thds.tabularasa 0.14.3__py3-none-any.whl → 0.14.4__py3-none-any.whl

thds/tabularasa/__main__.py

@@ -44,7 +44,6 @@ from thds.tabularasa.schema.util import all_predecessors, all_successors
 try:
     from bourbaki.application.cli import CommandLineInterface, cli_spec
 except ImportError:
-
     # stand-in decorators
     def noop_decorator(obj):
         return obj
@@ -84,7 +83,6 @@ else:
 try:
     from ruamel.yaml import YAML
 except ImportError:
-
     import yaml
 
     load_yaml = yaml.safe_load
@@ -578,9 +576,9 @@ class ReferenceDataManager:
                 package=self.package,
             )
         elif target == CompilationTarget.attrs_sqlite:
-            assert (
-                self.sqlite_db_path is not None
-            ), "Must specify sqlite db path in build options to generate sqlite interface"
+            assert self.sqlite_db_path is not None, (
+                "Must specify sqlite db path in build options to generate sqlite interface"
+            )
             renderer = partial(
                 render_attrs_sqlite_schema,
                 package=self.package,
@@ -595,9 +593,9 @@ class ReferenceDataManager:
     @output_handler(print_file_hashes_status)
     def check_hashes(self) -> Dict[str, DataFileHashes]:
         """Check actual hashes of on-disk built data files against those documented in the schema"""
-        assert (
-            self.package_data_dir is not None and self.transient_data_dir is not None
-        ), "Can't check hashes without package data dirs"
+        assert self.package_data_dir is not None and self.transient_data_dir is not None, (
+            "Can't check hashes without package data dirs"
+        )
         hashes = {}
         for table in self.schema.build_time_package_tables:
             name = table.name
@@ -750,9 +748,9 @@ class ReferenceDataManager:
           ensure hashes embedded in source code are up-to-date. By default, this runs when any hashes are
           updated in the config file.
         """
-        assert (
-            self.package_data_dir is not None and self.transient_data_dir is not None
-        ), "Can't update hashes without package data dirs"
+        assert self.package_data_dir is not None and self.transient_data_dir is not None, (
+            "Can't update hashes without package data dirs"
+        )
         hashes_updated = []
         tables_to_update = (
             [self.schema.tables[t] for t in tables] if tables else self.schema.build_time_package_tables

thds/tabularasa/data_dependencies/build.py

@@ -181,15 +181,15 @@ class ReferenceDataBuildCommand(setuptools.command.build_py.build_py):
 
         if self.options.sqlite_data:
             # now initialize database and load reference data into database
-            assert (
-                self.options.sqlite_db_path is not None
-            ), "No sqlite_db_path specified; can't populate db"
-            assert (
-                self.options.package_data_dir is not None
-            ), "No package_data_dir specified; can't populate db"
-            assert (
-                self.options.transient_data_dir is not None
-            ), "No transient_data_dir specified; can't populate db"
+            assert self.options.sqlite_db_path is not None, (
+                "No sqlite_db_path specified; can't populate db"
+            )
+            assert self.options.package_data_dir is not None, (
+                "No package_data_dir specified; can't populate db"
+            )
+            assert self.options.transient_data_dir is not None, (
+                "No transient_data_dir specified; can't populate db"
+            )
             populate_sqlite_db(
                 self.schema,
                 db_package=self.package_name,

thds/tabularasa/diff/schema.py

@@ -28,7 +28,9 @@ class NullabilityDiff(enum.IntEnum):
         return (
             NullabilityDiff.NO_CHANGE
             if nullable_before == nullable_after
-            else NullabilityDiff.NOT_NULL if nullable_before else NullabilityDiff.NULL
+            else NullabilityDiff.NOT_NULL
+            if nullable_before
+            else NullabilityDiff.NULL
         )
 
 
@@ -47,7 +49,9 @@ class OrderedDiff(enum.IntEnum):
         return (
             OrderedDiff.NO_CHANGE
             if ordered_before == ordered_after
-            else OrderedDiff.UNORDERED if ordered_before else OrderedDiff.ORDERED
+            else OrderedDiff.UNORDERED
+            if ordered_before
+            else OrderedDiff.ORDERED
         )
 
