A framework for quick web archiving
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

98 line
3.8 KiB

  1. import argparse
  2. import asyncio
  3. import importlib.util
  4. import logging
  5. import os.path
  6. import qwarc
  7. import qwarc.utils
  8. import qwarc.version
  9. import sys
  10. def setup_logging(logFilename, logLevel, logLevelStderr):
  11. if logLevelStderr is None:
  12. logLevelStderr = logLevel
  13. rootLogger = logging.getLogger()
  14. rootLogger.handlers = []
  15. rootLogger.setLevel(min(logLevel, logLevelStderr))
  16. formatter = qwarc.utils.LogFormatter()
  17. fileHandler = logging.FileHandler(logFilename)
  18. fileHandler.setFormatter(formatter)
  19. fileHandler.setLevel(logLevel)
  20. rootLogger.addHandler(fileHandler)
  21. stderrHandler = logging.StreamHandler()
  22. stderrHandler.setFormatter(formatter)
  23. stderrHandler.setLevel(logLevelStderr)
  24. rootLogger.addHandler(stderrHandler)
  25. def check_files(specFilename, logFilename):
  26. success = True
  27. if not os.path.isfile(specFilename):
  28. print(f'Error: "{specFilename}" does not exist or is not a regular file', file = sys.stderr)
  29. success = False
  30. if os.path.exists(logFilename):
  31. print(f'Error: "{logFilename}" already exists', file = sys.stderr)
  32. success = False
  33. if os.path.exists('STOP'):
  34. print('Error: "STOP" exists', file = sys.stderr)
  35. success = False
  36. return success
  37. def main():
  38. parser = argparse.ArgumentParser(formatter_class = argparse.ArgumentDefaultsHelpFormatter)
  39. parser.add_argument('--version', action = 'version', version = f'qwarc {qwarc.version.__version__}')
  40. parser.add_argument('--log', metavar = 'LOGFILE', default = './qwarc.log')
  41. parser.add_argument('--loglevel', metavar = 'LEVEL', help = 'verbosity of the log file', choices = ('INFO', 'DEBUG'), default = 'INFO')
  42. parser.add_argument('--loglevelstderr', metavar = 'LEVEL', help = 'verbosity of the log messages on stderr; set equal to the log file level if None', choices = (None, 'CRITICAL', 'ERROR', 'WARNING', 'INFO', 'DEBUG'), default = None)
  43. parser.add_argument('--database', metavar = 'DBFILE', default = './qwarc.db')
  44. parser.add_argument('--warc', metavar = 'PREFIX', help = 'prefix for the WARC filenames', default = './qwarc')
  45. parser.add_argument('--concurrency', type = int, default = 1)
  46. parser.add_argument('--memorylimit', metavar = 'LIMIT', type = int, help = 'pause when less than LIMIT bytes memory is free; disable if 0', default = 0)
  47. parser.add_argument('--disklimit', metavar = 'LIMIT', type = int, help = 'pause when less than LIMIT bytes disk space is free; disable if 0', default = 0)
  48. parser.add_argument('--warcsplit', metavar = 'SIZE', type = int, help = 'split WARCs into files of SIZE bytes; disable if 0', default = 0)
  49. parser.add_argument('--warcdedupe', action = 'store_true', help = 'enable deduplication of WARC records')
  50. parser.add_argument('specfile')
  51. args = parser.parse_args()
  52. args.loglevel = getattr(logging, args.loglevel)
  53. args.loglevelstderr = getattr(logging, args.loglevelstderr) if args.loglevelstderr is not None else args.loglevel
  54. if not check_files(args.specfile, args.log):
  55. sys.exit(1)
  56. setup_logging(args.log, args.loglevel, args.loglevelstderr)
  57. spec = importlib.util.spec_from_file_location('spec', args.specfile)
  58. specMod = importlib.util.module_from_spec(spec)
  59. spec.loader.exec_module(specMod)
  60. specDependencies = specMod.__dict__.get('specDependencies', qwarc.utils.SpecDependencies())
  61. a = qwarc.QWARC(
  62. itemClasses = set(qwarc.Item.get_subclasses()),
  63. warcBasePath = args.warc,
  64. dbPath = args.database,
  65. command = sys.argv,
  66. specFile = args.specfile,
  67. specDependencies = specDependencies,
  68. logFilename = args.log,
  69. concurrency = args.concurrency,
  70. memoryLimit = args.memorylimit,
  71. minFreeDisk = args.disklimit,
  72. warcSizeLimit = args.warcsplit,
  73. warcDedupe = args.warcdedupe,
  74. )
  75. loop = asyncio.get_event_loop()
  76. try:
  77. loop.run_until_complete(a.run(loop))
  78. except (Exception, KeyboardInterrupt) as e:
  79. logging.exception('Unhandled error')
  80. finally:
  81. loop.run_until_complete(a.close())
  82. loop.close()