Handle HTML escape characters.

Convert “&…” codes to corresponding characters.

ocaml/simpleHttp.ml

@@ -1,7 +1,8 @@
 open Printf
 open Http_client
 open Https_client
-open Nethtml;;
+open Nethtml
+open Netencoding;;
 
 Ssl.init();
 Convenience.configure_pipeline
@@ -19,7 +20,7 @@ let extract_string_value document =
 
 
 let find_string_value = function
-  |  Some s -> true
+  | Some s -> true
   | None -> false
 
 
@@ -57,11 +58,18 @@ let get_http_document body_str =
   let ch = new Netchannels.input_string body_str in
   Nethtml.parse ch
 
+let decode_esc_char str =
+  Html.decode ~in_enc:`Enc_utf8
+              ~out_enc:`Enc_utf8
+              ~entity_base:`Html () str
 
 let get_http_title body =
   let ch = new Netchannels.input_string body in
   let doc = Nethtml.parse ch in
-  get_title_element_from_list doc
+  let title = get_title_element_from_list doc in
+  match title with
+  | Some s -> Some (decode_esc_char s)
+  | None -> None
 
 
 (* TODO: Log errors *)

ocaml/test.ml

@@ -15,6 +15,7 @@ let test_get_title test_ctxt =
   let title = SimpleHttp.get_http_title body in
   assert_equal (Some "Fabien") title
 
+
 let test_remove_spaces test_ctxt =
   let str = "
         
@@ -26,11 +27,18 @@ let test_remove_spaces test_ctxt =
   assert_equal (Some "foo bar  baz") cleaned_str
 
 
+let test_escape_characters text_ctx = 
+  let html_str = "I'll" in
+  let expected = "I'll" in
+  assert_equal expected (SimpleHttp.decode_esc_char html_str)
+
+
 (* Name the test cases and group them together *)
 let suite =
 "suite">:::
  ["test get_title">:: test_get_title;
-  "test remove_spaces">:: test_remove_spaces]
+  "test remove_spaces">:: test_remove_spaces;
+  "test HTML character escaping">:: test_escape_characters]
 
 let () =
   run_test_tt_main suite