open-webui: Add missing datasets dependency for local "transformers" text-to-speech (TTS) (#509699)

pkgs/by-name/op/open-webui/package.nix

@@ -110,6 +110,7 @@ python3Packages.buildPythonApplication (finalAttrs: {
       chardet
       chromadb
       cryptography
+      datasets_3
       ddgs
       docx2txt
       einops

pkgs/development/python-modules/datasets/3.nix

@@ -0,0 +1,85 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  # build-system
+  setuptools,
+
+  # dependencies
+  dill,
+  filelock,
+  fsspec,
+  httpx,
+  huggingface-hub,
+  multiprocess,
+  numpy,
+  pandas,
+  pyarrow,
+  pyyaml,
+  requests,
+  tqdm,
+  xxhash,
+}:
+buildPythonPackage (finalAttrs: {
+  pname = "datasets";
+  version = "3.6.0";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "huggingface";
+    repo = "datasets";
+    tag = finalAttrs.version;
+    hash = "sha256-/xhu0cDKfCEwrp9IzKd0+AeQky1198f9sba/pdutvAk=";
+  };
+
+  build-system = [
+    setuptools
+  ];
+
+  dependencies = [
+    dill
+    filelock
+    fsspec
+    httpx
+    huggingface-hub
+    multiprocess
+    numpy
+    pandas
+    pyarrow
+    pyyaml
+    requests
+    tqdm
+    xxhash
+  ]
+  ++ fsspec.optional-dependencies.http;
+
+  pythonRelaxDeps = [
+    # https://github.com/huggingface/datasets/blob/a256b85cbc67aa3f0e75d32d6586afc507cf535b/setup.py#L117
+    # "pin until dill has official support for determinism"
+    "dill"
+    # https://github.com/huggingface/datasets/blob/4.5.0/setup.py#L127
+    "multiprocess"
+    # https://github.com/huggingface/datasets/blob/4.5.0/setup.py#L130
+    "fsspec"
+  ];
+
+  # Tests require pervasive internet access
+  doCheck = false;
+
+  # Module import will attempt to create a cache directory
+  postFixup = "export HF_MODULES_CACHE=$TMPDIR";
+
+  pythonImportsCheck = [ "datasets" ];
+
+  meta = {
+    description = "Open-access datasets and evaluation metrics for natural language processing";
+    mainProgram = "datasets-cli";
+    homepage = "https://github.com/huggingface/datasets";
+    changelog = "https://github.com/huggingface/datasets/releases/tag/${finalAttrs.src.tag}";
+    license = lib.licenses.asl20;
+    maintainers = with lib.maintainers; [
+      osbm
+      malteneuss
+    ];
+  };
+})

pkgs/development/python-modules/datasets/default.nix

@@ -20,7 +20,7 @@
   tqdm,
   xxhash,
 }:
-buildPythonPackage rec {
+buildPythonPackage (finalAttrs: {
   pname = "datasets";
   version = "4.5.0";
   pyproject = true;
@@ -28,7 +28,7 @@ buildPythonPackage rec {
   src = fetchFromGitHub {
     owner = "huggingface";
     repo = "datasets";
-    tag = version;
+    tag = finalAttrs.version;
     hash = "sha256-K8JqIbYz3ZfT1t1h5dRGCo9kBQp0E+kElqzaw2InaOI=";
   };
 
@@ -75,8 +75,8 @@ buildPythonPackage rec {
     description = "Open-access datasets and evaluation metrics for natural language processing";
     mainProgram = "datasets-cli";
     homepage = "https://github.com/huggingface/datasets";
-    changelog = "https://github.com/huggingface/datasets/releases/tag/${src.tag}";
+    changelog = "https://github.com/huggingface/datasets/releases/tag/${finalAttrs.src.tag}";
     license = lib.licenses.asl20;
     maintainers = with lib.maintainers; [ osbm ];
   };
-}
+})

pkgs/top-level/python-packages.nix

@@ -3634,6 +3634,8 @@ self: super: with self; {
 
   datasets = callPackage ../development/python-modules/datasets { };
 
+  datasets_3 = callPackage ../development/python-modules/datasets/3.nix { };
+
   datasette = callPackage ../development/python-modules/datasette { };
 
   datasette-publish-fly = callPackage ../development/python-modules/datasette-publish-fly { };