Revert "nltk-data: make searchable, add all downloadables" (#409843)

Author: winterqt

nixos/modules/services/web-apps/mealie.nix

@@ -76,7 +76,7 @@ in
         API_PORT = toString cfg.port;
         BASE_URL = "http://localhost:${toString cfg.port}";
         DATA_DIR = "/var/lib/mealie";
-        NLTK_DATA = pkgs.nltk-data.averaged-perceptron-tagger-eng;
+        NLTK_DATA = pkgs.nltk-data.averaged_perceptron_tagger_eng;
       } // (builtins.mapAttrs (_: val: toString val) cfg.settings);
 
       serviceConfig = {

pkgs/by-name/me/mealie/package.nix

@@ -109,7 +109,7 @@ pythonpkgs.buildPythonApplication rec {
 
   # Needed for tests
   preCheck = ''
-    export NLTK_DATA=${nltk-data.averaged-perceptron-tagger-eng}
+    export NLTK_DATA=${nltk-data.averaged_perceptron_tagger_eng}
   '';
 
   disabledTestPaths = [

pkgs/by-name/pa/paperless-ngx/package.nix

@@ -305,8 +305,8 @@ python.pkgs.buildPythonApplication rec {
       tesseract5
       ;
     nltkData = with nltk-data; [
-      punkt-tab
-      snowball-data
+      punkt_tab
+      snowball_data
       stopwords
     ];
     tests = { inherit (nixosTests) paperless; };

pkgs/by-name/un/unstructured-api/package.nix

@@ -152,7 +152,7 @@ let
 
     paths = [
       nltk-data.punkt
-      nltk-data.averaged-perceptron-tagger
+      nltk-data.averaged_perceptron_tagger
     ];
   };
 in

pkgs/development/python-modules/aider-chat/default.nix

@@ -125,7 +125,7 @@ let
   aider-nltk-data = symlinkJoin {
     name = "aider-nltk-data";
     paths = [
-      nltk-data.punkt-tab
+      nltk-data.punkt_tab
       nltk-data.stopwords
     ];
   };

pkgs/development/python-modules/ingredient-parser-nlp/default.nix

@@ -44,7 +44,7 @@ buildPythonPackage rec {
 
   # Needed for tests
   preCheck = ''
-    export NLTK_DATA=${nltk-data.averaged-perceptron-tagger-eng}
+    export NLTK_DATA=${nltk-data.averaged_perceptron_tagger_eng}
   '';
 
   meta = {

pkgs/development/python-modules/type-infer/default.nix

@@ -24,7 +24,7 @@ let
     name = "nltk-test-data";
     paths = [
       nltk-data.punkt
-      nltk-data.punkt-tab
+      nltk-data.punkt_tab
       nltk-data.stopwords
     ];
   };

