Recently we got one requirement to load the csv file data into the table using SSIS. Loading CSV file data in SSIS is very easy task when we are getting the all column. But the issue when we are not sure that when we will get the correct file. Client was agree that he will share the csv files that may have different number of column.
Let’s see the example.
For this demo we are taking pre
poll survey (Opinion poll) csv file. In this CSV file few columns fixed but
some of the column where may be we will get the data or may not get the data.
For this we have created the csv file which is having first six columns are
fixed I means for each records we will get these value and for the remaining
ten columns are not fixed. For one record we may get remaining 2 column or 4 or
6 or 8 or all column. We are getting in the sequence. See the below file.
In this file Sno 1002 & 1008 having all column. Sno 1005& 1009 missing 4 columns. Sno 1001, 1004 & 1006 missing 6 columns. Sno 1007 missing 8 columns and Sno 1010 missing 10 column.
See this file in note pad.
Here we do not have the comma for the missing column.
To load this file we can use C# or SQL Scrubber script. In
this example we will use SQL Scrubber.
First of all we need to load the CSV file in to the table in
one column. Creating a stage table to load the row data.
|
CREATE TABLE stg_RowData( id int IDENTITY(1,1) NOT NULL, RowData varchar(max) NULL ) |
Table created successfully.
Now opening the SSDT and creating a package.
Creating Oledb and flat file connection.
Taking Execute SQL task to truncate the stage table.
Now taking for each loop container to load more than one csv
files. Inside the container we are taking the Dataflow task.
Taking source as Flat file and destination as oledb
destination.
Doing the mapping.
Let’s load the data.
Data loaded successfully into the stage table and row data
loaded successfully.
Now we need to scrub these data and load it into the staging
table. Below is the staging and main table.
|
CREATE TABLE Stg_Survey( SNo INT NOT NULL, Voter_Name VARCHAR(100) NULL, Voter_Add VARCHAR(100) NULL, Constituency_Name VARCHAR(50) NULL, Constituency_State VARCHAR(50) NULL, Pre_Poll_date VARCHAR(50)NULL, Question_1 VARCHAR(200) NULL, Answer_1 VARCHAR(200) NULL, Question_2 VARCHAR(200) NULL, Answer_2 VARCHAR(200) NULL, Question_3 VARCHAR(200) NULL, Answer_3 VARCHAR(200) NULL, Question_4 VARCHAR(200) NULL, Answer_4 VARCHAR(200) NULL, Question_5 VARCHAR(200) NULL, Answer_5 VARCHAR(200) NULL ) CREATE TABLE Survey( SNo INT NOT NULL, Voter_Name VARCHAR(100) NULL, Voter_Add VARCHAR(100) NULL, Constituency_Name VARCHAR(50) NULL, Constituency_State VARCHAR(50) NULL, Pre_Poll_date DATE NULL, Question_1 VARCHAR(200) NULL, Answer_1 VARCHAR(200) NULL, Question_2 VARCHAR(200) NULL, Answer_2 VARCHAR(200) NULL, Question_3 VARCHAR(200) NULL, Answer_3 VARCHAR(200) NULL, Question_4 VARCHAR(200) NULL, Answer_4 VARCHAR(200) NULL, Question_5 VARCHAR(200) NULL, Answer_5 VARCHAR(200) NULL ) |
Table created successfully.
In the row staging data we have stored into comma separated.
We need to split these rows into columns. To split the row we are using user defined
function and giving the column sequence number for each columns.
Read here:
DelimitedSplit8K function to split the delimited text
https://bageshkumarbagi-msbi.blogspot.com/2024/08/delimitedsplit8k-function-to-split.html
See below
Sno 1001 has 10
columns and Sno 1002 has 16 columns.
Now we are writing the sp to load the data into stage table
and from stage to main table.
Below SP is used to load the data.
|
CREATE PROCEDURE Usp_load_survey_data AS BEGIN SET DATEFORMAT DMY; CREATE TABLE #temp_rowdata ( sno BIGINT, item VARCHAR(256) NULL, colsequence INT ); INSERT INTO #temp_rowdata (sno, item,
colsequence) SELECT LEFT(t.rowdata, Charindex(',', t.rowdata) - 1) AS sno, s.item, Row_number() OVER ( partition BY LEFT(t.rowdata, Charindex(',', t.rowdata) - 1) ORDER BY LEFT(t.rowdata, Charindex(',', t.rowdata) - 1) ASC ) AS ColSequence FROM stg_rowdata
AS t CROSS apply dbo.Delimitedsplit8k(t.rowdata, ',') AS s; INSERT INTO stg_survey (sno,
voter_name,
voter_add,
constituency_name,
constituency_state,
pre_poll_date,
question_1,
answer_1,
question_2,
answer_2,
question_3,
answer_3,
question_4,
answer_4,
