Excel to Text file

Question

Excel to Text file

Anamika

SSCertifiable

Points: 6202
More actions
December 6, 2009 at 8:12 pm

#137410

Hi,
I'm working on a package where i'm transferring data from excel to a flat file.
My problem is the date column sometimes consist of character data which i have to send it as a error output. If the give the following data,
date1date2amount
str12/5/2009abcd
str22/6/200925000
str325/6/0952000
str4abcdxyz
str5efghrqp
str6ijklabcd
25/02/082/2/2009
I'm getting
str1,,abcd
str2,,
str3,25/6/09,
str4,abcd,xyz
str5,efgh,rqp
str6,ijkl,abcd
25/02/08,,
to get the mixed set of data from Excel, we can include IMEX in the excel connection after doing that i got the result as
str1,39849,abcd
str2,39850,25000
str3,25/6/09,52000
str4,abcd,xyz
str5,efgh,rqp
str6,ijkl,abcd
25/02/08,39846,
In the first case i didn't get the date values and in the second case i'm getting as some numbers like '39850'. '39846' etc.
but i want the values how ever it is in the excel sheet, so that the same value i can return to the error output and name the error also. (example "invalid date")
how to achieve this
thanks
regards
anamika,

Viewing 10 posts - 1 through 9 (of 9 total)

You must be logged in to reply to this topic. Login to reply

Phil Parkin SSC Guru Points: 246960 More actions · Answer 1

I would guess that those 'numbers' you are referring to are the number of days since 1/1/1900 (or 1899, or whatever the base date is) - which is how Excel (and SQL Server) store dates internally.

The remaining dates (ie, those items which still look like dates) have not been converted to numbers because they are invalid (according to the American mm/dd/yyyy format).

As your dates are in UK format, that implies that all of the dates are being misinterpreted and read as m/d/y.

The absence of evidence is not evidence of absence
- Martin Rees
The absence of consumable DDL, sample data and desired results is, however, evidence of the absence of my response
- Phil Parkin

Ron McCullough SSC Guru Points: 63877 More actions · Answer 2

To gain an understanding of how Excel stores dates read this blog entry by Brian Jones. It contains an excellent discussion of how and why date information is actuall stored by Excel

http://blogs.msdn.com/brian_jones/archive/2006/10/25/spreadsheetml-dates.aspx

If everything seems to be going well, you have obviously overlooked something.

Ron

Please help us, help you -before posting a question please read[/url]
Before posting a performance problem please read[/url]

Anamika SSCertifiable Points: 6202 More actions · Answer 3

Hi,

I would guess that those 'numbers' you are referring to are the number of days since 1/1/1900 (or 1899, or whatever the base date is) - which is how Excel (and SQL Server) store dates internally.

The remaining dates (ie, those items which still look like dates) have not been converted to numbers because they are invalid (according to the American mm/dd/yyyy format).

As your dates are in UK format, that implies that all of the dates are being misinterpreted and read as m/d/y.

The 'numbers' could be referring the dates stored internally. but does it change from system to system, depending on the base date of the system? or is there a way to get into a exact date? yes we follow dd/mm/yy format apart from this format if anything comes i should send it to error output.

To gain an understanding of how Excel stores dates read this blog entry by Brian Jones. It contains an excellent discussion of how and why date information is actuall stored by Excel

http://blogs.msdn.com/brian_jones/archive/2006/10/25/spreadsheetml-dates.aspx

If everything seems to be going well, you have obviously overlooked something.

Thanks for the link.

by the way, it's not going well, our testing team put all junk characters and test the ETL. is it the correct way to test? or is there a mechanism / method to test? they put character data in place of numeric, numeric in place of date etc... please suggest some good testing methodology to test dates in Excel.

Thanks,

Regards,

Anamika

henrik.litsne SSC Enthusiast Points: 115 More actions · Answer 4

When we had a similar problem we used a macro to convert our excel files to CSV format.

Then we used BULK INSERT to import them to temporary tables.

Then we selected the valid rows from the temporary tables.

This is more of a hack than anything else, so I'm not sure how easy it is to implement

in production standard SSIS package.

Anamika SSCertifiable Points: 6202 More actions · Answer 5

How do you convert to flat file? in my project itz been placed in a folder. the package should pick and do the job automatically.

the files will be uploaded by the clients. so how do you employ a macro in it?

or how to go about it?

thanks,

sabyasm SSCarpal Tunnel Points: 4324 More actions · Answer 6

[font="Comic Sans MS"]

Anamika - one way to achive this is enforce a strong template for the excel sheet you are using.

It shows how you set the property of a particular column (string - currency - date etc).. you can restrict by selecting the column and do a 'format cell'.. you can format it to be a 'text' and it will be the same as how the test team has entered the data..

[/font]

[font="Comic Sans MS"]--
Sabya[/font]

Anamika SSCertifiable Points: 6202 More actions · Answer 7

Hi Sabya,

thatz what not happening. Eventhough i set the columns as text, as soon as they paste the 'date' data into it the properties are becoming date. and when you insert a text value in between, it becomes null or empty while coming to SSIS. it is the behaviour of Excel.

the solution could be saving the file in csv format. but how to do this programmatically, without manual intervention.

any guidance, links are highly appreciated.

thanks,

regards

anamika

shanmugavel_1984 Valued Member Points: 65 More actions · Answer 8

Hi friend. the solution is please change the excel registry.

Steps.

go to run.

type regedit.

to find "TypeGuessRows"

to change the value 8 to 0

please try now.

by Shanmugavel N.

shanmugavel_1984@yahoo.co.in

Chennai.

ph 9994045200

Anamika SSCertifiable Points: 6202 More actions · Answer 9

Hi,

I tried your suggestion. amazing it works well. but still there is one concern.

the date values it is giving as numbers like

31810

39866,,39480

39532,,39115

39866,,40211

for example i entered 2/2/2010 in excel and itz been interpreted as 40211. if i try to get the date in sql server using

select datepart(dd,40211),datepart(mm,40211),datepart(yy,40211)

i'm getting

(No column name)(No column name)(No column name)

422007

which is not correct.

how to get the correct dates?

Thanks,

Regards,

Anamika