mismo 0.2.7.dev1__tar.gz → 0.2.7.dev3__tar.gz

{mismo-0.2.7.dev1 → mismo-0.2.7.dev3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: mismo
-Version: 0.2.7.dev1
+Version: 0.2.7.dev3
 Summary: The SQL/Ibis powered sklearn of record linkage.
 Keywords: record linkage,entity resolution,fuzzy linking,machine learning,ibis,sql,splink,duckdb
 Author: Nick Crews

{mismo-0.2.7.dev1 → mismo-0.2.7.dev3}/mismo/_datasets.py

@@ -5,6 +5,8 @@ from typing import Callable, Iterable, Mapping, Sequence
 import ibis
 from ibis.expr import types as ir
 
+from mismo import _util
+
 
 class Datasets:
     """An ordered, dict-like collection of tables of records.
@@ -73,7 +75,7 @@ class Datasets:
     def map(self, f: ibis.Deferred | Callable[[str, ir.Table], ir.Table]) -> ir.Table:
         """Return a new Datasets with all tables transformed by `f`."""
         if isinstance(f, ibis.Deferred):
-            return self.__class__({name: f.bind(t) for name, t in self.items()})
+            return self.__class__({name: _util.bind(f, t) for name, t in self.items()})
         else:
             return self.__class__({name: f(name, t) for name, t in self.items()})
 

{mismo-0.2.7.dev1 → mismo-0.2.7.dev3}/mismo/_resolve.py

@@ -32,7 +32,7 @@ class DeferredResolver(ValueResolver):
 
     def __call__(self, t: ibis.Table) -> ibis.Column:
         raw = self.deferred.resolve(**{self.name: t})
-        return _resolve(t, raw)
+        return _util.bind_one(t, raw)
 
     def __repr__(self) -> str:
         if self.name == "_":
@@ -49,7 +49,7 @@ class LiteralResolver(ValueResolver):
 
     def __call__(self, t: ibis.Table) -> ibis.Column:
         """Resolve a literal value."""
-        resolved = t.bind(self.value)
+        resolved = _util.bind(t, self.value)
         if len(resolved) != 1:
             raise ValueError(
                 f"Expected 1 column, got {len(resolved)} from {self.value}"
@@ -95,13 +95,6 @@ class FuncResolver(ValueResolver):
         return f"FuncResolver({self.func!r})"
 
 
-def _resolve(t: ir.Table, spec) -> ir.Value:
-    values = t.bind(spec)
-    if len(values) != 1:
-        raise ValueError(f"Expected 1 column, got {len(values)} from {spec}")
-    return values[0]
-
-
 def value_resolver(spec: ibis.Value | Deferred | str) -> ValueResolver:
     """
     Given a spec, return a ValueResolver that resolves to a single column.

{mismo-0.2.7.dev1 → mismo-0.2.7.dev3}/mismo/_util.py

@@ -3,7 +3,7 @@ from __future__ import annotations
 import base64
 from collections.abc import Sequence
 from contextlib import contextmanager
-from typing import Any, Callable, Iterable, Literal, Mapping, TypeVar
+from typing import Any, Callable, Iterable, Literal, Mapping, TypeVar, overload
 import uuid
 import warnings
 
@@ -89,12 +89,38 @@ def cases(
         return builder.else_(else_).end()
 
 
-def bind(t: ir.Table, ref: Any) -> tuple[ir.Value]:
-    """Reference into a table to get Columns and Scalars."""
-    # ibis's bind() can't handle bind(_, "column_name")
-    if isinstance(ref, str):
-        return (t[ref],)
-    return t.bind(ref)
+@overload
+def bind(t: ibis.Table, ref: Any, /) -> tuple[ibis.Column, ...]: ...
+@overload
+def bind(t: ibis.Deferred, ref: Any, /) -> tuple[ibis.Deferred]: ...
+
+
+def bind(t: ibis.Deferred | ibis.Table, ref: Any) -> tuple[ibis.Column, ...]:
+    """Reference into a table to get Columns and Scalars.
+
+    ibis._.bind(ref) does not work because it returns another Deferred.
+
+    Also, per https://github.com/ibis-project/ibis/pull/11746,
+    in some versions of ibis, .bind() returns a generator, not a tuple,
+    so this function always returns a tuple.
+    """
+    if isinstance(t, ibis.Table):
+        return tuple(t.bind(ref))
+    return (t[ref],)
+
+
+@overload
+def bind_one(t: ibis.Table, ref: Any, /) -> ibis.Column: ...
+@overload
+def bind_one(t: ibis.Deferred, ref: Any, /) -> ibis.Deferred: ...
+
+
+def bind_one(t: ibis.Deferred | ibis.Table, ref: Any) -> ibis.Column | ibis.Deferred:
+    """Like bind(), but ensure that exactly one column is returned."""
+    cols = bind(t, ref)
+    if len(cols) != 1:
+        raise ValueError(f"Expected 1 column, got {len(cols)} from {ref}")
+    return cols[0]
 