question_5, answer_5) SELECT RowData.item AS sno, RowData2.item AS Voter_Name, RowData3.item AS Voter_Add, RowData4.item AS Constituency_Name, RowData5.item AS Constituency_State, RowData6.item AS Pre_Poll_date, Question_1.item AS Question_1, Answer_1.item AS Answer_1, Question_2.item AS Question_2, Answer_2.item AS Answer_2, Question_3.item AS Question_3, Answer_3.item AS Answer_3, Question_4.item AS Question_4, Answer_4.item AS Answer_4, Question_5.item AS Question_5, Answer_5.item AS Answer_5 FROM
#temp_rowdata AS RowData INNER JOIN (SELECT sno,
item,
colsequence FROM
#temp_rowdata) AS RowData2 ON RowData2.sno = RowData.sno AND RowData2.colsequence = 2 INNER JOIN (SELECT sno,
item,
colsequence FROM
#temp_rowdata) AS RowData3 ON RowData3.sno = RowData.sno AND RowData3.colsequence = 3 INNER JOIN (SELECT sno,
item,
colsequence FROM
#temp_rowdata) AS RowData4 ON RowData4.sno = RowData.sno AND RowData4.colsequence = 4 INNER JOIN (SELECT sno,
item,
colsequence FROM
#temp_rowdata) AS RowData5 ON RowData5.sno = RowData.sno AND RowData5.colsequence = 5 INNER JOIN (SELECT sno,
item,
colsequence FROM
#temp_rowdata) AS RowData6 ON RowData6.sno = RowData.sno AND RowData6.colsequence = 6 OUTER apply (SELECT sno,
item,
colsequence FROM
#temp_rowdata WHERE colsequence = 7
AND sno = RowData.sno) AS Question_1 OUTER apply (SELECT sno,
item, colsequence FROM
#temp_rowdata WHERE colsequence = 8
AND sno = RowData.sno) AS Answer_1 OUTER apply (SELECT sno,
item,
colsequence FROM
#temp_rowdata WHERE colsequence = 9
AND sno = RowData.sno) AS Question_2 OUTER apply (SELECT sno,
item,
colsequence FROM
#temp_rowdata WHERE colsequence = 10
AND sno = RowData.sno) AS Answer_2 OUTER apply (SELECT sno,
item,
colsequence FROM
#temp_rowdata WHERE colsequence = 11 AND sno = RowData.sno) AS Question_3 OUTER apply (SELECT sno,
item,
colsequence FROM
#temp_rowdata WHERE colsequence = 12
AND sno = RowData.sno) AS Answer_3 OUTER apply (SELECT sno,
item,
colsequence FROM #temp_rowdata WHERE colsequence = 13
AND sno = RowData.sno) AS Question_4 OUTER apply (SELECT sno,
item,
colsequence FROM
#temp_rowdata WHERE colsequence = 14
AND sno = RowData.sno) AS Answer_4 OUTER apply (SELECT sno,
item,
colsequence FROM
#temp_rowdata WHERE colsequence = 15
AND sno = RowData.sno) AS Question_5 OUTER apply (SELECT sno,
item,
colsequence FROM
#temp_rowdata WHERE colsequence = 16
AND sno = RowData.sno) AS Answer_5 WHERE RowData.colsequence = 1; --update the existing records in the main table. UPDATE sur SET Voter_Name
= stg.voter_name, Voter_Add = stg.voter_add,
Constituency_Name = stg.constituency_name,
Constituency_State = stg.constituency_state, Pre_Poll_date
= stg.pre_poll_date, Question_1 = stg.question_1, Answer_1 = stg.answer_1, Question_2 = stg.question_2, Answer_2 = stg.answer_2, Question_3 = stg.question_3, Answer_3 = stg.answer_3, Question_4 = stg.question_4, Answer_4 = stg.answer_4, Question_5 = stg.question_5, Answer_5 = stg.answer_5 FROM survey sur JOIN stg_survey stg ON stg.sno = sur.sno WHERE Isnull(sur.voter_name, '') <> Isnull(stg.voter_name, '') OR Isnull(sur.voter_add, '') <> Isnull(stg.voter_add, '') OR Isnull(sur.constituency_name, '') <> Isnull(stg.constituency_name, '') OR Isnull(sur.constituency_state, '') <> Isnull(stg.constituency_state, '') OR Isnull(sur.pre_poll_date, '') <> Isnull(stg.pre_poll_date, '') OR Isnull(sur.question_1, '') <> Isnull(stg.question_1, '') OR Isnull(sur.answer_1, '') <> Isnull(stg.answer_1, '') OR Isnull(sur.question_2, '') <> Isnull(stg.question_2, '') OR Isnull(sur.answer_2, '') <> Isnull(stg.answer_2, '') OR Isnull(sur.question_3, '') <> Isnull(stg.question_3, '') OR Isnull(sur.answer_3, '') <> Isnull(stg.answer_3, '') OR Isnull(sur.question_4, '') <> Isnull(stg.question_4, '') OR Isnull(sur.answer_4, '') <> Isnull(stg.answer_4, '') OR Isnull(sur.question_5, '') <> Isnull(stg.question_5, '') OR Isnull(sur.answer_5, '') <> Isnull(stg.answer_5, ''); --inserting the records INSERT INTO survey (sno,
voter_name,
voter_add,
constituency_name,
constituency_state,
pre_poll_date, question_1,
answer_1,
question_2,
answer_2,
question_3,
answer_3,
question_4,
answer_4,
question_5,
answer_5) SELECT stg.sno, stg.voter_name, stg.voter_add, stg.constituency_name, stg.constituency_state, stg.pre_poll_date, stg.question_1, stg.answer_1, stg.question_2, stg.answer_2, stg.question_3, stg.answer_3, stg.question_4, stg.answer_4, stg.question_5, stg.answer_5 FROM stg_survey
stg LEFT JOIN survey sur ON sur.sno = stg.sno WHERE sur.sno IS NULL; DROP TABLE #temp_rowdata; END; go |
Now package is ready to run.
Running this package.
Package executed successfully.
See the data into the main table.
Data loaded successfully.
No comments:
Post a Comment
If you have any doubt, please let me know.