components/domain_reliability/bake_in_configs.py

   1 #!/usr/bin/env python
   2 # Copyright 2014 The Chromium Authors. All rights reserved.
   3 # Use of this source code is governed by a BSD-style license that can be
   4 # found in the LICENSE file.
   5
   6
   7 """Takes the JSON files in components/domain_reliability/baked_in_configs and
   8 encodes their contents as an array of C strings that gets compiled in to Chrome
   9 and loaded at runtime."""
  10
  11
  12 import ast
  13 import json
  14 import os
  15 import sys
  16
  17
  18 # A whitelist of domains that the script will accept when baking configs in to
  19 # Chrome, to ensure incorrect ones are not added accidentally. Subdomains of
  20 # whitelist entries are also allowed (e.g. maps.google.com, ssl.gstatic.com).
  21 DOMAIN_WHITELIST = (
  22   '2mdn.net',
  23   'admob.biz',
  24   'admob.co.in',
  25   'admob.co.kr',
  26   'admob.co.nz',
  27   'admob.co.uk',
  28   'admob.co.za',
  29   'admob.com',
  30   'admob.com.br',
  31   'admob.com.es',
  32   'admob.com.fr',
  33   'admob.com.mx',
  34   'admob.com.pt',
  35   'admob.de',
  36   'admob.dk',
  37   'admob.es',
  38   'admob.fi',
  39   'admob.fr',
  40   'admob.gr',
  41   'admob.hk',
  42   'admob.ie',
  43   'admob.in',
  44   'admob.it',
  45   'admob.jp',
  46   'admob.kr',
  47   'admob.mobi',
  48   'admob.no',
  49   'admob.ph',
  50   'admob.pt',
  51   'admob.sg',
  52   'admob.tw',
  53   'admob.us',
  54   'admob.vn',
  55   'dartmotif.com',
  56   'doubleclick.com',
  57   'doubleclick.ne.jp',
  58   'doubleclick.net',
  59   'doubleclickusercontent.com',
  60   'g.co',
  61   'ggpht.com',
  62   'gmodules.com',
  63   'goo.gl',
  64   'google-analytics.com',
  65   'google-syndication.com',
  66   'google.ac',
  67   'google.ad',
  68   'google.ae',
  69   'google.af',
  70   'google.ag',
  71   'google.al',
  72   'google.am',
  73   'google.as',
  74   'google.at',
  75   'google.az',
  76   'google.ba',
  77   'google.be',
  78   'google.bf',
  79   'google.bg',
  80   'google.bi',
  81   'google.bj',
  82   'google.bs',
  83   'google.bt',
  84   'google.by',
  85   'google.ca',
  86   'google.cat',
  87   'google.cc',
  88   'google.cd',
  89   'google.cf',
  90   'google.cg',
  91   'google.ch',
  92   'google.ci',
  93   'google.cl',
  94   'google.cm',
  95   'google.cn',
  96   'google.co.ao',
  97   'google.co.bw',
  98   'google.co.ck',
  99   'google.co.cr',
 100   'google.co.hu',
 101   'google.co.id',
 102   'google.co.il',
 103   'google.co.im',
 104   'google.co.in',
 105   'google.co.je',
 106   'google.co.jp',
 107   'google.co.ke',
 108   'google.co.kr',
 109   'google.co.ls',
 110   'google.co.ma',
 111   'google.co.mz',
 112   'google.co.nz',
 113   'google.co.th',
 114   'google.co.tz',
 115   'google.co.ug',
 116   'google.co.uk',
 117   'google.co.uz',
 118   'google.co.ve',
 119   'google.co.vi',
 120   'google.co.za',
 121   'google.co.zm',
 122   'google.co.zw',
 123   'google.com',
 124   'google.com.af',
 125   'google.com.ag',
 126   'google.com.ai',
 127   'google.com.ar',
 128   'google.com.au',
 129   'google.com.bd',
 130   'google.com.bh',
 131   'google.com.bn',
 132   'google.com.bo',
 133   'google.com.br',
 134   'google.com.by',
 135   'google.com.bz',
 136   'google.com.cn',
 137   'google.com.co',
 138   'google.com.cu',
 139   'google.com.cy',
 140   'google.com.do',
 141   'google.com.ec',