 

thds/tabularasa/diff/summary.py

@@ -244,6 +244,8 @@ def markdown_dataframe_diff_summary(
                             )
                             col_heading = True
                         yield markdown_heading(heading_level + 3, f"{kind}:")
-                        yield counts.to_frame("count").reset_index().to_markdown(
-                            index=False, tablefmt=tablefmt
+                        yield (
+                            counts.to_frame("count")
+                            .reset_index()
+                            .to_markdown(index=False, tablefmt=tablefmt)
                         )

thds/tabularasa/loaders/util.py

@@ -44,8 +44,10 @@ from .parquet_util import (
 # used here but they're imported for backward compatibility with
 # existing generated code, which expects it to be importable from
 # here.  They were moved to sqlite_util to reduce the size of this file.
-from .sqlite_util import AttrsSQLiteDatabase  # noqa: F401
-from .sqlite_util import sqlite_constructor_for_record_type  # noqa: F401
+from .sqlite_util import (  # noqa: F401
+    AttrsSQLiteDatabase,
+    sqlite_constructor_for_record_type,
+)
 
 T = TypeVar("T")
 K = TypeVar("K", bound=PyType)
@@ -197,12 +199,12 @@ class _PackageDataOrFileInterface:
                     f"with local path {self.data_path}"
                 )
             else:
-                assert (
-                    self.md5 is not None
-                ), f"No md5 defined for {self.data_path}; can't safely sync blob"
+                assert self.md5 is not None, (
+                    f"No md5 defined for {self.data_path}; can't safely sync blob"
+                )
                 target_local_path = self.file_path(sync=False)
                 getLogger(__name__).info(
-                    f"Syncing blob with hash {self.md5}" f" to {target_local_path}" if link else ""
+                    f"Syncing blob with hash {self.md5} to {target_local_path}" if link else ""
                 )
                 remote_data_spec = self.blob_store.data_spec(self.md5)
                 local_files = sync_adls_data(remote_data_spec)

thds/tabularasa/schema/compilation/_format.py

@@ -1,48 +1,48 @@
 import logging
 import os.path
+import subprocess
 import tempfile
 from functools import lru_cache
-from typing import Any, Callable, List, Optional, Tuple
 from warnings import warn
 
 
 @lru_cache
-def __autoformat_imports() -> Tuple[Optional[Any], Optional[Callable[[List[str]], int]]]:
+def _ruff_available() -> bool:
     try:
-        import black
-    except ImportError:
+        subprocess.run(["ruff", "--version"], capture_output=True, check=True)
+        return True
+    except (FileNotFoundError, subprocess.CalledProcessError):
         warn(
-            "`black` is unavailable; generated python code will not be auto-formatted. "
+            "`ruff` is unavailable; generated python code will not be auto-formatted. "
             "Specify the 'cli' extra to ensure this dependency is present."
         )
-        black = None  # type: ignore
-    try:
-        from isort.main import main as isort_main  # type: ignore
-    except ImportError:
-        warn(
-            "`isort` is unavailable; imports in generated python code will not be automatically sorted. "
-            "Specify the 'cli' extra to ensure this dependency is present."
-        )
-        isort_main = None  # type: ignore
-    return black, isort_main  # type: ignore
+        return False
 
