2017-02-07 56 views
0

我正在使用Groovy Sql.withBatch来处理CSV文件并在我的Postgres数据库中加载所有数据。Groovy Sql WithBatch在DB中缺少记录

这里我的方法:

def processCSV() { 
    def logger = Logger.getLogger('groovy.sql') 
    logger.level = Level.FINE 
    logger.addHandler(new ConsoleHandler(level: Level.FINE)) 

    def fileName = "file.csv" 
    def resource = this.getClass().getResource('/csv/' + fileName) 

    File file = new File(resource.path) 

    String year = '2016' 

    char separator = ',' 

    def lines = CSV 
      .separator(separator) 
      .skipLines(1) 
      .quote(CSVParser.DEFAULT_QUOTE_CHARACTER) 
      .escape(CSVParser.DEFAULT_ESCAPE_CHARACTER) 
      .charset('UTF-8') 
      .create() 
      .reader(file) 
      .readAll() 

    def totalLines = lines.size() 

    Sql sql = getDatabaseInstance() 

    println("Delete existing rows for " + year + " if exists") 
    String dQuery = "DELETE FROM table1 WHERE year = ?" 
    sql.execute(dQuery, [year]) 

    def statement = 'INSERT INTO table1 (column1, column2, column3, coulmn4, year) VALUES (?, ?, ?, ?, ?)' 

    println("Total lines in the CSV files: " + totalLines) 

    def batches = [] 

    sql.withBatch(BATCH_SIZE, statement) { ps -> 
     lines.each { fields -> 
      String coulmn1 = fields[0] 
      String coulmn2 = fields[1] 
      String column3 = fields[2] 
      String column4 = fields[3] 

      def params = [column1, coulmn2, column3, column4, year] 

      def batch = ['params': params, 'error': false] 
      try { 
       ps.addBatch(params) 
      } 
      catch (all) { 
       batch['error'] = true 
       throw all 
      } 

      batches << batch 
     } 
    } 

    def recordsAddedInDB = sql.firstRow("SELECT count(*) FROM " + tableName + " WHERE year = ?", year)[0] 

    sql.close() 

    println("") 
    println("Processed lines: " + line) 
    println("Batches: " + batches.size()) 
    println("Batches in error: " + batches.findAll{ it.error }.size()) 
    println("Record in DB for " + year + ": " + recordsAddedInDB) 
} 

在CSV文件中的行(exclusing标题行)是23758. 这种方法的输出是下面的:

Delete existing rows for 2016 if exists 
Total lines in the CSV files: 23758 
Processed lines: 23758 
Batches: 23758 
Batches in error: 0 
Record in DB for 2016 year: 23580 

如果启用在日志中,使用BATCH_SIZE为500,我可以看到:

  • 47次,句子“成功LY执行批次500命令(多个)”
  • 1时间句子‘成功执行批次258命令(多个)’

这意味着23758插入语句已经被处理。

任何人都知道为什么数据库中的行数少于已处理的行数?

+0

只适用于exatra forensics:可能值得看看sql.withBatch的返回值,例如'def counts = sql.withBatch {} .sum()'。在ps.addBatch之后直接添加一个'ps.executeBatch()'也可能很有用,以查看你得到的结果。 –

+0

sql.withBatch {} .sum()返回23580 ...这与我在数据库中的记录数相同。如果我添加了ps.executeBatch(),那么每个插入语句都是单独执行的,sql.withBatch {} .sum()返回0,并且在数据库中我仍然有相同数量的记录。 – Bagbyte

+0

您是否有可能在indata中为id列重复值?即您可能会插入相同的行,它们会在数据库中相互覆盖。 –

回答

0

已解决。 INSERT语句有一个子查询,当子查询没有返回值时,INSERT语句被忽略。