hwdb/ids_parser.py

   1 #!/usr/bin/env python3
   2
   3 import re
   4 import sys
   5 from pyparsing import (Word, White, Literal, Regex,
   6                        LineEnd, SkipTo,
   7                        ZeroOrMore, OneOrMore, Combine, Optional, Suppress,
   8                        Group,
   9                        stringEnd, pythonStyleComment)
  10
  11 EOL = LineEnd().suppress()
  12 NUM1 = Word('0123456789abcdefABCDEF', exact=1)
  13 NUM2 = Word('0123456789abcdefABCDEF', exact=2)
  14 NUM3 = Word('0123456789abcdefABCDEF', exact=3)
  15 NUM4 = Word('0123456789abcdefABCDEF', exact=4)
  16 NUM6 = Word('0123456789abcdefABCDEF', exact=6)
  17 TAB = White('\t', exact=1).suppress()
  18 COMMENTLINE = pythonStyleComment + EOL
  19 EMPTYLINE = LineEnd()
  20 text_eol = lambda name: Regex(r'[^\n]+')(name) + EOL
  21
  22 def klass_grammar():
  23     klass_line = Literal('C ').suppress() + NUM2('klass') + text_eol('text')
  24     subclass_line = TAB + NUM2('subclass') + text_eol('text')
  25     protocol_line = TAB + TAB + NUM2('protocol') + text_eol('name')
  26     subclass = (subclass_line('SUBCLASS') -
  27                 ZeroOrMore(Group(protocol_line)('PROTOCOLS*')
  28                            ^ COMMENTLINE.suppress()))
  29     klass = (klass_line('KLASS') -
  30              ZeroOrMore(Group(subclass)('SUBCLASSES*')
  31                         ^ COMMENTLINE.suppress()))
  32     return klass
  33
  34 def usb_ids_grammar():
  35     vendor_line = NUM4('vendor') + text_eol('text')
  36     device_line = TAB + NUM4('device') + text_eol('text')
  37     vendor = (vendor_line('VENDOR') +
  38               ZeroOrMore(Group(device_line)('VENDOR_DEV*') ^ COMMENTLINE.suppress()))
  39
  40     klass = klass_grammar()
  41
  42     other_line = (Literal('AT ') ^ Literal('HID ') ^ Literal('R ')
  43                   ^ Literal('PHY ') ^ Literal('BIAS ') ^ Literal('HUT ')
  44                   ^ Literal('L ') ^ Literal('VT ') ^ Literal('HCC ')) + text_eol('text')
  45     other_group = (other_line - ZeroOrMore(TAB + text_eol('text')))
  46
  47     commentgroup = OneOrMore(COMMENTLINE).suppress() ^ EMPTYLINE.suppress()
  48     grammar = OneOrMore(Group(vendor)('VENDORS*')
  49                         ^ Group(klass)('CLASSES*')
  50                         ^ other_group.suppress() ^ commentgroup) + stringEnd()
  51
  52     grammar.parseWithTabs()
  53     return grammar
  54
  55 def pci_ids_grammar():
  56     vendor_line = NUM4('vendor') + text_eol('text')
  57     device_line = TAB + NUM4('device') + text_eol('text')
  58     subvendor_line = TAB + TAB + NUM4('a') + White(' ') + NUM4('b') + text_eol('name')
  59
  60     device = (device_line('DEVICE') +
  61               ZeroOrMore(Group(subvendor_line)('SUBVENDORS*') ^ COMMENTLINE.suppress()))
  62     vendor = (vendor_line('VENDOR') +
  63               ZeroOrMore(Group(device)('DEVICES*') ^ COMMENTLINE.suppress()))
  64
  65     klass = klass_grammar()
  66
  67     commentgroup = OneOrMore(COMMENTLINE).suppress() ^ EMPTYLINE.suppress()
  68     grammar = OneOrMore(Group(vendor)('VENDORS*')
  69                         ^ Group(klass)('CLASSES*')
  70                         ^ commentgroup) + stringEnd()
  71
  72     grammar.parseWithTabs()
  73     return grammar
  74
  75 def sdio_ids_grammar():
  76     vendor_line = NUM4('vendor') + text_eol('text')
  77     device_line = TAB + NUM4('device') + text_eol('text')
  78     vendor = (vendor_line('VENDOR') +
  79               ZeroOrMore(Group(device_line)('DEVICES*') ^ COMMENTLINE.suppress()))
  80
  81     klass = klass_grammar()
  82
  83     commentgroup = OneOrMore(COMMENTLINE).suppress() ^ EMPTYLINE.suppress()
  84     grammar = OneOrMore(Group(vendor)('VENDORS*')
  85                         ^ Group(klass)('CLASSES*')