 
 def autoformat(py_code: str) -> str:
     _LOGGER = logging.getLogger(__name__)
+    if not _ruff_available():
+        return py_code
+
     try:
-        black, isort_main = __autoformat_imports()
-        if black is not None:
-            _LOGGER.info("Applying `black` formatting to auto-generated code")
-            py_code = black.format_str(py_code, mode=black.FileMode())
-        if isort_main is not None:
-            _LOGGER.info("Applying `isort` formatting to auto-generated code")
-            with tempfile.TemporaryDirectory() as d:
-                outfile = os.path.join(d, "tmp.py")
-                with open(outfile, "w") as f:
-                    f.write(py_code)
-                isort_main([outfile, "--profile", "black"])
-                with open(outfile, "r") as f_:
-                    py_code = f_.read()
+        with tempfile.TemporaryDirectory() as d:
+            outfile = os.path.join(d, "tmp.py")
+            with open(outfile, "w") as f:
+                f.write(py_code)
+
+            _LOGGER.info("Applying `ruff` import sorting to auto-generated code")
+            subprocess.run(
+                ["ruff", "check", "--select", "I", "--fix", outfile],
+                capture_output=True,
+                check=False,
+            )
+
+            _LOGGER.info("Applying `ruff` formatting to auto-generated code")
+            subprocess.run(["ruff", "format", outfile], capture_output=True, check=True)
+
+            with open(outfile, "r") as f_:
+                py_code = f_.read()
+
         return py_code
     except Exception as ex:
         print(f"{repr(ex)} when attempting to format code:")

thds/tabularasa/schema/compilation/pyarrow.py

@@ -63,7 +63,7 @@ def pyarrow_field_literal(field: pa.Field) -> str:
     else:
         return (
             f'{pa.__name__}.field(\n    "{field.name}",\n'
-            f'{indent(pyarrow_type_literal(field.type), "    ")},\n    nullable={field.nullable!r},\n)'
+            f"{indent(pyarrow_type_literal(field.type), '    ')},\n    nullable={field.nullable!r},\n)"
         )
 
 

thds/tabularasa/schema/compilation/sqlite.py

@@ -31,7 +31,7 @@ def render_sql_index_schema(table: metaschema.Table) -> Optional[str]:
     if table.primary_key:
         table_constraints = (
             f"CREATE UNIQUE INDEX {index_name(table.snake_case_name, *table.primary_key)} ON "
-            f'{table.snake_case_name}({", ".join(table.primary_key)});'
+            f"{table.snake_case_name}({', '.join(table.primary_key)});"
         )
         index_defs.append(table_constraints)
 
@@ -39,7 +39,7 @@ def render_sql_index_schema(table: metaschema.Table) -> Optional[str]:
         unique = "UNIQUE " if frozenset(index) in unique_constraints else ""
         index_def = (
             f"CREATE {unique}INDEX {index_name(table.snake_case_name, *index)} "
-            f'ON {table.snake_case_name}({", ".join(index)});'
+            f"ON {table.snake_case_name}({', '.join(index)});"
         )
         index_defs.append(index_def)
 

thds/tabularasa/schema/compilation/util.py

@@ -42,7 +42,7 @@ def _dict_literal(named_exprs: Iterable[Tuple[str, str]], linebreak: bool = True
     except StopIteration:
         return "{}"
     else:
-        return f"dict({start}{sep.join(map(keyval, itertools.chain((peek,), named_exprs)))}{end})" ""
+        return f"dict({start}{sep.join(map(keyval, itertools.chain((peek,), named_exprs)))}{end})"
 
 
 def _indent(expr: str, level: int = 1, first_line: bool = False) -> str:

thds/tabularasa/schema/constraints.py

@@ -255,7 +255,9 @@ class MatchesRegex(StrConstraint):
 
             def check_fn(s: pd.Series):
                 return s.str.fullmatch(
-                    re.compile(self.matches.pattern), case=self.case_sensitive, na=True  # type: ignore[arg-type]
+                    re.compile(self.matches.pattern),  # type: ignore[arg-type]
+                    case=self.case_sensitive,
+                    na=True,
                 )
 
         else:
@@ -301,7 +303,7 @@ class EnumConstraint(ColumnConstraint):
         return pa.Check.isin(self.enum)
 
     def sqlite_check_expr(self, colname: str) -> str:
-        return f'{colname} IN ({", ".join(map(repr, self.enum))})'
+        return f"{colname} IN ({', '.join(map(repr, self.enum))})"
 
     def applies_to(self, dtype: DType) -> bool:
         if not self.enum:

thds/tabularasa/schema/metaschema.py

@@ -297,7 +297,7 @@ class UniqueColumnsConstraint(BaseModel, extra=Extra.forbid):
 
