chore: relock deps to newer versions and fix now-failing tests

Author: NickCrews

mismo/conftest.py

@@ -11,6 +11,20 @@
 from ibis.expr import types as ir
 import pytest
 
+try:
+    # Workaround for https://github.com/explosion/thinc/pull/965
+    import thinc.util
+
+    _original_fix_random_seed = thinc.util.fix_random_seed
+
+    def _patched_fix_random_seed(seed: int = 0) -> None:
+        seed = seed % (2**32)
+        _original_fix_random_seed(seed)
+
+    thinc.util.fix_random_seed = _patched_fix_random_seed
+except ImportError:
+    pass
+
 # we want to have pytest assert introspection in the helpers
 pytest.register_assert_rewrite("mismo.tests.util")
 

mismo/lib/geo/_census.py

@@ -129,7 +129,9 @@ def us_census_geocode(
             if col != "__row_number"
         }
     )
-    re_joined = t.inner_join(unduped, "__row_number").drop("__row_number")
+    input_cols = [c for c in t.columns if c != "__row_number"]
+    geocode_cols = [c for c in unduped.columns if c != "__row_number"]
+    re_joined = t.inner_join(unduped, "__row_number").select(*input_cols, *geocode_cols)
     return re_joined
 
 

mismo/types/_linked_table.py

@@ -185,7 +185,8 @@ def with_linked_values(
             t,
             self_id_to_other_array_vals,
             uname,
-        ).drop(uname)
+        )
+        with_other = with_other.select(*self.columns, *value_names)
         return self.__class__(with_other, other=self.other_, links=self.links_)
 
     def with_n_links(self, /, *, name: str = "n_links") -> _typing.Self:

pyproject.toml

@@ -48,7 +48,9 @@ viz = [
   "anywidget>=0.9.18",
 ]
 postal = [
-  "postal>=1.1.7; sys_platform == 'linux' or sys_platform == 'darwin'",
+  # pin to below 1.1.11 due to breaking changes in the build system
+  # https://github.com/openvenues/pypostal/commits/df602af33f28c2c9d9dd198de8764244d063734f/
+  "postal>=1.1.7,<1.1.11; sys_platform == 'linux' or sys_platform == 'darwin'",
 ]
 spacy = [
   "en-us-address-ner-sm",