package.py

# Copyright 2013-2020 Lawrence Livermore National Security, LLC and other
# Spack Project Developers. See the top-level COPYRIGHT file for details.
#
# SPDX-License-Identifier: (Apache-2.0 OR MIT)

from spack.package import *


class NeurodamusModels(CMakePackage):
    """Neuroscientific models to be used with Neurodamus"""

    homepage = "https://github.com/BlueBrain/neurodamus-models"
    git = "https://github.com/BlueBrain/neurodamus-models.git"

    version("develop", branch="main")
    version("2.4.1", tag="2.4.1")
    version("2.4.0", tag="2.4.0")
    version("2.3.0", tag="2.3.0")
    version("2.2.0", tag="2.2.0")
    version("2.1.0", tag="2.1.0")
    version("2.0.0", tag="2.0.0")

    variant("caliper", default=False, description="Enable Caliper instrumentation")
    variant("coreneuron", default=False, description="Enable CoreNEURON support")

    variant(
        "model",
        default="neocortex",
        values=("hippocampus", "neocortex", "thalamus"),
        multi=False,
        description="Which brain region mechanism to enable",
    )
    variant("ngv", default=False, when="model=neocortex", description="Enable NGV mechanisms")
    variant(
        "metabolism",
        default=False,
        when="model=neocortex",
        description="Enable metabolism mechanisms",
    )
    variant(
        "plasticity",
        default=False,
        when="model=neocortex",
        description="Enable plasticity related mechanisms",
    )
    variant("v5", default=True, when="model=neocortex", description="Enable V5 circuit mechanisms")

    depends_on("neuron", type=("build", "link", "run"))
    depends_on("neuron+coreneuron", when="+coreneuron", type=("build", "link", "run"))
    depends_on("neuron+caliper", when="+caliper", type=("build", "link", "run"))

    depends_on("py-neurodamus", type=("build", "run"))
    depends_on("libsonata-report")

    def cmake_args(self):
        args = [
            self.define("NEURODAMUS_CORE_DIR", self.spec["py-neurodamus"].package.datadir),
            self.define_from_variant("NEURODAMUS_MECHANISMS", "model"),
            self.define_from_variant("NEURODAMUS_ENABLE_CORENEURON", "coreneuron"),
            self.define_from_variant("NEURODAMUS_NCX_METABOLISM", "metabolism"),
            self.define_from_variant("NEURODAMUS_NCX_NGV", "ngv"),
            self.define_from_variant("NEURODAMUS_NCX_PLASTICITY", "plasticity"),
            self.define_from_variant("NEURODAMUS_NCX_V5", "v5"),
        ]
        return args

    def setup_build_environment(self, env):
        env.unset("LC_ALL")

    def setup_run_environment(self, env):
        mech = self.spec.variants["model"].value
        env.set(f"NEURODAMUS_{mech.upper()}_ROOT", self.prefix)
        env.set("HOC_LIBRARY_PATH", self.prefix.share.join(f"neurodamus_{mech}").hoc)
        if "+coreneuron" in self.spec:
            env.set("CORENEURONLIB", self.prefix.lib + "/libcorenrnmech.so")
        for libnrnmech_name in find(self.prefix.lib, "libnrnmech*.so", recursive=False):
            env.set("NRNMECH_LIB_PATH", libnrnmech_name)
            env.set("BLUECELLULAB_MOD_LIBRARY_PATH", libnrnmech_name)

        # ENV variables to enable Caliper with certain settings
        if "+caliper" in self.spec:
            env.set("NEURODAMUS_CALI_ENABLED", "true")  # Needed for slurm.taskprolog
            env.set("CALI_MPIREPORT_FILENAME", "/dev/null")  # Prevents 'stdout' output
            env.set("CALI_CHANNEL_FLUSH_ON_EXIT", "false")
            env.set(
                "CALI_MPIREPORT_LOCAL_CONFIG",
                "SELECT sum(sum#time.duration), \
                                                        inclusive_sum(sum#time.duration) \
                                                    GROUP BY prop:nested",
            )
            env.set(
                "CALI_MPIREPORT_CONFIG",
                "SELECT annotation, \
                        mpi.function, \
                        min(sum#sum#time.duration) as exclusive_time_rank_min, \
                        max(sum#sum#time.duration) as exclusive_time_rank_max, \
                        avg(sum#sum#time.duration) as exclusive_time_rank_avg, \
                        min(inclusive#sum#time.duration) AS inclusive_time_rank_min, \
                        max(inclusive#sum#time.duration) AS inclusive_time_rank_max, \
                        avg(inclusive#sum#time.duration) AS inclusive_time_rank_avg, \
                        percent_total(sum#sum#time.duration) AS exclusive_time_pct, \
                        inclusive_percent_total(sum#sum#time.duration) AS inclusive_time_pct \
                    GROUP BY prop:nested FORMAT json",
            )
            env.set("CALI_SERVICES_ENABLE", "aggregate,event,mpi,mpireport,timestamp")
            env.set("CALI_MPI_BLACKLIST", "MPI_Comm_rank,MPI_Comm_size,MPI_Wtick,MPI_Wtime")