 142   'google.com.eg',
 143   'google.com.et',
 144   'google.com.fj',
 145   'google.com.ge',
 146   'google.com.gh',
 147   'google.com.gi',
 148   'google.com.gr',
 149   'google.com.gt',
 150   'google.com.hk',
 151   'google.com.iq',
 152   'google.com.jm',
 153   'google.com.jo',
 154   'google.com.kh',
 155   'google.com.kw',
 156   'google.com.lb',
 157   'google.com.ly',
 158   'google.com.mm',
 159   'google.com.mt',
 160   'google.com.mx',
 161   'google.com.my',
 162   'google.com.na',
 163   'google.com.nf',
 164   'google.com.ng',
 165   'google.com.ni',
 166   'google.com.np',
 167   'google.com.nr',
 168   'google.com.om',
 169   'google.com.pa',
 170   'google.com.pe',
 171   'google.com.pg',
 172   'google.com.ph',
 173   'google.com.pk',
 174   'google.com.pl',
 175   'google.com.pr',
 176   'google.com.py',
 177   'google.com.qa',
 178   'google.com.ru',
 179   'google.com.sa',
 180   'google.com.sb',
 181   'google.com.sg',
 182   'google.com.sl',
 183   'google.com.sv',
 184   'google.com.tj',
 185   'google.com.tn',
 186   'google.com.tr',
 187   'google.com.tw',
 188   'google.com.ua',
 189   'google.com.uy',
 190   'google.com.vc',
 191   'google.com.ve',
 192   'google.com.vn',
 193   'google.cv',
 194   'google.cz',
 195   'google.de',
 196   'google.dj',
 197   'google.dk',
 198   'google.dm',
 199   'google.dz',
 200   'google.ee',
 201   'google.es',
 202   'google.fi',
 203   'google.fm',
 204   'google.fr',
 205   'google.ga',
 206   'google.ge',
 207   'google.gg',
 208   'google.gl',
 209   'google.gm',
 210   'google.gp',
 211   'google.gr',
 212   'google.gy',
 213   'google.hk',
 214   'google.hn',
 215   'google.hr',
 216   'google.ht',
 217   'google.hu',
 218   'google.ie',
 219   'google.im',
 220   'google.info',
 221   'google.iq',
 222   'google.ir',
 223   'google.is',
 224   'google.it',
 225   'google.it.ao',
 226   'google.je',
 227   'google.jo',
 228   'google.jobs',
 229   'google.jp',
 230   'google.kg',
 231   'google.ki',
 232   'google.kz',
 233   'google.la',
 234   'google.li',
 235   'google.lk',
 236   'google.lt',
 237   'google.lu',
 238   'google.lv',
 239   'google.md',
 240   'google.me',
 241   'google.mg',
 242   'google.mk',
 243   'google.ml',
 244   'google.mn',
 245   'google.ms',
 246   'google.mu',
 247   'google.mv',
 248   'google.mw',
 249   'google.ne',
 250   'google.ne.jp',
 251   'google.net',
 252   'google.ng',
 253   'google.nl',
 254   'google.no',
 255   'google.nr',
 256   'google.nu',
 257   'google.off.ai',
 258   'google.org',
 259   'google.pk',
 260   'google.pl',
 261   'google.pn',
 262   'google.ps',
 263   'google.pt',
 264   'google.ro',
 265   'google.rs',
 266   'google.ru',
 267   'google.rw',
 268   'google.sc',
 269   'google.se',
 270   'google.sh',
 271   'google.si',
 272   'google.sk',
 273   'google.sm',
 274   'google.sn',
 275   'google.so',
 276   'google.sr',
 277   'google.st',
 278   'google.td',
 279   'google.tg',
 280   'google.tk',
 281   'google.tl',
 282   'google.tm',
 283   'google.tn',
 284   'google.to',
 285   'google.tt',
 286   'google.us',
 287   'google.uz',
 288   'google.vg',
 289   'google.vu',
 290   'google.ws',
 291   'googleadservices.com',
 292   'googleadsserving.cn',
 293   'googlealumni.com',
 294   'googleapis.com',
 295   'googleapps.com',
 296   'googlecbs.com',
 297   'googlecommerce.com',
 298   'googledrive.com',
