{ lib, buildPythonPackage, fetchFromGitHub, python, requests }: let json-ld = fetchFromGitHub { owner = "json-ld"; repo = "json-ld.org"; rev = "843a70e4523d7cd2a4d3f5325586e726eb1b123f"; sha256 = "05j0nq6vafclyypxjj30iw898ig0m32nvz0rjdlslx6lawkiwb2a"; }; normalization = fetchFromGitHub { owner = "json-ld"; repo = "normalization"; rev = "aceeaf224b64d6880189d795bd99c3ffadb5d79e"; sha256 = "125q5rllfm8vg9mz8hn7bhvhv2vqpd86kx2kxlk84smh33l8kbyl"; }; in buildPythonPackage rec { pname = "pyld"; version = "1.0.5"; format = "setuptools"; src = fetchFromGitHub { owner = "digitalbazaar"; repo = pname; rev = version; sha256 = "0z2vkllw8bvzxripwb6l757r7av5qwhzsiy4061gmlhq8z8gq961"; }; propagatedBuildInputs = [ requests ]; # Unfortunately PyLD does not pass all testcases in the JSON-LD corpus. We # check for at least a minimum amount of successful tests so we know it's not # getting worse, at least. checkPhase = '' ok_min=401 if ! ${python.interpreter} tests/runtests.py -d ${json-ld}/test-suite 2>&1 | tee test.out; then ok_count=$(grep -F '... ok' test.out | wc -l) if [[ $ok_count -lt $ok_min ]]; then echo "Less than $ok_min tests passed ($ok_count). Failing the build." exit 1 fi fi ${python.interpreter} tests/runtests.py -d ${normalization}/tests ''; meta = with lib; { description = "Python implementation of the JSON-LD API"; homepage = "https://github.com/digitalbazaar/pyld"; license = licenses.bsd3; maintainers = with maintainers; [ apeschar ]; }; }