From cb85c5d92d80edc60ce1dcfbf7abc249c7312edc Mon Sep 17 00:00:00 2001
From: Ryan Kavanagh <rak@rak.ac>
Date: Mon, 1 Jun 2020 12:55:43 -0400
Subject: Mime type parsing

---
 gemini/mimeType.ml | 43 ++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 40 insertions(+), 3 deletions(-)

(limited to 'gemini')

diff --git a/gemini/mimeType.ml b/gemini/mimeType.ml
index 0559168..1682416 100644
--- a/gemini/mimeType.ml
+++ b/gemini/mimeType.ml
@@ -24,9 +24,46 @@ struct
 
   exception ParseError of string
 
-  let from_string _ = { mime_type = "text";
-                        mime_subtype = "plain";
-                        mime_param = None }
+  let create_mimetype mime_type mime_subtype mime_param =
+    { mime_type; mime_subtype; mime_param; }
+
+  let from_string (s : string) =
+    (* RFC 2045 § 5.1 *)
+    (* RFC 6838 § 4.2 *)
+    (* FIXME: parameter does not handle quoted strings and
+       does not try to ensure that everything is well-formed *)
+    let open Angstrom in
+    let lift_or p q b = p b || q b in
+    let is_letter = function
+      | 'a' .. 'z' | 'A' .. 'Z' -> true
+      | _ -> false in
+    let is_digit = function
+      | '0' .. '9' -> true
+      | _ -> false in
+    let is_symbol = function
+      | '!' | '#' | '$' | '&' | '-' | '^' | '_' | '.' | '+' -> true
+      | _ -> false in
+    let is_space = function
+      | ' ' | '\t' -> true
+      | _ -> false in
+    let take_all = take_while (fun _ -> true) in
+    let maybe p =
+      option None (lift (fun s -> Some s) p) in
+    let restricted_name =
+      lift2 (fun c -> fun s -> (Char.escaped c) ^ s)
+        (satisfy (lift_or is_letter is_digit))
+        (take_while (lift_or is_letter (lift_or is_digit is_symbol))) in
+    let parameter = lift2 (fun p -> fun v -> (p, v))
+        (take_till (fun c -> c = '='))
+        (char '=' *> take_all) <?> "parameter" in
+    let parse = lift3 create_mimetype
+        (restricted_name <?> "type")
+        ((char '/' *> restricted_name) <?> "subtype")
+        (maybe (char ';' *> skip_while is_space *> parameter))
+        <?> "mimetype"
+    in match Angstrom.parse_string ~consume:All parse s with
+    | Ok x -> x
+    | Error msg -> raise (ParseError msg)
 
   let to_type m = m.mime_type
 
-- 
cgit v1.2.3