seppo/test/t_tag.ml

open Seppo_lib

let test_diff () =
  (match Tag.diff String.compare [] [] with
   | [], [], [] -> assert true
   | _ -> assert false);
  (let s, p, m = Tag.diff String.compare [ "b"; "d"; "f" ] [ "a"; "d"; "e" ] in
   s |> String.concat " " |> Assrt.equals_string __LOC__ "d";
   p |> String.concat " " |> Assrt.equals_string __LOC__ "a e";
   m |> String.concat " " |> Assrt.equals_string __LOC__ "b f");
  assert true

let assert_cat msg exp res =
  Assrt.equals_int msg (List.length exp) (List.length res);
  Assrt.equals_string msg
    (exp |> String.concat " ")
    (res |> List.map (fun (Tag.Tag t) -> t) |> String.concat " ")

let test_map () =
  (* https://v2.ocaml.org/api/Map.Make.html *)
  let m = Tag.(Tmap.(empty |> add_tag_list [ Tag "#Ä"; Tag "#à" ])) in
  m |> Tag.Tmap.cardinal |> Assrt.equals_int __LOC__ 1;
  let (Tag t) = m |> Tag.Tmap.find_opt "#a" |> Option.get in
  t |> Assrt.equals_string __LOC__ "#Ä";
  assert true

(* Example from Uuseg
   let test_segment_word () =
     let ret = Tag.utf_8_segments `Word "Uhu" in
     ret |> List.length |> Assrt.equals_int __LOC__ 1;
     let ret = Tag.utf_8_segments `Word "Hello World!" in
     ret |> List.length |> Assrt.equals_int __LOC__ 4;
     let ret = Tag.utf_8_segments `Word "Hello, #World!" in
     ret |> List.length |> Assrt.equals_int __LOC__ 6;
     let ret = Tag.utf_8_segments `Word "Hello, #🌍World!" in
     ret |> List.length |> Assrt.equals_int __LOC__ 7;
     let ret = Tag.utf_8_segments `Word "Hello, #🌍🎉World!" in
     ret |> List.length |> Assrt.equals_int __LOC__ 8;
     assert true
*)

let test_of_string () =
  Tag.of_string "Hello, #World!" |> assert_cat "of_string 1" [ "#World" ];
  Tag.of_string "abc#def" |> assert_cat "of_string 1.1" [];
  Tag.of_string "abc #def" |> assert_cat "of_string 1.2" [ "#def" ];
  Tag.of_string "abc.#def" |> assert_cat "of_string 1.3" [ "#def" ];
  Tag.of_string "ab #@uu @cd@ef #gh ij"
  |> assert_cat "of_string 1.4" [ "#@uu"; "#gh" ];
  Tag.of_string "H #Uhu un #🐫 d #So ja"
  |> assert_cat "of_string 2" [ "#Uhu"; "#So" ];
  Tag.of_string "H #Uhu un 🐫 d #So ja"
  |> assert_cat "of_string 3" [ "#Uhu"; "#So" ];
  (* TODO the flag: *)
  Tag.of_string
    "#† #RIP #AD2021 📻 🇦🇹 Hannes Leopoldseder ist tot - ooe.ORF.at\n\n\
     via https://twitter.com/wasbuxton/status/1361797119871508485\n\
     via https://twitter.com/jnd1er\n\
     auch https://ooe.orf.at/stories/3090120/"
  |> assert_cat "of_string 4" [ "#†"; "#RIP"; "#AD2021" ];
  Tag.of_string
    "#† #RIP #AD2021 Dr. Brad J. Cox Ph.D. Obituary - Manassas, VA | SCNow\n\n\
     \"… Dr. Brad J. Cox Ph. DDr. Brad Cox, Ph. D of Manassas, Virginia, died \
     on January\n\
     2, 2021 at his residence. Dr. Cox was a computer scientist known mostly for\n\
     creating the Objective – C programming language …\"\n\n\
     via https://news.ycombinator.com/item?id=25876767 #objc"
  |> assert_cat "of_string 5" [ "#†"; "#RIP"; "#AD2021"; "#objc" ];
  assert true

let test_zero_width_space () =
  Tag.of_string "Hello, #World​s!" |> assert_cat "zero_width_space 1" [ "#World" ];
  assert true

let test_sift_channel () =
  let ic = open_in "t_tag.1.txt" in
  let ret = Tag.sift_channel ic in
  close_in ic;
  (match ret with
   | Ok v -> v |> assert_cat "comp" [ "#World" ]
   | Error v -> v |> Assrt.equals_string __LOC__ "#Hello");
  assert true

(* https://codeberg.org/mro/ShaarliGo/src/branch/master/t_tags.go#L57 *)
let test_fold () =
  let tst msg exp src =
    Tag.Tag src |> Tag.fold |> Assrt.equals_string msg exp
  in
  tst "fold 1" "hallo wyrld!" "Hälló wÿrld!";
  tst "fold 2" "demaiziere" "DeMaizière";
  tst "fold 3" "cegłowski!" "Cegłowski!";
  tst "fold 3" "iieeean" "íìéèêäñ";
  assert true

let test_normalize () =
  let sh, lo, ts =
    Tag.normalise "#Uhu" "Aha\n#more" [ Tag "#Uhu"; Tag "#less" ] Tag.Tmap.empty
  in
  sh |> Assrt.equals_string __LOC__ "#Uhu";
  lo |> Assrt.equals_string __LOC__ "Aha\n#more\n#less";
  ts |> assert_cat "comp" [ "#Uhu"; "#less"; "#more" ];
  let sh, lo, ts =
    Tag.normalise "#Uh/u" "Aha\n#mo.re" [ Tag "#Uhu"; Tag "#less" ] Tag.Tmap.empty
  in
  sh |> Assrt.equals_string __LOC__ "#Uh/u";
  lo |> Assrt.equals_string __LOC__ "Aha\n#mo.re\n#Uhu #less";
  ts |> assert_cat "comp" [ "#Uh"; "#Uhu"; "#less"; "#mo" ];
  assert true

let test_cdb () =
  let fn = "tmp/tag.cdb" in
  Unix.(try unlink fn with Unix_error (ENOENT, "unlink", _) -> ());
  File.touch fn;
  let db = Mapcdb.Cdb "tmp/tag.cdb" in 
  let db = Mapcdb.add_string "#a" "#Ä" db in
  let (Mapcdb.Cdb db') = db in
  db' |> Assrt.equals_string __LOC__ fn;
  let (s,l,ts) = Tag.cdb_normalise "uhu #á" "aha #B" [] db in
  s |> Assrt.equals_string __LOC__ "uhu #á";
  l |> Assrt.equals_string __LOC__ "aha #B";
  ts |> List.length |> Assrt.equals_int __LOC__ 2;
  Mapcdb.find_string_opt "#a" db |> Option.get |> Assrt.equals_string __LOC__ "#Ä";
  Mapcdb.find_string_opt "#b" db |> Option.get |> Assrt.equals_string __LOC__ "#B";
  assert true

let () =
  Unix.chdir "../../../test/";
  (* test_segment_word (); *)
  test_map ();
  test_diff ();
  test_of_string ();
  test_zero_width_space ();
  test_sift_channel ();
  test_fold ();
  test_normalize ();
  test_cdb ();
  assert true