# Copyright (C) 2009 Google Inc. All rights reserved. # Copyright (C) 2010 Chris Jerdonek (chris.jerdonek@gmail.com) # Copyright (C) 2010 ProFUSION embedded systems # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions are # met: # # * Redistributions of source code must retain the above copyright # notice, this list of conditions and the following disclaimer. # * Redistributions in binary form must reproduce the above # copyright notice, this list of conditions and the following disclaimer # in the documentation and/or other materials provided with the # distribution. # * Neither the name of Google Inc. nor the names of its # contributors may be used to endorse or promote products derived from # this software without specific prior written permission. # # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS # "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT # LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR # A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT # OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, # SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT # LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, # DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY # THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT # (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. """Front end of some style-checker modules.""" import logging import os.path import sys from checkers.common import categories as CommonCategories from checkers.common import CarriageReturnChecker from checkers.cpp import CppChecker from checkers.python import PythonChecker from checkers.test_expectations import TestExpectationsChecker from checkers.text import TextChecker from checkers.xml import XMLChecker from error_handlers import DefaultStyleErrorHandler from filter import FilterConfiguration from optparser import ArgumentParser from optparser import DefaultCommandOptionValues from webkitpy.style_references import configure_logging as _configure_logging _log = logging.getLogger("webkitpy.style.checker") # These are default option values for the command-line option parser. _DEFAULT_MIN_CONFIDENCE = 1 _DEFAULT_OUTPUT_FORMAT = 'emacs' # FIXME: For style categories we will never want to have, remove them. # For categories for which we want to have similar functionality, # modify the implementation and enable them. # # Throughout this module, we use "filter rule" rather than "filter" # for an individual boolean filter flag like "+foo". This allows us to # reserve "filter" for what one gets by collectively applying all of # the filter rules. # # The base filter rules are the filter rules that begin the list of # filter rules used to check style. For example, these rules precede # any user-specified filter rules. Since by default all categories are # checked, this list should normally include only rules that begin # with a "-" sign. _BASE_FILTER_RULES = [ '-build/endif_comment', '-build/include_what_you_use', # for std::string '-build/storage_class', # const static '-legal/copyright', '-readability/multiline_comment', '-readability/braces', # int foo() {}; '-readability/fn_size', '-readability/casting', '-readability/function', '-runtime/arrays', # variable length array '-runtime/casting', '-runtime/sizeof', '-runtime/explicit', # explicit '-runtime/virtual', # virtual dtor '-runtime/printf', '-runtime/threadsafe_fn', '-runtime/rtti', '-whitespace/blank_line', '-whitespace/end_of_line', '-whitespace/labels', # List Python pep8 categories last. # # Because much of WebKit's Python code base does not abide by the # PEP8 79 character limit, we ignore the 79-character-limit category # pep8/E501 for now. # # FIXME: Consider bringing WebKit's Python code base into conformance # with the 79 character limit, or some higher limit that is # agreeable to the WebKit project. '-pep8/E501', ] # The path-specific filter rules. # # This list is order sensitive. Only the first path substring match # is used. See the FilterConfiguration documentation in filter.py # for more information on this list. # # Each string appearing in this nested list should have at least # one associated unit test assertion. These assertions are located, # for example, in the test_path_rules_specifier() unit test method of # checker_unittest.py. _PATH_RULES_SPECIFIER = [ # Files in these directories are consumers of the WebKit # API and therefore do not follow the same header including # discipline as WebCore. ([# TestNetscapePlugIn has no config.h and uses funny names like # NPP_SetWindow. "Tools/DumpRenderTree/TestNetscapePlugIn/", # The API test harnesses have no config.h and use funny macros like # TEST_CLASS_NAME. "Tools/WebKitAPITest/", "Tools/TestWebKitAPI/"], ["-build/include", "-readability/naming"]), ([# The