wip: Add medical/American hunspell dict

homes/desktop.nix

@@ -1,4 +1,10 @@
-{ pkgs, custom-pkgs, ... }: {
+{
+  pkgs,
+  lib,
+  custom-pkgs,
+  ...
+}:
+{
 
   imports = [
     ./../programs/firefox.nix
@@ -31,6 +37,7 @@
       woodpecker-cli
       (texlive.combine { inherit (texlive) scheme-basic latex-bin latexmk; })
       custom-pkgs.ark
+      custom-pkgs.hunspell-en-med
       custom-pkgs.jlfmt
       custom-pkgs.runic
     ];
@@ -48,5 +55,21 @@
       '';
     };
   };
-  services = { syncthing.enable = true; };
+  services = {
+    syncthing.enable = true;
+  };
+
+  # Symlink the merged en_MED Hunspell dictionary into the user dictionary
+  # directory that Qt/Sonnet (and macOS native spellcheck) scan at startup.
+  home.file = lib.mkIf pkgs.stdenv.isDarwin {
+    "Library/Spelling/en_MED.dic".source = "${custom-pkgs.hunspell-en-med}/share/hunspell/en_MED.dic";
+    "Library/Spelling/en_MED.aff".source = "${custom-pkgs.hunspell-en-med}/share/hunspell/en_MED.aff";
+  };
+
+  # On Linux, Qt/Sonnet resolves QStandardPaths::GenericDataLocation + "/hunspell"
+  # which maps to ~/.local/share/hunspell (managed via xdg.dataFile).
+  xdg.dataFile = lib.mkIf pkgs.stdenv.isLinux {
+    "hunspell/en_MED.dic".source = "${custom-pkgs.hunspell-en-med}/share/hunspell/en_MED.dic";
+    "hunspell/en_MED.aff".source = "${custom-pkgs.hunspell-en-med}/share/hunspell/en_MED.aff";
+  };
 }

pkgs/default.nix

@@ -1,6 +1,8 @@
 { pkgs, ... }:
-with pkgs; {
+with pkgs;
+{
   ark = callPackage ./ark.nix { };
+  hunspell-en-med = callPackage ./hunspell-en-med.nix { };
   jlfmt = callPackage ./jlfmt.nix { };
   runic = callPackage ./runic.nix { };
   sc4pac = callPackage ./sc4pac.nix { };

pkgs/hunspell-en-med.nix

@@ -0,0 +1,55 @@
+{ pkgs, lib }:
+
+let
+  # Hunspell dictionary of English medical terms by glutanimate.
+  # Merges OpenMedSpel (R. Robinson) and MTH-Med-Spel-Chek (Rajasekharan N.),
+  # covering ~90,000 terms (anatomy, pharmacology, DSM-IV, ICD-9, etc.).
+  # Note: word list is current to April 2014; GPL-3.0 licensed.
+  # Source: https://github.com/glutanimate/hunspell-en-med-glut
+  medDic = pkgs.fetchurl {
+    url = "https://raw.githubusercontent.com/glutanimate/hunspell-en-med-glut/master/en_med_glut.dic";
+    # Run `nix build` once with this placeholder to get the real hash from the
+    # error message, then replace it here.
+    hash = "sha256-voNMOdaB3QED0Lut3KujWXhr0Sa+sdDAi27ZMoVJDBs=";
+  };
+
+in
+pkgs.stdenv.mkDerivation {
+  pname = "hunspell-en-med";
+  # Date of last upstream commit to the dictionary file
+  version = "2017-01-04";
+
+  dontUnpack = true;
+
+  installPhase = ''
+    runHook preInstall
+    mkdir -p $out/share/hunspell
+
+    # Strip the word-count header from both .dic files, concatenate,
+    # de-duplicate, then prepend the new count to form a valid .dic file.
+    {
+      tail -n +2 "${pkgs.hunspellDicts.en_US}/share/hunspell/en_US.dic"
+      tail -n +2 ${medDic}
+    } | sort -u > words.txt
+
+    count=$(wc -l < words.txt)
+    { printf '%s\n' "$count"; cat words.txt; } > $out/share/hunspell/en_MED.dic
+
+    # Reuse en_US affix rules; the medical word list carries no morphology.
+    cp "${pkgs.hunspellDicts.en_US}/share/hunspell/en_US.aff" $out/share/hunspell/en_MED.aff
+
+    runHook postInstall
+  '';
+
+  meta = {
+    description = "Merged US English + medical terms Hunspell dictionary (en_MED)";
+    longDescription = ''
+      Combines the standard en_US Hunspell dictionary with the glutanimate
+      medical terms word list (hunspell-en-med-glut), giving a single
+      en_MED dictionary that covers everyday English and ~90,000 medical terms.
+    '';
+    homepage = "https://github.com/glutanimate/hunspell-en-med-glut";
+    license = lib.licenses.gpl3;
+    platforms = lib.platforms.all;
+  };
+}