I'm trying run BULK INSERT in a loop. Loop through each file in some directory ends with no of particular file. Below is my solution
DECLARE #startFlag INT
DECLARE #endFlag INT
DECLARE #fileName varchar(50)
SET #startFlag = 1
SET #endFlag = 10
WHILE (#startFlag <= #endFlag)
BEGIN
SET #fileName = 'c:\path to file\filename_' + cast(#startFlag as varchar) + '.csv'
BULK
INSERT dbo.Intraday
FROM #fileName
WITH
(
FIELDTERMINATOR = '|',
ROWTERMINATOR = '\n'
)
SET #startFlag = #startFlag + 1
END
GO
but seems don't work. Is there anything I've overlooked or another missing stuff I can fix this issue?
You can't use variables or expressions all the places you might like in TSQL. You'll have to use dynamic SQL:
declare #fileName nvarchar(2000) = 'foo.csv'
SET #fileName = 'foo'
declare #sql nvarchar(max) = N'
BULK
INSERT dbo.Intraday
FROM '''+#fileName+'''
WITH
(
FIELDTERMINATOR = ''|'',
ROWTERMINATOR = ''\n''
)';
exec (#sql);
you can not use veritable name after From. you have to provide the name of file after from clause not variable. so you need to make complete bulk insert statement dynamically. please refer below sample code -
declare #sql nvarchar(max)
DECLARE #fileName varchar(50)
set #fileName ='C:\Input.txt'
set #sql = 'BULK
INSERT dbo.Intraday
FROM ''' + #fileName + '''
WITH
(
FIELDTERMINATOR = ''|'',
ROWTERMINATOR = ''\n''
)'
exec(#sql)
Related
I cobbled together the code below. I can write file paths into a table but I can't use the paths in the table to bulk load CSV files in a folder. Can some expert here take a look and let me know what's wrong? TIA.
IF OBJECT_ID('tempdb..#DirectoryTree') IS NOT NULL
DROP TABLE #DirectoryTree;
CREATE TABLE #DirectoryTree (
id int IDENTITY(1,1)
,subdirectory nvarchar(512)
,depth int
,isfile bit);
INSERT #DirectoryTree (subdirectory,depth,isfile)
EXEC master.sys.xp_dirtree 'C:\my_path\CSV Files\',1,1;
SELECT * FROM #DirectoryTree
WHERE isfile = 1 AND RIGHT(subdirectory,4) = '.csv'
ORDER BY id;
GO
DROP TABLE ALLFILENAMES
--CREATE TABLE ALLFILENAMES(id VARCHAR(999),subdirectory VARCHAR(255),depth VARCHAR(1),isfile VARCHAR(1))
Select * INTO ALLFILENAMES
From #DirectoryTree
--code above is fine; problems start here
--cursor loop
--bulk insert won't take a variable name, so make a sql and execute it instead:
Declare #sql varchar(8000)
set #sql = 'BULK INSERT BULKACT FROM ''' + 'ALLFILENAMES.subdirectory' + ''' '
+ ' WITH (
DATAFILETYPE = ''char'',
FIELDTERMINATOR = '','',
ROWTERMINATOR = ''\n'',
FIRSTROW = 2
) '
print #sql
exec (#sql)
The problem is with the Bulk Insert. Here is the error message that I get: Msg 4860, Level 16, State 1, Line 28
Cannot bulk load. The file "ALLFILENAMES.subdirectory" does not exist.
So, 'ALLFILENAMES' is the name of the table and 'subdirectory' is the name of the field that contains all paths to all CSV files.
you need to do a select from that ALLFILENAMES table. You can't just specified the tablename + column name like that and expect it to work
also you need to specify the full path in the FROM file name
and you may use temp table for ALLFILENAMES instead of permanent tble
Declare #sql varchar(max)
select #sql = isnull(#sql , '')
+ 'BULK INSERT BULKACT FROM ''C:\my_path\CSV Files\' + ALLFILENAMES.subdirectory + ''' '
+ ' WITH (
DATAFILETYPE = ''char'',
FIELDTERMINATOR = '','',
ROWTERMINATOR = ''\n'',
FIRSTROW = 2
); ' + char(13)
from ALLFILENAMES
print #sql
and there is a WITH (FORMAT = 'CSV'); option for bulk insert from CSV file
Thanks for the help, Sqiurrel. I got the code below to work, and add some comments. It's really ugly though. I guess SQL Server really isn't designed for these kinds of things...
