Merge ea2924f into b3efd1e

dedupe/__init__.py

@@ -1,7 +1,15 @@
-#!/usr/bin/python
-# -*- coding: utf-8 -*-
-from pkgutil import extend_path
-
-__path__ = extend_path(__path__, __name__)
-
-from dedupe._init import *  # noqa
+from dedupe.api import (  # noqa: F401
+    Dedupe,
+    Gazetteer,
+    RecordLink,
+    StaticDedupe,
+    StaticGazetteer,
+    StaticRecordLink,
+)
+from dedupe.convenience import (  # noqa: F401
+    canonicalize,
+    console_label,
+    training_data_dedupe,
+    training_data_link,
+)
+from dedupe.serializer import read_training, write_training  # noqa: F401

dedupe/datamodel.py

@@ -1,22 +1,18 @@
 from __future__ import annotations
 
 import copyreg
-import pkgutil
+import importlib
 import types
 from typing import TYPE_CHECKING, cast
 
 import numpy
+import pluggy
 
-import dedupe.variables
+import dedupe.hookspecs
 from dedupe.variables.base import FieldType as FieldVariable
 from dedupe.variables.base import MissingDataType, Variable
 from dedupe.variables.interaction import InteractionType
 
-for _, module, _ in pkgutil.iter_modules(  # type: ignore
-    dedupe.variables.__path__, "dedupe.variables."
-):
-    __import__(module)
-
 if TYPE_CHECKING:
     from typing import Generator, Iterable, Sequence
 
@@ -28,7 +24,26 @@
     )
     from dedupe.predicates import Predicate
 
-VARIABLE_CLASSES = {k: v for k, v in FieldVariable.all_subclasses() if k}
+
+DEFAULT_VARIABLES = [
+    "dedupe.variables.base",
+    "dedupe.variables.string",
+    "dedupe.variables.categorical_type",
+    "dedupe.variables.exists",
+    "dedupe.variables.exact",
+    "dedupe.variables.latlong",
+    "dedupe.variables.interaction",
+    "dedupe.variables.price",
+    "dedupe.variables.set",
+]
+
+pm = pluggy.PluginManager("dedupe")
+pm.add_hookspecs(dedupe.hookspecs)
+pm.load_setuptools_entrypoints("dedupe")
+
+for plugin in DEFAULT_VARIABLES:
+    mod = importlib.import_module(plugin)
+    pm.register(mod, plugin)
 
 
 class DataModel(object):
@@ -38,6 +53,7 @@ def __init__(self, variable_definitions: Iterable[VariableDefinition]):
         variable_definitions = list(variable_definitions)
         if not variable_definitions:
             raise ValueError("The variable definitions cannot be empty")
+
         all_variables: list[Variable]
         self.primary_variables, all_variables = typify_variables(variable_definitions)
         self._derived_start = len(all_variables)
@@ -145,6 +161,11 @@ def __setstate__(self, d):
 def typify_variables(
     variable_definitions: Iterable[VariableDefinition],
 ) -> tuple[list[FieldVariable], list[Variable]]:
+
+    variable_types = {}
+    for variable_type in pm.hook.register_variable():
+        variable_types.update(variable_type)
+
     primary_variables: list[FieldVariable] = []
     all_variables: list[Variable] = []
     only_custom = True
@@ -181,11 +202,11 @@ def typify_variables(
             ]
 
         try:
-            variable_class = VARIABLE_CLASSES[variable_type]
+            variable_class = variable_types[variable_type]
         except KeyError:
             raise KeyError(
                 "Field type %s not valid. Valid types include %s"
-                % (definition["type"], ", ".join(VARIABLE_CLASSES))
+                % (definition["type"], ", ".join(variable_types))
             )
 
         variable_object = variable_class(definition)

dedupe/hookspecs.py

@@ -0,0 +1,9 @@
+import pluggy
+
+hookimpl = pluggy.HookimplMarker("dedupe")
+hookspec = pluggy.HookspecMarker("dedupe")
+
+
+@hookspec
+def register_variable():
+    """Register a variable for use in a datamodel"""

dedupe/variables/__init__.py

@@ -1,3 +0,0 @@
-from pkgutil import extend_path
-
-__path__ = extend_path(__path__, __name__)

dedupe/variables/base.py

@@ -3,9 +3,10 @@
 from typing import TYPE_CHECKING
 
 from dedupe import predicates
+from dedupe.hookspecs import hookimpl
 
 if TYPE_CHECKING:
-    from typing import Any, ClassVar, Generator, Iterable, Optional, Sequence, Type
+    from typing import Any, ClassVar, Iterable, Sequence, Type
 
     from dedupe._typing import Comparator, PredicateFunction, VariableDefinition
 
@@ -47,15 +48,6 @@ def __getstate__(self) -> dict[str, Any]:
 
         return odict
 
-    @classmethod
-    def all_subclasses(
-        cls,
-    ) -> Generator[tuple[Optional[str], Type["Variable"]], None, None]:
-        for q in cls.__subclasses__():
-            yield getattr(q, "type", None), q
-            for p in q.all_subclasses():
-                yield p
-
 
 class DerivedType(Variable):
     type = "Derived"
@@ -135,3 +127,8 @@ def indexPredicates(
             index_predicates.append(predicate(threshold, field))
 
     return index_predicates
+
+
+@hookimpl
+def register_variable():
+    return {CustomType.type: CustomType}

dedupe/variables/categorical_type.py

@@ -4,6 +4,7 @@
 
 from dedupe import predicates
 from dedupe._typing import PredicateFunction, VariableDefinition
+from dedupe.hookspecs import hookimpl
 from dedupe.variables.base import DerivedType, FieldType
 