     @property
     def sqlite(self) -> str:
-        return f'UNIQUE ({", ".join(self.unique)})'
+        return f"UNIQUE ({', '.join(self.unique)})"
 
     @staticmethod
     def make_pandera_check_expr(unique: IdTuple) -> str:

thds/tabularasa/schema/validation.py

@@ -1,6 +1,7 @@
 import io
 import itertools
 import os
+from _warnings import warn
 from collections import Counter
 from functools import lru_cache
 from pathlib import Path
@@ -9,7 +10,6 @@ from typing import Any, Collection, Dict, List, Mapping, Optional, Set, Tuple, T
 import networkx as nx
 import pkg_resources
 import yaml
-from _warnings import warn
 
 from .. import git_util
 from .constraints import AnyColumnConstraint, EnumConstraint

{thds_tabularasa-0.14.3.dist-info → thds_tabularasa-0.14.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: thds.tabularasa
-Version: 0.14.3
+Version: 0.14.4
 Summary: Trilliant Health reference data build system.
 Author-email: Trilliant Health <info@trillianthealth.com>
 Project-URL: Repository, https://github.com/TrilliantHealth/ds-monorepo

{thds_tabularasa-0.14.3.dist-info → thds_tabularasa-0.14.4.dist-info}/RECORD

@@ -1,5 +1,5 @@
 thds/tabularasa/__init__.py,sha256=jc6w1WD868MQ2t4wkRNYvRssojwXvPDcNyC8V5gwbl0,169
-thds/tabularasa/__main__.py,sha256=DlaUfXu03tbBVucRuMw7354LeBs8d5tRCuAprZs0XYs,47778
+thds/tabularasa/__main__.py,sha256=ooLtYWjH30oGHhOg1UFTpwTICG30_9h88JnRE0j0QBs,47776
 thds/tabularasa/compat.py,sha256=j0313TPIXtkbfvRI0AH4if8GLrjQSrDJ9heayCIl9w8,1037
 thds/tabularasa/git_util.py,sha256=fBFhaCPi_5W2BpG2B3WiPcAWJvuVI_pG47rt73wLO6E,1388
 thds/tabularasa/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -8,41 +8,41 @@ thds/tabularasa/sqlite_from_parquet.py,sha256=yJatUIAbgErHUOL5dhchWJwzKZCrDrx93S
 thds/tabularasa/to_sqlite.py,sha256=5lcEUh38MNebxAJdLp2XGWOP_WQDIADtL1fyhOvi9UU,1715
 thds/tabularasa/data_dependencies/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 thds/tabularasa/data_dependencies/adls.py,sha256=vJAuc5Key-vO1N6DGo5dj9fIx_4hMALAVC17qhvkT7Y,3257
-thds/tabularasa/data_dependencies/build.py,sha256=6iYgw93sOF2Nlnb6WSmA9NvPNwOf_Yyi2wXUQpRVkJM,23382
+thds/tabularasa/data_dependencies/build.py,sha256=sFxlMtjluje5t2TP3oYVV4TdMOv85gdJSBQ4E7R0JuE,23382
 thds/tabularasa/data_dependencies/sqlite.py,sha256=sMP_NInBEDoH5SScIRYxtOvcPUi9WXfE3_jCoOBduGo,12825
 thds/tabularasa/data_dependencies/tabular.py,sha256=oq9wFse235ikLEv8Zvol59ptRRojZbkbzXJyQeFfC9o,6529
