# -*- coding: utf-8 -*-
"""Ethertype IEEE 802 Numbers
================================
.. module:: pcapkit.vendor.reg.ethertype
This module contains the vendor crawler for **Ethertype IEEE 802 Numbers**,
which is automatically generating :class:`pcapkit.const.reg.ethertype.EtherType`.
"""
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 Optional
__all__ = ['EtherType']
[docs]
class EtherType(Vendor):
"""Ethertype IEEE 802 Numbers"""
#: Value limit checker.
FLAG = 'isinstance(value, int) and 0x0000 <= value <= 0xFFFF'
#: Link to registry.
LINK = 'https://www.iana.org/assignments/ieee-802-numbers/ieee-802-numbers-1.csv'
#: Row labels that the historical Neil Sembower list itself marks as
#: carrying no real assignment -- a company holding the block but naming
#: nothing (``DEC Unassigned``), or a range the list says is dead/invalid
#: outright -- rather than a proprietary protocol's real name. The owner's
#: original ruling on :issue:`775`/:issue:`841` held these two out as the only rows to
#: convert to
#: :meth:`~pcapkit.corekit.enum.EnumRegistry._unregistered_member`, on the
#: theory that a proprietary protocol's company name IS
#: the final concrete name for every other row. :issue:`775`'s final round
#: converts every row alike, so this set does not decide *whether* a
#: row registers -- only *how its name is spelled*: these two keep the
#: bare label with no hex suffix, since the label itself already says
#: nothing was assigned, while every other row keeps the hex-suffixed
#: name it always rendered.
UNASSIGNED_ROW_NAMES = frozenset({
'DEC Unassigned',
'Old Xerox Experimental values. Invalid as an Ethertype since 1983.',
})
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[4]),
filter(lambda item: len(item[1].split('-')) != 2, reader)))
def rename(self, name: 'str', code: 'str', *, original: 'Optional[str]' = None) -> 'str': # pylint: disable=redefined-outer-name
"""Rename duplicated fields.
Args:
name: Field name.
code: Field code.
Keyword Args:
original: Original field name (extracted from CSV records).
Returns:
Revised field name.
"""
if self.record[self.safe_name(name)] > 1:
name = f'{name}_0x{code}'
return self.safe_name(name)
@staticmethod
def _insert_range(ranges: 'list[tuple[int, int, list[str]]]', entry: 'tuple[int, int, list[str]]') -> 'None':
"""Insert a range block ahead of the first already-placed range that fully contains it.
The generated :meth:`~EtherType._missing_` tests these ``if`` blocks in
list order and returns on the first match, so a range that is fully
contained within an earlier, wider one would never be reached -- see
GitHub issue :issue:`862`, where the source table lists ``0x0101-0x01FF``
after the ``0x0000-0x05DC`` range that wholly contains it.
The IANA CSV is not otherwise sorted by range, so this only reorders
the pairs where containment would otherwise shadow one of them;
every other pair keeps the source table's row order, appended in the
order it was read.
Args:
ranges: Range blocks already placed for emission, as
``(start, stop, lines)``.
entry: The ``(start, stop, lines)`` tuple to place.
"""
lo, hi, _ = entry
for index, (rlo, rhi, _) in enumerate(ranges):
if rlo <= lo and hi <= rhi:
ranges.insert(index, entry)
return
ranges.append(entry)
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]
ranges = [] # type: list[tuple[int, int, list[str]]]
for item in reader:
name = item[4]
rfcs = item[5]
temp = [] # type: list[str]
for rfc in self.split_references(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('_', ' '))
tmp1 = re.sub(r'( )( )*', ' ', f"{''.join(temp)}".replace('\n', ' ')) if rfcs else ''
tmp2 = re.sub(r'\r*\n', ' ', tmp1, flags=re.MULTILINE)
tmp3 = name.replace('\n', ' ')
desc = self.wrap_comment(f"{tmp3} {tmp2}")
try:
code, _ = item[1], int(item[1], base=16)
renm = re.sub(r'( )( )*', ' ', self.rename(name, code).replace('\n', ' '))
pres = f'{renm} = 0x{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[1].split('-')
block = [f'if 0x{start} <= value <= 0x{stop}:', f' #: {desc}']
if name in self.UNASSIGNED_ROW_NAMES:
block.append(f" return cls._unregistered_member(value, '{self.safe_name(name)}')")
else:
block.append(f" return cls._unregistered_member(value, '{self.safe_name(name)}_0x%s' % hex(value)[2:].upper().zfill(4))") # pylint: disable=line-too-long
self._insert_range(ranges, (int(start, base=16), int(stop, base=16), block))
miss = [line for _, _, block in ranges for line in block] # type: list[str]
return enum, miss
if __name__ == '__main__':
sys.exit(EtherType()) # type: ignore[arg-type]