feat: Work on declarations

src/glxml.gleam

@@ -7,7 +7,12 @@ import gleam/result
 import gleam/string
 
 pub type Declaration {
-  Declaration(versioninfo: String, encoding: String, standalone: Bool)
+  XMLDecl(versioninfo: String, encoding: String, standalone: Bool)
+  GEntityDecl
+  PEntityDecl(name: String, decl: String)
+  ElementDecl
+  AttListDecl
+  NotationDecl
 }
 
 pub type Entity {
@@ -16,8 +21,17 @@ pub type Entity {
   PublicExternalEntity(literal: String, pubidliteral: String)
 }
 
+pub type ExternalID {
+  SystemID(system_literal: String)
+  PublicID(system_literal: String, public_literal: String)
+}
+
 pub type DocType {
-  DocType(name: String, entities: dict.Dict(String, Entity))
+  DocType(
+    name: String,
+    external_id: Option(ExternalID),
+    entities: dict.Dict(String, Entity),
+  )
 }
 
 pub type Document {
@@ -387,7 +401,7 @@ fn process_reference(
   doctype: Option(DocType),
 ) -> Result(String, Nil) {
   case doctype {
-    Some(DocType(_, entities)) -> {
+    Some(DocType(_, _, entities)) -> {
       get_reference(entities, ref)
     }
     None -> {
@@ -431,13 +445,249 @@ fn do_parse_name(doc: String, name: String) -> Result(#(String, String), Nil) {
 fn parse_prolog(
   doc: String,
 ) -> Result(#(Declaration, Option(DocType), String), Nil) {
-  let #(decl, doc) = case parse_decl(doc) {
-    Ok(#(decl, doc)) -> #(decl, doc)
-    _ -> #(Declaration("1.0", "UTF-8", False), doc)
-  }
+  let #(decl, doc) =
+    parse_decl(doc) |> result.unwrap(#(XMLDecl("1.0", "UTF-8", False), doc))
+
   let doc = parse_misc(doc)
 
