feat: Added comment parsing

src/glxml.gleam

@@ -1,5 +1,6 @@
 import gleam/option.{type Option, None}
 import gleam/result
+import gleam/string
 
 pub type Declaration {
   Declaration(versioninfo: String, encoding: String, standalone: Bool)
@@ -14,7 +15,10 @@ pub type Document {
 }
 
 pub fn main() {
-  parse_document("<?xml version=\"1.1\" encoding='UTF-8'?>") |> echo
+  parse_document(
+    "<?xml version=\"1.1\" encoding='UTF-8'?>\r\n   <!-- hello-world -->   \n",
+  )
+  |> echo
 }
 
 fn parse_document(doc: String) -> Result(Document, Nil) {
@@ -30,7 +34,14 @@ fn parse_prolog(
     Ok(#(decl, doc)) -> #(decl, doc)
     _ -> #(Declaration("1.0", "UTF-8", False), doc)
   }
-
+  let #(comment, doc) =
+    parse_multiple_optional(
+      doc,
+      try_parsers([parse_comment, parse_space], _),
+      "",
+    )
+  comment |> echo
+  doc |> echo
   Ok(#(decl, None, doc))
 }
 
@@ -237,6 +248,67 @@ fn do_parse_alpha(doc: String) -> Result(#(String, String), Nil) {
   }
 }
 
+fn parse_comment(doc: String) -> Result(#(String, String), Nil) {
+  case doc {
+    "<!--" <> tail -> {
+      let #(comment, doc) = do_parse_comment(tail)
+      case doc {
+        "-->" <> tail -> Ok(#(comment, tail))
+        _ -> Error(Nil)
+      }
+    }
+    _ -> Error(Nil)
+  }
+}
+
+fn do_parse_comment(doc: String) -> #(String, String) {
+  parse_multiple_optional(
+    doc,
+    try_parsers(
+      [
+        parse_char_except_dash,
+        fn(doc) {
+          case doc {
+            "-" <> tail -> {
+              use #(char, doc) <- result.try(parse_char_except_dash(tail))
+              Ok(#("-" <> char, doc))
+            }
+            _ -> Error(Nil)
+          }
+        },
+      ],
+      _,
+    ),
+    "",
+  )
+}
+
+fn parse_char_except_dash(doc: String) -> Result(#(String, String), Nil) {
+  case doc {
+    "-" <> _ -> Error(Nil)
+    _ -> parse_char(doc)
+  }
+}
+
+fn parse_char(doc: String) -> Result(#(String, String), Nil) {
+  case string.pop_grapheme(doc) {
+    Ok(#("\r\n", tail)) -> Ok(#("\r\n", tail))
+    Ok(#("\t", tail)) -> Ok(#("\t", tail))
+    Ok(#("\n", tail)) -> Ok(#("\n", tail))
+    Ok(#("\r", tail)) -> Ok(#("\r", tail))
+    Ok(#(char, tail)) -> {
+      let assert [codepoint] = string.to_utf_codepoints(char)
+      case string.utf_codepoint_to_int(codepoint) {
+        i if i >= 0x20 && i <= 0xD7FF -> Ok(#(char, tail))
+        i if i >= 0xE000 && i <= 0xFFFD -> Ok(#(char, tail))
+        i if i >= 0x10000 && i <= 0x10FFFF -> Ok(#(char, tail))
+        _ -> Error(Nil)
+      }
+    }
+    Error(_) -> Error(Nil)
+  }
+}
+
 fn trim_space(doc: String) -> String {
   case parse_space(doc) {
     Ok(#(_, doc)) -> trim_space(doc)