 thds/tabularasa/data_dependencies/util.py,sha256=FQ9G1nIpqKh00z2lXOt0Y2R1mLQsEb-BC6Tka1z2egc,8489
 thds/tabularasa/diff/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 thds/tabularasa/diff/data.py,sha256=J1YK9thlXRn4WYQjCKG5kgkGPPRp2wu9rPjkLbyIi3E,11680
-thds/tabularasa/diff/schema.py,sha256=I-RC7yQuLWj5DqxWMWzQgMecT6LanJ2k1bH9-IJELcg,8855
-thds/tabularasa/diff/summary.py,sha256=gENtDwhSrDYeN-8fWr6Ug2zgdp584b0pZF9UBYzKFlc,10039
+thds/tabularasa/diff/schema.py,sha256=obd3HwjMB8P6WXk9ETui7HSvJ7xb4r5sOGXEMCl_PZk,8903
+thds/tabularasa/diff/summary.py,sha256=5D5ZsLiJpsiz7spqOjqFJEzG8C-6DD8e-KjJnfzKrPU,10099
 thds/tabularasa/loaders/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 thds/tabularasa/loaders/lazy_adls.py,sha256=jrWy5tTKDQfWEv6aHQ3UJhFzLrDPOlSGsArv9zcl1g8,1375
 thds/tabularasa/loaders/parquet_util.py,sha256=u75j3PkMSakO2zfq4zksWzXLYnaO--WizAgXTcSpXRY,13354
 thds/tabularasa/loaders/sqlite_util.py,sha256=etLiV0h1l7NQHaEdNjz0QtqYaRBSm5L6EVWOwnCRv2w,11185
-thds/tabularasa/loaders/util.py,sha256=XmsGkDdL6O8R6B4667Iqi5HoRgq0YMs6LP3VvPIqPVU,21369
+thds/tabularasa/loaders/util.py,sha256=2hBUEtforxBxwhpb_xqFRZI2u4zcAYnYbFnkqPWTZTM,21340
 thds/tabularasa/schema/__init__.py,sha256=bowvNXrrDrWB3TAmwDxCeEAvVEe9z7iRfqRaNg1Qmo4,440
-thds/tabularasa/schema/constraints.py,sha256=V2vh01BhYR8OVQvgdujqSi0l_fMJvFKYSlBvWExZFG0,9744
+thds/tabularasa/schema/constraints.py,sha256=hdRpdWdzIZxS7XVKjyA69BPwWgWRxFE6XsO-2oSxfCY,9785
 thds/tabularasa/schema/dtypes.py,sha256=wHoDXiQoZ5_wC5K6b_TnPkmhtkrSMYmjZQ8TNault0I,4816
 thds/tabularasa/schema/extract_from_parquet.py,sha256=WBOr0Vr2BFShfIyE3q7yXzgGhtTTyFYi3WV4lqr4MME,4730
 thds/tabularasa/schema/files.py,sha256=gvDaeM9a0W7SpOBICw2WvWzepaRumu6c5ijAiXFeY_8,7604
-thds/tabularasa/schema/metaschema.py,sha256=dPcyfVoeFOooRdOm-oliznU9NUkgkbNhwgFUXd2--eU,36997
+thds/tabularasa/schema/metaschema.py,sha256=yt4ByweXfqHa2dYF8izG3BAhiBw8iQ1Dmmu_UrTdxhM,36997
 thds/tabularasa/schema/util.py,sha256=3Rv3Mf1-Tw0obmq9vWoc2bQr9sy0viqn48a41c-RK84,4265
-thds/tabularasa/schema/validation.py,sha256=5ESI1SCj8wgYiRaDP3oxtLgdbyhOShN_wwD5zSng7js,32193
+thds/tabularasa/schema/validation.py,sha256=Ek3L2r2aSm-X4KiLuLT9RFPaVU_p_iNbLCQJIy4q2zU,32193
 thds/tabularasa/schema/compilation/__init__.py,sha256=kOX2fFyPsWfidxW1lYs1-oiKH_DsA1nPKalppf7gLWQ,636
-thds/tabularasa/schema/compilation/_format.py,sha256=RayqDj-SXTtEFfPLyYvTRC1sL82LNVcOol40w5s8Qz8,1854
