Conditional XQuery: How to select a desirable node when it occurs multiple times

I would very much appreicate if someone could help me with the following

Return CountryCodes node based on the following rules:

(1) Ignore <AlternativeState> completely

(2) When <CurrentEvent>MarketSize</CurrentEvent> get CountryCodes from <MarketSize> node only

(3) When <CurrentEvent>MarketShare</CurrentEvent> get CountryCodes from <OtherEvents> node only

(4) When <CurrentEvent> doesn't exist then xml would have only one CountryCodes; get that node

I have come up with the following so far which is far from what is desirable

SELECT UsageID, Countries.Code.query('

for $CountryCode in .

return data($CountryCode)

') AS CountryCodes

FROM UsageAnalysis

CROSS APPLY xmlState.nodes('//*[not(self::AlternativeState)]/*/CountryCodes') AS Countries(Code)

GO

Please keep in mind xml comes from a table column.

The following are three possible simplified cases

Case 1

<State>

<StatsState>

<CurrentState>

<MarketSize>

<CountryCodes>KT,LC,VG,SU,TT,UY,VE</CountryCodes>

</MarketSize>

<CurrentEvent>MarketSize</CurrentEvent>

</CurrentState>

</StatsState>

</State>

Case 2

<State>

<DefinitionState>

<CountryCodes>BR</CountryCodes>

</DefinitionState>

</State>

Case 3

<State>

<StatsState>

<CurrentState>

<OtherEvents>

<CountryCodes>FR</CountryCodes>

<AlternativeState>

<OtherEvents>

<CountryCodes>FR</CountryCodes>

</OtherEvents>

<MarketSize>

<CountryCodes>FR,FP,FG</CountryCodes>

</MarketSize>

<CurrentEvent>MarketShare</CurrentEvent>

</AlternativeState>

</OtherEvents>

<CurrentEvent>MarketShare</CurrentEvent>

<MarketSize>

<CountryCodes>,FR</CountryCodes>

</MarketSize>

</CurrentState>

</StatsState>

</State>

Hope this solve your problem:

Code Snippet

declare @.x xml

set @.x =

'<State>

<StatsState>

<CurrentState>

<MarketSize>

<CountryCodes>KT,LC,VG,SU,TT,UY,VE</CountryCodes>

</MarketSize>

<CurrentEvent>MarketSize</CurrentEvent>

</CurrentState>

</StatsState>

</State>

<State>

<DefinitionState>

<CountryCodes>BR</CountryCodes>

</DefinitionState>

</State>

<State>

<StatsState>

<CurrentState>

<OtherEvents>

<CountryCodes>FR</CountryCodes>

<AlternativeState>

<OtherEvents>

<CountryCodes>FR</CountryCodes>

</OtherEvents>

<MarketSize>

<CountryCodes>FR,FP,FG</CountryCodes>

</MarketSize>

<CurrentEvent>MarketShare</CurrentEvent>

</AlternativeState>

</OtherEvents>

<CurrentEvent>MarketShare</CurrentEvent>

<MarketSize>

<CountryCodes>,FR</CountryCodes>

</MarketSize>

</CurrentState>

</StatsState>

</State>'

select @.x.query('

for $s in /State

return

if (data(($s/StatsState/CurrentState/CurrentEvent)[1]) = "MarketSize")

then $s/StatsState/CurrentState/MarketSize/CountryCodes

else (

if (data(($s/StatsState/CurrentState/CurrentEvent)[1]) = "MarketShare")

then $s/StatsState/CurrentState/OtherEvents/CountryCodes

else $s//CountryCodes

)

')

|||

Should this also be returned?

<CountryCodes>,FR</CountryCodes>

Please excuse me because I am rather new to the XML sector. I am confused by the question and the answer. I coded this up:

declare @.x xml
set @.x =
'<State>
<StatsState>
<CurrentState>
<MarketSize>
<CountryCodes>KT,LC,VG,SU,TT,UY,VE</CountryCodes>
</MarketSize>
<CurrentEvent>MarketSize</CurrentEvent>
</CurrentState>
</StatsState>
</State>
<State>
<DefinitionState>
<CountryCodes>BR</CountryCodes>
</DefinitionState>
</State>
<State>
<StatsState>
<CurrentState>
<OtherEvents>
<CountryCodes>FR</CountryCodes>
<AlternativeState>
<OtherEvents>
<CountryCodes>FR</CountryCodes>
</OtherEvents>
<MarketSize>
<CountryCodes>FR,FP,FG</CountryCodes>
</MarketSize>
<CurrentEvent>MarketShare</CurrentEvent>
</AlternativeState>
</OtherEvents>
<CurrentEvent>MarketShare</CurrentEvent>
<MarketSize>
<CountryCodes>,FR</CountryCodes>
</MarketSize>
</CurrentState>
</StatsState>
</State>'

select coalesce (
nullif(t.c.query('./StatsState/CurrentState/MarketSize/CountryCodes').value('.','varchar(20)'), ''),
nullif(t.c.query('./StatsState/CurrentState/OtherEvents/CountryCodes').value('.','varchar(20)'),''),
t.c.query('./DefinitionState/CountryCodes').value('.','varchar(20)')

)
as CountryCodes
from @.x.nodes('State') t(c)

and received this result:

/*
CountryCodes
--
KT,LC,VG,SU,TT,UY,VE
BR
,FR
*/

Do the correct results need to include the markup such that the results should look more like this:

/*
CountryCodes
--
<CountryCodes>KT,LC,VG,SU,TT,UY,VE</CountryCodes><CountryCodes>BR</CountryCodes><CountryCodes>,FR</CountryCodes>
*/

(Trying to learn what is going on -- and I'm a bit confused.)