 
@@ -36,3 +37,8 @@ def __init__(self, definition: VariableDefinition):
 
     def __len__(self) -> int:
         return len(self.higher_vars)
+
+
+@hookimpl
+def register_variable():
+    return {CategoricalType.type: CategoricalType}

dedupe/variables/exact.py

@@ -1,6 +1,7 @@
 from typing import Any
 
 from dedupe import predicates
+from dedupe.hookspecs import hookimpl
 from dedupe.variables.base import FieldType
 
 
@@ -14,3 +15,8 @@ def comparator(field_1: Any, field_2: Any) -> int:
             return 1
         else:
             return 0
+
+
+@hookimpl
+def register_variable():
+    return {ExactType.type: ExactType}

dedupe/variables/exists.py

@@ -5,6 +5,7 @@
 from categorical import CategoricalComparator
 
 from dedupe._typing import PredicateFunction, VariableDefinition
+from dedupe.hookspecs import hookimpl
 from dedupe.variables.base import DerivedType
 from dedupe.variables.categorical_type import CategoricalType
 
@@ -37,3 +38,8 @@ def comparator(self, field_1: Any, field_2: Any) -> list[int]:
     # This flag tells fieldDistances in dedupe.core to pass
     # missing values (None) into the comparator
     comparator.missing = True  # type: ignore
+
+
+@hookimpl
+def register_variable():
+    return {ExistsType.type: ExistsType}

dedupe/variables/interaction.py

@@ -4,6 +4,7 @@
 from typing import Mapping
 
 from dedupe._typing import VariableDefinition
+from dedupe.hookspecs import hookimpl
 from dedupe.variables.base import FieldType as FieldVariable
 from dedupe.variables.base import Variable
 
@@ -77,3 +78,8 @@ def atomicInteractions(
                 atomic_interactions.append(field)
 
         return atomic_interactions
+
+
+@hookimpl
+def register_variable():
+    return {InteractionType.type: InteractionType}

dedupe/variables/latlong.py

@@ -5,6 +5,7 @@
 from haversine import haversine
 
 from dedupe import predicates
+from dedupe.hookspecs import hookimpl
 from dedupe.variables.base import FieldType
 
 
@@ -16,3 +17,8 @@ class LatLongType(FieldType):
     @staticmethod
     def comparator(x: tuple[float, float], y: tuple[float, float]) -> float:
         return sqrt(haversine(x, y))
+
+
+@hookimpl
+def register_variable():
+    return {LatLongType.type: LatLongType}

dedupe/variables/price.py

@@ -3,6 +3,7 @@
 import numpy
 
 from dedupe import predicates
+from dedupe.hookspecs import hookimpl
 from dedupe.variables.base import FieldType
 
 
@@ -22,3 +23,8 @@ def comparator(price_1: int | float, price_2: int | float) -> float:
             return numpy.nan
         else:
             return abs(numpy.log10(price_1) - numpy.log10(price_2))
+
+
+@hookimpl
+def register_variable():
+    return {PriceType.type: PriceType}

dedupe/variables/set.py

@@ -2,6 +2,7 @@
 
 from dedupe import predicates
 from dedupe._typing import VariableDefinition
+from dedupe.hookspecs import hookimpl
 from dedupe.variables.base import FieldType
 
 
@@ -31,3 +32,8 @@ def __init__(self, definition: VariableDefinition):
             definition["corpus"] = []
 
         self.comparator = CosineSetSimilarity(definition["corpus"])  # type: ignore[assignment]
+
+
+@hookimpl
+def register_variable():
+    return {SetType.type: SetType}

dedupe/variables/string.py

@@ -6,6 +6,7 @@
 
 from dedupe import predicates
 from dedupe._typing import VariableDefinition
+from dedupe.hookspecs import hookimpl
 from dedupe.variables.base import FieldType, indexPredicates
 
 crfEd = CRFEditDistance()
@@ -105,3 +106,12 @@ def __init__(self, definition: VariableDefinition):
             definition["corpus"] = []
 
         self.comparator = CosineTextSimilarity(definition["corpus"])  # type: ignore[assignment]
+
+
+@hookimpl
+def register_variable():
+    return {
+        ShortStringType.type: ShortStringType,
+        StringType.type: StringType,
+        TextType.type: TextType,
+    }

pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "dedupe"
 description = "A python library for accurate and scaleable data deduplication and entity-resolution"
-version = "2.0.19"
+version = "3.0.0"
 readme = "README.md"
 requires-python = ">=3.7"
 license = {file = "LICENSE"}
@@ -38,6 +38,7 @@ dependencies = [
     "zope.index",
     "Levenshtein_search==1.4.5",
     "typing_extensions",
+    "pluggy",
 ]
 
 [project.urls]
@@ -51,7 +52,7 @@ MailingList = "https://groups.google.com/forum/#!forum/open-source-deduplication
 
 
 [build-system]
-requires = ["setuptools==63",
+requires = ["setuptools",
             "wheel",
             "cython"]
 build-backend = "setuptools.build_meta"