EFL APIs use EFL naming style, which includes # both lower-cased and camel-cased, underscore-sparated # values. "Source/WebKit/efl/ewk/", # There is no clean way to avoid "yy_*" names used by flex. "Source/WebCore/css/CSSParser.cpp", # Qt code uses '_' in some places (such as private slots # and on test xxx_data methos on tests) "Source/JavaScriptCore/qt/", "Source/WebKit/qt/Api/", "Source/WebKit/qt/tests/", "Source/WebKit/qt/declarative/", "Source/WebKit/qt/examples/"], ["-readability/naming"]), ([# Qt's MiniBrowser has no config.h "Tools/MiniBrowser/qt"], ["-build/include"]), ([# The GTK+ APIs use GTK+ naming style, which includes # lower-cased, underscore-separated values. # Also, GTK+ allows the use of NULL. "Source/WebCore/bindings/scripts/test/GObject", "WebKit/gtk/webkit/", "Tools/DumpRenderTree/gtk/"], ["-readability/naming", "-readability/null"]), ([# Header files in ForwardingHeaders have no header guards or # exceptional header guards (e.g., WebCore_FWD_Debugger_h). "/ForwardingHeaders/"], ["-build/header_guard"]), ([# assembler has lots of opcodes that use underscores, so # we don't check for underscores in that directory. "/Source/JavaScriptCore/assembler/"], ["-readability/naming"]), ([# JITStubs has an usual syntax which causes false alarms for a few checks. "JavaScriptCore/jit/JITStubs.cpp"], ["-readability/parameter_name", "-whitespace/parens"]), # WebKit2 rules: # WebKit2 and certain directories have idiosyncracies. ([# NPAPI has function names with underscores. "Source/WebKit2/WebProcess/Plugins/Netscape"], ["-readability/naming"]), ([# The WebKit2 C API has names with underscores and whitespace-aligned # struct members. Also, we allow unnecessary parameter names in # WebKit2 APIs because we're matching CF's header style. "Source/WebKit2/UIProcess/API/C/", "Source/WebKit2/WebProcess/InjectedBundle/API/c/"], ["-readability/naming", "-readability/parameter_name", "-whitespace/declaration"]), # For third-party Python code, keep only the following checks-- # # No tabs: to avoid having to set the SVN allow-tabs property. # No trailing white space: since this is easy to correct. # No carriage-return line endings: since this is easy to correct. # (["webkitpy/thirdparty/"], ["-", "+pep8/W191", # Tabs "+pep8/W291", # Trailing white space "+whitespace/carriage_return"]), ] _CPP_FILE_EXTENSIONS = [ 'c', 'cpp', 'h', ] _PYTHON_FILE_EXTENSION = 'py' _TEXT_FILE_EXTENSIONS = [ 'ac', 'cc', 'cgi', 'css', 'exp', 'flex', 'gyp', 'gypi', 'html', 'idl', 'in', 'js', 'mm', 'php', 'pl', 'pm', 'pri', 'pro', 'rb', 'sh', 'txt', 'wm', 'xhtml', 'y', ] _XML_FILE_EXTENSIONS = [ 'vcproj', 'vsprops', ] # Files to skip that are less obvious. # # Some files should be skipped when checking style. For example, # WebKit maintains some files in Mozilla style on purpose to ease # future merges. _SKIPPED_FILES_WITH_WARNING = [ "gtk2drawing.c", # WebCore/platform/gtk/gtk2drawing.c "gtkdrawing.h", # WebCore/platform/gtk/gtkdrawing.h "Source/WebKit/gtk/tests/", # Soup API that is still being cooked, will be removed from WebKit # in a few months when it is merged into soup proper. The style # follows the libsoup style completely. "Source/WebCore/platform/network/soup/cache/", ] # Files to skip that are more common or obvious. # # This list should be in addition to files with FileType.NONE. Files # with FileType.NONE are automatically skipped without warning. _SKIPPED_FILES_WITHOUT_WARNING = [ "LayoutTests/", ] # Extensions of files which are allowed to contain carriage returns. _CARRIAGE_RETURN_ALLOWED_FILE_EXTENSIONS = [ 'vcproj', 'vsprops', ] # The maximum number of errors to report per file, per category. # If a category is not a key, then it has no maximum. _MAX_REPORTS_PER_CATEGORY = { "whitespace/carriage_return": 1 } def _all_categories(): """Return the set of all categories used by check-webkit-style.""" # Take the union across all checkers. categories = CommonCategories.union(CppChecker.categories) categories = categories.union(TestExpectationsChecker.categories) # FIXME: Consider adding all of the pep8 categories. Since they # are not too meaningful for documentation purposes, for # now we add only the categories needed for the unit tests # (which validate the consistency of the configuration # settings against the known categories, etc). categories = categories.union(["pep8/W191", "pep8/W291", "pep8/E501"]) return categories def _check_webkit_style_defaults(): """Return the default command-line options for check-webkit-style.""" return DefaultCommandOptionValues(min_confidence=_DEFAULT_MIN_CONFIDENCE, output_format=_DEFAULT_OUTPUT_FORMAT) # This function assists in optparser not having to import from checker. def