Reorganize checkers

neurots/generate/algorithms/tmdgrower.py

@@ -27,6 +27,7 @@
 from neurots.generate.algorithms.common import section_data
 from neurots.morphmath import sample
 from neurots.morphmath.utils import norm
+from neurots.preprocess.relevancy_checkers import check_min_bar_length
 
 L = logging.getLogger(__name__)
 
@@ -65,19 +66,7 @@ def __init__(
         self.persistence_length = self.barcode.get_persistence_length()
         # Validate parameters and distributions
         if not skip_validation:
-            self.check_min_bar_length(params, input_data)
-
-    @staticmethod
-    def check_min_bar_length(params, distrs):
-        """Consistency check between parameters - persistence diagram."""
-        barSZ = distrs["min_bar_length"]
-        stepSZ = params["step_size"]["norm"]["mean"]
-        if stepSZ >= barSZ:
-            L.warning(
-                "Selected step size %f is too big for bars of size %f",
-                stepSZ,
-                barSZ,
-            )
+            check_min_bar_length(params, input_data, start_point, context)
 
     def select_persistence(self, input_data, random_generator=np.random):
         """Select the persistence.

neurots/preprocess/__init__.py

@@ -1,4 +1,22 @@
 """Functions used for preprocessing."""
-from neurots.preprocess import checkers  # noqa
+
+# Copyright (C) 2022  Blue Brain Project, EPFL
+#
+# 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/>.
+
+from neurots.preprocess import relevancy_checkers  # noqa
+from neurots.preprocess import validity_checkers  # noqa
 from neurots.preprocess.utils import preprocess_inputs  # noqa
 from neurots.preprocess.utils import register_preprocess  # noqa
+from neurots.preprocess.utils import register_validator  # noqa

neurots/preprocess/preprocessors.py

@@ -0,0 +1,28 @@
+"""Functions to modify the given parameters and distributions before running the algorithm.
+
+The functions used as preprocesses should have a name like 'preprocess_*' and have the following
+signature: `check_something(params, distrs)`. These functions should be very generic and should
+not depend on any context in which the related grower is used.
+
+The preprocesses should be registered to be executed in the preprocess step using the
+`@register_preprocess` decorator.
+"""
+
+# Copyright (C) 2022  Blue Brain Project, EPFL
+#
+# 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/>.
+
+# TODO: remove the next line and the 'noqa' when a preprocess is added
+# pylint: disable=unused-import
+from neurots.preprocess.utils import register_preprocess  # noqa

neurots/preprocess/relevancy_checkers.py

@@ -0,0 +1,39 @@
+"""Functions to check that the given parameters and distributions will given relevant results.
+
+The functions used as relevancy checkers should have a name like 'check_*' and have the following
+signature: `check_something(params, distrs, start_point=None, context=None)`. The `start_point` and
+`context` parameters should always be optional, as they will not be known during the preprocessing
+step.
+"""
+
+# Copyright (C) 2022  Blue Brain Project, EPFL
+#
+# 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/>.
+
+import logging
+
+L = logging.getLogger(__name__)
+
+
+def check_min_bar_length(params, distrs, start_point=None, context=None):
+    """Consistency check between parameters - persistence diagram."""
+    # pylint: disable=unused-argument
+    barSZ = distrs["min_bar_length"]
+    stepSZ = params["step_size"]["norm"]["mean"]
+    if stepSZ >= barSZ:
+        L.warning(
+            "Selected step size %f is too big for bars of size %f",
+            stepSZ,
+            barSZ,
+        )

neurots/preprocess/utils.py

@@ -1,19 +1,50 @@
 """Some utils for preprocessing."""
+
+# Copyright (C) 2022  Blue Brain Project, EPFL
+#
+# 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/>.
+
 from collections import defaultdict
 from copy import deepcopy
+from itertools import chain
 
 from neurots.validator import validate_neuron_distribs
 from neurots.validator import validate_neuron_params
 
-_PREPROCESS_FUNCTIONS = defaultdict(set)
+_PREPROCESS_FUNCTIONS = {
+    "preprocess": defaultdict(set),
+    "validator": defaultdict(set),
+}
 
 
 def register_preprocess(*growth_methods):
     """Register a preprocess function."""
 
     def inner(func):
         for i in growth_methods:
-            _PREPROCESS_FUNCTIONS[i].add(func)
+            _PREPROCESS_FUNCTIONS["preprocess"][i].add(func)
+        return func
+
+    return inner
+
+
+def register_validator(*growth_methods):
+    """Register a validation function."""
+
+    def inner(func):
+        for i in growth_methods:
+            _PREPROCESS_FUNCTIONS["validator"][i].add(func)
         return func
 
     return inner
@@ -28,7 +59,10 @@ def preprocess_inputs(params, distrs):
 
     for grow_type in params["grow_types"]:
         growth_method = params[grow_type]["growth_method"]
-        for preprocess_func in _PREPROCESS_FUNCTIONS[growth_method]:
+        for preprocess_func in chain(
+            _PREPROCESS_FUNCTIONS["validator"][growth_method],
+            _PREPROCESS_FUNCTIONS["preprocess"][growth_method],
+        ):
             preprocess_func(params[grow_type], distrs[grow_type])
 
     return params, distrs

neurots/preprocess/validity_checkers.py

@@ -0,0 +1,54 @@
+"""Functions to check that the given parameters and distributions will not break the algorithm.
+
+The functions used as validity checkers should have a name like 'check_*' and have the following
+signature: `check_something(params, distrs)`. These functions should be very generic and should
+not depend on any context in which the related grower is used.
+
+The checkers should be registered to be executed in the preprocess step using the
+`@register_validator` decorator.
+"""
+
+# Copyright (C) 2022  Blue Brain Project, EPFL
+#
+# 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/>.
+
+from neurots.preprocess.relevancy_checkers import check_min_bar_length
+from neurots.preprocess.utils import register_validator
+
+
+@register_validator("trunk")
+def check_num_seg(params, distrs):
+    """Check that params contains a 'num_seg' entry."""
+    # pylint: disable=unused-argument
+    if "num_seg" not in params:
+        raise KeyError(
+            "The parameters must contain a 'num_seg' entry when the "
+            "'growth_method' entry in parameters is 'trunk'."
+        )
+
+
+@register_validator("tmd", "tmd_apical", "tmd_gradient")
+def check_bar_length(params, distrs):
+    """Check consistency between parameters and persistence diagram."""
+    if "min_bar_length" not in distrs:
+        raise KeyError(
+            "The distributions must contain a 'min_bar_length' entry when the "
+            "'growth_method' entry in parameters is in ['tmd', 'tmd_apical', 'tmd_gradient']."
+        )
+    if "mean" not in params.get("step_size", {}).get("norm", {}):
+        raise KeyError(
+            "The parameters must contain a 'step_size' entry when the "
+            "'growth_method' entry in parameters is in ['tmd', 'tmd_apical', 'tmd_gradient']."
+        )
+    check_min_bar_length(params, distrs)