pkgs/tools/text/nltk-data/default.nix

@@ -10,16 +10,12 @@ let
     version = "0-unstable-2024-07-29";
     nativeBuildInputs = [ unzip ];
     dontBuild = true;
-    dontFixup = true;
     meta = with lib; {
       description = "NLTK Data";
       homepage = "https://github.com/nltk/nltk_data";
       license = licenses.asl20;
       platforms = platforms.all;
-      maintainers = with maintainers; [
-        bengsparks
-        happysalada
-      ];
+      maintainers = with maintainers; [ happysalada ];
     };
   };
   makeNltkDataPackage =
@@ -54,212 +50,41 @@ let
         '';
       }
     );
-
-  makeChunker =
-    pname:
-    makeNltkDataPackage {
-      inherit pname;
-      location = "chunkers";
-      hash = "sha256-kemjqaCM9hlKAdMw8oVJnp62EAC9rMQ50dKg7wlAwEc=";
-    };
-
-  makeCorpus =
-    pname:
-    makeNltkDataPackage {
-      inherit pname;
-      location = "corpora";
-      hash = "sha256-8lMjW5YI8h6dHJ/83HVY2OYGDyKPpgkUAKPISiAKqqk=";
-    };
-
-  makeGrammar =
-    pname:
-    makeNltkDataPackage {
-      inherit pname;
-      location = "grammars";
-      hash = "sha256-pyLEcX3Azv8j1kCGvVYonuiNgVJxtWt7veU0S/yNbIM=";
-    };
-
-  makeHelp =
-    pname:
-    makeNltkDataPackage {
-      inherit pname;
-      location = "help";
-      hash = "sha256-97mYLNES5WujLF5gD8Ul4cJ6LqSzz+jDzclUsdBeHNE=";
-    };
-
-  makeMisc =
-    pname:
-    makeNltkDataPackage {
-      inherit pname;
-      location = "misc";
-      hash = "sha256-XtizfEsc8TYWqvvC/eSFdha2ClC5/ZiJM8nue0vXLb4=";
-    };
-
-  makeModel =
-    pname:
-    makeNltkDataPackage {
-      inherit pname;
-      location = "models";
-      hash = "sha256-iq3weEgCci6rgLW2j28F2eRLprJtInGXKe/awJPSVG4=";
-    };
-
-  makeTagger =
-    pname:
-    makeNltkDataPackage {
-      inherit pname;
-      location = "taggers";
-      hash = "sha256-tl3Cn2okhBkUtTXvAmFRx72Brez6iTGRdmFTwFmpk3M=";
-    };
-
-  makeTokenizer =
-    pname:
-    makeNltkDataPackage {
-      inherit pname;
-      location = "tokenizers";
-      hash = "sha256-OzMkruoYbFKqzuimOXIpE5lhHz8tmSqOFoLT+fjdTVg=";
-    };
-
-  makeStemmer =
-    pname:
-    makeNltkDataPackage {
-      inherit pname;
-      location = "stemmers";
-      hash = "sha256-mNefwOPVJGz9kXV3LV4DuV7FJpNir/Nwg4ujd0CogEk=";
-    };
 in
 lib.makeScope newScope (self: {
-  ## Chunkers
-  maxent-ne-chunker = makeChunker "maxent_ne_chunker";
-  maxent-ne-chunker-tab = makeChunker "maxent_ne_chunker_tab";
-
-  ## Corpora
-  abc = makeCorpus "abc";
-  alpino = makeCorpus "alpino";
-  bcp47 = makeCorpus "bcp47";
-  biocreative-ppi = makeCorpus "biocreative_ppi";
-  brown = makeCorpus "brown";
-  brown-tei = makeCorpus "brown_tei";
-  cess-cat = makeCorpus "cess_cat";
-  cess-esp = makeCorpus "cess_esp";
-  chat80 = makeCorpus "chat80";
-  city-database = makeCorpus "city_database";
-  cmudict = makeCorpus "cmudict";
-  comparative-sentences = makeCorpus "comparative_sentences";
-  comtrans = makeCorpus "comtrans";
-  conll2000 = makeCorpus "conll2000";
-  conll2002 = makeCorpus "conll2002";
-  conll2007 = makeCorpus "conll2007";
-  crubadan = makeCorpus "crubadan";
-  dependency-treebank = makeCorpus "dependency_treebank";
-  dolch = makeCorpus "dolch";
-  europarl-raw = makeCorpus "europarl_raw";
-  extended-omw = makeCorpus "extended_omw";
-  floresta = makeCorpus "floresta";
-  framenet-v15 = makeCorpus "framenet_v15";
-  framenet-v17 = makeCorpus "framenet_v17";
-  gazetteers = makeCorpus "gazetteers";
-  genesis = makeCorpus "genesis";
-  gutenberg = makeCorpus "gutenberg";
-  ieer = makeCorpus "ieer";
-  inaugural = makeCorpus "inaugural";
-  indian = makeCorpus "indian";
-  jeita = makeCorpus "jeita";
-  kimmo = makeCorpus "kimmo";
-  knbc = makeCorpus "knbc";
-  lin-thesaurus = makeCorpus "lin_thesaurus";
-  mac-morpho = makeCorpus "mac_morpho";
-  machado = makeCorpus "machado";
-  masc-tagged = makeCorpus "masc_tagged";
-  movie-reviews = makeCorpus "movie_reviews";
-  mte-teip5 = makeCorpus "mte_teip5";
-  names = makeCorpus "names";
-  nombank-1-0 = makeCorpus "nombank.1.0";
-  nonbreaking-prefixes = makeCorpus "nonbreaking_prefixes";
-  nps-chat = makeCorpus "nps_chat";
-  omw = makeCorpus "omw";
-  omw-1-4 = makeCorpus "omw-1.4";
-  opinion-lexicon = makeCorpus "opinion_lexicon";
-  panlex-swadesh = makeCorpus "panlex_swadesh";
