Add option to run linter on complete directory

lib/lg_linter/scripts/init_lg_linter

@@ -8,18 +8,22 @@ chmod +x "$REPO_DIR"/.git/hooks/pre-commit
 PRE_COMMIT_SCRIPT="#!/usr/bin/env python3
 
 import os
-import pathlib
+from pathlib import Path
 
-from lg_linter.src.lg_linter import lint
+from lg_linter.src.lg_linter import LinterConfig, lint
 
 
 def main():
-    repo_path = pathlib.Path(__file__).parent.parent.parent.resolve()
-    if lint(repo_path):
-      return 0
-    else:
-      return 1
+    repo_path = Path(__file__).parent.parent.parent.resolve()
 
+    config = LinterConfig(repo_path, False)
+
+    linted_successfully = lint(config)
+
+    if not linted_successfully:
+        return 1
+
+    return 0
 
 if __name__ == '__main__':
     exit(main())

lib/lg_linter/scripts/lint_repo

@@ -1,18 +1,27 @@
 #!/usr/bin/env python3
 import argparse
-import pathlib
 import sys
+from pathlib import Path
 sys.path.append('../src')
 
-from lg_linter import lint
+from lg_linter.src.lg_linter import LinterConfig, lint
 
 
 def main():
     parser = argparse.ArgumentParser(description='Lint all staged files')
-    parser.add_argument('repo_path')
+    parser.add_argument('directory_path')
+    parser.add_argument(
+        '-a',
+        '--all',
+        action='store_true',
+        help='Lint all files (default is to only lint staged files).')
     args = parser.parse_args()
 
-    if not lint(pathlib.Path(args.repo_path)):
+    config = LinterConfig(Path(args.directory_path), args.all)
+
+    linted_successfully = lint(config)
+
+    if not linted_successfully:
         return 1
 
     return 0

lib/lg_linter/src/file_linters.py

@@ -1,8 +1,9 @@
 import os
 import subprocess
+from pathlib import Path
 
 
-def lint_cpp_file(file_path):
+def lint_cpp_file(file_path: Path) -> (bool, str):
     os.system(f'clang-format -i {file_path}')
     os.system(f'git add {file_path}')
     completed_process = subprocess.run(['cpplint', file_path],
@@ -17,7 +18,7 @@ def lint_cpp_file(file_path):
     return False, completed_process.stderr.decode('utf-8')
 
 
-def lint_python_file(file_path):
+def lint_python_file(file_path: Path) -> (bool, str):
     os.system(f'yapf -i {file_path}')
     os.system(f'git add {file_path}')
     completed_process = subprocess.run(['pylint', '--reports=n', file_path],
@@ -30,7 +31,7 @@ def lint_python_file(file_path):
     return False, completed_process.stdout.decode('utf-8')
 
 
-def lint_shell_file(file_path):
+def lint_shell_file(file_path: Path) -> (bool, str):
     completed_process = subprocess.run(['shellcheck', file_path],
                                        check=False,
                                        stdout=subprocess.PIPE)

lib/lg_linter/src/lg_linter.py

@@ -1,4 +1,6 @@
 import os
+from pathlib import Path
+from dataclasses import dataclass
 
 import git
 
@@ -9,52 +11,85 @@
                       '========================')
 
 
-def print_file_names(files):
-    for elem in files:
-        print(f'- {elem.a_path}')
+@dataclass
+class LinterConfig:
+    directory_path: Path
+    lint_all: bool = False
 
 
-def files_were_modified_after_staging(modified_files, staged_files):
-    for modified_file in modified_files:
-        if modified_file.a_path in [x.a_path for x in staged_files]:
-            return True
-
-    return False
+def lint(config: LinterConfig):
+    """
+    Run complete linter.
+    """
+    print(DASHED_DOUBLE_LINE)
+    print('Running linter...')
 
+    if config.lint_all:
+        files = get_all_files(config.directory_path)
+    else:
+        files = get_staged_files(config.directory_path)
 
-def get_file_type(file_path):
-    EXTENSIONS = {
-        'cpp': ['.h', '.hpp', '.cc', '.cpp'],
-        'python': ['.py'],
-        'shell': ['.sh', '.bash']
-    }
+    success = True
+    outputs = []
 
-    SHEBANG_KEYWORDS = {
-        'cpp': [],
-        'python': ['python'],
-        'shell': ['sh', 'bash']
-    }
+    for file in files:
+        print(f'Linting {file}... ', end='')
+        file_is_ok, stdout = lint_file(file)
+        if file_is_ok:
+            print('[\033[92mOK\033[0m]')
+        else:
+            print('[\033[91mFAIL\033[0m]')
+            success = False
+            outputs.append(stdout)
 
-    _, extension = os.path.splitext(file_path)
+    if success:
+        print('\nHurray linted succesfully!')
 
