Merge pull request 'replace omd with cmarkit' (#167) from cmarkit into main

Reviewed-on: #167

builder-web.opam

@@ -45,7 +45,7 @@ depends: [
   "cmdliner" {>= "1.1.0"}
   "uri"
   "fmt" {>= "0.8.7"}
-  "omd" {>= "2.0.0~alpha3"}
+  "cmarkit"
   "tar"
   "owee"
   "solo5-elftool" {>= "0.3.0"}

lib/dune

@@ -11,7 +11,7 @@
   caqti-lwt
   opamdiff
   ptime.clock.os
-  omd
+  cmarkit
   tar
   owee
   solo5-elftool

lib/utils.ml

@@ -45,109 +45,29 @@ let compare_pkgs p1 p2 =
   in
   diff_map (parse_pkgs p1) (parse_pkgs p2)
 
-module Omd = struct
+let md_to_html ?adjust_heading ?(safe = true) data =
-
+  let open Cmarkit in
-  let make_safe omd =
+  let doc = Doc.of_string ~heading_auto_ids:true data in
-    let rec safe_block = function
+  let doc =
-      | Omd.Paragraph (attr, inline) ->
+    Option.fold ~none:doc
-        safe_inline inline
+      ~some:(fun lvl ->
-        |> Option.map (fun inline -> Omd.Paragraph (attr, inline))
+          let block _m = function
-      | Omd.List (attr, typ, spacing, blocks) ->
+            | Block.Heading (h, meta) ->
-        let blocks = List.filter_map (fun b ->
+              let open Block.Heading in
-          let b = List.filter_map safe_block b in
+              let level = level h
-          if b = [] then None else Some b)
+              and id = id h
-          blocks
+              and layout = layout h
+              and inline = inline h
               in
-        if blocks = [] then None else
+              let h' = make ?id ~layout ~level:(level + lvl) inline in
-          Some (Omd.List (attr, typ, spacing, blocks))
+              Mapper.ret (Block.Heading (h', meta))
-      | Omd.Blockquote (attr, blocks) ->
+            | _ -> Mapper.default
-        let blocks = List.filter_map safe_block blocks in
-        if blocks = [] then None else
-          Some (Omd.Blockquote (attr, blocks))
-      | Omd.Heading (attr, level, inline) ->
-        safe_inline inline
-        |> Option.map (fun inline -> Omd.Heading (attr, level, inline))
-      | Omd.Html_block _ -> None
-      | Omd.Definition_list (attr, def_elts) ->
-        let def_elts = List.filter_map safe_def_elts def_elts in
-        if def_elts = [] then None else
-          Some (Omd.Definition_list (attr, def_elts))
-      | Omd.Code_block _
-      | Omd.Thematic_break _ as v -> Some v
-      | Omd.Table (attr, header_row, rows) ->
-        let header_row =
-          List.fold_left (fun acc (cell, alignment) ->
-              match acc with
-              | None -> None
-              | Some xs ->
-                Option.map (fun cell -> xs @ [ cell, alignment ])
-                  (safe_inline cell))
-            (Some []) header_row
           in
-        Option.map
+          let mapper = Mapper.make ~block () in
-          (fun header_row ->
+          Mapper.map_doc mapper doc)
-             let rows =
+      adjust_heading
-               List.filter_map (fun row ->
-                   List.fold_left (fun acc cell ->
-                       match acc with
-                       | None -> None
-                       | Some xs -> Option.map (fun cell -> xs @ [ cell ])
-                                      (safe_inline cell))
-                     (Some []) row)
-                 rows
   in
-             Omd.Table (attr, header_row, rows))
+  Cmarkit_html.of_doc ~safe doc
-          header_row
-    and safe_def_elts { term ; defs } =
-      let defs = List.filter_map safe_inline defs in
-      safe_inline term
-      |> Option.map (fun term -> { Omd.term ; defs })
-    and safe_inline = function
-      | Concat (attr, inline) ->
-        Some (Concat (attr, List.filter_map safe_inline inline))
-      | Emph (attr, inline) ->
-        safe_inline inline
-        |> Option.map (fun inline -> Omd.Emph (attr, inline))
-      | Strong (attr, inline) ->
-        safe_inline inline
-        |> Option.map (fun inline -> Omd.Strong (attr, inline))
-      | Link (attr, link) ->
-        begin match safe_link link with
-          | `No_label | `Relative -> safe_inline link.Omd.label
-          | `Link l -> Some (Omd.Link (attr, l))
-        end
-      | Image (attr, link) ->
-        begin match safe_link link with
-          | `No_label | `Relative -> None
-          | `Link l -> Some (Omd.Image (attr, l))
-        end
-      | Html _ -> None
-      | Text _
-      | Code _
-      | Hard_break _
-      | Soft_break _ as v -> Some v
-    and safe_link ({ label ; destination ; _ } as l) =
-      let absolute_link =
-        String.(length destination >= 2 && equal (sub destination 0 2) "//") ||
-        String.(length destination >= 7 && equal (sub destination 0 7) "http://") ||
-        String.(length destination >= 8 && equal (sub destination 0 8) "https://")
-      in
-      if absolute_link then
-        match safe_inline label with
-        | None -> `No_label
-        | Some label -> `Link { l with label }
-      else
-        `Relative
-    in
-    List.filter_map safe_block omd
-
-  let html_of_string markdown =
-    markdown
-    |> Omd.of_string
-    |> make_safe
-    |> Omd.to_html
-
-end
 
