Commit
+39 -51 +/-1 browse
| 1 | diff --git a/lib/note.ml b/lib/note.ml |
| 2 | index edb619f..7c817b8 100644 |
| 3 | --- a/lib/note.ml |
| 4 | +++ b/lib/note.ml |
| 5 | @@ -45,56 +45,6 @@ let get_tags t = |
| 6 | |
| 7 | let get_path t = Slug.get_path t.slug |
| 8 | |
| 9 | - let tokenize t = |
| 10 | - let rec _tokenize markdown = |
| 11 | - List.fold |
| 12 | - ~init:([] : string list) |
| 13 | - ~f:(fun accm entry -> |
| 14 | - match entry with |
| 15 | - | Omd.Text text -> accm @ String.split_on_chars ~on:[ ' ' ] text |
| 16 | - | Omd.H1 header -> accm @ _tokenize header |
| 17 | - | Omd.H2 header -> accm @ _tokenize header |
| 18 | - | Omd.H3 header -> accm @ _tokenize header |
| 19 | - | Omd.H4 header -> accm @ _tokenize header |
| 20 | - | Omd.H5 header -> accm @ _tokenize header |
| 21 | - | Omd.H6 header -> accm @ _tokenize header |
| 22 | - | Omd.Paragraph paragraph -> accm @ _tokenize paragraph |
| 23 | - | Omd.Emph text -> accm @ _tokenize text |
| 24 | - | Omd.Bold text -> accm @ _tokenize text |
| 25 | - | Omd.Ul markdown_list -> |
| 26 | - let inner = |
| 27 | - List.fold ~init:[] |
| 28 | - ~f:(fun accm entry -> accm @ _tokenize entry) |
| 29 | - markdown_list |
| 30 | - in |
| 31 | - accm @ inner |
| 32 | - | Omd.Ol markdown_list -> |
| 33 | - let inner = |
| 34 | - List.fold ~init:[] |
| 35 | - ~f:(fun accm entry -> accm @ _tokenize entry) |
| 36 | - markdown_list |
| 37 | - in |
| 38 | - accm @ inner |
| 39 | - | Omd.Ulp markdown_list -> |
| 40 | - let inner = |
| 41 | - List.fold ~init:[] |
| 42 | - ~f:(fun accm entry -> accm @ _tokenize entry) |
| 43 | - markdown_list |
| 44 | - in |
| 45 | - accm @ inner |
| 46 | - | Omd.Olp markdown_list -> |
| 47 | - let inner = |
| 48 | - List.fold ~init:[] |
| 49 | - ~f:(fun accm entry -> accm @ _tokenize entry) |
| 50 | - markdown_list |
| 51 | - in |
| 52 | - accm @ inner |
| 53 | - | _ -> accm) |
| 54 | - markdown |
| 55 | - in |
| 56 | - |
| 57 | - _tokenize t.markdown |
| 58 | - |
| 59 | let get_data t = |
| 60 | let data = |
| 61 | List.filter_map |
| 62 | @@ -159,6 +109,43 @@ let of_string ~data slug = |
| 63 | let markdown = Omd.of_string data in |
| 64 | { frontmatter; markdown; slug } |
| 65 | |
| 66 | + module Util = struct |
| 67 | + let split_words str = |
| 68 | + List.filter_map |
| 69 | + ~f:(fun x -> |
| 70 | + match String.strip ~drop:(fun x -> Char.equal x ' ') x with |
| 71 | + | "" -> None |
| 72 | + | _ -> Some x) |
| 73 | + (String.split ~on:' ' str) |
| 74 | + |
| 75 | + let rec to_words markdown = |
| 76 | + match markdown with |
| 77 | + | [] -> [] |
| 78 | + | hd :: tl -> |
| 79 | + ( match hd with |
| 80 | + | Omd.Text s -> split_words s |
| 81 | + | Omd.H1 v |
| 82 | + | Omd.H2 v |
| 83 | + | Omd.H3 v |
| 84 | + | Omd.H4 v |
| 85 | + | Omd.H5 v |
| 86 | + | Omd.H6 v |
| 87 | + | Omd.Blockquote v |
| 88 | + | Omd.Bold v |
| 89 | + | Omd.Emph v |
| 90 | + | Omd.Paragraph v -> |
| 91 | + to_words v |
| 92 | + | Omd.Url (_, inner, title) -> split_words title @ to_words inner |
| 93 | + | Omd.Ref (_, _, title, _) -> split_words title |
| 94 | + | Omd.Ol l | Omd.Olp l | Omd.Ul l | Omd.Ulp l -> |
| 95 | + List.fold |
| 96 | + ~init:([] : string list) |
| 97 | + ~f:(fun accm elem -> accm @ to_words elem) |
| 98 | + l |
| 99 | + | _ -> [] ) |
| 100 | + @ to_words tl |
| 101 | + end |
| 102 | + |
| 103 | module Encoding = struct |
| 104 | let to_string ~style t = |
| 105 | match style with |
| 106 | @@ -227,7 +214,8 @@ module Display = struct |
| 107 | let title = (get_title note, [ Reset ]) in |
| 108 | let tags = (String.concat ~sep:"|" (get_tags note), [ Reset ]) in |
| 109 | let word_count = |
| 110 | - (Core.sprintf "%d" (List.length (tokenize note)), [ Reset ]) |
| 111 | + ( Core.sprintf "%d" (List.length (Util.to_words note.markdown)), |
| 112 | + [ Reset ] ) |
| 113 | in |
| 114 | let slug = (Slug.to_string note.slug, [ Reset ]) in |
| 115 | accm @ [ [ title; tags; word_count; slug ] ]) |