XPath node predicates and functions Sarah Stanley Elisa Beshero-Bondar TEI Summer School, University of Graz, Fri, 13 Sep 19 Women Writers Project (via website)
url:mailto:wwp@northeastern.edu

Copyright 2018 Sarah Stanley, Martin D. Holmes, and the Women Writers Project

This TEI-encoded XML file is available under the terms of the Creative Commons Attribution-ShareAlike 3.0 (Unported) license.

Born digital

 Updated for 2019 Graz Summer School — mostly minor tweaks Added comments and suggestions, especially to show how to return a count of the number of times Hamlet speaks. Adapted slides from MDH and DJB Node Predicates and Ten Functions slide sets.
XPath Functions Functions in XPath work just like functions in other programming languages. A function is called by name, with following parentheses, containing parameters:functionName( param1, param2 ) Functions can wrap around the entire XPath expression or can be used within an expression to further refine results. Parameters are things you want the function to work on.
Some Essential Functions name() //sp/*/name() returns a list of all of the names of elements that occur within sp distinct-values() distinct-values( //sp/*/name() ) returns a list of all of the element types that occur inside speeches count() count( //sp/p ) gives us the number of paragraphs that occur as a child of speech
Some more functions contains() //stage[ contains( .,'exit') ] Shows us all the stage directions that contain the string exit . means search within the current element (self) and the string to be searched for is within quotes after the comma Note: straight quotes, never curly Question: why single quotes instead of double? substring-after() //sp/substring-after( @who,'#') Returns the part of param1 that is after param2 Useful if you want to find everything after a specific, consistent prefix Thus id( substring-after( ancestor::sp/@who,'#') ) gets you the role element of whoever is speaking at the moment (i.e., where the cursor is in oXygen), except fails silently when there is more than one speaker
XPath Predicates You use paths and axes in XPath to arrive at specific nodes in your XML. You use predicates to further filter or test those nodes. Only nodes which satisfy the predicate will be selected. Predicates follow the step they apply to, and use square brackets. /TEI/text/body/div means all the div nodes which are children of body. /TEI/text/body/div[ @type eq 'act'] means only those div nodes which are children of body and also have a type attribute with the value act.
Some Essential Predicates //div[1], //div/p[3], //list/item[12] Numbers as predicates allow you to search for the Nth child Axes (e.g., descendant::, parent::, and ancestor::) can be used in predicates //div[descendant::persName] searches for divisions that have descendants of persName— what you get back is a sequence of divs (as opposed to //div//persName, which searches for the persName elements that are descendants of div— what you get back is a sequence of persNames)
XPath Predicates: some examples //div[head] = all div nodes which have a direct child head element. //sp//stage[contains(.,"Ghost")] = all stage nodes that are inside an sp and contain the text Ghost. /TEI/teiHeader/fileDesc/titleStmt/editor[2] = the second editor node in the titleStmt. (//div)[position() = last()] = the last div node in the document. Any guesses about why (//div) is wrapped in parentheses here?
XPath Predicates can be chained //div[ @type eq 'scene'][ count(descendant::l) gt 200 ] = all scenes containing more than 200 lines. //div[ contains( head, 'Scene') ][ descendant::sp/@who eq '#Horatio'] = all scenes in which Horatio speaks. //sp[ position() eq last() ][ @who eq '#Hamlet'] = speeches by Hamlet which are the last speeches in their scene.

The Hamlet file actually has lots of ab elements in speeches as well as l elements, so this line count stuff is not a true reflection of the length of speeches. It might be worth telling the students there's something wrong with it, and asking them to look at the XML source to see if they can figure out what it is. Then you could amend the first item in this slide, and some of the ones in the next slide, to account for this.

The Hamlet file actually has lots of ab elements in speeches as well as l elements, so this line count stuff is not a true reflection of the length of speeches. It might be worth telling the students there's something wrong with it, and asking them to look at the XML source to see if they can figure out what it is. Then you could amend the first item in this slide, and some of the ones in the next slide, to account for this.
XPath Predicates can be nested //sp[ descendant::l[ contains(., "Ophelia") ] ] = all speeches which contain metrical lines which mention Ophelia. //div[ contains( head, "Scene") ][ count( descendant::sp[ @who eq '#Hamlet']) gt 30 ] = all scenes in which Hamlet has more than 30 speeches. //sp[ count( descendant::l[ contains( .,'love') ] ) gt 2] = speeches with three or more lines that mention love. //sp[ count( l[contains( .,'love') ] ) gt 2] How does this differ from the XPath above? Under what circumstances would this difference matter?
More fun with XPath Node Predicates and Functions What is the last speech that Hamlet gives in any given scene? - //sp[@who eq '#Hamlet'][ position() eq last() ] How do we modify this to find Hamlet's last speech overall? What happens if we switch those two predicates?
Let's try some exercises together How many speeches contain prose (encoded with ab)? How many lines of poetry are there in speeches within Hamlet? What elements occur within poetic lines in speeches? Prose within speeches? Find all of the exits where multiple people leave the stage (hint: exeunt is usually usedin stage directions where multiple characters leave the stage) How many times does Hamlet speak in the play? Find all of the scenes where Hamlet has more than 10 speeches.