tools/make-bluetooth.py

   1 #!/usr/bin/env python3
   2 #
   3 # Wireshark - Network traffic analyzer
   4 # By Gerald Combs <gerald@wireshark.org>
   5 # Copyright 1998 Gerald Combs
   6 #
   7 # SPDX-License-Identifier: GPL-2.0-or-later
   8 #
   9 '''
  10 make-bluetooth - Generate value_strings containing bluetooth uuids and company identifiers.
  11 It makes use of the databases from
  12 The Bluetooth SIG Repository: https://bitbucket.org/bluetooth-SIG/public/src/main/assigned_numbers/
  13 and processes the YAML into human-readable strings to go into packet-bluetooth.c.
  14 '''
  15
  16 import sys
  17 import urllib.request, urllib.error, urllib.parse
  18 import yaml
  19
  20 base_url = "https://bitbucket.org/bluetooth-SIG/public/raw/HEAD/assigned_numbers/"
  21
  22 MIN_UUIDS = 1400       # 1424 as of 31-12-2023
  23 MIN_COMPANY_IDS = 3400 # 3405 as of 31-12-2023
  24
  25 ##
  26 ## UUIDs
  27 ##
  28
  29 '''
  30 List of all YAML files to retrieve, the lists of UUIDs to put into the value_string
  31 and other information.
  32 Unfortunately the encoding of the names among the YAML files is inconsistent,
  33 to say the least. This will need post-processing.
  34 Also the previous value_string contained additional uuids, which are not currently
  35 present in the databases. Prepare the lists with these uuids so they are not lost.
  36 When they do appear in the databases they must be removed here.
  37 '''
  38
  39 uuids_sources = [
  40 {   # 0x0001
  41     "yaml": "protocol_identifiers.yaml",
  42     "description": "Protocol Identifiers",
  43     "unCamelCase": True,
  44     "unlist": [],
  45     "list": [
  46         { "uuid": 0x001D, "name": "UDI C-Plane" },
  47     ]
  48 },
  49 {   # 0x1000
  50     "yaml": "service_class.yaml",
  51     "description": "Service Class",
  52     "unCamelCase": True,
  53     "unlist": [],
  54     "list": [
  55         # Then we have this weird one stuck in between "Service Class"
  56         # from browse_group_identifiers.yaml
  57         { "uuid": 0x1002, "name": "Public Browse Group" },
  58         # And some from other sources
  59         { "uuid": 0x1129, "name": "Video Conferencing GW" },
  60         { "uuid": 0x112A, "name": "UDI MT" },
  61         { "uuid": 0x112B, "name": "UDI TA" },
  62         { "uuid": 0x112C, "name": "Audio/Video" },
  63     ]
  64 },
  65 {   # 0x1600
  66     "yaml": "mesh_profile_uuids.yaml",
  67     "description": "Mesh Profile",
  68     "unCamelCase": False,
  69     "unlist": [],
  70     "list": []
  71 },
  72 {   # 0x1800
  73     "yaml": "service_uuids.yaml",
  74     "description": "Service",
  75     "unCamelCase": False,
  76     "unlist": [],
  77     "list": []
  78 },
  79 {   # 0x2700
  80     "yaml": "units.yaml",
  81     "description": "Units",
  82     "unCamelCase": False,
  83     "unlist": [],
  84     "list": []
  85 },
  86 {   # 0x2800
  87     "yaml": "declarations.yaml",
  88     "description": "Declarations",
  89     "unCamelCase": False,
  90     "unlist": [],
  91     "list": []
  92 },
  93 {   # 0x2900
  94     "yaml": "descriptors.yaml",
  95     "description": "Descriptors",
  96     "unCamelCase": False,
  97     "unlist": [],
  98     "list": []
  99 },
 100 {   # 0x2a00
 101     "yaml": "characteristic_uuids.yaml",
 102     "description": "Characteristics",
 103     "unCamelCase": False,
 104     "unlist": [],
 105     "list": [
 106         # Then we have these weird ones stuck in between "Characteristics"
 107         # from object_types.yaml