-  paradigms = makeCorpus "paradigms";
-  pe08 = makeCorpus "pe08";
-  pil = makeCorpus "pil";
-  pl196x = makeCorpus "pl196x";
-  ppattach = makeCorpus "ppattach";
-  problem-reports = makeCorpus "problem_reports";
-  product-reviews-1 = makeCorpus "product_reviews_1";
-  product-reviews-2 = makeCorpus "product_reviews_2";
-  propbank = makeCorpus "propbank";
-  pros-cons = makeCorpus "pros_cons";
-  ptb = makeCorpus "ptb";
-  qc = makeCorpus "qc";
-  reuters = makeCorpus "reuters";
-  rte = makeCorpus "rte";
-  semcor = makeCorpus "semcor";
-  senseval = makeCorpus "senseval";
-  sentence-polarity = makeCorpus "sentence_polarity";
-  sentiwordnet = makeCorpus "sentiwordnet";
-  shakespeare = makeCorpus "shakespeare";
-  sinica-treebank = makeCorpus "sinica_treebank";
-  smultron = makeCorpus "smultron";
-  state-union = makeCorpus "state_union";
-  stopwords = makeCorpus "stopwords";
-  subjectivity = makeCorpus "subjectivity";
-  swadesh = makeCorpus "swadesh";
-  switchboard = makeCorpus "switchboard";
-  timit = makeCorpus "timit";
-  toolbox = makeCorpus "toolbox";
-  treebank = makeCorpus "treebank";
-  twitter-samples = makeCorpus "twitter_samples";
-  udhr = makeCorpus "udhr";
-  udhr2 = makeCorpus "udhr2";
-  unicode-samples = makeCorpus "unicode_samples";
-  universal-treebanks-v20 = makeCorpus "universal_treebanks_v20";
-  verbnet = makeCorpus "verbnet";
-  verbnet3 = makeCorpus "verbnet3";
-  webtext = makeCorpus "webtext";
-  wordnet = makeCorpus "wordnet";
-  wordnet-ic = makeCorpus "wordnet_ic";
-  wordnet2021 = makeCorpus "wordnet2021";
-  wordnet2022 = makeCorpus "wordnet2022";
-  wordnet31 = makeCorpus "wordnet31";
-  words = makeCorpus "words";
-  ycoe = makeCorpus "ycoe";
-
-  ## Grammars
-  basque-grammars = makeGrammar "basque_grammars";
-  book-grammars = makeGrammar "book_grammars";
-  large-grammars = makeGrammar "large_grammars";
-  sample-grammars = makeGrammar "sample_grammars";
-  spanish-grammars = makeGrammar "spanish_grammars";
-
-  ## Help
-  tagsets-json = makeHelp "tagsets_json";
-
-  ## Misc
-  mwa-ppdb = makeMisc "mwa_ppdb";
-  perluniprops = makeMisc "perluniprops";
-
-  ## Models
-  bllip-wsj-no-aux = makeModel "bllip_wsj_no_aux";
-  moses-sample = makeModel "moses_sample";
-  wmt15-eval = makeModel "wmt15_eval";
-  word2vec-sample = makeModel "word2vec_sample";
-
-  ## Taggers
-  averaged-perceptron-tagger = makeTagger "averaged_perceptron_tagger";
-  averaged-perceptron-tagger-eng = makeTagger "averaged_perceptron_tagger_eng";
-  averaged-perceptron-tagger-ru = makeTagger "averaged_perceptron_tagger_ru";
-  averaged-perceptron-tagger-rus = makeTagger "averaged_perceptron_tagger_rus";
-  maxent-treebank-pos-tagger = makeTagger "maxent_treebank_pos_tagger";
-  maxent-treebank-pos-tagger-tab = makeTagger "maxent_treebank_pos_tagger_tab";
-  universal-tagset = makeTagger "universal_tagset";
-
-  ## Tokenizers
-  punkt = makeTokenizer "punkt";
-  punkt-tab = makeTokenizer "punkt_tab";
-
-  ## Stemmers
-  porter-test = makeStemmer "porter_test";
-  rslp = makeStemmer "rslp";
-  snowball-data = makeStemmer "snowball_data";
+  punkt = makeNltkDataPackage {
+    pname = "punkt";
+    location = "tokenizers";
+    hash = "sha256-OzMkruoYbFKqzuimOXIpE5lhHz8tmSqOFoLT+fjdTVg=";
+  };
+  punkt_tab = makeNltkDataPackage {
+    pname = "punkt_tab";
+    location = "tokenizers";
+    hash = "sha256-OzMkruoYbFKqzuimOXIpE5lhHz8tmSqOFoLT+fjdTVg=";
+  };
+  averaged_perceptron_tagger = makeNltkDataPackage {
+    pname = "averaged_perceptron_tagger";
+    location = "taggers";
+    hash = "sha256-tl3Cn2okhBkUtTXvAmFRx72Brez6iTGRdmFTwFmpk3M=";
+  };
+  averaged_perceptron_tagger_eng = makeNltkDataPackage {
+    pname = "averaged_perceptron_tagger_eng";
+    location = "taggers";
+    hash = "sha256-tl3Cn2okhBkUtTXvAmFRx72Brez6iTGRdmFTwFmpk3M=";
+  };
+  snowball_data = makeNltkDataPackage {
+    pname = "snowball_data";
+    location = "stemmers";
+    hash = "sha256-mNefwOPVJGz9kXV3LV4DuV7FJpNir/Nwg4ujd0CogEk=";
+  };
+  stopwords = makeNltkDataPackage {
+    pname = "stopwords";
+    location = "corpora";
+    hash = "sha256-8lMjW5YI8h6dHJ/83HVY2OYGDyKPpgkUAKPISiAKqqk=";
+  };
+  wordnet = makeNltkDataPackage {
+    pname = "wordnet";
+    location = "corpora";
+    hash = "sha256-8lMjW5YI8h6dHJ/83HVY2OYGDyKPpgkUAKPISiAKqqk=";
+  };
 })

