Script to parse US Code files

parse-content.wls

Map[
 Cases[
   Import[#, "XML"],
   XMLElement[tag_,
     {___, "identifier" -> id_}, {
      ___,
      XMLElement[
       Pattern[container,
        "content"
         | "heading"
         | "chapeau"
        ], _, {Pattern[content, XMLElement[_, _, {text_}]], ___}],
      ___
      }]
    :> {
     "id" -> 
      StringReplace[StringJoin[id, "-", tag, "-", container], 
       "/" -> "-"],
     "categories" -> StringSplit[id, "/"],
     "tag" -> tag,
     "container" -> container,
     "text" -> text
     }
   , \[Infinity]] &,
 FileNames[__ ~~ ".xml", "/home/sguzman/Code/docs/us-code"]]