I appreciate the help.

|||

Jinghao, thanks very much. Your provided snippet does exactly what I have been trying to achieve. The only change I decided to introduce is to use data() so that I could get the scalar values for country codes as follows:

select @.x.query('

for $s in /State

return

if (data(($s/StatsState/CurrentState/CurrentEvent)[1]) = "MarketSize")

then data($s/StatsState/CurrentState/MarketSize/CountryCodes)

else (

if (data(($s/StatsState/CurrentState/CurrentEvent)[1]) = "MarketShare")

then data($s/StatsState/CurrentState/OtherEvents/CountryCodes)

else data($s//CountryCodes)

)

')

/*

Result set from your query:

<CountryCodes>KT,LC,VG,SU,TT,UY,VE</CountryCodes>

<CountryCodes>BR</CountryCodes>

<CountryCodes>FR</CountryCodes>

Results after introducing data()

KT,LC,VG,SU,TT,UY,VE BR FR

*/

Now I could use a function call to return a list of country codes.

Thanks again for your help.

|||

Kent,

I must say that it took me a while to fully understand the solution you suggested by clever use of COALESCE. It did exactly what I was trying to achieve. i.e get a list of selected country codes.

/*

KT,LC,VG,SU,TT,UY,VE

BR

FR

*/

I just wanted to have a list of countries, without having any markups. i.e. just the scalar values of <countryCodes>

Your response has shown me another use of COALESCE function and I very much appreciate your help

Conditional split on date ?

Hi,

I have a DT_DATE column. I'd like to achieve a conditional split to ignore all records for which the date is below a specific hardcoded date (eg: 2007-03-01).

I'm having a hard time trying to express this using the conditional split transform.

What is the correct syntax to express a DT_DATE literal ?

eg:
[date] < (DT_DATE) "2007-03-01"

regards

Thibaut

What you have should work fine. I built a little test package to verify, and each of these worked as expected:

Code Snippet

HireDate < (DT_DATE)"1998-01-30"

Code Snippet

[HireDate] < (DT_DATE)"1998-01-30"

Code Snippet

HireDate < (DT_DATE)"01/30/1998"

Code Snippet

[HireDate] < (DT_DATE)"01/30/1998"

What behavior are you experiencing that prompts you to ask the question?

|||Are you sure [date] is a DT_DATE column and not a DT_DBTIMESTAMP column? That is, does it contain a time component?

Just double checking.

Conditional split error message

Getting the below error msg on my conditional split. I changed the error output to ignore errors and that keeps the error msg from appearing (and everything seems to work normally), but why would it evaluate to NULL?

Thanks

[Conditional Split - Find rows with balances [3412]] Error: The expression "FINDSTRING(Column0,"OPENING",1) > 0 || FINDSTRING(Column0,"CLOSING",1) > 0" on "output "Balance Rows" (3415)" evaluated to NULL, but the "component "Conditional Split - Find rows with balances" (3412)" requires a Boolean results. Modify the error row disposition on the output to treat this result as False (Ignore Failure) or to redirect this row to the error output (Redirect Row). The expression results must be Boolean for a Conditional Split. A NULL expression result is an error.

Can Column0 be NULL?|||

I honestly don't know how.

The data file has between 4 and 6 rows on any given day - 4 of those rows always have "CLOSING" or "OPENING" in them.

So the conditional split should ignore the other rows, right?

That's why I don't understand why it's finding a null?

|||Try this expression:

FINDSTRING((ISNULL(Column0) ? "" : Column0),"OPENING",1) > 0 || FINDSTRING((ISNULL(Column0) ? "" : Column0),"CLOSING",1) > 0|||The problem may be that when evaluating Column0, some of the columns are NULL, and hence when the conditional split tries to evaluate the statement I provided to you a few days ago, it may fail. Using the new statement I just posted, we "trap" the fact that if the column is NULL, we set its contents to "" and continue on with the FINDSTRING statement.|||

So it's throwing the "NULL" error if finds any null columns (regardless of whether the row has OPENING or CLOSING in it, because it has to evaluate ALL rows?)

Such as, for example:

1234, ,1234 intstead of 1234," ",1234 ?

Is the first case above considered a null?

(Although I looked through my file, and I do not see any null fields at all)

|||Well, yes, that'd be true. ALL rows pass through the conditional split. The output path that gets chosen depends on which condition evaluates to true.|||

So to answer the other question,

a blank between file delimiters is considered a null value to ssis?

Such as 1234, ,1234 ?

I want to clarify this because it's got me concerned, as this situation is also causing problems with another file where it can't convert a value "without loss of data", because a numeric field is blank.

I was told to convert the value to string first, then convert it back to a numeric. Is this considered a best practice for working with numeric values?

Thanks

|||

Yes, it will be trated as NULL if your Flat File Source is configured to parse it that way. There is the property on the source adapter to control that.

It is not best the practice to convert numeric data to strings and back, but you need to make sure your numeric data is really numeric. NULLs should be fine if you can handle them downstream. It looks like your conditional split was not prepared for them.

Thanks,

~Bob