pkgs/top-level/aliases.nix

@@ -1354,10 +1354,6 @@ mapAliases {
   # When the nixops_unstable alias is removed, nixops_unstable_minimal can be renamed to nixops_unstable.
 
   nixosTest = testers.nixosTest; # Added 2022-05-05
-  nltk-data.averaged_perceptron_tagger = nltk-data.averaged-perceptron-tagger; # Added 2025-05-21
-  nltk-data.averaged_perceptron_tagger_eng = nltk-data.averaged-perceptron-tagger-eng; # Added 2025-05-21
-  nltk-data.punkt_tab = nltk-data.punkt-tab; # Added 2025-05-21
-  nltk-data.snowball_data = nltk-data.snowball-data; # Added 2025-05-21
   nmap-unfree = throw "'nmap-unfree' has been renamed to/replaced by 'nmap'"; # Converted to throw 2024-10-17
   noah = throw "'noah' has been removed because it was broken and its upstream archived"; # Added 2025-05-10
   nodejs_18 = throw "Node.js 18.x has reached End-Of-Life and has been removed"; # Added 2025-04-23

pkgs/top-level/all-packages.nix

@@ -2344,7 +2344,7 @@ with pkgs;
 
   mpd-sima = python3Packages.callPackage ../tools/audio/mpd-sima { };
 
-  nltk-data = lib.recurseIntoAttrs (callPackage ../tools/text/nltk-data { });
+  nltk-data = callPackage ../tools/text/nltk-data { };
 
   seabios-coreboot = seabios.override { ___build-type = "coreboot"; };
   seabios-csm = seabios.override { ___build-type = "csm"; };