.linter, re.sub(r"\s\s+", " ", message.description).strip()
+    )
+
+
 def make_linter_env(root: Path, revision: str) -> Dict[str, str]:
     """Populate environment variables for running linters
 
@@ -375,25 +409,32 @@ def run_linters(
     return _print_new_linter_messages(baseline, messages, diff_line_mapping, use_color)
 
 
+def _identity_line_processor(message: LinterMessage) -> LinterMessage:
+    """Default line processor which doesn't modify the line at all"""
+    return message
+
+
 def _get_messages_from_linters(
     linter_cmdlines: Iterable[Union[str, List[str]]],
     root: Path,
     paths: Collection[Path],
     env: Dict[str, str],
+    line_processor: Callable[[LinterMessage], LinterMessage] = _identity_line_processor,
 ) -> Dict[MessageLocation, List[LinterMessage]]:
     """Run given linters for the given directory and return linting errors
 
     :param linter_cmdlines: The command lines for running the linters
     :param root: The common root of all files to lint
     :param paths: Paths of files to check, relative to ``root``
     :param env: The environment variables to pass to the linter
+    :param line_processor: Pre-processing callback for linter output lines
     :return: Linter messages
 
     """
     result = defaultdict(list)
     for cmdline in linter_cmdlines:
         for message_location, message in run_linter(cmdline, root, paths, env).items():
-            result[message_location].append(message)
+            result[message_location].append(line_processor(message))
     return result
 
 
@@ -419,7 +460,7 @@ def _print_new_linter_messages(
         is_modified_line = old_location == NO_MESSAGE_LOCATION
         old_messages: List[LinterMessage] = baseline.get(old_location, [])
         for message in messages:
-            if not is_modified_line and message in old_messages:
+            if not is_modified_line and normalize_whitespace(message) in old_messages:
                 # Only hide messages when
                 # - they occurred previously on the corresponding line
                 # - the line hasn't been modified
@@ -460,6 +501,7 @@ def _get_messages_from_linters_for_baseline(
             clone_root,
             paths,
             make_linter_env(root, rev1_commit),
+            normalize_whitespace,
         )
 
 

src/darker/tests/test_linting.py

@@ -73,6 +73,18 @@ def test_diff_line_mapping_ignores_column(
     assert result == expect
 
 
+def test_normalize_whitespace():
+    """Whitespace runs and leading/trailing whitespace is normalized"""
+    description = "module.py:42:  \t  indented message,    trailing spaces and tabs \t "
+    message = LinterMessage("mylinter", description)
+
+    result = linting.normalize_whitespace(message)
+
+    assert result == LinterMessage(
+        "mylinter", "module.py:42: indented message, trailing spaces and tabs"
+    )
+
+
 @pytest.mark.kwparametrize(
     dict(
         line="module.py:42: Just a line number\n",