7.7.3. PUGIXML-03 — XPath Queries

This tutorial demonstrates querying XML documents with XPath, using both convenience wrappers and compiled queries in pugixml/PUGIXML_boost.

The tutorial uses an inline catalog XML for most examples, then queries books.xml at the end.

7.7.3.1. select_text — first match text

select_text runs an XPath query and returns the text content of the first matching node. Returns a default string if nothing matches:

parse_xml(CATALOG_XML) <| $(doc, ok) {
    if (!ok) { return; }
    let root = doc.document_element

    let first_name = select_text(root, "product[1]/name")
    print("first product: {first_name}\n")
    // first product: Wireless Mouse

    let missing = select_text(root, "product/description", "N/A")
    print("description: {missing}\n")
    // description: N/A
}

7.7.3.2. select_value — attribute or text

select_value returns the string value of the first XPath match — either an attribute’s value or an element’s text content:

let id = select_value(root, "product[1]/@id")
// "A1"

let cat = select_value(root, "product[@id='B1']/@category")
// "books"

let price = select_value(root, "product[@id='A2']/price")
// "79.99"

let missing = select_value(root, "product/@color", "none")
// "none"

7.7.3.3. for_each_select — iterate matches

for_each_select invokes a block for every matching xpath_node. Use .node or .attribute to access the underlying XML node or attribute:

// Iterate element results
root |> for_each_select("product/name") <| $(xn) {
    let n = xn.node
    print("  {n.text as string}\n")
}

// Iterate attribute results
root |> for_each_select("product/@id") <| $(xn) {
    let a = xn.attribute
    print("  {a as string}\n")
}

This is the most common pattern for processing multiple XPath results.

7.7.3.4. Compiled XPath with with_xpath

with_xpath pre-compiles an XPath expression once and frees it when the block ends. Use evaluate_node_set to run the compiled query:

with_xpath("product/name") <| $(query) {
    var ns = evaluate_node_set(query, root)
    print("Product count: {ns.size}\n")

    ns |> for_each() <| $(xn) {
        let n = xn.node
        print("  {n.text as string}\n")
    }
    unsafe { delete ns; }
}

7.7.3.5. Low-level XPath API

select_node returns the first match. select_nodes returns all matches as an xpath_node_set, which is directly iterable in a for loop (via the each iterator):

// First match
let result = select_node(root, "product[price > 40]/name")
if (result.ok) {
    print("first expensive: {result.node.text as string}\n")
}

// All matches — iterate directly
var ns = select_nodes(root, "product[price < 40]/@id")
print("cheap products: {ns.size}\n")
for (xn in ns) {
    let a = xn.attribute
    print("  {a as string}\n")
}
unsafe { delete ns; }

7.7.3.6. Practical example — querying books.xml

The tutorial ends by querying the real books.xml sample file:

open_xml("tutorials/dasPUGIXML/books.xml") <| $(doc, ok) {
    if (!ok) { return; }
    let root = doc.document_element

    var en_books = select_nodes(root, "book[@lang='en']")
    print("English books: {en_books.size}\n")
    unsafe { delete en_books; }

    let cheapest = select_text(root, "book[not(price > ../book/price)]/title")
    print("cheapest: {cheapest}\n")

    root |> for_each_select("book/author") <| $(xn) {
        print("  {xn.node.text as string}\n")
    }
}

See also

Full source: tutorials/dasPUGIXML/03_xpath.das

Previous tutorial: PUGIXML-02 — Building XML

Next tutorial: PUGIXML-04 — Struct Serialization