 299   'googleenterprise.com',
 300   'googlegoro.com',
 301   'googlehosted.com',
 302   'googlepayments.com',
 303   'googlesource.com',
 304   'googlesyndication.com',
 305   'googletagmanager.com',
 306   'googletagservices.com',
 307   'googleusercontent.com',
 308   'googlevideo.com',
 309   'gstatic.com',
 310   'gvt1.com',
 311   'gvt2.com',
 312   'withgoogle.com',
 313   'youtu.be',
 314   'youtube-3rd-party.com',
 315   'youtube-nocookie.com',
 316   'youtube.ae',
 317   'youtube.al',
 318   'youtube.am',
 319   'youtube.at',
 320   'youtube.az',
 321   'youtube.ba',
 322   'youtube.be',
 323   'youtube.bg',
 324   'youtube.bh',
 325   'youtube.bo',
 326   'youtube.ca',
 327   'youtube.cat',
 328   'youtube.ch',
 329   'youtube.cl',
 330   'youtube.co',
 331   'youtube.co.ae',
 332   'youtube.co.at',
 333   'youtube.co.hu',
 334   'youtube.co.id',
 335   'youtube.co.il',
 336   'youtube.co.in',
 337   'youtube.co.jp',
 338   'youtube.co.ke',
 339   'youtube.co.kr',
 340   'youtube.co.ma',
 341   'youtube.co.nz',
 342   'youtube.co.th',
 343   'youtube.co.ug',
 344   'youtube.co.uk',
 345   'youtube.co.ve',
 346   'youtube.co.za',
 347   'youtube.com',
 348   'youtube.com.ar',
 349   'youtube.com.au',
 350   'youtube.com.az',
 351   'youtube.com.bh',
 352   'youtube.com.bo',
 353   'youtube.com.br',
 354   'youtube.com.by',
 355   'youtube.com.co',
 356   'youtube.com.do',
 357   'youtube.com.ee',
 358   'youtube.com.eg',
 359   'youtube.com.es',
 360   'youtube.com.gh',
 361   'youtube.com.gr',
 362   'youtube.com.gt',
 363   'youtube.com.hk',
 364   'youtube.com.hr',
 365   'youtube.com.jm',
 366   'youtube.com.jo',
 367   'youtube.com.kw',
 368   'youtube.com.lb',
 369   'youtube.com.lv',
 370   'youtube.com.mk',
 371   'youtube.com.mt',
 372   'youtube.com.mx',
 373   'youtube.com.my',
 374   'youtube.com.ng',
 375   'youtube.com.om',
 376   'youtube.com.pe',
 377   'youtube.com.ph',
 378   'youtube.com.pk',
 379   'youtube.com.pt',
 380   'youtube.com.qa',
 381   'youtube.com.ro',
 382   'youtube.com.sa',
 383   'youtube.com.sg',
 384   'youtube.com.tn',
 385   'youtube.com.tr',
 386   'youtube.com.tw',
 387   'youtube.com.ua',
 388   'youtube.com.uy',
 389   'youtube.com.ve',
 390   'youtube.cz',
 391   'youtube.de',
 392   'youtube.dk',
 393   'youtube.ee',
 394   'youtube.es',
 395   'youtube.fi',
 396   'youtube.fr',
 397   'youtube.ge',
 398   'youtube.gr',
 399   'youtube.gt',
 400   'youtube.hk',
 401   'youtube.hr',
 402   'youtube.hu',
 403   'youtube.ie',
 404   'youtube.in',
 405   'youtube.is',
 406   'youtube.it',
 407   'youtube.jo',
 408   'youtube.jp',
 409   'youtube.kr',
 410   'youtube.lk',
 411   'youtube.lt',
 412   'youtube.lv',
 413   'youtube.ma',
 414   'youtube.md',
 415   'youtube.me',
 416   'youtube.mk',
 417   'youtube.mx',
 418   'youtube.my',
 419   'youtube.ng',
 420   'youtube.nl',
 421   'youtube.no',
 422   'youtube.pe',
 423   'youtube.ph',
 424   'youtube.pk',
 425   'youtube.pl',
 426   'youtube.pr',
 427   'youtube.pt',
 428   'youtube.qa',
 429   'youtube.ro',
 430   'youtube.rs',
 431   'youtube.ru',
 432   'youtube.sa',
 433   'youtube.se',
 434   'youtube.sg',
 435   'youtube.si',
 436   'youtube.sk',
 437   'youtube.sn',
 438   'youtube.tn',
 439   'youtube.ua',
 440   'youtube.ug',
 441   'youtube.uy',
 442   'youtube.vn',
 443   'youtubeeducation.com',
 444   'youtubemobilesupport.com',
 445   'ytimg.com'
 446 )
