hive.py 4.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123
  1. import os
  2. from app.core.datasource.datasource import DataSourceBase
  3. from pyhive import hive
  4. from pyhive.exc import DatabaseError
  5. from app.utils.get_kerberos import get_kerberos_to_local
  6. from configs.logging import logger
  7. from utils import flat_map
  8. class HiveDS(DataSourceBase):
  9. type = 'hive'
  10. def __init__(self, host, port,database_name,\
  11. username=None, password=None, kerberos=0, \
  12. keytab=None, krb5config=None, kerberos_service_name=None, \
  13. principal=None, type='hive'):
  14. DataSourceBase.__init__(self, host, port, username, password, database_name, type)
  15. self.host = host
  16. self.port = port
  17. self.username = username
  18. self.password = password
  19. self.database_name = 'default' if not database_name else database_name
  20. self.kerberos = int(kerberos)
  21. self.keytab = keytab
  22. self.krb5config = krb5config
  23. self.kerberos_service_name = kerberos_service_name
  24. self.principal = principal
  25. @property
  26. def jdbc_url(self):
  27. return f'jdbc:hive2://{self.host}:{self.port}/{self.database_name}'
  28. @property
  29. def jdbc_driver_class(self):
  30. return 'org.apache.hive.jdbc.HiveDriver'
  31. @property
  32. def connection_str(self):
  33. pass
  34. def _execute_sql(self, sqls):
  35. conn = None
  36. res = []
  37. try:
  38. if self.kerberos == 0:
  39. conn = hive.Connection(host=self.host, port=self.port, username=self.username, database=self.database_name)
  40. else:
  41. get_kerberos_to_local(self.keytab)
  42. file_name = './assets/kerberos/'+self.keytab.split("/")[-1]
  43. os.system(f'kinit -kt {file_name} {self.principal}')
  44. conn = hive.Connection(host=self.host, database=self.database_name, port=self.port, auth="KERBEROS", kerberos_service_name=self.kerberos_service_name)
  45. cursor = conn.cursor()
  46. for sql in sqls:
  47. cursor.execute(sql)
  48. res.append(cursor.fetchall())
  49. # logger.info(res)
  50. except Exception as e:
  51. logger.error(e)
  52. finally:
  53. if conn is not None:
  54. conn.close()
  55. return res
  56. def is_connect(self):
  57. sql = 'select 1'
  58. res = self._execute_sql([sql])
  59. logger.info(res)
  60. if res:
  61. return True
  62. else:
  63. return False
  64. def get_preview_data(self, table_name, limit=100, page = 0):
  65. sql1 = f'describe {self.database_name}.{table_name}'
  66. sql2 = f"SELECT * FROM {table_name} LIMIT {page},{limit}"
  67. res = self._execute_sql([sql1, sql2])
  68. logger.info(res)
  69. return {
  70. 'header': flat_map(lambda x: [':'.join(x[:2])], res[0]),
  71. 'content': res[1]
  72. }
  73. def get_data_num(self, table_name):
  74. sql2 = f"SELECT 1 FROM {table_name}"
  75. res = self._execute_sql([sql2])
  76. return len(res[0])
  77. def list_tables(self):
  78. sql = f'show tables'
  79. res = self._execute_sql([sql])
  80. return flat_map(lambda x: x, res[0])
  81. def get_table_schema(self, table_name):
  82. logger.info(self.database_name)
  83. sql1 = f'show columns in {self.database_name}.{table_name}'
  84. res = self._execute_sql([sql1])
  85. if res:
  86. columns = list(map(lambda x: x[0],res[0]))
  87. # logger.info(columns)
  88. else:
  89. raise Exception(f'{table_name} no columns')
  90. ans = []
  91. for i, col in enumerate(columns):
  92. sql = f'describe {self.database_name}.{table_name} {col}'
  93. try:
  94. res = self._execute_sql([sql])
  95. if res:
  96. # print(res[0])
  97. res = [[str(i), *x] for x in filter(lambda x: x[0] != '', res[0])]
  98. ans.append(''.join(flat_map(lambda x: ':'.join(x[:3]), res)))
  99. else:
  100. raise Exception('table not found')
  101. except Exception:
  102. return ans
  103. return ans