----------------------------------------------------------
-- Create 5 tables and all fields in tables
DECLARE #intFlag INT
SET #intFlag = 1
WHILE (#intFlag <=5)
BEGIN
declare #cmd nvarchar(1000),
#MyTableName nvarchar(100)
print str(#intFlag)
set #MyTableName = 'CSV' + replace(str(#intFlag),' ','')
print #MyTableName
set #cmd = 'CREATE TABLE dbo.' + quotename(#MyTableName, '[') + '(Name varchar(255), Address varchar(255), Age varchar(255), Work varchar(255));';
print #cmd
exec(#cmd)
SET #intFlag = #intFlag + 1
END
GO
----------------------------------------------------------
IF OBJECT_ID('tempdb..#DirectoryTree') IS NOT NULL
DROP TABLE #DirectoryTree;
CREATE TABLE #DirectoryTree (
id int IDENTITY(1,1)
,subdirectory nvarchar(512)
,depth int
,isfile bit);
INSERT #DirectoryTree (subdirectory,depth,isfile)
EXEC master.sys.xp_dirtree 'C:\your_path_here\',1,1;
SELECT * FROM #DirectoryTree
WHERE isfile = 1 AND RIGHT(subdirectory,4) = '.csv'
ORDER BY id;
GO
DROP TABLE ALLFILENAMES
Select * INTO ALLFILENAMES
From #DirectoryTree
----------------------------------------------------------
--cursor loop
--bulk insert won't take a variable name, so make a sql and execute it instead:
Declare #sql varchar(max)
select #sql = isnull(#sql , '')
+ 'BULK INSERT ' + ALLFILENAMES.subdirectory + ' FROM ''C:\your_path_here\' + ALLFILENAMES.subdirectory + ''' '
+ ' WITH (
DATAFILETYPE = ''char'',
FIELDTERMINATOR = '','',
ROWTERMINATOR = ''\n'',
FIRSTROW = 2
); ' + char(13)
from ALLFILENAMES
print #sql
exec (#sql)
----------------------------------------
I want to insert multiple csv file to sqlserver table. I have a procedure for doing that. It works but my all files have same column name. So instead of importing all records it just imports column name which is the first row. If I manually delete first row then it imports the other records, I have thousands of files so I cant work manually. I am posting store procedure here. Please tell me if I can change something to make this work.
ALTER procedure [dbo].[usp_ImportMultipleFiles] #filepath varchar(500),
#pattern varchar(100), #TableName varchar(128)
as
set quoted_identifier off
declare #query varchar(1000)
declare #max1 int
declare #count1 int
Declare #filename varchar(100)
set #count1 =0
create table #x (name varchar(200))
set #query ='master.dbo.xp_cmdshell "dir '+#filepath+#pattern +' /b"'
insert #x exec (#query)
delete from #x where name is NULL
select identity(int,1,1) as ID, name into #y from #x
drop table #x
set #max1 = (select max(ID) from #y)
--print #max1
--print #count1
While #count1 <= #max1
begin
set #count1=#count1+1
set #filename = (select name from #y where [id] = #count1)
set #Query ='BULK INSERT '+ #Tablename + ' FROM '''+ #Filepath+#Filename+'''
WITH ( FIELDTERMINATOR = '','',ROWTERMINATOR = ''\n'')'
--print #query
exec (#query)
insert into logtable (query) select #query
end
drop table #y
You can use the First Row option in your bulk insert statement something like...