  86                         ^ commentgroup) + stringEnd()
  87
  88     grammar.parseWithTabs()
  89     return grammar
  90
  91 def oui_grammar(type):
  92     prefix_line = (Combine(NUM2 - Suppress('-') - NUM2 - Suppress('-') - NUM2)('prefix')
  93                    - Literal('(hex)') -  text_eol('text'))
  94     if type == 'small':
  95         vendor_line = (NUM3('start') - '000-' - NUM3('end') - 'FFF'
  96                        - Literal('(base 16)') - text_eol('text2'))
  97     elif type == 'medium':
  98         vendor_line = (NUM1('start') - '00000-' - NUM1('end') - 'FFFFF'
  99                        - Literal('(base 16)') - text_eol('text2'))
 100     else:
 101         assert type == 'large'
 102         vendor_line = (NUM6('start')
 103                        - Literal('(base 16)') - text_eol('text2'))
 104
 105     extra_line = TAB - TAB - TAB - TAB - SkipTo(EOL)
 106     vendor = prefix_line + vendor_line + ZeroOrMore(extra_line) + Optional(EMPTYLINE)
 107
 108     grammar = (Literal('OUI') + text_eol('header')
 109                + text_eol('header') + text_eol('header') + EMPTYLINE
 110                + OneOrMore(Group(vendor)('VENDORS*')) + stringEnd())
 111
 112     grammar.parseWithTabs()
 113     return grammar
 114
 115
 116 def header(file, *sources):
 117     print('''\
 118 # This file is part of systemd.
 119 #
 120 # Data imported from:{}{}'''.format(' ' if len(sources) == 1 else '\n#   ',
 121                                     '\n#   '.join(sources)),
 122           file=file)
 123
 124 def add_item(items, key, value):
 125     if key in items:
 126         print(f'Ignoring duplicate entry: {key} = "{items[key]}", "{value}"')
 127     else:
 128         items[key] = value
 129
 130 def usb_vendor_model(p):
 131     items = {}
 132
 133     for vendor_group in p.VENDORS:
 134         vendor = vendor_group.vendor.upper()
 135         text = vendor_group.text.strip()
 136         add_item(items, (vendor,), text)
 137
 138         for vendor_dev in vendor_group.VENDOR_DEV:
 139             device = vendor_dev.device.upper()
 140             text = vendor_dev.text.strip()
 141             add_item(items, (vendor, device), text)
 142
 143     with open('20-usb-vendor-model.hwdb', 'wt') as out:
 144         header(out, 'http://www.linux-usb.org/usb.ids')
 145
 146         for key in sorted(items):
 147             if len(key) == 1:
 148                 p, n = 'usb:v{}*', 'VENDOR'
 149             else:
 150                 p, n = 'usb:v{}p{}*', 'MODEL',
 151             print('', p.format(*key),
 152                   f' ID_{n}_FROM_DATABASE={items[key]}', sep='\n', file=out)
 153
 154     print(f'Wrote {out.name}')
 155
 156 def usb_classes(p):
 157     items = {}
 158
 159     for klass_group in p.CLASSES:
 160         klass = klass_group.klass.upper()
 161         text = klass_group.text.strip()
 162
 163         if klass != '00' and not re.match(r'(\?|None|Unused)\s*$', text):
 164             add_item(items, (klass,), text)
 165
 166         for subclass_group in klass_group.SUBCLASSES:
 167             subclass = subclass_group.subclass.upper()
 168             text = subclass_group.text.strip()
 169             if subclass != '00' and not re.match(r'(\?|None|Unused)\s*$', text):
 170                 add_item(items, (klass, subclass), text)
 171
 172             for protocol_group in subclass_group.PROTOCOLS:
 173                 protocol = protocol_group.protocol.upper()
 174                 text = protocol_group.name.strip()
 175                 if klass != '00' and not re.match(r'(\?|None|Unused)\s*$', text):
 176                     add_item(items, (klass, subclass, protocol), text)
 177
 178     with open('20-usb-classes.hwdb', 'wt') as out:
 179         header(out, 'http://www.linux-usb.org/usb.ids')
