I have the following code:
CONVERT(VARCHAR(20), TimeCard_Date, 101) <
CONVERT(VARCHAR(20), dateadd(dd,-3,getdate()), 101)
The Original TimeCard_Date value = 2018-06-01
The GetDate() return = 11/14/2017
Can anyone assist as to why it thinks the Timecard_Date value set for June 2018 is less than the GetDate() minus 3 days value?
When you convert, it converts to a varchar datatype. 06/01/2018 is less than 11/14/2017 as a varchar since it is an alphabetical (or by number?) comparison. If you compare by date, the comparison is by the date datatype, which is as you expect.
You can change your code to:
TimeCard_Date < dateadd(dd,-3,getdate())
You don't need to convert DATETIME to VARCHAR in order to compare dates. Just use:
TimeCard_Date < DATEADD(dd,-3,GETDATE())
On the other hand, if you ever have to convert them to do it, you have to standarize the format (yyyyMMdd). You can check the FORMAT function https://learn.microsoft.com/en-us/sql/t-sql/functions/format-transact-sql
Related
I have a date in format dd/mm/yyyy. I want to subtract one month from it.
I am using this code but the output is "09/10/2020" I don't know why my code does the subtraction of the year -2 also.
This is my request
SELECT
FORMAT(CONVERT (DATE, DATEADD(MONTH, -1, CONVERT(char(9), GETDATE()))), 'dd/MM/yyyy')
you need to change it to:
select format(CONVERT (date,DATEADD(MONTH, -1,GETDATE())), 'dd/MM/yyyy' )
but as Larnu stated. it seems like you need to change the column.
Your current code doesn't work as expected because:
SELECT CONVERT(char(9), GETDATE());
Returns this (at least in my language):
Nov 9 20
Which is, unfortunately, and again in my language, a valid date (but in {20}20, not {20}22).
Even in the right style (103), char(9) would yield 10/11/202 tomorrow, since 9 digits is only enough if either the day or month is a single digit.
Don't know why you are converting GETDATE() to a string. Just perform date math on it and then format it if you need to (using a specific style number, e.g. 103 for d/m/y):
SELECT CONVERT(char(10), DATEADD(MONTH, -1, GETDATE()), 103);
I really wouldn't use FORMAT() for such simple output, as the CLR overhead really isn't worth it. Ideally you leave it as a date/time type until presentation time - surely your presentation layer can present your date as d/m/y if that's really a wise idea.
And if you are storing or passing dates as strings (and worse, in regional formats like d/m/y) you really should consider fixing that.
First of all,
You should be storing your Date as a string for easier manipulation. If you don't want to change the column, you can always convert from Date to Varchar and then (re)convert it.
Example:
First, convert Date to varchar using the style code '112' ISO for formatting as yyyyMMdd:
DECLARE #date DATE = GETDATE();
DECLARE #dateConverted as VARCHAR (8) = (SELECT CONVERT(VARCHAR, #date, 112));
Then you just subtract the month using DATEADD():
DECLARE #previousMonth AS VARCHAR (8) = (SELECT FORMAT(DATEADD(month, -1, #dateConverted), 'yyyyMMdd'));
Finally, convert varchar do Date again:
DECLARE #previousMonthConverted AS DATE = (SELECT CONVERT(CHAR(10), CONVERT(date, #previousMonth), 120));
I've a varchar column with a value of 42978.6736458333 that I want to convert back to a proper datetime. I've searched quite a bit and tried many of the suggestions but I cannot seem to find one that works with the data I have.
I got this value from excel when I did a CONCATENATE of all the fields in the sheet to do an insert to my DB; the output of the datetime column looks like "42978.6736458333" (the cell originally contained "2017-08-31 16:10:03"). I tried formatting and various things in excel to no avail.
Here are a few examples of what I've tried:
Select
convert(varchar(23), date, 112) DATE1,
convert(datetime, '20160805') DATE2,
convert(datetime, '2011-09-28 18:01:00', 120) DATE3,
dateadd(second, 42978.6736458333 * 24*24*60, '1899-12-31') DATE4
From
[dbo].[trainingLog]
Results:
DATE 1 = 42978.6736458333
DATE 2 = 2016-08-05 00:00:00.000
DATE 3 = 2011-09-28 18:01:00.000
DATE 4 = 1947-01-25 11:16:01.000
For every result. DATE 2/3/4 don't count up even though the original datetime varchar increments.
For example, here are more varchar values:
42981.5092361111
42982.7187615741
42983.8171527778
The above attempts return a value/date, but it's the same date even though my varchar value increments.
I expect any datetime format. I really only want the month/day/year in any format.
Try
SELECT CAST(42978.6736458333 AS DATETIME)
returns
2017-09-02 16:10:03.000
However SQL server uses 01/01/1900 as the epoch whereas your excel uses 30/12/1899 or 31/12/1899 as the epoch so it looks like you need to subtract 2 days off after you cast.
e.g.
