Digital Humanities

• XPath stands for "XML path language"
• A query language used to search for data within a larger XML system
• It is standard way to find specific XML elements
• If you imagine the XML document as a tree, you can use XPath to navigate the tree and get to the leaves you need
• The expressions are called paths - imagine you are telling the computer the path you want to take to get to the elements you are searching for

​​​

<div type="dedicatoryEpistle">      <p> Hi </p> </div>

• XML (eXtensible Markup Language) is a way to mark up or annotate text in a way that can easily be expanded (extensible)
• Elements in XML are mostly structured within each other
• An element is anything that is contained by a tag like this: <tag> I am an element </tag>
• A tag begins with its name surrounded by <> and ends with its name surrounded by </>

• We can use XPath to find the elements we want based on the relationship elements have with each other and the data associated with them

• div is an element node
• type would be an attribute node

<root>      <div type="dedicatoryEpistle">           <salute> Hi </salute>           <p> Hi </p>      </div>      <p> Hello </p> </root>

• Start expressions with / or // and have a tag name after it - you can chain them together

• Starting the expression with / means it will find elements starting from the root - the root is simply the element that contains every other element (it can be called anything)
  • If search for salute with /salute, we won't find it because it's not directly under the root
  • We need to do /root/div/salute and then we will find it - exact location
  • This is called the absolute path (the exact location relative to the root)

• Starting with // means it will locate elements no matter where they are in the document - better for searching
  • //salute will find salute no matter where
  • //p will get both the <p> inside and outside the div
  • This is called the relative path because it can be relative to anything

• Elements can be related to each other in many ways when using XML
• An element can be a:
  • Parent - The element that directly contains the current element (1 level up)
  • Ancestor - Anything that is containing the current element or its parent (many levels up)
  • Child - Anything that the current element directly contains (1 level within)
  • Sibling - Any element next to the current element (or is contained by the same parent as the current element)
  • Descendant - Anything that the current element or its children contain (many levels within)

Parent:

<root>      <div type="dedicatoryEpistle">           <salute> Hi </salute>           <p> Hi </p>      </div>      <p> Hello </p> </root>

• div is the parent of salute
• It is one level above salute

Ancestor:

<root>      <div type="dedicatoryEpistle">           <salute> Hi </salute>           <p> Hi </p>      </div>      <p> Hello </p> </root>

• root and div are both ancestors of salute
• They are one or more levels above salute and root contains the parents of salute
• Ancestors can be thought of the parents + its parents and so on

Child:

<root>      <div type="dedicatoryEpistle">           <salute> Hi </salute>           <p> Hi </p>      </div>      <p> Hello </p> </root>

• div is the child of the root
• It is one level within the root

Sibling:

<root>      <div type="dedicatoryEpistle">           <salute> Hi </salute>           <p> Hi </p>      </div>      <p> Hello </p> </root>

• div and p are siblings because they are next to each other
• They also share the same parent, the root

Descendant:

<root>      <div type="dedicatoryEpistle">           <salute> Hi </salute>           <p> Hi </p>      </div>      <p> Hello </p> </root>

• div, p, and everything inside them are descendants of the root
• salute, and the other p are all children of the children and are multiple levels within
• Descendants can be thought of as children + their children and so on

• We can use the relationships elements have with each other to tell XPath exactly what elements we're looking at
• These ways to specify are called XPath axes
• Each relationship has one or more axis associated with it

 

• We write axes like this - axis::target-elements
  • axis is the type of relationship we want to look at
  • target-elements are the elements we want to select which have that relationship
  • Use "::" in between them
  • They go in the middle of the normal path expressions

• parent - parent (can be abbreviated with ".." in some cases)
• child - child
• ancestor
  • ancestor - ignore the current element
  • ancestor-or-self - include it
• descendent
  • descendent - ignore the current element
  • descendent-or-self - include it
• sibling
  • preceding-sibling - only the siblings before an element
  • following-sibling - only the siblings after an element
• There are a few more but they are not part of the main relationships and can be called on using different and more convenient notation

• //persName/ancestor::p
  • This will search the whole document for <persName> tags and then select all the ancestors that are <p> tags
• //p/child::note
  • This will find all the <p> tags and then select only the <note> children
  • You don't need to use this expression, of course there is a simpler way to get the note elements (//note)

• Many of these expressions give you multiple results (a list of results)
• How can we find the specific one we are looking for if we need only one?
• We can target elements with certain properties by using [ ] to the right of a tag name
• In the square brackets, we can specify a number starting with 1 to pick the element from the list if we know which we want
  • /TEI[1]/text[1]/body[1]/div[1]/p[18]/persName[5]
  • This is the absolute location of a <persName> tag by specifying that it's inside the 1st <TEI>, the 1st <text>, the 1st <body>, the 1st <div>, the 18th <p> and it will be the 5th <persName>

• We can also specify further which elements we want by picking certain attribute values if we don't know the exact numbers

• We specify that we are talking about an attribute in the square brackets [ ] by using the @ symbol

  • //p[18]/persName[@type="hist"]
  • This will find the 18th <p> then pick all the <persName> tags that have a "type" attribute with the value "hist"
  • In other words, all the historical figures in the 18th paragraph

• The square brackets can be chained to further refine the result

  • //p[18]/persName[@type="hist"][1]
  • This will take the list above and pick the first <persName> in it
  • So, the first historical figure in that paragraph

• You can do this with any attribute and value

• Those are all the main techniques needed to use XPath
• We recommend using these techniques to design better queries to get the data you need for research!

• Download the Discouerie of Witchcraft XML file and experiment with XPath:
  • https://tapasproject.org/kit-marlowe/files/discouerie-witchcraft
• Or use your own text!

1. W3Schools XPath guide - a good overview of everything: https://www.w3schools.com/xml/xpath_intro.asp
2. Axes guide - describes the relationships and axes we talked about: https://youtu.be/aAWvwGFkySI
3. Mozilla XPath axes - list of all the axes and what they do by the developers of Firefox: https://developer.mozilla.org/en-US/docs/Web/XPath/Axes
4. Mozilla XPath functions guide - we didn't talk about this, but lists functions which can be used in the square brackets: https://developer.mozilla.org/en-US/docs/Web/XPath/Functions

CitizenK, Cuimingda, Dria, ExE-Boss, Fredchat, Jonnyq, mete0r, Mgjbot, Nickolay, Ptak82, Sheppy, & SphinxKnight. (n.d.). Axes. Axes - XPath | MDN. Retrieved December 13, 2021, from https://developer.mozilla.org/en-US/docs/Web/XPath/Axes.

CitizenK, Cuimingda, Dria, ExE-Boss, Fredchat, Jonnyq, mete0r, Mgjbot, Nickolay, Ptak82, Sheppy, & SphinxKnight. (n.d.). Functions. Functions - XPath | MDN. Retrieved December 13, 2021, from https://developer.mozilla.org/en-US/docs/Web/XPath/Functions.

W3Schools. (n.d.). XPath Tutorial. Retrieved December 13, 2021, from https://www.w3schools.com/xml/xpath_intro.asp.

"XPath Axes - ancestor, parent, following-sibling, preceding-sibling, child, descendant." YouTube, uploaded by H Y R Tutorials, December 13, 2021, https://youtu.be/aAWvwGFkySI

Fall 2021 Honors Digital Humanities Class: Nick Ribeiro and Ellie Lynch

Spring 2022 Library Digital Humanities Intern: Meeghan Bresnahan