check_webkit_style_parser(): all_categories = _all_categories() default_options = _check_webkit_style_defaults() return ArgumentParser(all_categories=all_categories, base_filter_rules=_BASE_FILTER_RULES, default_options=default_options) def check_webkit_style_configuration(options): """Return a StyleProcessorConfiguration instance for check-webkit-style. Args: options: A CommandOptionValues instance. """ filter_configuration = FilterConfiguration( base_rules=_BASE_FILTER_RULES, path_specific=_PATH_RULES_SPECIFIER, user_rules=options.filter_rules) return StyleProcessorConfiguration(filter_configuration=filter_configuration, max_reports_per_category=_MAX_REPORTS_PER_CATEGORY, min_confidence=options.min_confidence, output_format=options.output_format, stderr_write=sys.stderr.write) def _create_log_handlers(stream): """Create and return a default list of logging.Handler instances. Format WARNING messages and above to display the logging level, and messages strictly below WARNING not to display it. Args: stream: See the configure_logging() docstring. """ # Handles logging.WARNING and above. error_handler = logging.StreamHandler(stream) error_handler.setLevel(logging.WARNING) formatter = logging.Formatter("%(levelname)s: %(message)s") error_handler.setFormatter(formatter) # Create a logging.Filter instance that only accepts messages # below WARNING (i.e. filters out anything WARNING or above). non_error_filter = logging.Filter() # The filter method accepts a logging.LogRecord instance. non_error_filter.filter = lambda record: record.levelno < logging.WARNING non_error_handler = logging.StreamHandler(stream) non_error_handler.addFilter(non_error_filter) formatter = logging.Formatter("%(message)s") non_error_handler.setFormatter(formatter) return [error_handler, non_error_handler] def _create_debug_log_handlers(stream): """Create and return a list of logging.Handler instances for debugging. Args: stream: See the configure_logging() docstring. """ handler = logging.StreamHandler(stream) formatter = logging.Formatter("%(name)s: %(levelname)-8s %(message)s") handler.setFormatter(formatter) return [handler] def configure_logging(stream, logger=None, is_verbose=False): """Configure logging, and return the list of handlers added. Returns: A list of references to the logging handlers added to the root logger. This allows the caller to later remove the handlers using logger.removeHandler. This is useful primarily during unit testing where the caller may want to configure logging temporarily and then undo the configuring. Args: stream: A file-like object to which to log. The stream must define an "encoding" data attribute, or else logging raises an error. logger: A logging.logger instance to configure. This parameter should be used only in unit tests. Defaults to the root logger. is_verbose: A boolean value of whether logging should be verbose. """ # If the stream does not define an "encoding" data attribute, the # logging module can throw an error like the following: # # Traceback (most recent call last): # File "/System/Library/Frameworks/Python.framework/Versions/2.6/... # lib/python2.6/logging/__init__.py", line 761, in emit # self.stream.write(fs % msg.encode(self.stream.encoding)) # LookupError: unknown encoding: unknown if logger is None: logger = logging.getLogger() if is_verbose: logging_level = logging.DEBUG handlers = _create_debug_log_handlers(stream) else: logging_level = logging.INFO handlers = _create_log_handlers(stream) handlers = _configure_logging(logging_level=logging_level, logger=logger, handlers=handlers) return handlers # Enum-like idiom class FileType: NONE = 0 # FileType.NONE evaluates to False. # Alphabetize remaining types CPP = 1 PYTHON = 2 TEXT = 3 XML = 4 class CheckerDispatcher(object): """Supports determining whether and how to check style, based on path.""" def _file_extension(self, file_path): """Return the file extension without the leading dot.""" return os.path.splitext(file_path)[1].lstrip(".") def should_skip_with_warning(self, file_path): """Return whether the given file should be skipped with a warning.""" for skipped_file in _SKIPPED_FILES_WITH_WARNING: if file_path.find(skipped_file) >= 0: return True return False def should_skip_without_warning(self, file_path): """Return whether the given file should be skipped without a warning.""" if not self._file_type(file_path): # FileType.NONE. return True # Since "LayoutTests" is in _SKIPPED_FILES_WITHOUT_WARNING, make # an exception to prevent files like "LayoutTests/ChangeLog" and # "LayoutTests/ChangeLog-2009-06-16" from being skipped. # Files like 'test_expectations.txt' and 'drt_expectations.txt' # are also should not be skipped. # # FIXME: Figure out a good way to avoid having