postgresql_reader.py 3.7 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576
  1. # -*- coding:utf-8 -*-
  2. import re
  3. from configparser import ConfigParser
  4. from dw_base.datax.datax_constants import *
  5. from dw_base.datax.plugins.reader.reader import Reader
  6. # PostgreSQL reader
  7. from dw_base.datax.plugins.writer.postgresql_writer import POSTGRE_SQL_WRITER_PARAMETER_COLUMN, \
  8. POSTGRE_SQL_WRITER_PARAMETER_CONNECTION, POSTGRE_SQL_WRITER_PARAMETER_DATABASE, POSTGRE_SQL_WRITER_PARAMETER_TABLE
  9. POSTGRE_SQL_READER_NAME = 'postgresqlreader'
  10. POSTGRE_SQL_READER_PARAMETER_CONNECTION = 'connection'
  11. POSTGRE_SQL_READER_PARAMETER_DATABASE = 'database'
  12. POSTGRE_SQL_READER_PARAMETER_FETCH_SIZE = 'fetchSize'
  13. POSTGRE_SQL_READER_PARAMETER_QUERY_SQL = 'querySql'
  14. POSTGRE_SQL_READER_PARAMETER_TABLE = 'table'
  15. POSTGRE_SQL_READER_PARAMETER_COLUMN = 'column'
  16. POSTGRE_SQL_READER_PARAMETER_WHERE = 'where'
  17. POSTGRE_SQL_READER_PARAMETER_SPLIT_PK = 'splitPk'
  18. class PostgreSQLReader(Reader):
  19. def __init__(self, base_dir: str, config_parser: ConfigParser, start_date: str = None, stop_date: str = None):
  20. super(PostgreSQLReader, self).__init__(base_dir, config_parser, start_date, stop_date)
  21. self.plugin_name = POSTGRE_SQL_READER_NAME
  22. def load_others(self):
  23. start_date = self.start_date
  24. stop_date = self.stop_date
  25. database = self.config_parser.get(self.plugin_type, POSTGRE_SQL_WRITER_PARAMETER_DATABASE)
  26. self.check_config(POSTGRE_SQL_WRITER_PARAMETER_DATABASE, database)
  27. table = self.config_parser.get(self.plugin_type, POSTGRE_SQL_WRITER_PARAMETER_TABLE)
  28. self.check_config(POSTGRE_SQL_WRITER_PARAMETER_TABLE, table)
  29. fetch_size = self.config_parser.get(self.plugin_type, POSTGRE_SQL_READER_PARAMETER_FETCH_SIZE) or '1000'
  30. self.parameter[POSTGRE_SQL_READER_PARAMETER_FETCH_SIZE] = fetch_size
  31. split_pk = self.config_parser.get(self.plugin_type, POSTGRE_SQL_READER_PARAMETER_SPLIT_PK)
  32. self.parameter[POSTGRE_SQL_READER_PARAMETER_SPLIT_PK] = split_pk
  33. where = self.config_parser.get(self.plugin_type, POSTGRE_SQL_READER_PARAMETER_WHERE)
  34. where = where.replace('${start_date}', start_date)
  35. where = where.replace('${start-date}', start_date)
  36. where = where.replace('${stop_date}', stop_date)
  37. where = where.replace('${stop-date}', stop_date)
  38. self.parameter[POSTGRE_SQL_READER_PARAMETER_WHERE] = where
  39. jdbc_url: str = self.parameter[DS_POSTGRE_SQL_JDBC_URL]
  40. matcher = re.search('jdbc:postgresql://(.+?)/(.+)', jdbc_url)
  41. if matcher:
  42. if database:
  43. jdbc_url = jdbc_url.replace(matcher.group(2), database)
  44. elif jdbc_url.endswith('/'):
  45. jdbc_url = f'{jdbc_url}{database}'
  46. else:
  47. jdbc_url = f'{jdbc_url}/{database}'
  48. query_sql = self.config_parser.get(self.plugin_type, POSTGRE_SQL_READER_PARAMETER_QUERY_SQL)
  49. query_sql = query_sql.replace('${start_date}', start_date)
  50. query_sql = query_sql.replace('${start-date}', start_date)
  51. query_sql = query_sql.replace('${stop_date}', stop_date)
  52. query_sql = query_sql.replace('${stop-date}', stop_date)
  53. if query_sql:
  54. connection = {
  55. DS_POSTGRE_SQL_JDBC_URL: jdbc_url.split(','),
  56. POSTGRE_SQL_READER_PARAMETER_QUERY_SQL: query_sql
  57. }
  58. else:
  59. connection = {
  60. DS_POSTGRE_SQL_JDBC_URL: jdbc_url.split(','),
  61. POSTGRE_SQL_READER_PARAMETER_TABLE: table.split(',')
  62. }
  63. self.parameter[POSTGRE_SQL_WRITER_PARAMETER_CONNECTION] = [connection]
  64. del self.parameter[DS_POSTGRE_SQL_JDBC_URL]
  65. def load_column(self):
  66. columns = self.config_parser.get(self.plugin_type, POSTGRE_SQL_WRITER_PARAMETER_COLUMN).split(',')
  67. self.parameter[POSTGRE_SQL_WRITER_PARAMETER_COLUMN] = columns