SELECT dateadd(d, -2, CAST(42978.6736458333 AS DATETIME) )
returns
2017-08-31 16:10:03.000
From you comment I am not sure how you get your value as running
SELECT CAST(42981.5092361111 AS DATETIME)
returns for me
2017-09-05 12:13:18.000
This looks like an Excel time. So:
select dateadd(minute, 42978.6736458333 * 24*60, '1899-12-31')
If you try to use seconds, you will get an overflow error. If seconds are important, you can do:
select dateadd(day, floor(42978.6736458333), '1899-12-31') +
dateadd(second, (42978.6736458333 % 1)*24*60*60, 0)
I have this query and I tried converting it to every format, I mean the date time etc but it doesn't work and throws error:
Conversion failed when converting date and/or time from character string.
SELECT W.Organization_ID,
W.NIT_No,
W.SchemeID,
OpeningDate,
OpeningTime,
GETDATE(),
WorkNo,
CONVERT(decimal(10, 2), W.Cost) AS Cost,
WorkName,
W.ExpiryDate as ExpiryDate,
CONVERT(VARCHAR,OpeningDate,106),
CASE WHEN
CONVERT(DATETIME, CONVERT(VARCHAR(20),OpeningDate,106) + ' '
+ CONVERT(VARCHAR(20),OpeningTime,108))< GETDATE()
THEN 1
ELSE 0 END AS OpeningVaild
FROM Works W
the CASE part throws error.
OpeningDate is of type Varchar and OpeningTime is of type Time.
Why?
You are converting just a TIME datatype not a DATETIME so you don't need to specify the style:
DECLARE #T TIME = '08:05:06';
SELECT CONVERT(VARCHAR(8), #T) AS [Time];
SELECT CAST(#T AS VARCHAR(8)) AS [Time];
Or since you are using CONVERT()pick the right style for TIME datatype which is 108 or 114, instead of 106
SELECT CONVERT(VARCHAR(8), #T, 108) AS [Time];
Update:
According to the error Msg, your problem is in the CASE part.
That because you are trying to concat a DATETIME with a VARCHAR datatype, look at here to what you'r converting:
CASE WHEN
CONVERT(DATETIME, CONVERT(VARCHAR(20),OpeningDate,106) + ' '
+ CONVERT(VARCHAR(20),OpeningTime,108))< GETDATE()
THEN 1
ELSE 0 END AS OpeningVaild
Also the column OpeningDate -according to the error Msg- is VARCHAR so your are convert a VARCHAR to VARCHAR then convert it again to DATETIME then you try to concatinate the DATETIME with the VARCHAR returned from converting OpeningTime column from TIMEto VARCHAR, then try to compare them with GETDATE() which is DATETIME datatype.
So you CASE should look like:
CASE WHEN
(
CAST(OpeningDate AS DATETIME) + -- VARCHAR to DATETIME
CAST(OpeningTime AS DATETIME) -- TIME to DATETIME
) < GETDATE()
THEN 1
ELSE 0 END AS OpeningVaild
A beside note, here in this line
CONVERT(VARCHAR,OpeningDate,106),
You are trying to convert a VARCHAR to VARCHAR and without specify the lenght too, so this line should be:
CONVERT(VARCHAR(10),CAST(OpeningDate AS DATE),106),
Finally, don't ever ever store DATE as VARCHAR, the DATE/ TIME/ DATEIME are there for a reason, so use them and all other datatypes wisely.
Here is a demo represent your issue, and how to fix it.
You can simplify this big time by changing the expression a little.
This way you don't have to convert and concatenate.
SELECT
CASE WHEN OpeningDate < GETDATE() - OpeningTime
THEN 1
ELSE 0 END AS OpeningVaild
Note I am assuming that Openingdate has the format dd-mon-yyyy. Otherwise you still need to convert it, but still shorter:
SELECT
CASE WHEN Convert(date, OpeningDate, 106) < GETDATE() - OpeningTime
THEN 1
ELSE 0 END AS OpeningVaild
So I understand the problem is with this part:
CASE WHEN CONVERT(DATETIME, CONVERT(VARCHAR(20),OpeningDate,106) + ' ' + CONVERT(VARCHAR(20),OpeningTime,108))< GETDATE() THEN 1 ELSE 0 END AS OpeningVaild
Update
Since I've first posted my answer it turns out that you store the opening date as varchar instead of date.
First, you should stop doing that. Never store dates in anything other than a Date column (unless you need them with time as well, and then use DateTime2).
For more information, read Aaron Bertrand's Bad habits to kick : choosing the wrong data type.
Assuming the data type of the column can't change, you wrote in the comments to the question:
#ZoharPeled: this is the format of openingdate 2017-04-10
Illustrating one of the problems caused by storing dates as strings - How can I, or anyone else for that matter, know if that's the 10th of April or the 4th of October? The answer is we can't.
So, assuming it's the 10th of April, you can convert it to DateTime using convert with 126 as the style parameter:
CASE
WHEN CONVERT(DateTime, OpeningDate, 126) + CAST(OpeningTime As DateTime) < GETDATE() THEN
1
ELSE
0
END As OpeningVaild
First version:
Assuming that the data type of OpeningDate is Date and the data type of OpeningTime is Time, Seems like you are attempting to figure out if these columns combination into a DateTime is before the current DateTime.