+thds/tabularasa/schema/compilation/_format.py,sha256=ibAszFBpHESb1YjiUAYyDrEvUe4lLRKXr5yXKEf1U8k,1521
 thds/tabularasa/schema/compilation/attrs.py,sha256=u6sTIajs2uEfbsRJcst05C-ekBS_-bsvHwJK8l0Q_ZQ,7749
 thds/tabularasa/schema/compilation/attrs_sqlite.py,sha256=ayWOXgw7iuZtxkS9CHwjjtjUr4Wn8Vuus3tuiEbBiOk,10148
 thds/tabularasa/schema/compilation/io.py,sha256=ehhVP0On0XwIm6qWSCO5uo7r5CjpThpBRLYKAqrb40I,3199
 thds/tabularasa/schema/compilation/pandas.py,sha256=6yxXTTIlcTXsttVh9uRMccetty6OrvdbmRXgbvvHdyA,8176
-thds/tabularasa/schema/compilation/pyarrow.py,sha256=pcNQ3a6UPJT1PBj6xHOl99UvZftp1hqV5OFdFbT0OHU,3030
+thds/tabularasa/schema/compilation/pyarrow.py,sha256=G3ItYcK2R2bkBdHaIkml6zT9z72FtR9OJQAK8YnCwyE,3030
 thds/tabularasa/schema/compilation/sphinx.py,sha256=we5X-ZpCk6WH-8KCXAv6Nklg1JZmnkGPT3V2EHa2_rg,17491
-thds/tabularasa/schema/compilation/sqlite.py,sha256=wSrSlVCYeuTpOf9AOHAnp6gJHkjHZhx8UkgkYgfoQVw,2368
-thds/tabularasa/schema/compilation/util.py,sha256=YXFe1_yoBobED010hstKIoq-dwLHo6SBv1v1IAw6AYU,3886
+thds/tabularasa/schema/compilation/sqlite.py,sha256=-_xdz8cUkOTOczu3OYoIoL8TNqi6JJ4rAkrEdu6QR2w,2368
+thds/tabularasa/schema/compilation/util.py,sha256=kCJQxjOe_VnQtKFG7p3162HOZFzTTc1eVBhdrRsvYqY,3883
 thds/tabularasa/testing/__init__.py,sha256=XoLzB-DotxFw9KHt2vfH72k7pyAAFI2bW-qqq6nww1g,85
 thds/tabularasa/testing/mock_sqlite.py,sha256=xoV4w_GaDgtZf17iUux2-LA6Va1XRJdC2FU34dysh0o,4769
-thds_tabularasa-0.14.3.dist-info/METADATA,sha256=WfrgVRGa1Te2GYA4qFHRogd9-H8ZPJRsTNVOwhKbL1Q,26786
-thds_tabularasa-0.14.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-thds_tabularasa-0.14.3.dist-info/entry_points.txt,sha256=PX4ShRonjv6lMsVjrGu8RkFzpyyvgM9EnZlNfMomd9k,61
-thds_tabularasa-0.14.3.dist-info/top_level.txt,sha256=LTZaE5SkWJwv9bwOlMbIhiS-JWQEEIcjVYnJrt-CriY,5
-thds_tabularasa-0.14.3.dist-info/RECORD,,
+thds_tabularasa-0.14.4.dist-info/METADATA,sha256=8ZjGMMWzgAC0V37yG_EwzXE9qb3_pvScmkNXOQ4xZ4I,26786
+thds_tabularasa-0.14.4.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+thds_tabularasa-0.14.4.dist-info/entry_points.txt,sha256=PX4ShRonjv6lMsVjrGu8RkFzpyyvgM9EnZlNfMomd9k,61
+thds_tabularasa-0.14.4.dist-info/top_level.txt,sha256=LTZaE5SkWJwv9bwOlMbIhiS-JWQEEIcjVYnJrt-CriY,5
+thds_tabularasa-0.14.4.dist-info/RECORD,,

{thds_tabularasa-0.14.3.dist-info → thds_tabularasa-0.14.4.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.9.0)
+Generator: setuptools (80.10.2)
 Root-Is-Purelib: true
 Tag: py3-none-any