 108         { "uuid": 0x2ACA, "name": "Unspecified" },
 109         { "uuid": 0x2ACB, "name": "Directory Listing" },
 110         # And some from other sources
 111         { "uuid": 0x2A0B, "name": "Exact Time 100" },
 112         { "uuid": 0x2A10, "name": "Secondary Time Zone" },
 113         { "uuid": 0x2A15, "name": "Time Broadcast" },
 114         { "uuid": 0x2A1A, "name": "Battery Power State" },
 115         { "uuid": 0x2A1B, "name": "Battery Level State" },
 116         { "uuid": 0x2A1F, "name": "Temperature Celsius" },
 117         { "uuid": 0x2A20, "name": "Temperature Fahrenheit" },
 118         { "uuid": 0x2A2F, "name": "Position 2D" },
 119         { "uuid": 0x2A30, "name": "Position 3D" },
 120         { "uuid": 0x2A3A, "name": "Removable" },
 121         { "uuid": 0x2A3B, "name": "Service Required" },
 122         { "uuid": 0x2A3C, "name": "Scientific Temperature Celsius" },
 123         { "uuid": 0x2A3D, "name": "String" },
 124         { "uuid": 0x2A3E, "name": "Network Availability" },
 125         { "uuid": 0x2A56, "name": "Digital" },
 126         { "uuid": 0x2A57, "name": "Digital Output" },
 127         { "uuid": 0x2A58, "name": "Analog" },
 128         { "uuid": 0x2A59, "name": "Analog Output" },
 129         { "uuid": 0x2A62, "name": "Pulse Oximetry Control Point" },
 130         # These have somehow disappeared. We keep them for if they were used.
 131         { "uuid": 0x2BA9, "name": "Media Player Icon Object Type" },
 132         { "uuid": 0x2BAA, "name": "Track Segments Object Type" },
 133         { "uuid": 0x2BAB, "name": "Track Object Type" },
 134         { "uuid": 0x2BAC, "name": "Group Object Type" },
 135     ]
 136 },
 137 {   # 0xfxxx
 138     "yaml": "member_uuids.yaml",
 139     "description": "Members",
 140     "unCamelCase": False,
 141     "unlist": [],
 142     "list": [
 143         # This they really screwed up. The UUID was moved to sdo_uuids,
 144         # thereby breaking the range and ordering completely.
 145         { "uuid": 0xFCCC, "name": "Wi-Fi Easy Connect Specification" },
 146     ]
 147 },
 148 {   # 0xffef (and 0xfccc)
 149     "yaml": "sdo_uuids.yaml",
 150     "description": "SDO",
 151     "unCamelCase": False,
 152     "unlist": [ 0xFCCC,
 153     ],
 154     "list": []
 155 }]
 156
 157 '''
 158 Retrieve the YAML files defining the UUIDs and add them to the lists
 159 '''
 160 for uuids in uuids_sources:
 161     req_headers = { 'User-Agent': 'Wireshark make-bluetooth' }
 162     try:
 163         req = urllib.request.Request(base_url + 'uuids/' + uuids["yaml"], headers=req_headers)
 164         response = urllib.request.urlopen(req)
 165         lines = response.read().decode('UTF-8', 'replace')
 166     except Exception as e:
 167         print("Failed to get UUIDs at {url}, because of: {e}".format(url=base_url + 'uuids/' + uuids["yaml"], e=e), file=sys.stderr)
 168         sys.exit(1)
 169
 170     uuids_dir = yaml.safe_load(lines)
 171     for uuid in uuids_dir["uuids"]:
 172         if uuid["uuid"] not in uuids["unlist"]:
 173             uuids["list"].append(uuid)
 174
 175 '''
 176 Go through the lists and perform general and specific transforms.
 177 Several exceptional cases are addressed directly by their UUID, because of the inconsistent nature
 178 by which their name is constructed.
 179 When they appear more sensibly in the databases they must be removed here.
 180 When new inconsistent entries appear in the databases their transforms can be added here,
 181 but also add their UUID below.
