I have the following ColdFusion 9 code:
<cfloop from="1" to="#arrayLen(tagArray)#" index="i">
<cfquery name="qryGetSPFAQs" datasource="#application.datasource#">
EXEC searchFAQ '#tagArray[i]#'
</cfquery>
</cfloop>
The EXEC executes a stored procedure on the database server, which returns rows of data, depending on what the parameter is. What I am trying to do is combine the queries into one query object. In other words, if it loops 3 times and each loop returns 4 rows, I want a query object that has all 12 rows in one object. How do I acheive this?
You might want to take a different approach (modify your stored procedure to accept multiple arguments or use a list and fnSplit) and return the dataset all at once. However, to directly answer your question, this is how you could combine the queries as you're asking to:
You can use UNION in a Query of Queries to combine all of the datasets.
<cfloop from="1" to="#arrayLen(tagArray)#" index="i">
<cfquery name="qryGetSPFAQs#i#" datasource="#application.datasource#">
EXEC searchFAQ '#tagArray[i]#'
</cfquery>
</cfloop>
<cfquery name="combined" dbtype="query">
<cfloop from="1" to="#arrayLen(tagArray)#" index="i">
select * from qryGetSPFAQs#i#
<cfif i lt arrayLen(tagArray)>UNION</cfif>
</cfloop>
</cfquery>
A more direct way might be something like this:
<cfset bigQ = queryNew("column")>
<cfloop from="1" to="#arrayLen(tagArray)#" index="i">
<cfquery name="qryGetSPFAQs" datasource="#application.datasource#">
EXEC searchFAQ '#tagArray[i]#'
</cfquery>
<cfset queryAddRow(bigQ)>
<cfset querySetCell(bigQ, "column". qryGetSPFAQs)>
</cfloop>
You will need a querySetCell() assignment for each column. Check out the query functions in the live docs for more information.
Here is an out of the box solution, abandoning the StoredProc for a SQL View (I'll explain).
Disclaimer: without seeing the SP source code, I can't tell if my solution fits. I'm assuming that the SP is fairly basic, and I admit I usually prefer the compiled execution of an SP over a view, but the one-time execution of a SQL View should outperform the looping of the SP x times.
First make a view that looks like the SELECT statement in the SP (minus the parameterization, of course -- you'll cover that in a WHERE clause within the CFQUERY of your new view.
Second, set up your loop to do no more than build a data set we're going to use for the WHERE clause. You'll need to use ArrayToList and a little bit of string manipulation to tidy it up, with the end product being a string stored in a single CF variable looking like this:
('ValueOfArrayElement1','ValueOfArrayElement2','Value_And_So_On')
Building the string is pretty easy, using the delimeter attribute of ArrayToList, and after the loop is complete, append a Left Parenthesis & Single Quote to the Left most position of the string; and append a Single Quote & Right Parenthesis to the Right most position in the string.
Now, write the CFQUERY statement to SELECT the columns you need from your view (instead of executing your SP). And instead of passing a parameter to the SP, you're going to put a WHERE clause in the CFQUERY.
Oh, BTW, I am stating you need a SQL View, but the entire SELECT could be built in CFQUERY. Personally, when I have a multi-table JOIN, I like to define that in a SQL View where it's executed more quickly than a JOIN in CFQUERY. Ultimately a StoredProc is even faster, but our WHERE clause is much friendlier to code and read like this than it would be to send into StoredProc without looping in and out of SQL.
It's a good goal to make only one trip out to the database and back if possible. That's why we looped through the array to write a string equating to all the values in the dataset. This way, we'll only execute one query, one time.
SELECT Col1, Col2, Col_etc
FROM SQL_View_Name
WHERE ColumnName in #BigStringWeMadeFromArrayToList#
when our CFQUERY is rendered, the clause will look just like this in SQL:
WHERE ColumnName in
('ValueOfArrayElement1','ValueOfArrayElement2','Value_And_So_On')
So there you have it. Like I said, this is nice because it makes only one trip to the DB, and since we are building a view, the performance will still be pretty good -- better than running a StoredProc 4+ times. (no offense)
I'll must repeat... without having seen the SP code, I'm not sure if this is do-able. Plus, it's kind of odd to abandon a StoredProc for a SQL View, a "lesser" entity in the RDBMS, but I'm sure we will achieve greater performance and I think it's pretty readable, too.
