Source code for pcapkit.vendor.ipv6.extension_header

# -*- coding: utf-8 -*-
"""IPv6 Extension Header Types
=================================

.. module:: pcapkit.vendor.ipv6.extension_header

This module contains the vendor crawler for **IPv6 Extension Header Types**,
which is automatically generating :class:`pcapkit.const.ipv6.extension_header.ExtensionHeader`.

"""

import collections
import csv
import re
import sys
from typing import TYPE_CHECKING

from pcapkit.vendor.default import Vendor

if TYPE_CHECKING:
    from collections import Counter
    from typing import Callable

__all__ = ['ExtensionHeader']

LINE = lambda NAME, DOCS, ENUM, MODL: f'''\
# -*- coding: utf-8 -*-
# pylint: disable=line-too-long,consider-using-f-string
"""{(name := DOCS.split(' [', maxsplit=1)[0])}
{'=' * (len(name) + 6)}

.. module:: {MODL.replace('vendor', 'const')}

This module contains the constant enumeration for **{name}**,
which is automatically generated from :class:`{MODL}.{NAME}`.

"""

from aenum import IntEnum, extend_enum

__all__ = ['{NAME}']


class {NAME}(IntEnum):
    """[{NAME}] {DOCS}"""

    {ENUM}

    @staticmethod
    def get(key: 'int | str', default: 'int' = -1) -> '{NAME}':
        """Backport support for original codes.

        Args:
            key: Key to get enum item.
            default: Default value if not found.

        :meta private:
        """
        if isinstance(key, int):
            return {NAME}(key)
        return {NAME}[key]  # type: ignore[misc]
'''  # type: Callable[[str, str, str, str], str]


[docs] class ExtensionHeader(Vendor): """IPv6 Extension Header Types""" #: Link to registry. LINK = 'https://www.iana.org/assignments/protocol-numbers/protocol-numbers-1.csv' def count(self, data: 'list[str]') -> 'Counter[str]': """Count field records. Args: data: CSV data. Returns: Field recordings. """ reader = csv.reader(data) next(reader) # header return collections.Counter(map(lambda item: self.safe_name(item[1] or item[2]), filter(lambda item: len(item[0].split('-')) != 2, reader))) def process(self, data: 'list[str]') -> 'tuple[list[str], list[str]]': """Process CSV data. Args: data: CSV data. Returns: Enumeration fields and missing fields. """ reader = csv.reader(data) next(reader) # header enum = [] # type: list[str] miss = [] # type: list[str] for item in reader: flag = item[3] if flag != 'Y': continue name = item[1] rfcs = item[4] temp = [] # type: list[str] for rfc in filter(None, re.split(r'\[|\]', rfcs)): if 'RFC' in rfc and re.match(r'\d+', rfc[3:]): #temp.append(f'[{rfc[:3]} {rfc[3:]}]') temp.append(f'[:rfc:`{rfc[3:]}`]') else: temp.append(f'[{rfc}]'.replace('_', ' ')) lrfc = re.sub(r'( )( )*', ' ', f" {''.join(temp)}".replace('\n', ' ')) if rfcs else '' subd = re.sub(r'( )( )*', ' ', item[2].replace('\n', ' ')) tmp1 = f' {subd}' if item[2] else '' split = name.split(' (', 1) if len(split) == 2: name, cmmt = split[0], f" ({split[1]}" else: name, cmmt = name, '' # pylint: disable=self-assigning-variable if name: tmp1 = f',{tmp1}' if tmp1 else '' else: name, tmp1 = item[2], '' desc = self.wrap_comment(f'{name}{tmp1}{lrfc}{cmmt}') try: code, _ = item[0], int(item[0]) if not name: name, desc = item[2], '' renm = self.rename(name, code, original=item[1]) pres = f"{renm} = {code}" sufs = f"#: {desc}" #if len(pres) > 74: # sufs = f"\n{' '*80}{sufs}" #enum.append(f'{pres.ljust(76)}{sufs}') enum.append(f'{sufs}\n {pres}') except ValueError: start, stop = item[0].split('-') miss.append(f'if {start} <= value <= {stop}:') miss.append(f' #: {desc}') miss.append(f" return extend_enum(cls, '{self.safe_name(name)}_%d' % value, value)") return enum, miss def context(self, data: 'list[str]') -> 'str': """Generate constant context. Args: data: CSV data. Returns: Constant context. """ enum, _ = self.process(data) ENUM = '\n\n '.join(map(lambda s: s.rstrip(), enum)) return LINE(self.NAME, self.DOCS, ENUM, self.__module__)
if __name__ == '__main__': sys.exit(ExtensionHeader()) # type: ignore[arg-type]