datax-admin/app/core/datax/hdfs.py

from typing import List
from app.core.datax.base import ReaderBase, WriterBase
from app.models import JobJdbcDatasource
from app.schemas.datax_json import DataXJsonParam
from app.utils import *

from configs.settings import config
import json

{
    "writer": {
          "name": "hdfswriter",
          "parameter": {
            "defaultFS": "hdfs://192.168.199.107:9000",
            "fileType": "text",
            "path": "/usr/hive/warehouse/test_1",
            "fileName": "test_1",
            "writeMode": "append",
            "fieldDelimiter": "|",
            "column": [
              {
                "name": "id",
                "type": "int"
              },
              {
                "name": "ssn",
                "type": "varchar"
              },
              {
                "name": "test2",
                "type": "int"
              }
            ]
          }
        }
}


{
    "reader": {
        "name": "hdfsreader",
        "parameter": {
        "path": "/usr/hive/warehouse/grades/*",
        "defaultFS": "hdfs://192.168.199.107:9000",
        "fileType": "csv",
        "fieldDelimiter": ",",
        "column": [
            {
            "index": 0,
            "type": "long"
            },
            {
            "index": 3,
            "type": "string"
            },
            {
            "index": 5,
            "type": "long"
            }
        ]
        }
    }
}


def _build_hadoop_configs():
    hadoop = config.get('HADOOP_INNER', 'hadoop_config')
    kerberos = config.get('HADOOP_INNER', 'kerberos_config')
    return json.loads(hadoop), json.loads(kerberos)

class HdfsReader(WriterBase):
    def __init__(self, ds: JobJdbcDatasource):
        WriterBase.__init__(self, ds)
        if ds.datasource == 'hive':
            self.name = 'hdfsreader'
        else:
            raise Exception('Unimplemented HdfsReader')

    def _build_column(self, columns: List[dict]):
        res = []
        for column in columns:
            tmp = dict()
            index, name, type = column.split(':')
            tmp['index'] = index
            tmp['type'] = self._convert_type(type)
            res.append(tmp)
        if not res:
            raise Exception('No column found')
        return res

    def _convert_type(self, type):
        if 'int' in type.lower().strip():
            return 'long'
        elif 'varchar' in type.lower().strip():
            return 'string'
        return 'string'

    def build_parameter(self, param: DataXJsonParam, is_show=True):
        parameter = dict()
        # 需要kerberos验证的hive
        if str(self.ds.id) in config.get('HADOOP_INNER', 'datasource').split(','):
            parameter['defaultFS'] = config.get('HADOOP_INNER', 'default_fs')
            hadoop, kerberos = _build_hadoop_configs()
            parameter['hadoopConfig'] = hadoop
            parameter.update(kerberos)
        else:
            parameter['defaultFS'] = param.hive_reader.reader_default_fs
        parameter['path'] = param.hive_reader.reader_path.strip()
        parameter['fileType'] = param.hive_reader.reader_file_type
        parameter['fieldDelimiter'] = param.hive_reader.reader_field_delimiter.strip()
        parameter['column'] = self._build_column(param.reader_columns)
        return parameter


    def build(self, param: DataXJsonParam, is_show=True):
        reader = dict()
        parameter = self.build_parameter(param, is_show)
        reader['name'] = self.name
        reader['parameter'] = parameter
        return reader




class HdfsWriter(WriterBase):
    def __init__(self, ds: JobJdbcDatasource):
        WriterBase.__init__(self, ds)
        if ds.datasource == 'hive':
            self.name = 'hdfswriter'
        else:
            raise Exception('Unimplemented HdfsWriter')

    def _build_column(self, columns: List[dict]):
        res = []
        for column in columns:
            tmp = dict()
            _, name, type = column.split(':')
            tmp['name'] = name
            tmp['type'] = type
            res.append(tmp)
        if not res:
            raise Exception('No column found')
        return res



    def build_parameter(self, param: DataXJsonParam, is_show=True):
        parameter = dict()
        # 需要kerberos验证的hive
        if str(self.ds.id) in config.get('HADOOP_INNER', 'datasource').split(','):
            parameter['defaultFS'] = config.get('HADOOP_INNER', 'default_fs')
            hadoop, kerberos = _build_hadoop_configs()
            parameter['hadoopConfig'] = hadoop
            parameter.update(kerberos)
        else:
            parameter['defaultFS'] = param.hive_writer.writer_default_fs
        parameter['fileType'] = param.hive_writer.writer_file_type
        parameter['path'] = param.hive_writer.writer_path.strip()
        parameter['fileName'] = param.hive_writer.writer_filename
        parameter['writeMode'] = param.hive_writer.writer_mode
        parameter['fieldDelimiter'] = param.hive_writer.writer_field_delimiter.strip()
        parameter['column'] = self._build_column(param.writer_columns)
        return parameter

    def build(self, param: DataXJsonParam, is_show=True):
        writer = dict()
        parameter = self.build_parameter(param, is_show)
        writer['name'] = self.name
        writer['parameter'] = parameter
        return writer