SQL*Loader 无法加载一个字段中包含多行的 CSV 文件

Question

我有一个 CSV 文件，我正在尝试使用 SQL*Loader 将其加载到 Oracle 表中：

a, ab, abcdefg
b, bc, bcdefghij
c, cd, cdefghijk-lmnop
orstuv
wxyz
d, de, defghijk

在“c”行，第 3 列有多行，当我们使用 sqlldr 加载这一行时，它无法将该列加载到表中 - 它会将其放入错误文件中。

我试图为这样的某些列包含引号字符串

a, ab, "abcdefg"
b, bc, "bcdefghij"
c, cd, "cdefghijk-lmnop
orstuv
wxyz"
d, de, "defghijk"

我尝试了多种格式的控制文件，包括：

load data
CHARACTERSET TH8TISASCII
infile '/data/csv/test3.txt'
APPEND 
PRESERVE BLANKS
into table ITEM
fields terminated by ","
TRAILING NULLCOLS
( col1,
  col2,
  col3,
  col4)

和：

load data
CHARACTERSET TH8TISASCII
infile '/data/csv/test3.txt' "str '\n'"
APPEND 
PRESERVE BLANKS
into table ITEM
fields terminated by ","
TRAILING NULLCOLS
( col1,
  col2,
  col3,
  col4)

和：

load data
CHARACTERSET TH8TISASCII
infile '/data/csv/test3.txt'
APPEND 
PRESERVE BLANKS
into table ITEM
fields terminated by "," optionally enclosed by '"'   
TRAILING NULLCOLS
( col1,
  col2,
  col3,
  col4)

和：

load data
CHARACTERSET TH8TISASCII
infile '/data/csv/test3.txt'
APPEND 
PRESERVE BLANKS
into table ITEM
fields terminated by "," optionally enclosed by '\n' 
TRAILING NULLCOLS
( col1,
  col2,
  col3,
  col4)

但还是不行。

如何从该 CSV 格式加载数据？

---

根据 Gary_W 推荐 我尝试添加“str x'0D'”并且数据可以正确加载到列中 但它只加载 CSV 文件的第一行并成功返回日志，如下所示

load data
CHARACTERSET TH8TISASCII
infile '/data/csv/b.csv' "str x'0D'"
APPEND
PRESERVE BLANKS
into table ITEM2
fields terminated by "," optionally enclosed by '"' 
TRAILING NULLCOLS
( col1,
  col2,
  col3,
  col4
)


--Sample CSV file
c, cd, "cdefghijk-lmnop
orstuv
wxyz"
,c
a, ab, "abcdefg", a
b, bc, "bcdefghij", b
d, de, "defghijk",d


--sql loader command
sqlldr USER/PASSWORD readsize=2000000000 bindsize=2000000000 control=file_pointing2.ctl > /data/csv/sqlldr.log

###########################LOG########################
Table ITEM:
1 Row successfully loaded.
0 Rows not loaded due to data errors.
0 Rows not loaded because all WHEN clauses were failed.
0 Rows not loaded because all fields were null.

there have only 1 record.
###########################OUT_PUT########################
col1| col2|         col3            | col4
cd  | cd  |efghijk-lmnoporstuvwxyz  | c

but if I remove the option "str x'0D'", in a case of the normal CSV format all row can load currently expect the abnormal row.

load data
CHARACTERSET TH8TISASCII
infile '/data/csv/b.csv' "str x'0D'"
APPEND
PRESERVE BLANKS
into table ITEM2
fields terminated by "," optionally enclosed by '"' 
TRAILING NULLCOLS
( col1,
  col2,
  col3,
  col4
)
--Sample CSV file
c, cd, "cdefghijk-lmnop
orstuv
wxyz"
,c
a, ab, "abcdefg", a
b, bc, "bcdefghij", b
d, de, "defghijk",d



--sql loader command
sqlldr USER/PASSWORD readsize=2000000000 bindsize=2000000000 control=file_pointing2.ctl > /data/csv/sqlldr.log


there have 4 records.but the record 1(col3) is not correct
###########################OUT_PUT########################
col1| col2|         col3            | col4
c   | cd  |efghijk-lmnop            | c
a   | ab  |abcdefg                  | a
b   | bc  |bcdefghij                | b
d   | de  |defghijk                 | d

Answer 1

在第三列周围加上双引号，您需要使用 INFILE 行上的流式子句告诉 sqlldr 回车是记录字符的结尾，因此双引号内的换行符将被忽略：

infile '/data/csv/test3.txt' "str x'0D'" 

由于我已经在这篇文章中解释了所有内容，请查看此处了解更多信息：https://stackoverflow.com/a/37216660/2543416