AlkantarClanX12
Current Path : /opt/cloudlinux/venv/lib64/python3.11/site-packages/pylint/utils/ |
Current File : //opt/cloudlinux/venv/lib64/python3.11/site-packages/pylint/utils/utils.py |
# Licensed under the GPL: https://www.gnu.org/licenses/old-licenses/gpl-2.0.html # For details: https://github.com/PyCQA/pylint/blob/main/LICENSE # Copyright (c) https://github.com/PyCQA/pylint/blob/main/CONTRIBUTORS.txt from __future__ import annotations try: import isort.api import isort.settings HAS_ISORT_5 = True except ImportError: # isort < 5 import isort HAS_ISORT_5 = False import argparse import codecs import os import re import sys import textwrap import tokenize import warnings from collections.abc import Sequence from io import BufferedReader, BytesIO from typing import ( TYPE_CHECKING, Any, List, Pattern, TextIO, Tuple, TypeVar, Union, overload, ) from astroid import Module, modutils, nodes from pylint.constants import PY_EXTS from pylint.typing import OptionDict if sys.version_info >= (3, 8): from typing import Literal else: from typing_extensions import Literal if TYPE_CHECKING: from pylint.checkers.base_checker import BaseChecker from pylint.lint import PyLinter DEFAULT_LINE_LENGTH = 79 # These are types used to overload get_global_option() and refer to the options type GLOBAL_OPTION_BOOL = Literal[ "suggestion-mode", "analyse-fallback-blocks", "allow-global-unused-variables", ] GLOBAL_OPTION_INT = Literal["max-line-length", "docstring-min-length"] GLOBAL_OPTION_LIST = Literal["ignored-modules"] GLOBAL_OPTION_PATTERN = Literal[ "no-docstring-rgx", "dummy-variables-rgx", "ignored-argument-names", "mixin-class-rgx", ] GLOBAL_OPTION_PATTERN_LIST = Literal["exclude-too-few-public-methods", "ignore-paths"] GLOBAL_OPTION_TUPLE_INT = Literal["py-version"] GLOBAL_OPTION_NAMES = Union[ GLOBAL_OPTION_BOOL, GLOBAL_OPTION_INT, GLOBAL_OPTION_LIST, GLOBAL_OPTION_PATTERN, GLOBAL_OPTION_PATTERN_LIST, GLOBAL_OPTION_TUPLE_INT, ] T_GlobalOptionReturnTypes = TypeVar( "T_GlobalOptionReturnTypes", bool, int, List[str], Pattern[str], List[Pattern[str]], Tuple[int, ...], ) def normalize_text( text: str, line_len: int = DEFAULT_LINE_LENGTH, indent: str = "" ) -> str: """Wrap the text on the given line length.""" return "\n".join( textwrap.wrap( text, width=line_len, initial_indent=indent, subsequent_indent=indent ) ) CMPS = ["=", "-", "+"] # py3k has no more cmp builtin def cmp(a: int | float, b: int | float) -> int: return (a > b) - (a < b) def diff_string(old: int | float, new: int | float) -> str: """Given an old and new int value, return a string representing the difference. """ diff = abs(old - new) diff_str = f"{CMPS[cmp(old, new)]}{diff and f'{diff:.2f}' or ''}" return diff_str def get_module_and_frameid(node: nodes.NodeNG) -> tuple[str, str]: """Return the module name and the frame id in the module.""" frame = node.frame(future=True) module, obj = "", [] while frame: if isinstance(frame, Module): module = frame.name else: obj.append(getattr(frame, "name", "<lambda>")) try: frame = frame.parent.frame(future=True) except AttributeError: break obj.reverse() return module, ".".join(obj) def get_rst_title(title: str, character: str) -> str: """Permit to get a title formatted as ReStructuredText test (underlined with a chosen character). """ return f"{title}\n{character * len(title)}\n" def get_rst_section( section: str | None, options: list[tuple[str, OptionDict, Any]], doc: str | None = None, ) -> str: """Format an option's section using as a ReStructuredText formatted output.""" result = "" if section: result += get_rst_title(section, "'") if doc: formatted_doc = normalize_text(doc) result += f"{formatted_doc}\n\n" for optname, optdict, value in options: help_opt = optdict.get("help") result += f":{optname}:\n" if help_opt: assert