 
 def get_column(

{mismo-0.2.7.dev1 → mismo-0.2.7.dev3}/mismo/compare/_match_level.py

@@ -1,7 +1,8 @@
 from __future__ import annotations
 
 from abc import ABCMeta
-from typing import Iterable, Iterator, Literal, Type, TypeVar, overload
+from collections.abc import Iterable, Iterator
+from typing import Generic, Literal, Type, TypeVar, overload
 
 import ibis
 from ibis.expr import types as ir
@@ -320,7 +321,7 @@ def _is_stringy(v):
 MatchLevelT = TypeVar("MatchLevelT", bound=MatchLevel)
 
 
-class LevelComparer:
+class LevelComparer(Generic[MatchLevelT]):
     """
     Assigns a MatchLevel to record pairs based on one dimension, e.g. *name*
     """
@@ -373,7 +374,7 @@ class LevelComparer:
         if representation is None:
             representation = self.representation
 
-        cases = [(pairs.bind(c)[0], level) for c, level in self.cases]
+        cases = [(_util.bind_one(pairs, c), level) for c, level in self.cases]
         if representation == "string":
             cases = [(c, level.as_string()) for c, level in cases]
         elif representation == "integer":

{mismo-0.2.7.dev1 → mismo-0.2.7.dev3}/mismo/lib/geo/_address.py

@@ -72,7 +72,7 @@ def _featurize(
     # for every time it appears in the SQL. See https://github.com/duckdb/duckdb/discussions/14649.
     # So, if we did one .mutate(), we would end up with like literally 100 regex
     # evaluations in the SQL, which is 100x slower than evaluating the regex once.
-    input_column = t.bind(input_column)[0]
+    input_column = _util.bind_one(t, input_column)
     t = t.mutate(_parsed=parse_street1_re(input_column.street1), _cleaned=input_column)
     t = t.mutate(
         __address_featured=ibis.struct(

{mismo-0.2.7.dev1 → mismo-0.2.7.dev3}/mismo/tf/_tf.py

@@ -52,7 +52,7 @@ class ColumnStats:
         self,
         table: ibis.Table,
         *,
-        column: str | ibis.Deferred | ibis.Column = None,
+        column: str | ibis.Deferred | ibis.Column | None = None,
         name_as: str | None = None,
         default: Literal["1/N"] | int | float = "1/N",
     ) -> ibis.Table:
@@ -70,7 +70,7 @@ class ColumnStats:
         else:
             default = ibis.literal(default, "float64")
 
-        table_column = table.bind(column)[0]
+        table_column = _util.bind_one(table, column)
 
         unique_name = _util.unique_name("join_key")
         # TODO: this could be factored out into a join_lookup() function

{mismo-0.2.7.dev1 → mismo-0.2.7.dev3}/mismo/types/_linked_table.py

@@ -384,7 +384,11 @@ class LinkCountsTable(TableWrapper):
         else:
             subtitle = "eg 'there were 1000 records with 0 links, 500 with 1 link, 100 with 2 links, ...'"  # noqa: E501
 
-        frac_records = self.n_records / total_records if total_records > 0 else 0
+        frac_records: ir.StringValue = (
+            (self.n_records / total_records * 100).cast(int).cast(str)
+            if total_records > 0
+            else ibis.literal("0")
+        )
         t = self.mutate(
             frac_records=frac_records,
             explanation=(
@@ -392,7 +396,7 @@ class LinkCountsTable(TableWrapper):
                 + f"{total_records:_} records, "
                 + self.n_records.cast(str)
                 + " ("
-                + (frac_records * 100).cast(int).cast(str)
+                + frac_records
                 + "%) had "
                 + self.n_links.cast(str)
                 + " links."

{mismo-0.2.7.dev1 → mismo-0.2.7.dev3}/mismo/types/_updates.py

@@ -282,8 +282,8 @@ class Updates(TableWrapper):
 
     def is_changed(self, column: str, /) -> ibis.ir.BooleanColumn:
         """Is column.before different from column.after? Never returns NULL."""
-        (val,) = self.bind(column)
-        return is_changed(val)
+        resolved_col = _util.bind_one(self._t, column)
+        return is_changed(resolved_col)
 
     def filter(self, *args, **kwargs):
         return self.__class__(

{mismo-0.2.7.dev1 → mismo-0.2.7.dev3}/pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "mismo"
 description = "The SQL/Ibis powered sklearn of record linkage."
-version = "0.2.7.dev1"
+version = "0.2.7.dev3"
 readme = "README.md"
 requires-python = ">=3.10" # ibis>=9.1 requires python >=3.10
 license = { text = "LGPL-3.0-or-later" }