mrkdwn: remove old code and promote two tests

[commit_comment.mrkdwn_comment]
- add more MD elements to be tested (headings ordered lists, etc.)
- properly escape characters
- remove extraneous line after blockquotes and nested lists

Author: yasunariw

lib/mrkdwn.ml

@@ -1,8 +1,6 @@
 open Omd
 open Base
 
-let escape_url_chars = Staged.unstage @@ String.Escaping.escape ~escapeworthy:[ '<'; '>'; '|' ] ~escape_char:'\\'
-
 (* https://api.slack.com/reference/surfaces/formatting#escaping *)
 let escape_mrkdwn =
   String.concat_map ~f:(function
@@ -11,58 +9,6 @@ let escape_mrkdwn =
     | '&' -> "&amp;"
     | c -> String.make 1 c)
 
-(** Unescape markdown characters escaped because "any ASCII punctuation
-    character may be backslash-escaped"
-    https://spec.commonmark.org/0.30/#backslash-escapes
-
-    This pertains to '\\', '[', ']', '(', ')', '`', '*' unconditionally,
-    and '.', '-', '+', '!', '<', '>', '#' depending on chars before/after.
-    Argument escapeworthy_map can be left blank because escaped chars are
-    unescaped to themselves. *)
-let unescape_omd = Staged.unstage @@ String.Escaping.unescape_gen_exn ~escapeworthy_map:[] ~escape_char:'\\'
-
-(** Escape the `escape_char` '\\' for use with `unescape_omd` later *)
-let escape_omd = Staged.unstage @@ String.Escaping.escape_gen_exn ~escapeworthy_map:[] ~escape_char:'\\'
-
-let transform_text = escape_mrkdwn
-
-(** `Omd.to_markdown` escapes backslash (and other applicable chars) in
-    `Text` elements but not `Code` elements, so do the same for the latter so
-    that `unescape_omd` can apply uniformly to the whole mrkdwn string later *)
-let transform_code s = escape_omd @@ escape_mrkdwn s
-
-let rec transform_list = List.map ~f:transform
-
-and transform_flatten = List.map ~f:transform_list
-
-and surround s t =
-  let t = to_markdown @@ transform_list t in
-  Raw (Printf.sprintf "%s%s%s" s t s)
-
-(** massage markdown AST so that rendered result looks like slack mrkdwn *)
-and transform = function
-  | H1 t | H2 t | H3 t | H4 t | H5 t | H6 t -> Paragraph (transform_list [ Bold t ])
-  | Paragraph t -> Paragraph (transform_list t)
-  | Emph t -> surround "_" (transform_list t)
-  | Bold t -> surround "*" (transform_list t)
-  | Ul ts -> Ul (transform_flatten ts)
-  | Ol ts -> Ol (transform_flatten ts)
-  | Ulp ts -> Ulp (transform_flatten ts)
-  | Olp ts -> Olp (transform_flatten ts)
-  | Url (href, label, title) ->
-    let label = escape_url_chars @@ to_markdown @@ transform_list label in
-    let title = if String.length title > 0 then Printf.sprintf "%s - " @@ escape_url_chars title else title in
-    Raw (Printf.sprintf "<%s|%s%s>" href title label)
-  | Html _ as e -> Raw (Printf.sprintf "`%s`" @@ to_markdown [ e ])
-  | Html_comment _ -> Br
-  | Html_block _ as e -> Code_block ("", to_markdown [ e ])
-  | Blockquote t -> Blockquote (transform_list t)
-  | Img (alt, src, title) -> transform @@ Url (src, [ Text alt ], title)
-  | Code_block (_, str) -> Code_block ("", transform_code str)
-  | Code (_, str) -> Code ("", transform_code str)
-  | Text s -> Text (transform_text s)
-  | (Br | Hr | NL | Ref _ | Img_ref _ | Raw _ | Raw_block _ | X _) as e -> e
-
 (** Translates omd AST to a Slack mrkdwn string. Code heavily adapted
     from omd 1.3.1 source.
     https://github.com/ocaml/omd/blob/1.3.1/src/omd_backend.ml#L872
@@ -224,8 +170,7 @@ let rec mrkdwn_of_md md =
       loop list_indent s;
       Buffer.add_char b '>';
       loop list_indent tl
-    | H1 md' | H2 md' | H3 md' | H4 md' | H5 md' | H6 md' ->
-      loop list_indent (Paragraph (transform_list [ Bold md' ]) :: tl)
+    | H1 md' | H2 md' | H3 md' | H4 md' | H5 md' | H6 md' -> loop list_indent (Paragraph [ Bold md' ] :: tl)
     | Br ->
       (* the string "\\n" (backslash-newline) or end of line double-space renders Br *)
       nl b;
@@ -254,4 +199,4 @@ let rec mrkdwn_of_md md =
   end;
   Buffer.contents b
 