to add special logic # for this special case. basename = os.path.basename(file_path) if basename.startswith('ChangeLog'): return False elif basename == 'test_expectations.txt' or basename == 'drt_expectations.txt': return False for skipped_file in _SKIPPED_FILES_WITHOUT_WARNING: if file_path.find(skipped_file) >= 0: return True return False def should_check_and_strip_carriage_returns(self, file_path): return self._file_extension(file_path) not in _CARRIAGE_RETURN_ALLOWED_FILE_EXTENSIONS def _file_type(self, file_path): """Return the file type corresponding to the given file.""" file_extension = self._file_extension(file_path) if (file_extension in _CPP_FILE_EXTENSIONS) or (file_path == '-'): # FIXME: Do something about the comment below and the issue it # raises since cpp_style already relies on the extension. # # Treat stdin as C++. Since the extension is unknown when # reading from stdin, cpp_style tests should not rely on # the extension. return FileType.CPP elif file_extension == _PYTHON_FILE_EXTENSION: return FileType.PYTHON elif file_extension in _XML_FILE_EXTENSIONS: return FileType.XML elif (os.path.basename(file_path).startswith('ChangeLog') or (not file_extension and "Tools/Scripts/" in file_path) or file_extension in _TEXT_FILE_EXTENSIONS): return FileType.TEXT else: return FileType.NONE def _create_checker(self, file_type, file_path, handle_style_error, min_confidence): """Instantiate and return a style checker based on file type.""" if file_type == FileType.NONE: checker = None elif file_type == FileType.CPP: file_extension = self._file_extension(file_path) checker = CppChecker(file_path, file_extension, handle_style_error, min_confidence) elif file_type == FileType.PYTHON: checker = PythonChecker(file_path, handle_style_error) elif file_type == FileType.XML: checker = XMLChecker(file_path, handle_style_error) elif file_type == FileType.TEXT: basename = os.path.basename(file_path) if basename == 'test_expectations.txt' or basename == 'drt_expectations.txt': checker = TestExpectationsChecker(file_path, handle_style_error) else: checker = TextChecker(file_path, handle_style_error) else: raise ValueError('Invalid file type "%(file_type)s": the only valid file types ' "are %(NONE)s, %(CPP)s, and %(TEXT)s." % {"file_type": file_type, "NONE": FileType.NONE, "CPP": FileType.CPP, "TEXT": FileType.TEXT}) return checker def dispatch(self, file_path, handle_style_error, min_confidence): """Instantiate and return a style checker based on file path.""" file_type = self._file_type(file_path) checker = self._create_checker(file_type, file_path, handle_style_error, min_confidence) return checker # FIXME: Remove the stderr_write attribute from this class and replace # its use with calls to a logging module logger. class StyleProcessorConfiguration(object): """Stores configuration values for the StyleProcessor class. Attributes: min_confidence: An integer between 1 and 5 inclusive that is the minimum confidence level of style errors to report. max_reports_per_category: The maximum number of errors to report per category, per file. stderr_write: A function that takes a string as a parameter and serves as stderr.write. """ def __init__(self, filter_configuration, max_reports_per_category, min_confidence, output_format, stderr_write): """Create a StyleProcessorConfiguration instance. Args: filter_configuration: A FilterConfiguration instance. The default is the "empty" filter configuration, which means that all errors should be checked. max_reports_per_category: The maximum number of errors to report per category, per file. min_confidence: An integer between 1 and 5 inclusive that is the minimum confidence level of style errors to report. The default is 1, which reports all style errors. output_format: A string that is the output format. The supported output formats are "emacs" which emacs can parse and "vs7" which Microsoft Visual Studio 7 can parse. stderr_write: A function that takes a string as a parameter and serves as stderr.write. """ self._filter_configuration = filter_configuration self._output_format = output_format self.max_reports_per_category = max_reports_per_category self.min_confidence = min_confidence self.stderr_write = stderr_write def is_reportable(self, category, confidence_in_error, file_path): """Return whether an error is reportable. An error is reportable if both the confidence in the error is at least the minimum confidence level and the current filter says the category should be checked for the given path. Args: category: A string that is a style category. confidence_in_error: An integer between 1 and 5 inclusive that is the application's confidence in the error. A higher number means greater confidence. file_path: The path of the