Preserve non-generated files during cleanup

Co-authored-by: P4x-ng <P4X-ng@users.noreply.github.com>

Author: cursoragent

generator/generate.py

@@ -954,6 +954,25 @@ def parse(json_path, output_path):
     return domains
 
 
+def cleanup_generated_files(directory: Path,
+                            preserved_names: typing.Iterable[str] = (),
+                            generated_suffix: typing.Optional[str] = None) -> None:
+    '''
+    Remove generated files from a directory while preserving dotfiles and any
+    explicitly listed support files.
+    '''
+    for subpath in directory.iterdir():
+        if not subpath.is_file():
+            continue
+        if subpath.name.startswith('.'):
+            continue
+        if subpath.name in preserved_names:
+            continue
+        if generated_suffix is not None and subpath.suffix != generated_suffix:
+            continue
+        subpath.unlink()
+
+
 def generate_init(init_path, domains):
     '''
     Generate an ``__init__.py`` that exports the specified modules.
@@ -977,8 +996,7 @@ def generate_docs(docs_path, domains):
     logger.info('Generating Sphinx documents')
 
     # Remove generated documents
-    for subpath in docs_path.iterdir():
-        subpath.unlink()
+    cleanup_generated_files(docs_path, generated_suffix='.rst')
 
     # Generate document for each domain
     for domain in domains:
@@ -1030,9 +1048,12 @@ def main():
     output_path.mkdir(exist_ok=True)
 
     # Remove generated code
-    for subpath in output_path.iterdir():
-        if subpath.is_file() and subpath.name not in ('py.typed', 'util.py', 'connection.py', '__init__.py') and not subpath.name.startswith('.'):
-            subpath.unlink()
+    cleanup_generated_files(output_path, preserved_names=(
+        'py.typed',
+        'util.py',
+        'connection.py',
+        '__init__.py',
+    ))
 
     # Parse domains
     domains = list()

generator/test_generate.py

@@ -10,9 +10,17 @@
 codegen tests is almost always easier with the values displayed on stdout.
 '''
 
+from pathlib import Path
 from textwrap import dedent
 
-from generate import CdpCommand, CdpDomain, CdpEvent, CdpType, docstring
+from generate import (
+    CdpCommand,
+    CdpDomain,
+    CdpEvent,
+    CdpType,
+    cleanup_generated_files,
+    docstring,
+)
 
 
 def test_docstring():
@@ -977,3 +985,55 @@ def test_cdp_domain_sphinx():
     domain = CdpDomain.from_json(json_domain)
     actual = domain.generate_sphinx()
     assert expected == actual
+
+
+def test_cleanup_generated_files_preserves_hidden_and_support_files(tmp_path: Path):
+    generated = tmp_path / 'generated.py'
+    util = tmp_path / 'util.py'
+    connection = tmp_path / 'connection.py'
+    init_file = tmp_path / '__init__.py'
+    py_typed = tmp_path / 'py.typed'
+    hidden = tmp_path / '.bish-index'
+    subdir = tmp_path / 'nested'
+
+    generated.write_text('# generated\n')
+    util.write_text('# keep\n')
+    connection.write_text('# keep\n')
+    init_file.write_text('# keep\n')
+    py_typed.write_text('')
+    hidden.write_text('keep hidden metadata\n')
+    subdir.mkdir()
+
+    cleanup_generated_files(tmp_path, preserved_names=(
+        'py.typed',
+        'util.py',
+        'connection.py',
+        '__init__.py',
+    ))
+
+    assert not generated.exists()
+    assert util.exists()
+    assert connection.exists()
+    assert init_file.exists()
+    assert py_typed.exists()
+    assert hidden.exists()
+    assert subdir.exists()
+
+
+def test_cleanup_generated_files_only_removes_generated_docs(tmp_path: Path):
+    generated_doc = tmp_path / 'page.rst'
+    hidden = tmp_path / '.bish-index'
+    support_file = tmp_path / 'README.txt'
+    subdir = tmp_path / 'images'
+
+    generated_doc.write_text('generated docs\n')
+    hidden.write_text('keep hidden metadata\n')
+    support_file.write_text('keep support file\n')
+    subdir.mkdir()
+
+    cleanup_generated_files(tmp_path, generated_suffix='.rst')
+
+    assert not generated_doc.exists()
+    assert hidden.exists()
+    assert support_file.exists()
+    assert subdir.exists()