 182 '''
 183 for uuids in uuids_sources:
 184     for uuid in uuids["list"]:
 185         # Handle a few exceptional cases
 186         if uuid["uuid"] == 0x001E:
 187             uuid["name"] = "MCAP Control Channel"
 188         elif uuid["uuid"] == 0x001F:
 189             uuid["name"] = "MCAP Data Channel"
 190         elif uuid["uuid"] == 0x1102:
 191             uuid["name"] = "LAN Access Using PPP"
 192         elif uuid["uuid"] == 0x1104:
 193             uuid["name"] = "IrMC Sync"
 194         elif uuid["uuid"] == 0x1105:
 195             uuid["name"] = "OBEX Object Push"
 196         elif uuid["uuid"] == 0x1106:
 197             uuid["name"] = "OBEX File Transfer"
 198         elif uuid["uuid"] == 0x1107:
 199             uuid["name"] = "IrMC Sync Command"
 200         elif uuid["uuid"] == 0x1200:
 201             uuid["name"] = "PnP Information"
 202         elif uuid["uuid"] == 0x2B8C:
 203             uuid["name"] = "CO\u2082 Concentration"
 204         else:
 205         # And these in general
 206             uuid["name"] = uuid["name"].replace("_", " ")
 207             uuid["name"] = uuid["name"].replace('"', '\\"')
 208
 209 '''
 210 Go through the lists and, for those lists flagged as such, perform the unCamelCase transform
 211 on all the names in that list.
 212 Several exceptional cases were addressed directly by their UUID and must be excluded from this
 213 transform.
 214 When additional characters indicating a break in words appear in database entries they can be
 215 added to break_chars.
 216 '''
 217 for uuids in uuids_sources:
 218     if uuids["unCamelCase"]:
 219         for uuid in uuids["list"]:
 220             # if not a few exceptional cases (see above)
 221             if uuid["uuid"] not in [0x001E, 0x001F, 0x1102, 0x1104, 0x1105, 0x1106, 0x1107, 0x1200, 0x2B8C]:
 222                 # Parse through the names and look for capital letters; when
 223                 # not preceded by another capital letter or one of break_chars, insert a space
 224                 break_chars = [" ", "-", "+", "/", "(", ".", "0", "1", "2", "3", "4", "5", "6", "7", "8", "9"]
 225                 was_break = True # fake space at beginning of string
 226                 was_upper = False
 227                 name = ""
 228                 for character in uuid["name"]:
 229                     is_upper = True if character.isupper() else False
 230                     if is_upper and not was_break and not was_upper:
 231                         name += " "
 232                     name += character
 233                     was_break = True if character in break_chars else False
 234                     was_upper = is_upper
 235                 uuid["name"] = name
 236
 237 '''
 238 To be able to generate a value_string_ext array the entries need to be sorted.
 239 '''
 240 for uuids in uuids_sources:
 241     uuids_sorted = sorted(uuids["list"], key=lambda uuid: uuid["uuid"])
 242     uuids["list"] = uuids_sorted
 243
 244 '''
 245 Do a check on duplicate entries.
 246 While at it, do a count of the number of UUIDs retrieved.
 247 '''
 248 prev_uuid = 0
 249 uuid_count = 0
 250 for uuids in uuids_sources:
 251     for uuid in uuids["list"]:
 252         if uuid["uuid"] > prev_uuid:
 253             prev_uuid = uuid["uuid"]
 254         else:
 255             print("Duplicate UUID detected: 0x{uuid:04X}".format(uuid=uuid["uuid"]), file=sys.stderr)
 256             sys.exit(1)
 257     uuid_count += len(uuids["list"])
 258
 259 '''
 260 Sanity check to see if enough entries were retrieved
 261 '''
 262 if (uuid_count < MIN_UUIDS):
 263     print("There are fewer UUIDs than expected: got {count} but was expecting {minimum}".format(count=uuid_count, minimum=MIN_UUIDS), file=sys.stderr)
 264     sys.exit(1)
 265
 266 '''
 267 Finally output the annotated source code for the value_string
 268 '''
 269 print("const value_string bluetooth_uuid_vals[] = {")
 270
 271 for uuids in uuids_sources:
 272     print("    /* {description} - {base_url}uuids/{yaml} */".format(description=uuids["description"], base_url=base_url, yaml=uuids["yaml"]))
 273     for uuid in uuids["list"]:
 274         print("    {{ 0x{uuid:04X},   \"{name}\" }},".format(uuid=uuid["uuid"], name=uuid["name"]))
 275
 276 print("    {      0,   NULL }")
 277 print("};")
 278 print("value_string_ext bluetooth_uuid_vals_ext = VALUE_STRING_EXT_INIT(bluetooth_uuid_vals);")
 279 print("")
 280
 281 ##
 282 ## Company Identifiers
 283 ##
 284
 285 '''
 286 List of the YAML files to retrieve and the lists of values to put into the value_string.
 287 Also the previous value_string contained additional company IDs, which are not currently
 288 present in the databases. Prepare the lists with these company IDs so they are not lost.