file being checked """ if confidence_in_error < self.min_confidence: return False return self._filter_configuration.should_check(category, file_path) def write_style_error(self, category, confidence_in_error, file_path, line_number, message): """Write a style error to the configured stderr.""" if self._output_format == 'vs7': format_string = "%s(%s): %s [%s] [%d]\n" else: format_string = "%s:%s: %s [%s] [%d]\n" self.stderr_write(format_string % (file_path, line_number, message, category, confidence_in_error)) class ProcessorBase(object): """The base class for processors of lists of lines.""" def should_process(self, file_path): """Return whether the file at file_path should be processed. The TextFileReader class calls this method prior to reading in the lines of a file. Use this method, for example, to prevent the style checker from reading binary files into memory. """ raise NotImplementedError('Subclasses should implement.') def process(self, lines, file_path, **kwargs): """Process lines of text read from a file. Args: lines: A list of lines of text to process. file_path: The path from which the lines were read. **kwargs: This argument signifies that the process() method of subclasses of ProcessorBase may support additional keyword arguments. For example, a style checker's check() method may support a "reportable_lines" parameter that represents the line numbers of the lines for which style errors should be reported. """ raise NotImplementedError('Subclasses should implement.') class StyleProcessor(ProcessorBase): """A ProcessorBase for checking style. Attributes: error_count: An integer that is the total number of reported errors for the lifetime of this instance. """ def __init__(self, configuration, mock_dispatcher=None, mock_increment_error_count=None, mock_carriage_checker_class=None): """Create an instance. Args: configuration: A StyleProcessorConfiguration instance. mock_dispatcher: A mock CheckerDispatcher instance. This parameter is for unit testing. Defaults to a CheckerDispatcher instance. mock_increment_error_count: A mock error-count incrementer. mock_carriage_checker_class: A mock class for checking and transforming carriage returns. This parameter is for unit testing. Defaults to CarriageReturnChecker. """ if mock_dispatcher is None: dispatcher = CheckerDispatcher() else: dispatcher = mock_dispatcher if mock_increment_error_count is None: # The following blank line is present to avoid flagging by pep8.py. def increment_error_count(): """Increment the total count of reported errors.""" self.error_count += 1 else: increment_error_count = mock_increment_error_count if mock_carriage_checker_class is None: # This needs to be a class rather than an instance since the # process() method instantiates one using parameters. carriage_checker_class = CarriageReturnChecker else: carriage_checker_class = mock_carriage_checker_class self.error_count = 0 self._carriage_checker_class = carriage_checker_class self._configuration = configuration self._dispatcher = dispatcher self._increment_error_count = increment_error_count def should_process(self, file_path): """Return whether the file should be checked for style.""" if self._dispatcher.should_skip_without_warning(file_path): return False if self._dispatcher.should_skip_with_warning(file_path): _log.warn('File exempt from style guide. Skipping: "%s"' % file_path) return False return True def process(self, lines, file_path, line_numbers=None): """Check the given lines for style. Arguments: lines: A list of all lines in the file to check. file_path: The path of the file to process. If possible, the path should be relative to the source root. Otherwise, path-specific logic may not behave as expected. line_numbers: A list of line numbers of the lines for which style errors should be reported, or None if errors for all lines should be reported. When not None, this list normally contains the line numbers corresponding to the modified lines of a patch. """ _log.debug("Checking style: " + file_path) style_error_handler = DefaultStyleErrorHandler( configuration=self._configuration, file_path=file_path, increment_error_count=self._increment_error_count, line_numbers=line_numbers) carriage_checker = self._carriage_checker_class(style_error_handler) # Check for and remove trailing carriage returns ("\r"). if self._dispatcher.should_check_and_strip_carriage_returns(file_path): lines = carriage_checker.check(lines) min_confidence = self._configuration.min_confidence checker = self._dispatcher.dispatch(file_path, style_error_handler, min_confidence) if checker is None: raise AssertionError("File should not be checked: '%s'" % file_path) _log.debug("Using class: " + checker.__class__.__name__) checker.check(lines)