BULK INSERT Table_Name
FROM 'C:\FilePath'
WITH
(
FIELDTERMINATOR = ','
,ROWTERMINATOR = '\n'
,FIRSTROW = 2 --<-- This option here
)
Edit to your proc
DECLARE #Query NVARCHAR(MAX);
SET #Query = N'BULK INSERT '+ #Tablename +
N' FROM '''+ #Filepath+#Filename +
N''' WITH ( FIELDTERMINATOR = '',''
,ROWTERMINATOR = ''\n''
,FIRSTROW = 2
)'
I am trying to insert data from multiple files into SQL Server. This is the code I am using--
DECLARE #MyCounter int;
DECLARE #Fileprefix nvarchar, #Filesuffix nvarchar, #fullname nvarchar, #Counter_string nvarchar;
SET #MyCounter = 1;
SET #Fileprefix= 'C:\Arvind_gpd\patents\';
SET #Filesuffix='data_corrected.csv';
WHILE (#MyCounter < 10)
BEGIN;
Set #Counter_string= Cast(#MyCounter AS varchar(1) );
Set #fullname = (#Fileprefix+ #Counter_string + #Filesuffix );
BULK INSERT GooglePatentsIndividualDec2012.dbo.patent
FROM #fullname WITH ( DATAFILETYPE = 'char', FIELDTERMINATOR = '^', ROWTERMINATOR = '\n' );
SET #MyCounter = #MyCounter + 1;
END;
GO
However I am getting these error messages--
Incorrect syntax near #fullname. Expecting Integer, String, TEXT_LEX.....
Incorrect syntax near DATAFILETYPE. Expecting SELECT or '('
What am I doing wrong in the above query?
You cannot use a variable as a filename. According to the syntax described in MSDN it has to be a constant.
You may create the command and use EXEC to execute it.
set #Command = 'BULK INSERT GooglePatentsIndividualDec2012.dbo.patent '
set #Command = #Command + 'FROM '''+#fullname+''' WITH ( '
set #Command = #Command + 'DATAFILETYPE = ''char'', '
set #Command = #Command ΓΌ 'FIELDTERMINATOR = ''^'', ROWTERMINATOR = ''\n'' )'
EXEC (#Command)
You cannot use a variable as a table name. If you want to use a variable you have to create a string and use the EXEC function like this:
DECLARE #fullname nvarchar, #sql nvarchar
SET #sql = 'SELECT * FROM ' + #fullname+ ' WHERE id = 1'
EXEC(#sql)
i am trying to bulk insert into Db using sql server 2005
Below is the code.
declare #path varchar(500)
set #path = 'E:\Support\test.csv';
Create table #mytable( name varchar(max), class varchar(max), roll varchar(max) )
BULK INSERT #mytable FROM #path <-- Error line
WITH ( FIELDTERMINATOR = ',', ROWTERMINATOR = '\n' );
Go
select * from #mytable
drop table #mytable
Problem: issue is that my file path is dynamic and comes from a variable instead of hard coding which is not working
If i change the error line to below it works
BULK INSERT #mytable FROM 'E:\Support\test.csv';
Please advise how to fix this
Try to use Dynamic SQL:
declare #sql varchar(max)
set #sql = 'BULK INSERT #mytable FROM ''' + #path + ''' WITH ...
exec (#sql)
DECLARE #path varchar(50) = 'D:\ARQUIVOS_CARGAS\CABOS\FILE.prn'
DECLARE #SQL_BULK VARCHAR(MAX)
SET #SQL_BULK = 'BULK INSERT #TAB FROM ''' + #path + ''' WITH
(
CODEPAGE = ''ACP'',
FIRSTROW = 1,
FIELDTERMINATOR = ''tab'',
ROWTERMINATOR = ''0x0a'',
KEEPNULLS
)'
EXEC (#SQL_BULK)
I am writing a stored procedure which is supposed to take data from a csv file and insert into a table. My problem is that the number of columns in the csv file are not fixed(ie number of columns is variable). So I need some way to create a temporary table with exactly the same number of columns as in the csv file. So that I can use bulk insert.