-    type_name = None
-    if extension:
-        for (file_type, extensions_for_type) in EXTENSIONS.items():
-            if extension in extensions_for_type:
-                type_name = file_type
-                break
     else:
-        shebang = open(file_path).readline()
-        for (file_type, keywords_for_type) in SHEBANG_KEYWORDS.items():
-            for keyword in keywords_for_type:
-                if keyword in shebang:
-                    type_name = file_type
-                break
+        print('\nFound the following errors:')
+        print(DASHED_DOUBLE_LINE)
+        for output in outputs:
+            print(output)
+            print(DASHED_DOUBLE_LINE)
+
+    return success
 
-    return type_name
 
+def get_staged_files(repo_path: Path):
+    repo = git.Repo(repo_path)
+    assert not repo.bare
 
-def lint_file(file_path):
+    modified_files = repo.index.diff(None)
+    staged_files = repo.index.diff('HEAD')
+    if files_were_modified_after_staging(modified_files, staged_files):
+        raise RuntimeError(
+            'Found files which were modified after staging. Please'
+            ' stage all modifications first.')
+
+    for staged_file in staged_files:
+        # Do not lint if the file was deleted. It seems as if pythongit has
+        # a bug and has new_file and deleted_file confused.
+        if not staged_file.new_file:
+            yield repo_path / staged_file.a_path
+
+
+def get_all_files(dir_path: Path):
+    for root, _, filenames in os.walk(dir_path):
+        for filename in filenames:
+            yield Path(root) / filename
+
+
+def files_were_modified_after_staging(modified_files, staged_files):
+    for modified_file in modified_files:
+        if modified_file.a_path in [
+                staged_file.a_path for staged_file in staged_files
+        ]:
+            return True
+
+    return False
+
+
+def lint_file(file_path: Path) -> (bool, str):
     file_type = get_file_type(file_path)
 
     if file_type == 'cpp':
@@ -68,46 +103,47 @@ def lint_file(file_path):
                   'ignored.')
 
 
-def lint(repo_path):
-    """
-    Lint all the staged files, return true if successful.
-    """
-    print(DASHED_DOUBLE_LINE)
-    print('Running linter...')
+def get_file_type(file_path: Path):
+    _, extension = os.path.splitext(file_path)
 
-    repo = git.Repo(repo_path)
-    assert not repo.bare
+    if extension:
+        return get_file_type_from_extension(extension)
 
-    modified_files = repo.index.diff(None)
-    staged_files = repo.index.diff('HEAD')
-    if files_were_modified_after_staging(modified_files, staged_files):
-        print('Error: Found files which were modified after staging. Please'
-              ' stage all modifications first.')
-        return False
+    return get_file_type_from_shebang(file_path)
 
-    success = True
-    outputs = []
 
-    for staged_file in staged_files:
-        # Do not lint if the file was deleted. It seems as if pythongit has
-        # a bug and has new_file and deleted_file confused.
-        if not staged_file.new_file:
-            print(f'Linting {staged_file.a_path}... ', end='')
-            file_is_ok, stdout = lint_file(repo_path / staged_file.a_path)
-            if file_is_ok:
-                print('[\033[92mOK\033[0m]')
-            else:
-                print('[\033[91mFAIL\033[0m]')
-                success = False
-                outputs.append(stdout)
+def get_file_type_from_extension(extension: str) -> str:
+    EXTENSIONS = {
+        'cpp': ['.h', '.hpp', '.cc', '.cpp'],
+        'python': ['.py'],
+        'shell': ['.sh', '.bash']
+    }
 
-    if success:
-        print("\nHurray linted succesfully!")
-    else:
-        print("\nFound the following errors:")
-        print(DASHED_DOUBLE_LINE)
-        for output in outputs:
-            print(output)
-            print(DASHED_DOUBLE_LINE)
+    for (file_type, extensions_for_type) in EXTENSIONS.items():
+        if extension in extensions_for_type:
+            return file_type
 
-    return success
+    return None
+
+
+def get_file_type_from_shebang(file_path: Path) -> str:
+    SHEBANG_KEYWORDS = {
+        'cpp': [],
+        'python': ['python'],
+        'shell': ['sh', 'bash']
+    }
+
+    try:
+        first_line = open(file_path).readline()
+    except UnicodeDecodeError:
+        # Ignore binary files.
+        return None
+
+    if not '#!' in first_line:
+        return None
+
+    for (file_type, keywords_for_type) in SHEBANG_KEYWORDS.items():
+        for keyword in keywords_for_type:
+            if keyword in first_line:
+                return file_type
+    return None

setup.py

@@ -9,19 +9,20 @@
 
 setuptools.setup(
     name="lg-linter",
-    version="v0.1.10",
+    version="v0.1.11",
     author="Lionel Gulich",
     author_email="[email protected]",
     url="https://github.com/lgulich/lg-linter",
     description="A pre commit linter for cpp, python and sh.",
     long_description=LONG_DESCRIPTION,
     long_description_content_type="text/markdown",
-    download_url="https://github.com/lgulich/lg-linter/archive/v0.1.8.tar.gz",
+    download_url="https://github.com/lgulich/lg-linter/archive/v0.1.11.tar.gz",
     install_requires=[
+        'dataclasses',
         'gitpython',
+        'lg-cpplint',
         'pylint',
         'yapf',
-        'lg-cpplint',
     ],
     classifiers=[
         "Programming Language :: Python :: 3 :: Only",