 module Path = struct
 

lib/views.ml

@@ -285,7 +285,7 @@ have questions or suggestions.
 
   let make_header =
     [
-      H.Unsafe.data (Utils.Omd.html_of_string data);
+      H.Unsafe.data (Utils.md_to_html data);
       H.form ~a:H.[a_action "/hash"; a_method `Get] [
         H.label [
           H.txt "Search artifact by SHA256";
@@ -383,7 +383,7 @@ module Job = struct
         [
           H.h2 ~a:H.[a_id "readme"] [H.txt "README"];
           H.a ~a:H.[a_href "#builds"] [H.txt "Skip to builds"];
-          H.Unsafe.data (Utils.Omd.html_of_string data)
+          H.Unsafe.data (Utils.md_to_html ~adjust_heading:2 data)
         ]
     )
 

test/dune

@@ -6,7 +6,7 @@
 (test
  (name markdown_to_html)
  (modules markdown_to_html)
- (libraries builder_web alcotest))
+ (libraries builder_web cmarkit alcotest))
 
 (test
  (name router)

test/markdown_to_html.ml

@@ -1,14 +1,14 @@
-let markdown_to_html = Builder_web__Utils.Omd.html_of_string
+let markdown_to_html = Builder_web__Utils.md_to_html
 
 let test_simple () =
   let markdown = {|# Hello world|} in
   let html = markdown_to_html markdown in
-  Alcotest.(check string "simple html" "<h1 id=\"hello-world\">Hello world</h1>\n" html)
+  Alcotest.(check string "simple html" "<h1 id=\"hello-world\"><a class=\"anchor\" aria-hidden=\"true\" href=\"#hello-world\"></a>Hello world</h1>\n" html)
 
 let test_html_script () =
   let markdown = {|# <script>Hello world</script>|} in
   let html = markdown_to_html markdown in
-  Alcotest.(check string "html script header" "<h1 id=\"hello-world\">Hello world</h1>\n" html)
+  Alcotest.(check string "html script header" "<h1 id=\"hello-world\"><a class=\"anchor\" aria-hidden=\"true\" href=\"#hello-world\"></a><!-- CommonMark raw HTML omitted -->Hello world<!-- CommonMark raw HTML omitted --></h1>\n" html)
 
 let test_preserve_span_content () =
   let markdown = {|* <span id="myref">My ref</span>
@@ -16,10 +16,8 @@ let test_preserve_span_content () =
   let html = markdown_to_html markdown in
   Alcotest.(check string "html span content preserved"
               {|<ul>
-<li>My ref
+<li><!-- CommonMark raw HTML omitted -->My ref<!-- CommonMark raw HTML omitted --></li>
-</li>
+<li><a href="#myref">See my ref</a> for more information</li>
-<li>See my ref for more information
-</li>
 </ul>
 |}
               html)
@@ -27,20 +25,21 @@ let test_preserve_span_content () =
 let test_remove_script () =
   let markdown = {|<script>alert(1);</script>|} in
   let html = markdown_to_html markdown in
-  Alcotest.(check string "html script removed" "" html)
+  Alcotest.(check string "html script removed" "<!-- CommonMark HTML block omitted -->\n" html)
 
 let test_list_with_html_block_and_markdown () =
   let markdown = "* <div> Hello, World!</div> *this is not html*" in
   let html = markdown_to_html markdown in
   Alcotest.(check string "list with html block and markdown"
-              (*"<ul>\n<li><em>this is not html</em>\n</li>\n</ul>\n"*) ""
+              (*"<ul>\n<li><em>this is not html</em>\n</li>\n</ul>\n"*)
+              "<ul>\n<li>\n<!-- CommonMark HTML block omitted -->\n</li>\n</ul>\n"
               html)
 