Related
This seems like it should be an easy one, but CF seems to have different ways of accomplishing the same thing when it comes to Arrays, so I want to be sure I'm doing it the best way;
I have a simple cfoutput;
<cfoutput query="getusers">
Username: #username# ID:#fnid#
</cfoutput>
I'm also using cfwebsocket elsewhere on the page, and the value of the 'fnid' (which is a session.auth variable) is part of the subscription. A dump of wsGetSubscribers gives me the following;
What I'd like to do is during the output of the 'getusers' query, check to see if the FNID under subscriberinfo.fndid is there - the goal being to add a note to say that user is subscribed at that point.
I know I can do this using a cfloop, but that seems longwinded!
What I'm trying to avoid is having to use cfloop (on the array) within the cfoutput query
So you want a kind of look-up for fnId? How about:
<cfset subscribersById = StructNew()>
<cfloop array="#wsGetSubscribers()#" index="subsciption">
<cfset subscribersById[subsciption.subscriberInfo.fnId] = subsciption>
</cfloop>
<cfoutput query="getusers">
Username: #username# ID: #fnid# Subscribed: #StructKeyExists(subscribersById, fnid)#
</cfoutput>
CF10+ has arrayEach() as well, if you prefer this syntax over the <cfloop>.
I am moving one of our applications from ColdFusion 9.01 to ColdFusion 11 and encountered a situation where I cannot get the date formatted the way I want it using "DateFormat". I read through the docs since things have changed in CF versions, but I honestly can't figure out why this isn't working. It worked beautifully in CF 9. I know it's probably something very easy, but I am just not seeing it.
The query (Oracle DB) provides me a list of the last 30 days and the loop is simply to reformat the date output from "2014-07-01 00:00:00.0" to a more friendly looking display of 01-Jul-2014 except that I cannot get it to format as "dd-mmm-yyyy" it just spits back the original output from the query. I hard coded the date where normally there would be a cfquerparam. Any ideas?
<cfquery name="qryDateArray" datasource="#request.db#">
select trunc(to_date('07/01/2014', 'mm/dd/yyyy') + 1 - rownum) as ref_date
from dual connect by rownum <= 30
</cfquery>
<cfloop from="1" to="#qryDateArray.recordcount#" index="j">
<cfset qryDateArray.ref_date[j] = DateFormat(qryDateArray.ref_date[j], "dd-mmm-yyyy")>
</cfloop>
<cfoutput>
<cfdump var="#qryDateArray#">
</cfoutput>
I could not test this on CF11 since I do not have it handy. I did verify that your code though returns results as you explained when I ran it on my CF10 environment here. So what you can do is add a column to the query object and define it as a varchar and add your formatted data to that. This in turn dumped out the formatted dates.
<cfquery name="qryDateArray" datasource="#request.db#">
select trunc(to_date('07/01/2014', 'mm/dd/yyyy') + 1 - rownum) as ref_date
from dual connect by rownum <= 30
</cfquery>
<cfset aryData = [] />
<cfloop from="1" to="#qryDateArray.recordcount#" index="j">
<cfset ArrayAppend(aryData, DateFormat(qryDateArray.ref_date[j], "dd-mmm-yyyy")) />
</cfloop>
<cfset QueryAddColumn(qryDateArray, "STRDATE", "VarChar", aryData) />
<cfoutput>
<cfdump var="#qryDateArray#">
</cfoutput>
If dependent on the query column names then could use something like Ben's method explained here to do some renaming of the columns: http://www.bennadel.com/blog/357-ask-ben-changing-coldfusion-query-column-names.htm
It'd be great if you'd given us a portable test case rather than one that relies on your database, but I suspect it is because ColdFusion has become more rigid with its type management of query columns.
So CF considers your ref_date column to be of type date, so when you try to put the formatted string back into the query column, CF tries (and succeeds) to convert the string back into a date.
Aside:
I have to wonder why you don't format the data string in the DB from the outset, and just return it the way you need it, rather than returning something else, then looping over the thing to adjust it..?
Just wondering, given a query and output like so:
<cfoutput query="someItems" group="someColumnName">
... doing some stuff here ..
<cfoutput> doing stuff with some sub items </cfoutput>
</cfoutput>
if there's a way to change the order of elements in the 'inner' cfoutput ?
Can the query be both grouped and sorted by?
You will need to add ORDER BY clauses in your query for this to work, but you can nest cfoutput tags that use the group attribute.
<cfoutput query="someItems" group="someColumnName">
... doing some stuff here ..
