Fix failing tests

(On my machine in any case.)

tests/common.ml

@@ -0,0 +1,4 @@
+let normalize_html s =
+  String.trim s
+  |> Soup.parse
+  |> Soup.pretty_print

tests/dune

@@ -1,8 +1,14 @@
 (executable
  (name extract_tests)
- (libraries str lambdasoup)
+ (libraries str common)
  (modules extract_tests))
 
+;  Code shared between various parts of the testing apartus
+(library
+ (name common)
+ (libraries lambdasoup)
+ (modules common))
+
 ;  Generate and run tests for the core omd package
 (rule
  (with-stdout-to
@@ -16,7 +22,7 @@
 
 (executable
  (name omd)
- (libraries str omd omd_tyxml tyxml lambdasoup)
+ (libraries str omd omd_tyxml tyxml common)
  (modules omd))
 
 ;  Generate the rules for diff-based tests

tests/extract_tests.ml

@@ -2,26 +2,26 @@
 
 let disabled =
   [
-    164;
-    175;
-    184;
-    185;
-    334;
-    353;
-    410;
-    411;
-    414;
-    415;
-    416;
-    428;
-    468;
-    469;
-    486;
-    516;
-    536;
-    570;
-    519;
-    591;
+    (* 164;
+     * 175;
+     * 184;
+     * 185;
+     * 334;
+     * 353;
+     * 410;
+     * 411;
+     * 414;
+     * 415;
+     * 416;
+     * 428;
+     * 468;
+     * 469;
+     * 486;
+     * 516;
+     * 536;
+     * 570;
+     * 519;
+     * 591; *)
   ]
 
 let with_open_in fn f =
@@ -75,7 +75,7 @@ let parse_test_spec filename =
               let rec get_html () =
                 let line = input_line ic in
                 if begins_with line test_delim then
-                  let html = Buffer.contents buf in
+                  let html = Buffer.contents buf |> Common.normalize_html in
                   {filename; example; markdown; html}
                 else begin
                   add_line buf line;
@@ -117,18 +117,13 @@ let write_dune_file test_specs tests =
     "@[<v1>(alias@ (name runtest)@ @[<v1>(deps%t)@])@]@."
     (fun ppf -> List.iter (pp ppf) tests)
 
-let li_begin_re = Str.regexp_string "<li>\n"
-let li_end_re = Str.regexp_string "\n</li>"
-
-let normalize_html s = Soup.(parse s |> pretty_print)
-
 let generate_test_files tests =
   let f {filename; example; markdown; html} =
     let base = Filename.remove_extension filename in
     with_open_out (Printf.sprintf "%s-%03d.md" base example)
       (fun oc -> output_string oc markdown);
     with_open_out (Printf.sprintf "%s-%03d.html" base example)
-      (fun oc -> output_string oc (normalize_html html))
+      (fun oc -> output_string oc html)
   in
   List.iter f tests
 

tests/omd.ml

@@ -1,33 +1,13 @@
-(* let li_begin_re = Str.regexp_string "<li>\n"
- * let li_end_re = Str.regexp_string "\n</li>" *)
-
-(* let normalize_html s =
- *   Str.global_replace li_end_re "</li>"
- *     (Str.global_replace li_begin_re "<li>" s) *)
-
 let with_open_in fn f =
   let ic = open_in fn in
   Fun.protect ~finally:(fun () -> close_in_noerr ic)
     (fun () -> f ic)
 
 (* FIXME: Resolve preferred backend *)
 
-(* FIXME: This is getting rediculous. Probably better, imo, to programmatically
-   format the spec HTML and compare the ASTs of the HTMl instead of doing this
-   string munging *)
-
 let replacements =
   [ Str.regexp_string "<br>", "<br>\n"
-  (* ; Str.regexp_string ">\n\n</a>", ">\n</a>" *)
-  (* ; Str.regexp "\n\n$", "\n" *)
-  (*   Str.regexp "\"/>", "\" />"
-   * ; Str.regexp_string "<br/>", "<br />\n"
-   * ; Str.regexp_string "<ul>", "<ul>\n"
-   * ; Str.regexp_string "</li>", "</li>\n"
-   * ; Str.regexp_string "</p><ul>", "</p>\n<ul>"
-   * ; Str.regexp_string "</ul><p>", "</ul>\n<p>"
-   * ; Str.regexp_string "<ol><", "<ol>\n<"
-   * ; Str.regexp_string "<ul><", "<ul>\n<" *)
+  ; Str.regexp_string "</pre><pre ", "</pre>\n<pre "
   ]
 
 
@@ -39,8 +19,6 @@ let html_of_omd s =
   |> List.map (fun b -> b |> Omd_tyxml.of_block |> tyxml_elt_to_string)
   |> String.concat ""
 
-let normalize_html s = Soup.(parse s |> pretty_print)
-
 let denormalize_html str =
   List.fold_left (fun s (re, rep) -> Str.global_replace re rep s) str replacements
 
@@ -49,6 +27,6 @@ let () =
   ic
   |> Omd.of_channel
   |> html_of_omd
-  |> normalize_html
+  |> Common.normalize_html
   |> denormalize_html
   |> print_string