lightweight-table-diff 0.1.2__tar.gz → 0.1.4__tar.gz

{lightweight_table_diff-0.1.2 → lightweight_table_diff-0.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lightweight-table-diff
-Version: 0.1.2
+Version: 0.1.4
 Summary: Cell-level table diffing for Polars
 Requires-Python: >=3.11.1
 Description-Content-Type: text/markdown

{lightweight_table_diff-0.1.2 → lightweight_table_diff-0.1.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "lightweight-table-diff"
-version = "0.1.2"
+version = "0.1.4"
 description = "Cell-level table diffing for Polars"
 readme = "README.md"
 requires-python = ">=3.11.1"

{lightweight_table_diff-0.1.2 → lightweight_table_diff-0.1.4}/src/lightweight_table_diff/core.py

@@ -20,7 +20,7 @@ def diff_tbls(
     after: pl.LazyFrame,
     id_cols: list[str],
     compare_cols: list[str] | None = None,
-    join_type: str = "full",
+    join_type: str = "inner",
 ) -> pl.LazyFrame:
     if compare_cols is None:
         compare_cols = list(before.drop(*id_cols).collect_schema().keys())
@@ -36,8 +36,8 @@ def diff_tbls(
         pl.when(~pl.col(f"b__{c}").eq_missing(pl.col(f"a__{c}")))
         .then(
             pl.struct(
-                pl.col(f"b__{c}").alias("before_val"),
-                pl.col(f"a__{c}").alias("after_val"),
+                pl.col(f"b__{c}").cast(pl.String).alias("before_val"),
+                pl.col(f"a__{c}").cast(pl.String).alias("after_val"),
             )
         )
         .otherwise(None)
@@ -69,7 +69,7 @@ def batch_diff_tbls(
     id_cols: list[str],
     compare_cols: list[str] | None = None,
     batch_size: int = 50,
-    join_type: str = "full",
+    join_type: str = "inner",
 ) -> pl.LazyFrame:
     if compare_cols is None:
         compare_cols = list(before.drop(*id_cols).collect_schema().keys())

{lightweight_table_diff-0.1.2 → lightweight_table_diff-0.1.4}/src/lightweight_table_diff/runner.py

@@ -32,14 +32,14 @@ class ComparisonResult:
     added_cols: list[str]
 
 
-def run_comparison(job: dict, **context) -> ComparisonResult:
+def run_comparison(job: dict, **kwargs) -> ComparisonResult:
     job = copy.deepcopy(job)
     name = job.get("name", "unnamed")
     keys = job["join_keys"]
     logger.info("Running comparison: %s", name)
 
-    before = adapters.load(job["before"], **context)
-    after = adapters.load(job["after"], **context)
+    before = adapters.load(job["before"], **kwargs)
+    after = adapters.load(job["after"], **kwargs)
 
     removed_cols, added_cols = column_indels(before, after, keys)
     if removed_cols:
@@ -72,7 +72,7 @@ def run_comparison(job: dict, **context) -> ComparisonResult:
         keys,
         cols,
         batch_size=job.get("batch_size", 50),
-        join_type=job.get("join_type", "full"),
+        join_type="inner",
     )
 
     n = diff.select(pl.len()).collect().item()
@@ -137,11 +137,11 @@ def write_results(result: ComparisonResult, output_dir: str | Path) -> None:
     logger.info("  Wrote %s", summary_path)
 
 
-def run_config(config_path: str | Path, **context) -> list[ComparisonResult]:
+def run_config(config_path: str | Path, **kwargs) -> list[ComparisonResult]:
     jobs = load_config(config_path)
     results: list[ComparisonResult] = []
     for job in jobs:
-        result = run_comparison(job, **context)
+        result = run_comparison(job, **kwargs)
         write_results(result, job.get("output_dir", "./diff_output"))
         results.append(result)
     return results

{lightweight_table_diff-0.1.2 → lightweight_table_diff-0.1.4}/src/lightweight_table_diff.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lightweight-table-diff
-Version: 0.1.2
+Version: 0.1.4
 Summary: Cell-level table diffing for Polars
 Requires-Python: >=3.11.1
 Description-Content-Type: text/markdown