Well I tried solving the issue by writing a sp which will take the csv file path as parameter and create a table names as temptable with the same format as that of the number of columns in the csv. CSV file looks like
eid,ename,esalary,etemp
1,Mark,1000,
2,Peter,1000,
Stored Proc script
create proc createtable
#path nvarchar(50)
as
begin
declare #execSQL nvarchar(1000)
declare #tempstr varchar(1000)
declare #col varchar(1000)
declare #table nvarchar(1000)
-- Create a temp table to with one column to hold the first row of the csv file
CREATE TABLE #tbl (line VARCHAR(1000))
SET #execSQL =
'BULK INSERT #tbl
FROM ''' + #path + '''
WITH (
FIELDTERMINATOR =''\n'',
FIRSTROW = 1,
ROWTERMINATOR = ''\n'',
LASTROW = 1
)
'
EXEC sp_executesql #stmt=#execSQL
SET #col = ''
SET #tempstr = (SELECT TOP 1 RTRIM(REPLACE(Line, CHAR(9), ',')) FROM #tbl)
DROP TABLE #tbl
WHILE CHARINDEX(',',#tempstr) > 0
BEGIN
SET #col=#col + LTRIM(RTRIM(SUBSTRING(#tempstr, 1, CHARINDEX(',',#tempstr)-1))) + ' varchar(100),'
SET #tempstr = SUBSTRING(#tempstr, CHARINDEX(',',#tempstr)+1, len(#tempstr))
END
SET #col = #col + #tempstr + ' varchar(100)'
if object_id('temptable') is not null
drop table temptable
SET #table = 'create table temptable (' + #col + ')'
EXEC sp_executesql #stmt=#table
-- Load data from csv
SET #execSQL =
'BULK INSERT temptable
FROM ''' + #path + '''
WITH (
FIELDTERMINATOR ='','',
FIRSTROW = 2,
ROWTERMINATOR = ''\n''
)
'
EXEC sp_executesql #stmt=#execSQL
end
improved nadeems script... A little bit more robust.
This code is excelent for loading multiple CSV files without using the default wizzards.
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
CREATE proc [dbo].[importeer_csv_as_table]
#path nvarchar(255),
#new_table_name varchar(255),
#field_terminator varchar(255),
#row_terminator varchar(255)
as
begin
declare #execsql nvarchar(max)
declare #tempstr varchar(max)
declare #col varchar(max)
declare #table nvarchar(max)
declare #drop_table varchar(max)
-- Create a temp table to with one column to hold the first row of the csv file
create table #tbl (line varchar(1000))
set #execsql =
'bulk insert #tbl
from ''' + #path + '''
with (
fieldterminator =''' + #row_terminator + ''',
firstrow = 1,
rowterminator = ''' + #row_terminator + ''',
lastrow = 1
)
'
exec sp_executesql #stmt=#execsql
--replace field terminator with comma
update #tbl set line = replace(line, #field_terminator, ',')
set #col = ''
set #tempstr = (select top 1 rtrim(replace(line, char(9), ',')) from #tbl)
drop table #tbl
while charindex(',',#tempstr) > 0
begin
set #col=#col + '[' + ltrim(rtrim(substring(#tempstr, 1, charindex(',',#tempstr)-1))) + '] varchar(max),'
set #tempstr = substring(#tempstr, charindex(',',#tempstr)+1, len(#tempstr))
end
set #col = #col + '[' + #tempstr + '] varchar(max)'
if object_id(#new_table_name) is not null
begin
set #drop_table = 'drop table [' + #new_table_name + ']'
exec sp_executesql #stmt= #drop_table
end
set #table = 'create table [' + #new_table_name + '] (' + #col + ')'
--select #table
exec sp_executesql #stmt=#table
--Load data from csvle
set #execsql =
'bulk insert [' + #new_table_name + ']
from ''' + #path + '''
with (
fieldterminator =''' + #field_terminator + ''',
firstrow = 2,
rowterminator = ''' + #row_terminator + '''
)
'
exec sp_executesql #stmt=#execsql
end
GO
You could use Powershell to process the CSV file, there is an example here which you could probably adapt to take account of the variable number of fields. You can build the SQL to create a table and then issue a bulk load.