feat: Added PI Parsing

src/glxml.gleam

@@ -38,11 +38,13 @@ pub type Element {
   Text(content: String)
   Comment(content: String)
   CData(content: String)
+  PI(name: String, content: String)
+  Whitespace
 }
 
 pub fn main() {
   parse_document(
-    "<?xml version=\"1.1\" encoding='UTF-8'?>\r\n   <!-- hello-world -->   \n<b><a attr='ha&#x20;&#38;#38;ha' battr='baba' ref='&amp;'/><!-- ma comment --><![CDATA[Testing&&<haha>]]></b>",
+    "<?xml version=\"1.1\" encoding='UTF-8'?>\r\n   <!-- hello-world -->   \n<b><a attr='ha&#x20;&#38;#38;ha' battr='baba' ref='&amp;'/><!-- ma comment --><![CDATA[Testing&&<haha>]]><?test asuhashd ?></b>",
   )
   |> echo
 }
@@ -153,13 +155,9 @@ fn parse_content(
         try_parsers(
           [
             parse_element(_, doctype),
-            fn(doc) {
+            parse_comment,
-              case parse_comment(doc) {
-                Ok(#(comment, doc)) -> Ok(#(Comment(comment), doc))
-                Error(_) -> Error(Nil)
-              }
-            },
             parse_cdata,
+            parse_pi,
           ],
           doc,
         )
@@ -172,6 +170,38 @@ fn parse_content(
   }
 }
 
+fn parse_pi(doc: String) -> Result(#(Element, String), Nil) {
+  case doc {
+    "<?" <> tail -> {
+      use #(name, doc) <- result.try(parse_name(tail))
+      use <- bool.guard(
+        when: string.lowercase(name) == "xml",
+        return: Error(Nil),
+      )
+      let doc = trim_space(doc)
+      use #(content, doc) <- result.try(parse_pi_content(doc, ""))
+      case doc {
+        "?>" <> tail -> Ok(#(PI(name, content), tail))
+        _ -> Error(Nil)
+      }
+    }
+    _ -> Error(Nil)
+  }
+}
+
+fn parse_pi_content(doc: String, pi: String) -> Result(#(String, String), Nil) {
+  case doc {
+    "?>" <> _ -> Ok(#(pi, doc))
+    "" -> Error(Nil)
+    _ -> {
+      case parse_char(doc) {
+        Ok(#(char, doc)) -> parse_pi_content(doc, pi <> char)
+        Error(_) -> Ok(#(pi, doc))
+      }
+    }
+  }
+}
+
 fn parse_cdata(doc: String) -> Result(#(Element, String), Nil) {
   case doc {
     "<![CDATA[" <> tail -> {
@@ -404,13 +434,22 @@ fn parse_prolog(
 }
 
 fn parse_misc(doc: String) -> String {
-  let #(_, doc) =
+  case
-    parse_multiple_optional(
+    try_parsers(
+      [
+        parse_comment,
+        fn(doc) {
+          parse_space(doc)
+          |> result.map(fn(sp) { #(Whitespace, sp.1) })
+        },
+        parse_pi,
+      ],
       doc,
-      try_parsers([parse_comment, parse_space], _),
-      "",
     )
-  doc
+  {
+    Ok(#(_element, doc)) -> parse_misc(doc)
+    Error(Nil) -> doc
+  }
 }
 
 fn parse_decl(doc: String) -> Result(#(Declaration, String), Nil) {
@@ -640,12 +679,12 @@ fn parse_alpha(doc: String) -> Result(#(String, String), Nil) {
   }
 }
 
-fn parse_comment(doc: String) -> Result(#(String, String), Nil) {
+fn parse_comment(doc: String) -> Result(#(Element, String), Nil) {
   case doc {
     "<!--" <> tail -> {
       let #(comment, doc) = do_parse_comment(tail)
       case doc {
-        "-->" <> tail -> Ok(#(comment, tail))
+        "-->" <> tail -> Ok(#(Comment(comment), tail))
         _ -> Error(Nil)
       }
     }