Pamhyr2/src/Model/Ensembles/Ensemble.py

# Ensemble.py -- Pamhyr
# Copyright (C) 2026  INRAE
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <https://www.gnu.org/licenses/>.

# -*- coding: utf-8 -*-

import struct
import logging
import itertools

import numpy as np

from Model.Tools.PamhyrDB import SQLSubModel
from Model.Scenario import Scenario
from Model.Ensembles.Distribution import Uniform

logger = logging.getLogger()


class Ensemble(SQLSubModel):
    _sub_classes = []

    def __init__(self, id: int = -1, name: str = "",
                 distribution=None, params=[0., 100.],
                 data_type=None, target_data=None,
                 status=None, owner_scenario=-1):
        super(Ensemble, self).__init__(
            id=id, status=status,
            owner_scenario=owner_scenario
        )

        self._name = name
        self._data_type = data_type
        self._distribution = distribution
        self._params = params
        self._target_data = target_data

    @classmethod
    def _db_create(cls, execute, ext=""):
        execute(f"""
          CREATE TABLE ensemble{ext} (
            {cls.create_db_add_pamhyr_id()},
            deleted BOOLEAN NOT NULL DEFAULT FALSE,
            name TEXT NOT NULL,
            type VARCHAR(8) NOT NULL,
            distribution INTEGER NOT NULL,
            distribution_name VARCHAR(32) NOT NULL,
            params BLOB NOT NULL,
            params_len INTEGER NOT NULL,
            data_pid INTEGER NOT NULL,
            {Scenario.create_db_add_scenario()},
            {Scenario.create_db_add_scenario_fk()},
            PRIMARY KEY(pamhyr_id, scenario)
          )
        """)

        if ext == "_tmp":
            return True

        return cls._create_submodel(execute)

    @classmethod
    def _db_update(cls, execute, version, data=None):
        major, minor, release = version.strip().split(".")
        if major == "0" and int(minor) < 2:
            cls._db_create(execute)
        elif major == "0" and int(minor) == 2:
            if int(release) < 8:
                cls._db_create(execute)

        return cls._update_submodel(execute, version, data)

    @classmethod
    def _db_load(cls, execute, data=None):
        new = []
        scenario = data["scenario"]
        distributions = data["ens_distributions"]
        stricklers = data["stricklers"]
        loaded = data['loaded_pid']

        table = execute(
            "SELECT pamhyr_id, deleted, name, type, " +
            "distribution, distribution_name, params, params_len, data_pid, " +
            "scenario " +
            "FROM ensemble " +
            f"WHERE scenario = {scenario.id} " +
            f"AND pamhyr_id NOT IN ({', '.join(map(str, loaded))})"
        )

        for row in table:
            it = iter(row)

            id = next(it)
            deleted = (next(it) == 1)
            name = next(it)
            data_type = next(it)
            fid = next(it)
            fname = next(it)
            params = next(it)
            params_len = next(it)
            data_pid = next(it)
            owner_scenario = next(it)

            def fn(f): return f._name == fname
            if fid != -1:
                def fn(f): return f._pamhyr_id == fid

            distribution = next(
                filter(fn, distributions.lst),
                None
            )

            target_data = next(
                filter(
                    lambda s: s._pamhyr_id == data_pid,
                    stricklers.lst
                ),
                None
            )

            rdata = list(cls._decode_params(params, params_len))

            new_ensemble = cls(
                id, name=name,
                distribution=distribution,
                params=rdata,
                data_type=data_type,
                target_data=target_data,
                status=data["status"],
                owner_scenario=owner_scenario
            )
            if deleted:
                new_ensemble.set_as_deleted()

            loaded.add(id)
            new.append(new_ensemble)

        return new

    def _db_save(self, execute, data=None):
        if not self.must_be_saved():
            return True

        execute(
            "DELETE FROM ensemble " +
            f"WHERE pamhyr_id = {self.pamhyr_id} " +
            f"AND scenario = {self._status.scenario_id}"
        )

        bparams, length = self._encode_params()

        if self._distribution is None:
            fid = -1
            fname = "NULL"
        else:
            fid = self._distribution._pamhyr_id
            fname = self._distribution._name

            if self._distribution._type == "generic":
                fid = -1

        data_pid = -1
        if self._target_data is not None:
            data_pid = self._target_data._pamhyr_id

        execute(
            "INSERT INTO " +
            "ensemble(pamhyr_id, deleted, name, type, " +
            "  distribution, distribution_name, params, params_len, " +
            "  data_pid, scenario) " +
            "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
            self.pamhyr_id, self.is_deleted(),
            self._name, self._data_type,
            fid, fname, bparams, length,
            data_pid, self._status.scenario_id
        )

        return True

    def _encode_params(self):
        length = len(self._params)
        data_format = ">" + ''.join(itertools.repeat("d", length))

        return struct.pack(data_format, *self._params), length

    @classmethod
    def _decode_params(cls, bparams, length):
        data_format = ">" + ''.join(itertools.repeat("d", length))

        return struct.unpack(data_format, bparams)

    def random(self, nb):
        return np.nditer(
            self._distribution.random(
                self._params, nb
            )
        )

    @property
    def name(self):
        return self._name

    @name.setter
    def name(self, name):
        self._name = name
        self._status.modified()

    @property
    def distribution(self):
        return self._distribution

    @distribution.setter
    def distribution(self, distribution):
        self._distribution = distribution
        self._status.modified()

    @property
    def params(self):
        return self._params

    @params.setter
    def params(self, params):
        self._params = params
        self._status.modified()

    @property
    def data_type(self):
        return self._data_type

    @data_type.setter
    def data_type(self, data_type):
        self._data_type = data_type
        self._status.modified()

    @property
    def target_data(self):
        return self._target_data

    @target_data.setter
    def target_data(self, target_data):
        self._target_data = target_data
        self._status.modified()

    def get_data_dict(self, use_ids=False):
        target_data = self.target_data
        distribution = self.distribution

        if use_ids:
            if self.target_data:
                target_data = self.target_data.pamhyr_id
            if self.distribution:
                distribution = self.distribution.name

        return {
            "name": self.name,
            "data_type": self.data_type,
            "target_data": target_data,
            "distribution": distribution,
            "params": self.params,
        }

    def set_data_dict(self, data):
        self.name = data["name"]
        self.data_type = data["data_type"]
        self.target_data = data["target_data"]
        self.distribution = data["distribution"]
        self.params = data["params"]