-let mrkdwn_of_markdown str = unescape_omd @@ to_markdown @@ transform_list @@ of_string str
+let mrkdwn_of_markdown str = mrkdwn_of_md @@ of_string str

mock_payloads/commit_comment.mrkdwn_comment.json

@@ -32,7 +32,7 @@
     "created_at": "2020-06-07T15:11:50Z",
     "updated_at": "2020-06-07T15:11:50Z",
     "author_association": "OWNER",
-    "body": "**bold**, *italic*\n> blockquote\n\n* list-element 1\n* list-element 2\n  * list-element2.1\n  * [list-element2.2](www.google.com)\n\n<div>html block</div>\n```ocaml\nStdio.printf \"hello ocaml\"\n```\nescaped chars \\\\\\(\\)\\[\\]\\`\\*\n`keep \\`\n```\nkeep \\\n```"
+    "body": "# header 1\r\n## header 2\r\n### header 3\r\n#### header 4\r\n##### header 5\r\n###### header 6\r\n\r\n***\r\n\r\nparagraph 1 **bold**, _italic_, ~strikethrough~, **_bold italic_**, _**italic bold**_\r\n\r\nparagraph 2\r\nhard line break\r\n\r\n> blockquote\r\n\r\n> > > multi\r\n> >\r\n> > level\r\n>\r\n> blockquote\r\n\r\n* list-element 1\r\n* list-element 2\r\n  * list-element 2.1\r\n    * list-element 2.1.1\r\n  * [list-element 2.2](www.google.com)\r\n\r\n1. list-element 1\r\n1. list-element 2\r\n    1. list-element 2.1\r\n        1. list-element 2.1.1\r\n    1. [list-element 2.2](www.google.com)\r\n1. list-element 3\r\n\r\n[some link text](www.google.com \"link with title\")\r\n\r\n* > quote in list\r\n\r\n> * list in quote\r\n\r\n![image](https://via.placeholder.com/20x20)\r\n\r\n![image](https://via.placeholder.com/20x20 \"with title\")\r\n\r\n![](https://via.placeholder.com/20x20 \"no alt\")\r\n\r\n![](https://via.placeholder.com/20x20)\r\n\r\n<div>html block with <i>styling</i></div>\r\n\r\n```ocaml\r\nStdio.printf \"hello ocaml\"\r\n```\r\n\r\ninline `code block`\r\n\r\nescape these \\\\\\(\\)\\[\\]\\`\\*  `but keep \\`\r\n\r\n```\r\nand keep \\\r\n```\r\n"
   },
   "repository": {
     "id": 0,

test/slack_payloads.expected

@@ -95,7 +95,7 @@ will notify #all-push-events
       "pretext":
         "<https://github.com/xinyuluo/monorepo|[xinyuluo/monorepo]> *xinyuluo* commented on `<https://github.com/xinyuluo/monorepo/commit/cd5b85afa306840e0790b62e349ee1f828b2a3c2#commitcomment-0|cd5b85af>` add new line at EOF",
       "text":
-        "*bold*, _italic_\n\n> blockquote\n\n\n- list-element 1\n- list-element 2\n    - list-element2.1\n    - <www.google.com|list-element2.2>\n\n\n```\n<div>html block</div>\n```\n```\nStdio.printf \"hello ocaml\"\n```\nescaped chars \\()[]`*\n`keep \\`\n\n```\nkeep \\\n```",
+        "*header 1*\n\n*header 2*\n\n*header 3*\n\n*header 4*\n\n*header 5*\n\n*header 6*\n\n* * *\nparagraph 1 *bold*, _italic_, ~strikethrough~, *_bold italic_*, _*italic bold*_\n\nparagraph 2\nhard line break\n\n> blockquote\n\n> > > multi\n> > \n> > level\n> \n> blockquote\n\n- list-element 1\n- list-element 2\n    - list-element 2.1\n        - list-element 2.1.1\n    - <www.google.com|list-element 2.2>\n\n1. list-element 1\n2. list-element 2\n    1. list-element 2.1\n        1. list-element 2.1.1\n    2. <www.google.com|list-element 2.2>\n3. list-element 3\n\n<www.google.com|link with title - some link text>\n\n- > quote in list\n\n> - list in quote\n\n<https://via.placeholder.com/20x20|image>\n\n<https://via.placeholder.com/20x20|with title - image>\n\n<https://via.placeholder.com/20x20|no alt - >\n\n<https://via.placeholder.com/20x20|>\n\n```\n&lt;div&gt;html block with &lt;i&gt;styling&lt;/i&gt;&lt;/div&gt;\n```\n```\nStdio.printf \"hello ocaml\"\n```\ninline `code block`\n\nescape these \\()[]`*  `but keep \\`\n\n```\nand keep \\\n```",
       "footer":
         "New comment by xinyuluo in <https://github.com/xinyuluo/monorepo/commit/cd5b85afa306840e0790b62e349ee1f828b2a3c2#commitcomment-0|main.ml>"
     }
@@ -196,7 +196,7 @@ will notify #frontend-bot
       "pretext":
         "<https://github.com/ahrefs/monorepo|[ahrefs/monorepo]> *yasunariw* <https://github.com/ahrefs/monorepo/pull/6168#issuecomment-17123|commented> on #6168 <https://github.com/ahrefs/monorepo/pull/6168|Feature 1234>",
       "text":
-        "> I do not understand this\n\n\nDo you have access to the logs on the website?"
+        "> I do not understand this\n\nDo you have access to the logs on the website?"
     }
   ]
 }