 let test_list_with_inline_html_and_markdown () =
   let markdown = "* <span> Hello, World!</span> *this is not html*" in
   let html = markdown_to_html markdown in
   Alcotest.(check string "list with html block and markdown"
-              "<ul>\n<li> Hello, World! <em>this is not html</em>\n</li>\n</ul>\n"
+              "<ul>\n<li><!-- CommonMark raw HTML omitted --> Hello, World!<!-- CommonMark raw HTML omitted --> <em>this is not html</em></li>\n</ul>\n"
               html)
 
 let test_absolute_link () =
@@ -51,35 +50,59 @@ let test_absolute_link () =
 let test_relative_link () =
   let markdown = "[foo](../foo.jpg)" in
   let html = markdown_to_html markdown in
-  Alcotest.(check string "relative link" "<p>foo</p>\n" html)
+  Alcotest.(check string "relative link" "<p><a href=\"../foo.jpg\">foo</a></p>\n" html)
 
 let test_absolute_image () =
   let markdown = "![alttext](https://foo.com/bar.jpg)" in
   let html = markdown_to_html markdown in
   Alcotest.(check string "absolute image"
-              "<p><img src=\"https://foo.com/bar.jpg\" alt=\"alttext\" /></p>\n" html)
+              "<p><img src=\"https://foo.com/bar.jpg\" alt=\"alttext\" ></p>\n" html)
 
 let test_absolute_image_no_alt () =
   let markdown = "![](https://foo.com/bar.jpg)" in
   let html = markdown_to_html markdown in
   Alcotest.(check string "absolute image"
-              "<p><img src=\"https://foo.com/bar.jpg\" alt=\"\" /></p>\n" html)
+              "<p><img src=\"https://foo.com/bar.jpg\" alt=\"\" ></p>\n" html)
 
 let test_relative_image () =
   let markdown = "![](/bar.jpg)" in
   let html = markdown_to_html markdown in
-  Alcotest.(check string "relative image" "" html)
+  Alcotest.(check string "relative image" "<p><img src=\"/bar.jpg\" alt=\"\" ></p>\n" html)
 
 let test_absolute_image_script_alt () =
   let markdown = "![<script src=\"bla.js\"></script>](https://foo.com/bar.jpg)" in
   let html = markdown_to_html markdown in
   Alcotest.(check string "absolute image with script alt text"
-              "<p><img src=\"https://foo.com/bar.jpg\" alt=\"\" /></p>\n" html)
+              "<p><img src=\"https://foo.com/bar.jpg\" alt=\"\" ></p>\n" html)
 
 let test_fragment_link () =
   let markdown = "[fragment](#fragment)" in
   let html = markdown_to_html markdown in
-  Alcotest.(check string "fragment link" "<p>fragment</p>\n" html)
+  Alcotest.(check string "fragment link" "<p><a href=\"#fragment\">fragment</a></p>\n" html)
+
+let test_heading_adjustment () =
+  let markdown = {|# foo
+## bar
+# baz
+## bazbar
+### bazbarbar
+#### bazbarbarbar
+##### bazbarbarbarbar
+###### bazbarbarbarbarbar
+|}
+  in
+  let html = markdown_to_html ~adjust_heading:2 markdown in
+  (* NB: the maximum heading is 6 in cmarkit, thus we reduce the structure *)
+  let exp = {|<h3 id="foo"><a class="anchor" aria-hidden="true" href="#foo"></a>foo</h3>
+<h4 id="bar"><a class="anchor" aria-hidden="true" href="#bar"></a>bar</h4>
+<h3 id="baz"><a class="anchor" aria-hidden="true" href="#baz"></a>baz</h3>
+<h4 id="bazbar"><a class="anchor" aria-hidden="true" href="#bazbar"></a>bazbar</h4>
+<h5 id="bazbarbar"><a class="anchor" aria-hidden="true" href="#bazbarbar"></a>bazbarbar</h5>
+<h6 id="bazbarbarbar"><a class="anchor" aria-hidden="true" href="#bazbarbarbar"></a>bazbarbarbar</h6>
+<h6 id="bazbarbarbarbar"><a class="anchor" aria-hidden="true" href="#bazbarbarbarbar"></a>bazbarbarbarbar</h6>
+<h6 id="bazbarbarbarbarbar"><a class="anchor" aria-hidden="true" href="#bazbarbarbarbarbar"></a>bazbarbarbarbarbar</h6>
+|} in
+  Alcotest.(check string "header adjustment works fine" exp html)
 
 let markdown_tests = [
   Alcotest.test_case "Simple" `Quick test_simple;
@@ -95,6 +118,7 @@ let markdown_tests = [
   Alcotest.test_case "relative image" `Quick test_relative_image;
   Alcotest.test_case "absolute image with script alt" `Quick test_absolute_image_script_alt;
   Alcotest.test_case "fragment link" `Quick test_fragment_link;
+  Alcotest.test_case "heading adjustment" `Quick test_heading_adjustment;
 ]
 
 let () =