Xpaths of every element, in document order

The same result as xml2::xml_path(xml2::xml_find_all(doc, "//*")), in linear time. libxml2 builds each path by scanning the element's siblings, which is quadratic when one parent has many children (a 990-PF grant list with tens of thousands of entries took 4-7 minutes; returns of 250+ MB would take hours). Here each parent's children are named once: an element gets [k] only when a sibling has the same name, as libxml2 does.

Usage

get_xml_paths(doc)

Arguments

doc

An xml2 document (namespaces stripped).

Value

Character vector of xpaths, one per element, in document order.

Details

Documents with elements still in a namespace after xml_ns_strip() (see EF2-11) fall back to xml2::xml_path(), which writes the prefixes.