isinstance(help_opt, str) formatted_help = normalize_text(help_opt, indent=" ") result += f"{formatted_help}\n" if value and optname != "py-version": value = str(_format_option_value(optdict, value)) result += f"\n Default: ``{value.replace('`` ', '```` ``')}``\n" return result def decoding_stream( stream: BufferedReader | BytesIO, encoding: str, errors: Literal["strict"] = "strict", ) -> codecs.StreamReader: try: reader_cls = codecs.getreader(encoding or sys.getdefaultencoding()) except LookupError: reader_cls = codecs.getreader(sys.getdefaultencoding()) return reader_cls(stream, errors) def tokenize_module(node: nodes.Module) -> list[tokenize.TokenInfo]: with node.stream() as stream: readline = stream.readline return list(tokenize.tokenize(readline)) def register_plugins(linter: PyLinter, directory: str) -> None: """Load all module and package in the given directory, looking for a 'register' function in each one, used to register pylint checkers. """ imported = {} for filename in os.listdir(directory): base, extension = os.path.splitext(filename) if base in imported or base == "__pycache__": continue if ( extension in PY_EXTS and base != "__init__" or ( not extension and os.path.isdir(os.path.join(directory, base)) and not filename.startswith(".") ) ): try: module = modutils.load_module_from_file( os.path.join(directory, filename) ) except ValueError: # empty module name (usually Emacs auto-save files) continue except ImportError as exc: print(f"Problem importing module {filename}: {exc}", file=sys.stderr) else: if hasattr(module, "register"): module.register(linter) imported[base] = 1 @overload def get_global_option( checker: BaseChecker, option: GLOBAL_OPTION_BOOL, default: bool | None = ... ) -> bool: ... @overload def get_global_option( checker: BaseChecker, option: GLOBAL_OPTION_INT, default: int | None = ... ) -> int: ... @overload def get_global_option( checker: BaseChecker, option: GLOBAL_OPTION_LIST, default: list[str] | None = ..., ) -> list[str]: ... @overload def get_global_option( checker: BaseChecker, option: GLOBAL_OPTION_PATTERN, default: Pattern[str] | None = ..., ) -> Pattern[str]: ... @overload def get_global_option( checker: BaseChecker, option: GLOBAL_OPTION_PATTERN_LIST, default: list[Pattern[str]] | None = ..., ) -> list[Pattern[str]]: ... @overload def get_global_option( checker: BaseChecker, option: GLOBAL_OPTION_TUPLE_INT, default: tuple[int, ...] | None = ..., ) -> tuple[int, ...]: ... def get_global_option( checker: BaseChecker, option: GLOBAL_OPTION_NAMES, default: T_GlobalOptionReturnTypes | None = None, # pylint: disable=unused-argument ) -> T_GlobalOptionReturnTypes | None | Any: """DEPRECATED: Retrieve an option defined by the given *checker* or by all known option providers. It will look in the list of all options providers until the given *option* will be found. If the option wasn't found, the *default* value will be returned. """ warnings.warn( "get_global_option has been deprecated. You can use " "checker.linter.config to get all global options instead.", DeprecationWarning, stacklevel=2, ) return getattr(checker.linter.config, option.replace("-", "_")) def _splitstrip(string: str, sep: str = ",") -> list[str]: """Return a list of stripped string by splitting the string given as argument on `sep` (',' by default), empty strings are discarded. >>> _splitstrip('a, b, c , 4,,') ['a', 'b', 'c', '4'] >>> _splitstrip('a') ['a'] >>> _splitstrip('a,\nb,\nc,') ['a', 'b', 'c'] :type string: str or unicode :param string: a csv line :type sep: str or unicode :param sep: field separator, default to the comma (',') :rtype: str or unicode :return: the unquoted string (or the input string if it wasn't quoted) """ return [word.strip() for word in string.split(sep) if word.strip()] def _unquote(string: str) -> str: """Remove optional quotes (simple or double) from the string. :param string: an optionally quoted string :return: the unquoted string (or the input string if it wasn't quoted) """ if not string: return string if string[0] in "\"'": string = string[1:] if string[-1] in "\"'": string = string[:-1] return string def _check_csv(value: list[str] | tuple[str] | str) -> Sequence[str]: if isinstance(value, (list, tuple)): return value return _splitstrip(value) def _comment(string: str) -> str: """Return string as a comment.""" lines = [line.strip() for line in string.splitlines()] sep = "\n" return "# " + f"{sep}# ".join(lines) def _format_option_value(optdict: OptionDict, value: Any) -> str: """Return the user input's value from a 'compiled' value. TODO: 3.0: Remove deprecated function """ if optdict.get("type", None) == "py_version": value = ".".join(str(item) for item in value) elif isinstance(value, (list, tuple)): value = ",".join(_format_option_value(optdict, item) for item in value) elif isinstance(value, dict): value = ",".join(f"{k}:{v}" for k, v in value.items()) elif hasattr(value, "match"): # optdict.get('type') == 'regexp' # compiled regexp value = value.pattern elif optdict.get("type") == "yn": value = "yes" if value else "no" elif isinstance(value, str) and value.isspace(): value = f"'{value}'" return str(value) def format_section( stream: TextIO, section: str, options: list[tuple[str, OptionDict, Any]], doc: str | None = None, ) -> None: """Format an option's section using the INI format.""" warnings.warn( "format_section has been deprecated. It will be removed in pylint 3.0.", DeprecationWarning, stacklevel=2, ) if doc: print(_comment(doc), file=stream) print(f"[{section}]", file=stream) with warnings.catch_warnings(): warnings.filterwarnings("ignore", category=DeprecationWarning) _ini_format(stream, options) def _ini_format(stream: TextIO, options: list[tuple[str, OptionDict, Any]]) -> None: """Format options using the INI format.""" warnings.warn( "_ini_format has been deprecated. It will be removed in pylint 3.0.", DeprecationWarning, stacklevel=2, ) for optname, optdict, value in options: # Skip deprecated option if "kwargs" in optdict: assert isinstance(optdict["kwargs"], dict) if "new_names" in optdict["kwargs"]: continue value = _format_option_value(optdict, value) help_opt = optdict.get("help") if help_opt: assert isinstance(help_opt, str) help_opt = normalize_text(help_opt, indent="# ") print(file=stream) print(help_opt, file=stream) else: print(file=stream) if value in {"None", "False"}: print(f"#{optname}=", file=stream) else: value = str(value).strip() if re.match(r"^([\w-]+,)+[\w-]+$", str(value)): separator = "\n " + " " * len(optname) value = separator.join(x + "," for x in str(value).split(",")) # remove trailing ',' from last element of the list value = value[:-1] print(f"{optname}={value}", file=stream) class IsortDriver: """A wrapper around isort API that changed between versions 4 and 5.""" def __init__(self, config: argparse.Namespace) -> None: if HAS_ISORT_5: self.isort5_config = isort.settings.Config( # There is no typo here. EXTRA_standard_library is # what most users want. The option has been named # KNOWN_standard_library for ages in pylint, and we # don't want to break compatibility. extra_standard_library=config.known_standard_library, known_third_party=config.known_third_party, ) else: # pylint: disable-next=no-member self.isort4_obj = isort.SortImports( # type: ignore[attr-defined] file_contents="", known_standard_library=config.known_standard_library, known_third_party=config.known_third_party, ) def place_module(self, package: str) -> str: if HAS_ISORT_5: return isort.api.place_module(package, self.isort5_config) return self.isort4_obj.place_module(package) # type: ignore[no-any-return]