 289 When they do appear in the databases they must be removed here.
 290 '''
 291
 292 company_ids_sources = [
 293 {
 294     "yaml": "company_identifiers.yaml",
 295     "list": [
 296         # Some from other sources
 297         { "value": 0x0418, "name": "Alpine Electronics Inc." },
 298         { "value": 0x0943, "name": "Inovonics Corp." },
 299         { "value": 0xFFFF, "name": "For use in internal and interoperability tests" },
 300     ]
 301 }]
 302
 303 '''
 304 Retrieve the YAML files defining the company IDs and add them to the lists
 305 '''
 306 for company_ids in company_ids_sources:
 307     req_headers = { 'User-Agent': 'Wireshark make-bluetooth' }
 308     try:
 309         req = urllib.request.Request(base_url + 'company_identifiers/' + company_ids["yaml"], headers=req_headers)
 310         response = urllib.request.urlopen(req)
 311         lines = response.read().decode('UTF-8', 'replace')
 312     except Exception as e:
 313         print("Failed to get company IDs at {url}, because of: {e}".format(url=base_url + 'company_identifiers/' + company_ids["yaml"], e=e), file=sys.stderr)
 314         sys.exit(-1)
 315
 316     company_ids_dir = yaml.safe_load(lines)
 317     company_ids["list"].extend(company_ids_dir["company_identifiers"])
 318
 319 '''
 320 Go through the lists and perform general transforms.
 321 '''
 322 for company_ids in company_ids_sources:
 323     for company_id in company_ids["list"]:
 324         company_id["name"] = company_id["name"].replace('"', '\\"')
 325
 326 '''
 327 To be able to generate a value_string_ext array the entries need to be sorted.
 328 '''
 329 for company_ids in company_ids_sources:
 330     company_ids_sorted = sorted(company_ids["list"], key=lambda company_id: company_id['value'])
 331     company_ids["list"] = company_ids_sorted
 332
 333 '''
 334 Do a check on duplicate entries.
 335 While at it, do a count of the number of company IDs retrieved.
 336 '''
 337 prev_company_id = -1
 338 company_id_count = 0
 339 for company_ids in company_ids_sources:
 340     for company_id in company_ids["list"]:
 341         if company_id["value"] > prev_company_id:
 342             prev_company_id = company_id["value"]
 343         else:
 344             print("Duplicate company ID detected: 0x{company_id:04X}".format(company_id=company_id["value"]), file=sys.stderr)
 345             sys.exit(1)
 346     company_id_count += len(company_ids["list"])
 347
 348 '''
 349 Sanity check to see if enough entries were retrieved
 350 '''
 351 if company_id_count < MIN_COMPANY_IDS:
 352     print("There are fewer company IDs than expected: got {count} but was expecting {minimum}".format(count=company_id_count, minimum=MIN_COMPANY_IDS), file=sys.stderr)
 353     sys.exit(1)
 354
 355 '''
 356 Finally output the source code for the value_string
 357 '''
 358 print("/* Taken from {base_url}company_identifiers/{yaml} */".format(base_url=base_url, yaml=company_ids_sources[0]["yaml"]))
 359 print("static const value_string bluetooth_company_id_vals[] = {")
 360
 361 for company_ids in company_ids_sources:
 362     for company_id in company_ids["list"]:
 363         print("    {{ 0x{company_id:04X},   \"{name}\" }},".format(company_id=company_id["value"], name=company_id["name"]))
 364
 365 print("    {      0,   NULL }")
 366 print("};")
 367 print("value_string_ext bluetooth_company_id_vals_ext = VALUE_STRING_EXT_INIT(bluetooth_company_id_vals);")
 368