Instead of converting them into strings and back to DateTime, you can cast both to DateTime and simply add them together:
CASE
WHEN CAST(OpeningDate As DateTime) + CAST(OpeningTime As DateTime) < GETDATE() THEN
1
ELSE
0
END As OpeningVaild
Another option would be to use GETDATE() twice. I don't think it should matter in the select clause, but in the where clause it's important to use this option since the first one will make these columns non-seargable, meaning the database engine will not be able to use any indexes that might help the execution plan of the statement:
CASE
WHEN OpeningDate < CAST(GETDATE() AS DATE)
OR
(
OpeningDate = CAST(GETDATE() AS DATE)
AND OpeningTime <= CAST(GETDATE() AS TIME)
) THEN
1
ELSE
0
END AS OpeningVaild
That being said, your query also have CONVERT(VARCHAR,OpeningDate,106) - The 106 style returns a string representation of the date as dd mon yyyy - meaning 11 chars - so change that to CONVERT(CHAR(11),OpeningDate,106) Note that using varchar without specifying the length defaults to 30, which is not a problem in this case since it's more than he 11 chars you need, but it's a bad habit to not specify length and you should kick it.
I want to select from table where date column is equal to specific date which I sending as a string in format 'yyyy-mm-dd'. I need to convert that string and than to compare if I have that date in my table.
For now I am doing this:
select *
FROM table
where CONVERT(char(10), date_column,126) = convert(char(10), '2016-10-28', 126)
date_column is a date type in table and I need to get it from table in this format 'yyyy-mm-dd' and because that I use 126 format. I am just not sure with the other part where I converting string which is already in that format and do I need to convert it because I don't know is it good to use this:
CONVERT(varchar(10), date_column,126) = '2016-10-28'
You don't need to convert the column as well. In fact, you better not convert the column, because using functions on columns prevents sql server from using any indexes that might help the query plan on that column.
Also, you are converting a string to char(10) - better just convert it to date:
where date_column = convert(date, '2016-10-28', 126)
Also, if you are using a datetime data type and not date, you need to check that the datetime value is between the date you pass to the next date.
You can convert string to date as follows using the CONVERT() function by giving a specific format of the input parameter
declare #date date
set #date = CONVERT(date, '2016-10-28', 126)
select #date
You can find the possible format parameter values for SQL Convert date function here
You do not need to do that. yyyy-MM-dd is the default format.
Please note that you need to take into account the time as well, if there's a timestamp in date_column. In that case you should write something like this
... WHERE date_column >= '2016-10-28 00:00:00' AND date_column < '2016-10-29 00:00:00'
... WHERE date_column BETWEEN '2016-10-28 00:00:00' and '2016-10-29 00:00:00'
As I just learned that (other than I thought) BETWEEN actually includes the end timestamp and thus is not equivalent to the above >= ... < approach.
This should use indexes properly as well.
I've studied all similar questions here, none of them works, what can be the problem? I want to convert varchar to date
convert(date,substring(replace(ent.Value,' ',''),1,10),103) < '20140101'
It returns error
Conversion failed when converting date and/or time from character string.
ent.Value here is something like '28/02/2014' Actually it works, when I enter something like
convert(date,substring('28/02/2013',1,10),103) < '20140101'
, but it doesn't work directly from the table
Thanks
To convert varchar to datetime : CONVERT(Datetime, '28/02/2014', 103) return the datetime like February, 28 2014 00:00:00+0000
To convert datetime to varchar : CONVERT(VARCHAR(10), datetime_field, 103) return something like '28/02/2014'
In both case you cannot compare it to this kind of text : '20140101'
You can change your text so both informations are ISO like : CONVERT(VARCHAR(10), CONVERT(Datetime, ent.Value, 103), 103) < '01/01/2014'
Or REPLACE(ent.Value,'/','') < '01012014'
But I don't think it's good to compare varchar => you can also do something like this : CONVERT(Datetime, '20140101', 112) :
CONVERT(Datetime, ent.Value, 103) < CONVERT(Datetime, '20140101', 112)
So your both field are in Datetime. ex : SQLFIDDLE DEMO
Break it down a bit: The first conversion works fine with the given example date string, but then you try and compare a DATE to a VARCHAR which requires an implicit conversion. That's never a good sign. However that actually works, so that aside the chances are that you have a date stored in text format that isn't a valid date. Since your examples use the UK date format, perhaps you have an American date in there (02/28/2013 - mm/dd/yyyy). If your default date format is UK style (i.e. you're not using American English settings) then it will fail as being out of range. You might also have entries which aren't dates at all with text in them like wibble or badly formatted dates like 02 Fub 2013.
Basically, you have a duff date somewhere in your table which you'll need to track down. Dates stored as text are a curse and to be avoided at all costs (although I appreciate that's not always possible when you inherit a legacy system - been there myself).
EDIT
To find your possible bad dates try this:
SELECT * FROM MyTable WHERE ISDATE(MyDateAsTextCol) <> 1