Filter to avoid manually selecting packages/distributions multiple times

Signed-off-by: Xiaochao Dong (@damnever) <[email protected]>

Author: damnever

Diff for: pigar/core.py

@@ -10,7 +10,7 @@
 import importlib
 import importlib.util
 import importlib.machinery
-from typing import NamedTuple, List, Dict, Any, Optional, Tuple
+from typing import NamedTuple, List, Dict, Any, Optional, Tuple, Set, AnyStr
 import asyncio
 
 from .db import database
@@ -140,6 +140,7 @@ def __init__(self, project_root):
             distributions=self._installed_dists.values()
         )
         self._requirements = _LocatableRequirements()
+        self._choosed_manually = dict()
         self._uncertain_requirements = collections.defaultdict(
             _LocatableRequirements
         )  # Multiple requirements for same import name.
@@ -394,15 +395,15 @@ def write_requirements(
 
         if self._uncertain_requirements:
             stream.write(
-                '\nWARNING(pigar): some manual fixes are required since pigar has found duplicate requirements for the same import name.\n'
+                '\n# WARNING(pigar): some manual fixes might be required as pigar has detected duplicate requirements for the same import name (possibly for different submodules).\n'  # noqa: E501
             )
             uncertain_requirements = sorted(
                 self._uncertain_requirements.items(),
                 key=lambda item: item[0].lower()
             )
             for import_name, reqs in uncertain_requirements:
                 stream.write(
-                    f'# WARNING(pigar): the following duplicate requirements are for import name: {import_name}\n'
+                    f'# WARNING(pigar): the following duplicate requirements are for the import name: {import_name}\n'  # noqa: E501
                 )
                 with_ref_comments_once = with_ref_comments
                 for _, req in reqs.sorted_items():
@@ -461,6 +462,10 @@ def _maybe_filter_distributions_with_same_import_name(
     ):
         if dists_filter is None or len(distributions) <= 1:
             return distributions
+        # We can use `functools.cache` in later versions of Python.
+        existing = self._choosed_manually.get(import_name, None)
+        if existing is not None:
+            return existing
 
         assert (hasattr(distributions[0], 'name'))
 
@@ -481,7 +486,11 @@ def _maybe_filter_distributions_with_same_import_name(
             best_match = casefold_match
         if best_match is None and len(contains) == 1:
             best_match = contains[0]
-        return dists_filter(import_name, locations, distributions, best_match)
+        choosed = dists_filter(
+            import_name, locations, distributions, best_match
+        )
+        self._choosed_manually[import_name] = choosed
+        return choosed
 
 
 class LocalRequirementWithLatestVersion(NamedTuple):

Diff for: pigar/tests/test_cli.py

@@ -35,15 +35,14 @@ def test_generate(self):
                 cli, [
                     'gen', '--with-referenced-comments',
                     '--dont-show-differences', '--exclude-glob',
-                    '**/tests/data/*', '--exclude-glob',
-                    '**/_vendor/pip/*', '-f',
-                    generated_requirement_file, project_path
+                    '**/tests/data/*', '--exclude-glob', '**/_vendor/pip/*',
+                    '-f', generated_requirement_file, project_path
                 ]
             )
             self.assertEqual(result.exit_code, 0, result.output)
             expected = self._read_filelines(expected_requirements)
             actual = self._read_filelines(generated_requirement_file)
-            self.assertEqual(len(expected), len(actual))
+            self.assertEqual(len(expected), len(actual), actual)
             for idx, line in enumerate(expected):
                 line2 = actual[idx]
                 if not line or line.startswith('#') or line.startswith('\n'):