<cfoutput group="someOtherColumnName> doing stuff with some sub items </cfoutput>
</cfoutput>
This assumes that in your query you have something that looks like:
ORDER BY someColumnName, someOtherColumnName
Keep in mind that the group attribute of cfquery is not the same as the GROUP BY clause in a SQL statement. You can use the group attribute of cfoutput for ANY column that is in the ORDER BY clause in your query.
One solution is to restructure your code to use the query-of-queries approach. Here is a good example of doing so:
http://www.bennadel.com/blog/2211-ColdFusion-Query-Of-Queries-vs-The-Group-Attribute-In-CFOutput.htm
Basically, you pull out all the data you care about in one master query (probably the query you have already written). You add a second query (against your first query, not against the database) that does the group by and aggregation of data that you need at the top level loop. Inside the loop driven by your second query, you use the row data in the group as a parameter to yet another query (against your first query again, not against the database) to pull out all the data relating to the current row ordered however you desire.
This idea of querying your query seems odd at first, but I have not had performance problems with it and it gives you a lot of flexibility to do what you want in your inner loop. Good luck!
The dumping results for the following QoQ are perfectly fine:
<cfquery datasource = "XX.XX.X.XX" name="master2">
SELECT DATE(Timedetail) as FIRSTCONN
, count(Timedetail) as FIRSTOccurances
, EVENTS
FROM MyDatabase
WHERE EVENTS = "FIRST" GROUP BY FIRSTCONN
<!--- LIMIT 0,10 --->;
</cfquery>
<cfdump var="#master2#">
<cfquery dbtype="query" name="detail2">
SELECT *
FROM master2
WHERE FIRSTCONN >= <cfqueryparam value="#startdate#" cfsqltype="cf_sql_date">
AND FIRSTCONN < <cfqueryparam value="#enddate#" cfsqltype="cf_sql_date">;
</cfquery>
Dumping Result: <cfdump var="#detail2#"><cfabort>
However, when I try to use the following check on the QoQ:
Dumping Result: <cfdump var="#detail2.FIRSTCONN#"><cfabort>
I don't see the full list of FIRSTCONN values. Instead I just see one line:
Dumping Result: {ts '2013-06-29 00:00:00'}
Ideally I should see the list of all the FIRSTCONN in my browser, shouldn't I?
You are looking at the default behaviour of coldfusion. When you output or dump queryname.fieldname, and don't specify a row number, you get the value from the first row. If you want to see all the rows, your choices are:
Look at the value list
Output/dump the entire query
Do another q of q for just that column and cfdump it.
Use cfoutput with a query attribute and just output that field
If you are looking to produce the same structured output that cfdump produces when dumping a query, I have two suggestions:
My First Inclination:
<cfdump var="#ListToArray(ValueList(queryName.columnName))#" />
That one is obviously a, very, minor continuation on Dan's suggestion.
The second is available for CF 8+ and it is
Exactly What You Wanted:
<cfdump var="#queryName#" show="columnName"/>
You may specify either columns to display in the output via the show attributes or you can hide specific columns by assigning a value to the hide attribute.
CFAbort in CF Docs
Granted, this post is almost 18 months old but maybe this will help someone that stumbles onto this page.
This is a little off topic, but I'd like to point out that in my instance of CF2016 the cfdump function suppresses the output of columns (and makes them appear empty) that have lots of text in them (or have the option to have lots of text). I'm not sure if it's the nVarChar(max) setting of the table field or what, but in order to see the content of these big fields, I actually have to make a separate query that selects only this one field, and use a separate cfdump in order to see its contents. This is only for debugging purposes, but it will keep you from going crazy and rewriting your update and insert statements over and over (because they appear to not be working all the way)...
I am trying to write a custom tag that will iterate over a cfquery object in a special way. I found this page: http://www.zrinity.com/developers/mx/undocumentation/query.cfm outlining how to use the underlying java methods to navigate the result set, but it doesn't seem to be working in CF9.
I can call .next(), .previous(), .first(), and .last() just fine, and each method updates query.currentRow, but referencing query.columnName always returns the value from the first row, not currentRow.
Example:
<cfquery name="testQuery" datasource="source">
SELECT FooName FROM NumberedFoos
</cfquery>
<cfloop from="1" to="3" index="i">
<cfoutput>#testQuery.currentRow# => #testQuery.fooName#</cfoutput><br />
<cfset testQuery.next()>
</cfloop>
Produces:
1 => Foo 1
2 => Foo 1
3 => Foo 1
I know i could use something like testQuery.fooName[testQuery.currentRow], but that is pretty undesirable for the people I am making the custom tag for. Was the functionality described in the above link removed from CF9? If so is there an alternative?
