From c079b8399757d65b157713a63b2b12820298f26b Mon Sep 17 00:00:00 2001
From: Ben Darwin <bcdarwin@gmail.com>
Date: Thu, 7 Nov 2024 14:29:23 -0500
Subject: [PATCH] python312Packages.pandera: init at 0.20.4

---
 .../python-modules/pandera/default.nix        | 126 ++++++++++++++++++
 pkgs/top-level/python-packages.nix            |   2 +
 2 files changed, 128 insertions(+)
 create mode 100644 pkgs/development/python-modules/pandera/default.nix

diff --git a/pkgs/development/python-modules/pandera/default.nix b/pkgs/development/python-modules/pandera/default.nix
new file mode 100644
index 0000000000000..bccacdc23e3a6
--- /dev/null
+++ b/pkgs/development/python-modules/pandera/default.nix
@@ -0,0 +1,126 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  setuptools,
+  multimethod,
+  numpy,
+  packaging,
+  pandas,
+  pydantic,
+  typeguard,
+  typing-inspect,
+  wrapt,
+  # test
+  joblib,
+  pyarrow,
+  pytestCheckHook,
+  pytest-asyncio,
+  # optional dependencies
+  black,
+  dask,
+  fastapi,
+  geopandas,
+  hypothesis,
+  pandas-stubs,
+  polars,
+  pyyaml,
+  scipy,
+  shapely,
+}:
+
+buildPythonPackage rec {
+  pname = "pandera";
+  version = "0.20.4";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "unionai-oss";
+    repo = "pandera";
+    rev = "refs/tags/v${version}";
+    hash = "sha256-VetLfZlBWok7Mr1jxlHHjDH/D5xEsPFWQtX/hrvobgQ=";
+  };
+
+  build-system = [ setuptools ];
+
+  dependencies = [
+    multimethod
+    numpy
+    packaging
+    pandas
+    pydantic
+    typeguard
+    typing-inspect
+    wrapt
+  ];
+
+  optional-dependencies =
+    let
+      dask-dataframe = [ dask ] ++ dask.optional-dependencies.dataframe;
+      extras = {
+        strategies = [ hypothesis ];
+        hypotheses = [ scipy ];
+        io = [
+          pyyaml
+          black
+          #frictionless # not in nixpkgs
+        ];
+        # pyspark expression does not define optional-dependencies.connect:
+        #pyspark = [ pyspark ] ++ pyspark.optional-dependencies.connect;
+        # modin not in nixpkgs:
+        #modin = [
+        #  modin
+        #  ray
+        #  dask-dataframe
+        #];
+        #modin-ray = [
+        #  modin
+        #  ray
+        #];
+        #modin-dask = [
+        #  modin
+        #  dask-dataframe
+        #];
+        dask = [ dask-dataframe ];
+        mypy = [ pandas-stubs ];
+        fastapi = [ fastapi ];
+        geopandas = [
+          geopandas
+          shapely
+        ];
+        polars = [ polars ];
+      };
+    in
+    extras // { all = lib.concatLists (lib.attrValues extras); };
+
+  nativeCheckInputs = [
+    pytestCheckHook
+    pytest-asyncio
+    joblib
+    pyarrow
+  ] ++ optional-dependencies.all;
+
+  disabledTestPaths = [
+    "tests/fastapi/test_app.py" # tries to access network
+    "tests/core/test_docs_setting_column_widths.py" # tests doc generation, requires sphinx
+    "tests/modin" # requires modin, not in nixpkgs
+    "tests/mypy/test_static_type_checking.py" # some typing failures
+    "tests/pyspark" # requires spark
+  ];
+
+  pythonImportsCheck = [
+    "pandera"
+    "pandera.api"
+    "pandera.config"
+    "pandera.dtypes"
+    "pandera.engines"
+  ];
+
+  meta = {
+    description = "Light-weight, flexible, and expressive statistical data testing library";
+    homepage = "https://pandera.readthedocs.io";
+    changelog = "https://github.com/unionai-oss/pandera/releases/tag/v${version}";
+    license = lib.licenses.mit;
+    maintainers = with lib.maintainers; [ bcdarwin ];
+  };
+}
diff --git a/pkgs/top-level/python-packages.nix b/pkgs/top-level/python-packages.nix
index e31ff8139b6a0..82d9d10a0f2d3 100644
--- a/pkgs/top-level/python-packages.nix
+++ b/pkgs/top-level/python-packages.nix
@@ -9734,6 +9734,8 @@ self: super: with self; {
 
   pandas-datareader = callPackage ../development/python-modules/pandas-datareader { };
 
+  pandera = callPackage ../development/python-modules/pandera { };
+
   pandoc-attributes = callPackage ../development/python-modules/pandoc-attributes { };
 
   pandoc-latex-environment = callPackage ../development/python-modules/pandoc-latex-environment { };