我正在尝试将带有管道分隔符的csv加载到配置单元外部表。数据字段中出现的管道用引号括起来。数据中出现的双引号用\转义。当我配置外部表时,我看到带有双引号的数据没有得到正确解释。
test.csv
id|name
105|"Test | pipe delim in field"
107|" Test Escaped single double quote in HIVE
108|" Test Escaped enclosed double quote in HIVE "
109|\" Test Escaped enclosed double quote in HIVE "
110|\" Test Escaped enclosed double quote in HIVE \"
外部表创建语句
drop table test_schema.hive_test;
CREATE EXTERNAL TABLE test_schema.hive_test (id string, name string)
ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.OpenCSVSerde' WITH SERDEPROPERTIES
(
"separatorChar" = "|",
"quoteChar" = """,
"escapeChar" = "\"
)
LOCATION '/staging/test/hive'
tblproperties ("skip.header.line.count"="1");
输出
+---------------+-------------------------------------------------+
| hive_test.id | hive_test.name |
+---------------+-------------------------------------------------+
| 105 | Test | pipe delim in field |
| 107 | NULL |
| 108 | NULL |
| 109 | NULL |
| 110 | " Test Escaped enclosed double quote in HIVE |
+---------------+-------------------------------------------------+
预期输出
+---------------+-------------------------------------------------+
| hive_test.id | hive_test.name |
+---------------+-------------------------------------------------+
| 105 | Test | pipe delim in field |
| 107 | " Test Escaped single double quote in HIVE |
| 108 | " Test Escaped enclosed double quote in HIVE " |
| 109 | NULL |
| 110 | NULL |
+---------------+-------------------------------------------------+
打开CSV 2.3版
遗憾的是,这是不可能实现的,因为OpenCSV使用单个字符作为转义符,而实际上您正试图使用双反斜杠作为转义符(这将是string
(。在OpenCSVSerde类中,您可以发现,无论传递什么作为转义符,OpenCSVSerde都会获取字符串值的第一个字符https://github.com/apache/hive/blob/master/serde/src/java/org/apache/hadoop/hive/serde2/OpenCSVSerde.java#L98
这是当前代码作为参考
private char getProperty(final Properties tbl, final String property, final char def) {
final String val = tbl.getProperty(property);
if (val != null) {
return val.charAt(0);
}
return def;
}
我认为缺少一个警告,让用户在创建表时知道只支持单个字符。