-  Ok(#(decl, None, doc))
+  let #(doctype, doc) =
+    parse_doctype(doc)
+    |> result.map(fn(d) { #(Some(d.0), d.1) })
+    |> result.unwrap(#(None, doc))
+
+  Ok(#(decl, doctype, doc))
+}
+
+fn parse_doctype(doc: String) -> Result(#(DocType, String), Nil) {
+  case doc {
+    "<!DOCTYPE" <> tail -> {
+      let doc = trim_space(tail)
+      use #(name, doc) <- result.try(parse_name(doc))
+
+      let #(external_id, doc) =
+        parse_external_id(doc) |> result.unwrap(#(None, doc))
+
+      let doc = trim_space(doc)
+
+      let #(int_subset, doc) =
+        parse_int_subset(doc) |> result.unwrap(#([], doc))
+
+      case doc {
+        ">" <> tail -> Ok(#(DocType(name, external_id, dict.new()), tail))
+        _ -> Error(Nil)
+      }
+    }
+    _ -> Error(Nil)
+  }
+}
+
+fn parse_int_subset(doc: String) -> Result(#(List(Declaration), String), Nil) {
+  let doc = trim_space(doc)
+  case doc {
+    "[" <> tail -> {
+      use #(decl_list, doc) <- result.try(do_parse_int_subset(tail, []))
+      case doc {
+        "]" <> tail -> {
+          Ok(#(decl_list, tail))
+        }
+        _ -> Error(Nil)
+      }
+    }
+    _ -> Error(Nil)
+  }
+}
+
+fn do_parse_int_subset(
+  doc: String,
+  decl_list: List(Declaration),
+) -> Result(#(List(Declaration), String), Nil) {
+  let doc = trim_space(doc)
+  case doc {
+    "%" <> tail -> {
+      use #(name, doc) <- result.try(parse_name(tail))
+      case doc {
+        ";" <> tail -> {
+          case get_entity_replacement(name, decl_list) {
+            Some(decl) -> {
+              do_parse_int_subset(decl <> tail, decl_list)
+            }
+            _ -> Error(Nil)
+          }
+        }
+        _ -> Error(Nil)
+      }
+    }
+    _ -> {
+      todo
+    }
+  }
+}
+
+fn get_entity_replacement(
+  entity: String,
+  decl_list: List(Declaration),
+) -> Option(String) {
+  list.find_map(decl_list, fn(decl) {
+    case decl {
+      PEntityDecl(name, decl) if name == entity -> Ok(Some(decl))
+      _ -> Error(Nil)
+    }
+  })
+  |> result.unwrap(None)
+}
+
+fn parse_external_id(doc: String) -> Result(#(Option(ExternalID), String), Nil) {
+  let doc = trim_space(doc)
+  case doc {
+    "SYSTEM" <> tail -> {
+      let doc = trim_space(tail)
+      use #(system_literal, doc) <- result.try(parse_system_literal(
+        doc,
+        None,
+        "",
+      ))
+      Ok(#(Some(SystemID(system_literal:)), doc))
+    }
+    "PUBLIC" <> tail -> {
+      let doc = trim_space(tail)
+      use #(public_literal, doc) <- result.try(parse_public_literal(
+        doc,
+        None,
+        "",
+      ))
+      let doc = trim_space(doc)
+      use #(system_literal, doc) <- result.try(parse_system_literal(
+        doc,
+        None,
+        "",
+      ))
+      Ok(#(Some(PublicID(system_literal:, public_literal:)), doc))
+    }
+    _ -> Error(Nil)
+  }
+}
+
+fn parse_public_literal(
+  doc: String,
+  quote: Option(String),
+  literal: String,
+) -> Result(#(String, String), Nil) {
+  case doc, quote {
+    "\"" as q <> tail, None | "'" as q <> tail, None ->
+      parse_public_literal(tail, Some(q), "")
+    "", _ -> Error(Nil)
+    _, None -> Error(Nil)
+    "\"" <> tail, Some("\"") -> Ok(#(literal, tail))
+    "'" <> tail, Some("'") -> Ok(#(literal, tail))
+    " " as char <> tail, Some(_)
+    | "\r" as char <> tail, Some(_)
+    | "\n" as char <> tail, Some(_)
+    | "0" as char <> tail, Some(_)
+    | "1" as char <> tail, Some(_)
+    | "2" as char <> tail, Some(_)
+    | "3" as char <> tail, Some(_)
+    | "4" as char <> tail, Some(_)
+    | "5" as char <> tail, Some(_)
+    | "6" as char <> tail, Some(_)
+    | "7" as char <> tail, Some(_)
+    | "8" as char <> tail, Some(_)
+    | "9" as char <> tail, Some(_)
+    | "a" as char <> tail, Some(_)
+    | "b" as char <> tail, Some(_)
+    | "c" as char <> tail, Some(_)
+    | "d" as char <> tail, Some(_)
+    | "e" as char <> tail, Some(_)
+    | "f" as char <> tail, Some(_)