EDIT
To expand on the why, the client wants a custom tag that allows them to "assert" certain things about a query. The client has a pretty low level understanding of CF, but are pretty solid writing SQL. Their desired end result is something akin to:
<cfquery name="purchaseTotals">
SELECT PurchaseId, Total FROM Purchases
</cfquery>
<CF_ASSERT query="purchaseTotals">
purchaseTotals.Total gte 0
</CF_ASSERT>
The desired output would be a html table with each row being the row from the query that fails the assertion. So to me, the CF_ASSERT tag need to be able to update the current row.
Edit 2:
The main challenge is to allow html in the body of the tag, while still having query values substituted from the appropriate row:
<CF_ASSERT query="purchaseTotals">
<CF_CONDITION expression="purchaseTotals.Total gte 0">
<!---error message when expression is false--->
<cfoutput>
Purchase #purchaseTotals.purchaseId# has a negative total!
</cfoutput>
</CF_CONDITION>
<CF_CONDITION expression="purchaseTotals.Total eq ''">
#PurchaseTotals.purchaseId# has a null total, this may be caused by:
<ul>
<li>Edge Case 1</li>
<li>Edge Case 2</li>
</ul>
</CF_CONDITION>
<CF_ASSERT>
The output here would be something like:
Purchase 120 has a negative total!
Purchase 157 has a negative total!
Purchase 157 has a null total, this may be caused by:
Edge Case 1Edge Case 2
Was the functionality described in the above link removed from CF9?
The internal stuff has definitely changed since the article was written in 2006. But I suspect the exact functionality you are describing may not have existed in any mx version. A key difference between your code and the linked examples is the usage of <cfoutput query=".."> (not just a plain <cfoutput>). The query attribute obviously provides some extra context when evaluating the variables. Remove it (like in your example) and the results are "the value from the first row, not currentRow.". Even under MX6, which does not bode well for subsequent versions. That exact functionality probably was not removed. It just never worked to begin with.
If so is there an alternative?
Like I said earlier, the cleanest approach would be to use array notion ie #query.column[row]#. Given that you seem to have rejected that option, you are basically left with evaluate(). You would need to loop through the query within the parent tag. Then use evaluate to process the subtag expression and content. It is not particularly elegant or simple IMO. But I think that may be good as it gets without array notation, or a ritual sacrifice of some kind.
ASSERT.cfm
<cfparam name="attributes.query" type="string">
<cfif thisTag.ExecutionMode is 'start'>
<!--- validate attributes.query is a query object --->
<cfif not ( structKeyExists(caller, attributes.query) AND IsQuery(caller[attributes.query]) )>
<cfthrow message="Attributes.query [#attributes.query#] is undefined or not a query object">
</cfif>
</cfif>
<cfif thisTag.ExecutionMode is 'end'>
<cfset variables[attributes.query] = caller[attributes.query]>
<cfloop query="variables.#attributes.query#">
<cfloop array="#thisTag.assocAttribs#" index="subTag">
<cfset variables.matchFound = evaluate(subTag.expression)>
<cfif variables.matchFound>
<cfoutput>[#currentRow#] #evaluate(DE(subTag.Content))#</cfoutput><hr>
</cfif>
</cfloop>
</cfloop>
</cfif>
CONDITION.cfm
Note: Do NOT use <cfoutput> tags within the tag content.
<cfparam name="attributes.expression" type="string">
<cfif thisTag.ExecutionMode is "start">
<cfassociate baseTag="CF_ASSERT">
</cfif>
<cfif thisTag.ExecutionMode is "end">
<cfset attributes.content = thisTag.GeneratedContent>
<cfset thisTag.GeneratedContent = "">
</cfif>
client has a pretty low level understanding of CF, but are pretty
solid writing SQL
Having said all that, are things being implemented this way because it is the best approach or because it is most similar to writing SQL ie comfortable ?
Classic example of the inner platform effect.
I would advise you not to do this as you are trying to create a system which mimics built in functionality of the base or running system which ultimately becomes a poorly implemented version of the system in which it is running on / implemented in.
Sounds confusing, I know - but this is a well known anti-pattern to avoid.
See here for more information : http://en.wikipedia.org/wiki/Inner-platform_effect
P.s. what you are looking for (although I disagree with the implementation) is itteration of a query outside of a cfloop, simply use array syntax :
#queryName.fieldName[rowNumber]#
Using this you can itterate the query however you wish, certainly no need for underlying java. Notice we aren't using queryName.currentRow. For previous() next() functionality, you just change the rowNumber up or down.