So where do you fit in the xinc:node-expand? Is it better to do it like this or expand the nodes straight away. I¹m not sure the performance of xinc:node-expand is great?
import module namespace xinc = "http://marklogic.com/xinclude"; at "/MarkLogic/xinclude/xinclude.xqy"; let $doc := collection()[*/itemMeta/url/@href eq $public_url]/* return xinc:node-expand($doc) Thanks Jon On 11/01/2012 22:36, "Evan Lenz" <[email protected]> wrote: > Or slightly more simply: > > /*[itemMeta/url/@href eq $public_url] > > However, I gave this to xdmp:plan() and, for some reason, it's not fully > leveraging the indexes here, despite my assumption that it would. At index > resolution, the above query selects all the XML documents in the database > (same as /*). > > It turns out xdmp:plan() is pretty essential here to figure out a formulation > that works to fully leverage the indexes (even though they theoretically all > should work). > > The following reformulation is no better: > > xdmp:plan( > collection()[/*/itemMeta/url/@href eq $public_url]/* > ) > > However, strangely, the following (equivalent) reformulation hits the mark > (selects 0 fragments in my database, which has no <itemMeta> elements): > > xdmp:plan( > collection()[*/itemMeta/url/@href eq $public_url]/* > ) > > And it's the only one that gives me this reassuring message: Step 1 predicate > 1 contributed 3 constraints: */itemMeta/url/@href eq "some_string" > > Moral of the story is that xdmp:plan() is your friend. And the last query > above is the most efficient. > > Evan Lenz > Software Developer, Community > MarkLogic Corporation > http://developer.marklogic.com > > From: Geert Josten <[email protected]> > Reply-To: General MarkLogic Developer Discussion > <[email protected]> > Date: Wed, 11 Jan 2012 13:47:33 -0800 > To: General MarkLogic Developer Discussion <[email protected]> > Subject: Re: [MarkLogic Dev General] Are xml attributes indexed by default > > Re: [MarkLogic Dev General] Are xml attributes indexed by default > Hi Jon, > > Your original expression was like this: > > /*[ > /*/itemMeta/url[@href=$public_url] > ] > > That would return every root element in the database if there is any doc which > meets the predicate expression. I think writing it as follows might improve > things: > > /*[ > ./itemMeta/url[@href=$public_url] > ] > > (note the period before /itemMeta) > > You could also use cts functions to explicitly rely on indexes: > > cts:search(collection(), cts:element-query(xs:QName(³itemMeta²), > cts:element-attribute-value-query(xs:QName(³url²), xs:QName(³href²), > $public_url)))/* > > Using CQ or Query Console Profile features, or functions like xdmp:plan can > also help to pinpoint and analyze performance problems.. > > Kind regards, > Geert > > > Van: [email protected] > [mailto:[email protected]] Namens Jonathan Cook - FM&T > Verzonden: woensdag 11 januari 2012 15:02 > Aan: General MarkLogic Developer Discussion > Onderwerp: Re: [MarkLogic Dev General] Are xml attributes indexed by default > > Thanks for the reply, > > In words I want to select the document that matches a given href eg return the > whole document which matches itemMeta/url[@href=$public_url] > > Xml could look like this: > > <?xml version="1.0" encoding="UTF-8"?> > <story> > <itemMeta> > <url href="news/business-1234"/> > <assetTypeCode>STY</assetTypeCode> > <headline>Full Headline</headline> > <shortHeadline>Headline</shortHeadline> > <summary>Summary</summary> > <section id="1" name="Technology"/> > <firstCreated>2011-10-17T14:47:52+00:00</firstCreated> > <lastUpdated>2011-12-05T13:40:19+00:00</lastUpdated> > <publicationStatus>PUBLISHED</publicationStatus> > <language>en-GB</language> > <siteId>News v6</siteId> > <provider>cps</provider> > </itemMeta> > <pageOptions/> > <body/> > <media> > <images/> > </media> > <relatedGroups/> > </story> > > So you would want to return the whole story document. However other documents > could have a different root node. ItemMeta will always exist under the root > node with this structure so as you say I suspect there is a more efficient way > of doing it. > > I had a quick play with let $results := > xinc:node-expand(//bbc:itemMeta/bbc:url[@href=$public_url]) but that just > returns the actual url element not the whole document. > > Thanks > Jon > > > On 11/01/2012 13:39, "Geert Josten" <[email protected]> wrote: > Hi Jon, > > You always have the Word index (Word lexicon), which can be explicitly used > with cts:element-attribute-word-query (and the element counterpart > cts:element-word-query. You will have to enable wildcard switches on the > database to use wildcards within the searched values. If you want to lookup > exact values, you might be better off with a range index, which can be > explicitly used with cts:element-value-query and > cts:element-attribute-value-query. > > Whether or not XPath expressions are backed by these indexes and lexicons, > depends how well the query optimizer is able to match them. You might be > interested in this blog item from Evan, which explains how you can analyze > this: > > http://developer.marklogic.com/blog/learning-with-query-trace > > However, I think you are asking something different here, looking at your > code. To be honest, the argument of xinc:node-expand looks rather suspicious. > /* accesses the entire database, but you do it a second time in the predicate. > That is indeed very expensive. Can you put in words what you are trying to do? > We might be able to suggest better ways of handling this.. > > Kind regards, > Geert > > > Van: [email protected] > [mailto:[email protected]] Namens Jonathan Cook - FM&T > Verzonden: woensdag 11 januari 2012 12:56 > Aan: [email protected] > Onderwerp: [MarkLogic Dev General] Are xml attributes indexed by default > > Hi, > > Are attributes indexed by default in Mark Logic Server? The documentation says > the whole document structure is indexed but that doesn¹t answer the question > really. > > For example I want to ensure the href attribute is indexed for the following > xquery: > xquery version "1.0-ml"; > import module namespace xinc = "http://marklogic.com/xinclude"; at > "/MarkLogic/xinclude/xinclude.xqy"; > > let $public_url := "%s" > > let $results := xinc:node-expand(/*[ > /*/itemMeta/url[@href=$public_url] > ]) > > let $asset := $results[1] > return $asset > > Perhaps the above xquery isn¹t very efficient as I also read widcards couldn¹t > be used with indexes? > > This option is set to false but I suspect this is something different: > ³Index attribute value positions for faster near searches involving > element-attribute-value-query (slower document loads and larger database > files).² > > Thanks > Jon > > http://www.bbc.co.ukThis e-mail (and any attachments) is confidential and may > contain personal views which are not the views of the BBC unless specifically > stated.If you have received it in error, please delete it from your system.Do > not use, copy or disclose the information in any way nor act in reliance on it > and notify the sender immediately.Please note that the BBC monitors e-mails > sent or received.Further communication will signify your consent to this. > > > _______________________________________________ > General mailing list > [email protected] > http://developer.marklogic.com/mailman/listinfo/general > > http://www.bbc.co.uk/ > This e-mail (and any attachments) is confidential and may contain personal > views which are not the views of the BBC unless specifically stated. > If you have received it in error, please delete it from your system. > Do not use, copy or disclose the information in any way nor act in reliance on > it and notify the sender immediately. > Please note that the BBC monitors e-mails sent or received. > Further communication will signify your consent to this. > > > > _______________________________________________ > General mailing list > [email protected] > http://developer.marklogic.com/mailman/listinfo/general
_______________________________________________ General mailing list [email protected] http://developer.marklogic.com/mailman/listinfo/general