+    | "g" as char <> tail, Some(_)
+    | "h" as char <> tail, Some(_)
+    | "i" as char <> tail, Some(_)
+    | "j" as char <> tail, Some(_)
+    | "k" as char <> tail, Some(_)
+    | "l" as char <> tail, Some(_)
+    | "m" as char <> tail, Some(_)
+    | "n" as char <> tail, Some(_)
+    | "o" as char <> tail, Some(_)
+    | "p" as char <> tail, Some(_)
+    | "q" as char <> tail, Some(_)
+    | "r" as char <> tail, Some(_)
+    | "s" as char <> tail, Some(_)
+    | "t" as char <> tail, Some(_)
+    | "u" as char <> tail, Some(_)
+    | "v" as char <> tail, Some(_)
+    | "w" as char <> tail, Some(_)
+    | "x" as char <> tail, Some(_)
+    | "y" as char <> tail, Some(_)
+    | "z" as char <> tail, Some(_)
+    | "A" as char <> tail, Some(_)
+    | "B" as char <> tail, Some(_)
+    | "C" as char <> tail, Some(_)
+    | "D" as char <> tail, Some(_)
+    | "E" as char <> tail, Some(_)
+    | "F" as char <> tail, Some(_)
+    | "G" as char <> tail, Some(_)
+    | "H" as char <> tail, Some(_)
+    | "I" as char <> tail, Some(_)
+    | "J" as char <> tail, Some(_)
+    | "K" as char <> tail, Some(_)
+    | "L" as char <> tail, Some(_)
+    | "M" as char <> tail, Some(_)
+    | "N" as char <> tail, Some(_)
+    | "O" as char <> tail, Some(_)
+    | "P" as char <> tail, Some(_)
+    | "Q" as char <> tail, Some(_)
+    | "R" as char <> tail, Some(_)
+    | "S" as char <> tail, Some(_)
+    | "T" as char <> tail, Some(_)
+    | "U" as char <> tail, Some(_)
+    | "V" as char <> tail, Some(_)
+    | "W" as char <> tail, Some(_)
+    | "X" as char <> tail, Some(_)
+    | "Y" as char <> tail, Some(_)
+    | "Z" as char <> tail, Some(_)
+    | "-" as char <> tail, Some(_)
+    | "(" as char <> tail, Some(_)
+    | ")" as char <> tail, Some(_)
+    | "+" as char <> tail, Some(_)
+    | "," as char <> tail, Some(_)
+    | "." as char <> tail, Some(_)
+    | "/" as char <> tail, Some(_)
+    | ":" as char <> tail, Some(_)
+    | "=" as char <> tail, Some(_)
+    | "?" as char <> tail, Some(_)
+    | ";" as char <> tail, Some(_)
+    | "!" as char <> tail, Some(_)
+    | "*" as char <> tail, Some(_)
+    | "#" as char <> tail, Some(_)
+    | "@" as char <> tail, Some(_)
+    | "$" as char <> tail, Some(_)
+    | "_" as char <> tail, Some(_)
+    | "%" as char <> tail, Some(_)
+    | "'" as char <> tail, Some("\"")
+    -> {
+      parse_public_literal(tail, quote, literal <> char)
+    }
+    _, _ -> Error(Nil)
+  }
+}
+
+fn parse_system_literal(
+  doc: String,
+  quote: Option(String),
+  literal: String,
+) -> Result(#(String, String), Nil) {
+  case doc, quote {
+    "\"" as q <> tail, None | "'" as q <> tail, None ->
+      parse_system_literal(tail, Some(q), "")
+    "", _ -> Error(Nil)
+    _, None -> Error(Nil)
+    "\"" <> tail, Some("\"") -> Ok(#(literal, tail))
+    "'" <> tail, Some("'") -> Ok(#(literal, tail))
+    _, _ -> {
+      let assert Ok(#(char, tail)) = string.pop_grapheme(doc)
+
+      parse_system_literal(tail, quote, literal <> char)
+    }
+  }
 }
 
 fn parse_misc(doc: String) -> String {
@@ -463,18 +713,14 @@ fn parse_decl(doc: String) -> Result(#(Declaration, String), Nil) {
   case doc {
     "<?xml" <> tail -> {
       use #(versioninfo, doc) <- result.try(parse_versioninfo(tail))
-      let #(encoding, doc) = case parse_encodingdecl(doc) {
-        Ok(e) -> e
-        Error(_) -> #("", doc)
-      }
-      let #(standalone, doc) = case parse_standalone(doc) {
-        Ok(e) -> e
-        Error(_) -> #(False, doc)
-      }
+      let #(encoding, doc) =
+        parse_encodingdecl(doc) |> result.unwrap(#("", doc))
+      let #(standalone, doc) =
+        parse_standalone(doc) |> result.unwrap(#(False, doc))
 
       case trim_space(doc) {
         "?>" <> tail ->
-          Ok(#(Declaration(versioninfo:, encoding:, standalone:), tail))
+          Ok(#(XMLDecl(versioninfo:, encoding:, standalone:), tail))
         _ -> Error(Nil)
       }
     }