 447
 448
 449 CC_HEADER = """// AUTOGENERATED FILE. DO NOT EDIT.
 450 //
 451 // (Update configs in components/domain_reliability/baked_in_configs and list
 452 // configs in components/domain_reliability/baked_in_configs.gypi instead.)
 453
 454 #include "components/domain_reliability/baked_in_configs.h"
 455
 456 #include <stdlib.h>
 457
 458 namespace domain_reliability {
 459
 460 const char* const kBakedInJsonConfigs[] = {
 461 """
 462
 463
 464 CC_FOOTER = """  nullptr
 465 };
 466
 467 }  // namespace domain_reliability
 468 """
 469
 470
 471 def read_json_files_from_gypi(gypi_file):
 472   with open(gypi_file, 'r') as f:
 473     gypi_text = f.read()
 474   json_files = ast.literal_eval(gypi_text)['variables']['baked_in_configs']
 475   return json_files
 476
 477
 478 def domain_is_whitelisted(domain):
 479   return any(domain == e or domain.endswith('.' + e)  for e in DOMAIN_WHITELIST)
 480
 481
 482 def quote_and_wrap_text(text, width=79, prefix='  "', suffix='"'):
 483   max_length = width - len(prefix) - len(suffix)
 484   output = prefix
 485   line_length = 0
 486   for c in text:
 487     if c == "\"":
 488       c = "\\\""
 489     elif c == "\n":
 490       c = "\\n"
 491     elif c == "\\":
 492       c = "\\\\"
 493     if line_length + len(c) > max_length:
 494       output += suffix + "\n" + prefix
 495       line_length = 0
 496     output += c
 497     line_length += len(c)
 498   output += suffix
 499   return output
 500
 501
 502 def main():
 503   if len(sys.argv) != 4:
 504     print >> sys.stderr, (('Usage: %s <JSON pathname base directory> ' +
 505                            '<input .gypi file> <output .cpp file>') %
 506                           sys.argv[0])
 507     print >> sys.stderr, sys.modules[__name__].__doc__
 508     return 1
 509   json_path = sys.argv[1]
 510   gypi_file = sys.argv[2]
 511   cpp_file = sys.argv[3]
 512
 513   json_files = read_json_files_from_gypi(gypi_file)
 514   json_files = [ os.path.join(json_path, f) for f in json_files ]
 515   json_files = [ os.path.normpath(f) for f in json_files ]
 516
 517   cpp_code = CC_HEADER
 518   found_invalid_config = False
 519
 520   for json_file in json_files:
 521     with open(json_file, 'r') as f:
 522       json_text = f.read()
 523     try:
 524       config = json.loads(json_text)
 525     except ValueError, e:
 526       print >> sys.stderr, "%s: error parsing JSON: %s" % (json_file, e)
 527       found_invalid_config = True
 528       continue
 529     if 'monitored_domain' not in config:
 530       print >> sys.stderr, '%s: no monitored_domain found' % json_file
 531       found_invalid_config = True
 532       continue
 533     domain = config['monitored_domain']
 534     if not domain_is_whitelisted(domain):
 535       print >> sys.stderr, ('%s: monitored_domain "%s" not in whitelist' %
 536                             (json_file, domain))
 537       found_invalid_config = True
 538       continue
 539
 540     # Re-dump JSON to get a more compact representation.
 541     dumped_json_text = json.dumps(config, separators=(',', ':'))
 542
 543     cpp_code += "  // " + json_file + ":\n"
 544     cpp_code += quote_and_wrap_text(dumped_json_text) + ",\n"
 545     cpp_code += "\n"
 546
 547   cpp_code += CC_FOOTER
 548
 549   if found_invalid_config:
 550     return 1
 551
 552   with open(cpp_file, 'wb') as f:
 553     f.write(cpp_code)
 554
 555   return 0
 556
 557
 558 if __name__ == '__main__':
 559   sys.exit(main())