From fcd2e4bdaf974e6754407fe78b1f0ca404f7b3d5 Mon Sep 17 00:00:00 2001
From: Pol Dellaiera <pol.dellaiera@protonmail.com>
Date: Wed, 2 Apr 2025 19:20:08 +0200
Subject: [PATCH] python312Packages.swh-export: init at 1.8.0

---
 pkgs/by-name/sw/swh/package.nix               |  1 +
 .../python-modules/swh-export/default.nix     | 90 +++++++++++++++++++
 pkgs/top-level/python-packages.nix            |  2 +
 3 files changed, 93 insertions(+)
 create mode 100644 pkgs/development/python-modules/swh-export/default.nix

diff --git a/pkgs/by-name/sw/swh/package.nix b/pkgs/by-name/sw/swh/package.nix
index 96561b1c703b..95828a6665d4 100644
--- a/pkgs/by-name/sw/swh/package.nix
+++ b/pkgs/by-name/sw/swh/package.nix
@@ -3,6 +3,7 @@
   writeShellApplication,
   withSwhPythonPackages ? [
     python3Packages.swh-auth
+    python3Packages.swh-export
     python3Packages.swh-model
     python3Packages.swh-objstorage
     python3Packages.swh-scanner
diff --git a/pkgs/development/python-modules/swh-export/default.nix b/pkgs/development/python-modules/swh-export/default.nix
new file mode 100644
index 000000000000..6429bd8b475e
--- /dev/null
+++ b/pkgs/development/python-modules/swh-export/default.nix
@@ -0,0 +1,90 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitLab,
+  setuptools,
+  setuptools-scm,
+  boto3,
+  click,
+  tqdm,
+  pyorc,
+  plyvel,
+  python,
+  types-requests,
+  swh-core,
+  swh-journal,
+  swh-model,
+  swh-storage,
+  pytestCheckHook,
+  pytest-kafka,
+  pytest-mock,
+  tzdata,
+  pkgs,
+}:
+
+buildPythonPackage rec {
+  pname = "swh-export";
+  version = "1.8.0";
+  pyproject = true;
+
+  src = fetchFromGitLab {
+    domain = "gitlab.softwareheritage.org";
+    group = "swh";
+    owner = "devel";
+    repo = "swh-export";
+    tag = "v${version}";
+    hash = "sha256-n97MMYn7EmTrv411YSxUD1+zfbFB8KOSns44N3NqqV8=";
+  };
+
+  build-system = [
+    setuptools
+    setuptools-scm
+  ];
+
+  dependencies = [
+    boto3
+    click
+    tqdm
+    pyorc
+    plyvel
+    types-requests
+    swh-core
+    swh-journal
+    swh-model
+    swh-storage
+  ];
+
+  preCheck = ''
+    # provide timezone data, works only on linux
+    export TZDIR=${tzdata}/${python.sitePackages}/tzdata/zoneinfo
+  '';
+
+  pythonImportsCheck = [ "swh.export" ];
+
+  nativeCheckInputs = [
+    pytestCheckHook
+    pytest-kafka
+    pytest-mock
+    pkgs.zstd
+    pkgs.pv
+  ];
+
+  disabledTests = [
+    # I don't know how to fix the following error
+    # E       fixture 'kafka_server' not found
+    "test_parallel_journal_processor"
+    "test_parallel_journal_processor_origin"
+    "test_parallel_journal_processor_origin_visit_status"
+    "test_parallel_journal_processor_offsets"
+    "test_parallel_journal_processor_masked"
+    "test_parallel_journal_processor_masked_origin"
+    "test_parallel_journal_processor_masked_origin_visit_statuses"
+  ];
+
+  meta = {
+    description = "Software Heritage dataset tools";
+    homepage = "https://gitlab.softwareheritage.org/swh/devel/swh-export";
+    license = lib.licenses.gpl3Only;
+    maintainers = with lib.maintainers; [ drupol ];
+  };
+}
diff --git a/pkgs/top-level/python-packages.nix b/pkgs/top-level/python-packages.nix
index 312d2cfa7213..7af786bb5bab 100644
--- a/pkgs/top-level/python-packages.nix
+++ b/pkgs/top-level/python-packages.nix
@@ -16673,6 +16673,8 @@ self: super: with self; {
 
   swh-core = callPackage ../development/python-modules/swh-core { };
 
+  swh-export = callPackage ../development/python-modules/swh-export { };
+
   swh-journal = callPackage ../development/python-modules/swh-journal { };
 
   swh-model = callPackage ../development/python-modules/swh-model { };