 180
 181         for key in sorted(items):
 182             if len(key) == 1:
 183                 p, n = 'usb:v*p*d*dc{}*', 'CLASS'
 184             elif len(key) == 2:
 185                 p, n = 'usb:v*p*d*dc{}dsc{}*', 'SUBCLASS'
 186             else:
 187                 p, n = 'usb:v*p*d*dc{}dsc{}dp{}*', 'PROTOCOL'
 188             print('', p.format(*key),
 189                   f' ID_USB_{n}_FROM_DATABASE={items[key]}', sep='\n', file=out)
 190
 191     print(f'Wrote {out.name}')
 192
 193 def pci_vendor_model(p):
 194     items = {}
 195
 196     for vendor_group in p.VENDORS:
 197         vendor = vendor_group.vendor.upper()
 198         text = vendor_group.text.strip()
 199         add_item(items, (vendor,), text)
 200
 201         for device_group in vendor_group.DEVICES:
 202             device = device_group.device.upper()
 203             text = device_group.text.strip()
 204             add_item(items, (vendor, device), text)
 205
 206             for subvendor_group in device_group.SUBVENDORS:
 207                 sub_vendor = subvendor_group.a.upper()
 208                 sub_model = subvendor_group.b.upper()
 209                 sub_text = subvendor_group.name.strip()
 210                 if sub_text.startswith(text):
 211                     sub_text = sub_text[len(text):].lstrip()
 212                 if sub_text:
 213                     sub_text = f' ({sub_text})'
 214                 add_item(items, (vendor, device, sub_vendor, sub_model), text + sub_text)
 215
 216     with open('20-pci-vendor-model.hwdb', 'wt') as out:
 217         header(out, 'http://pci-ids.ucw.cz/v2.2/pci.ids')
 218
 219         for key in sorted(items):
 220             if len(key) == 1:
 221                 p, n = 'pci:v0000{}*', 'VENDOR'
 222             elif len(key) == 2:
 223                 p, n = 'pci:v0000{}d0000{}*', 'MODEL'
 224             else:
 225                 p, n = 'pci:v0000{}d0000{}sv0000{}sd0000{}*', 'MODEL'
 226             print('', p.format(*key),
 227                   f' ID_{n}_FROM_DATABASE={items[key]}', sep='\n', file=out)
 228
 229     print(f'Wrote {out.name}')
 230
 231 def pci_classes(p):
 232     items = {}
 233
 234     for klass_group in p.CLASSES:
 235         klass = klass_group.klass.upper()
 236         text = klass_group.text.strip()
 237         add_item(items, (klass,), text)
 238
 239         for subclass_group in klass_group.SUBCLASSES:
 240             subclass = subclass_group.subclass.upper()
 241             text = subclass_group.text.strip()
 242             add_item(items, (klass, subclass), text)
 243
 244             for protocol_group in subclass_group.PROTOCOLS:
 245                 protocol = protocol_group.protocol.upper()
 246                 text = protocol_group.name.strip()
 247                 add_item(items, (klass, subclass, protocol), text)
 248
 249     with open('20-pci-classes.hwdb', 'wt') as out:
 250         header(out, 'http://pci-ids.ucw.cz/v2.2/pci.ids')
 251
 252         for key in sorted(items):
 253             if len(key) == 1:
 254                 p, n = 'pci:v*d*sv*sd*bc{}*', 'CLASS'
 255             elif len(key) == 2:
 256                 p, n = 'pci:v*d*sv*sd*bc{}sc{}*', 'SUBCLASS'
 257             else:
 258                 p, n = 'pci:v*d*sv*sd*bc{}sc{}i{}*', 'INTERFACE'
 259             print('', p.format(*key),
 260                   f' ID_PCI_{n}_FROM_DATABASE={items[key]}', sep='\n', file=out)
 261
 262     print(f'Wrote {out.name}')
 263
 264 def sdio_vendor_model(p):
 265     items = {}
 266
 267     for vendor_group in p.VENDORS:
 268         vendor = vendor_group.vendor.upper()
 269         text = vendor_group.text.strip()
 270         add_item(items, (vendor,), text)
 271
 272         for device_group in vendor_group.DEVICES:
 273             device = device_group.device.upper()
 274             text = device_group.text.strip()
 275             add_item(items, (vendor, device), text)
 276
 277     with open('20-sdio-vendor-model.hwdb', 'wt') as out:
 278         header(out, 'hwdb/sdio.ids')
 279
 280         for key in sorted(items):
 281             if len(key) == 1:
 282                 p, n = 'sdio:c*v{}*', 'VENDOR'
 283             else:
 284                 p, n = 'sdio:c*v{}d{}*', 'MODEL'
 285             print('', p.format(*key),
 286                   f' ID_{n}_FROM_DATABASE={items[key]}', sep='\n', file=out)
 287
 288     print(f'Wrote {out.name}')
 289
 290 def sdio_classes(p):
 291     items = {}
 292
 293     for klass_group in p.CLASSES:
 294         klass = klass_group.klass.upper()
 295         text = klass_group.text.strip()
 296         add_item(items, klass, text)
 297
 298     with open('20-sdio-classes.hwdb', 'wt') as out:
 299         header(out, 'hwdb/sdio.ids')
 300
 301         for klass in sorted(items):
 302             print(f'',
 303                   f'sdio:c{klass}v*d*',
 304                   f' ID_SDIO_CLASS_FROM_DATABASE={items[klass]}', sep='\n', file=out)
 305
 306     print(f'Wrote {out.name}')
 307
 308 # MAC Address Block Large/Medium/Small
 309 # Large  MA-L 24/24 bit (OUI)
 310 # Medium MA-M 28/20 bit (OUI prefix owned by IEEE)
 311 # Small  MA-S 36/12 bit (OUI prefix owned by IEEE)
 312 def oui(p1, p2, p3):
 313     prefixes = set()
 314     items = {}
 315
 316     for p, check in ((p1, False), (p2, False), (p3, True)):
 317         for vendor_group in p.VENDORS:
 318             prefix = vendor_group.prefix.upper()
 319             if check:
 320                 if prefix in prefixes:
 321                     continue
 322             else:
 323                 prefixes.add(prefix)
 324             start = vendor_group.start.upper()
 325             end = vendor_group.end.upper()
 326
 327             if end and start != end:
 328                 print(f'{prefix:} {start} != {end}', file=sys.stderr)
 329             text = vendor_group.text.strip()
 330
 331             key = prefix + start if end else prefix
 332             add_item(items, key, text)
 333
 334     with open('20-OUI.hwdb', 'wt') as out:
 335         header(out,
 336                'https://services13.ieee.org/RST/standards-ra-web/rest/assignments/download/?registry=MA-L&format=txt',
 337                'https://services13.ieee.org/RST/standards-ra-web/rest/assignments/download/?registry=MA-M&format=txt',
 338                'https://services13.ieee.org/RST/standards-ra-web/rest/assignments/download/?registry=MA-S&format=txt')
 339
 340         for pattern in sorted(items):
 341             print(f'',
 342                   f'OUI:{pattern}*',
 343                   f' ID_OUI_FROM_DATABASE={items[pattern]}', sep='\n', file=out)
 344
 345     print(f'Wrote {out.name}')
 346
 347 if __name__ == '__main__':
 348     args = sys.argv[1:]
 349
 350     if not args or 'usb' in args:
 351         p = usb_ids_grammar().parseFile(open('usb.ids', errors='replace'))
 352         usb_vendor_model(p)
 353         usb_classes(p)
 354
 355     if not args or 'pci' in args:
 356         p = pci_ids_grammar().parseFile(open('pci.ids', errors='replace'))
 357         pci_vendor_model(p)
 358         pci_classes(p)
 359
 360     if not args or 'sdio' in args:
 361         p = pci_ids_grammar().parseFile(open('sdio.ids', errors='replace'))
 362         sdio_vendor_model(p)
 363         sdio_classes(p)
 364
 365     if not args or 'oui' in args:
 366         p = oui_grammar('small').parseFile(open('ma-small.txt'))
 367         p2 = oui_grammar('medium').parseFile(open('ma-medium.txt'))
 368         p3 = oui_grammar('large').parseFile(open('ma-large.txt'))
 369         oui(p, p2, p3)