sas2parquet 0.1.1__py3-none-any.whl → 0.1.5__py3-none-any.whl

sas2parquet/convert.py

@@ -317,7 +317,27 @@ def reconvert_file_ultimate(sas_path: Path, parquet_path: Path) -> bool:
             writer.close()
             print("   ✅ Conversion succeeded")
 
+            # ===== FULL PARQUET VALIDATION (WORKING) =====
+            print("   🔍 Full Parquet validation...")
+            try:
+                pf = pq.ParquetFile(parquet_path)
+                total_rows = 0
+                num_groups = pf.metadata.num_row_groups
+                batch_count = 0
+                
+                for batch in pf.iter_batches():
+                    total_rows += batch.num_rows
+                    batch_count += 1
+                
+                print(f"   ✅ Parquet fully validated: {total_rows:,} rows across {num_groups} groups ({batch_count} batches)")
+                pf.close()
+            except Exception as e:
+                print(f"   ❌ Parquet validation failed: {e}")
+                return False
+            # ===== END =====
+
             st, dt = compare_and_report_diffs(sas_path, parquet_path)
+
             print(f"   🔍 Validation: {st}")
             for d in dt:
                 print("     -", d.replace("\n", "\n       "))

{sas2parquet-0.1.1.dist-info → sas2parquet-0.1.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sas2parquet
-Version: 0.1.1
+Version: 0.1.5
 Summary: SAS → Parquet Hybrid Converter & Validator
 License-File: LICENSE
 Author: Zaman Ziabakhshganji
@@ -128,37 +128,6 @@ Each file undergoes 4-stage validation:
 3. Column order
 4. Value comparison
 
-## 💻 Development
-```bash
-git clone <repo>
-cd sas2parquet
-pip install -e .
-```
-
-
-## 🧪 Testing
-
-This project includes a comprehensive **pytest** test suite covering:
-
-- Schema inference and overrides
-- SAS ↔ Parquet validation logic
-- Error handling and retry behavior
-- Directory mirroring (`sasdata/ → parquetdata/`)
-- Logging and CLI execution paths
-
-### Run tests
-
-```bash
-pip install -e .[dev]
-pytest -q
-```
-
-## 📦 Dependencies
-- pandas>=2.0
-- polars>=0.20
-- pyarrow>=15.0
-- pyreadstat>=1.4
-
 ## 📄 License
 MIT License
 

sas2parquet-0.1.5.dist-info/RECORD

@@ -0,0 +1,8 @@
+sas2parquet/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+sas2parquet/cli.py,sha256=Gap4-lrHE3a-tOTRljqmRP6uX-epLFHnOYN0pdffU-g,879
+sas2parquet/convert.py,sha256=bHfdtaI1sSeCSHKqARuuvbwgrNfnH1G_odKSEmVbNrI,14736
+sas2parquet-0.1.5.dist-info/METADATA,sha256=nJNYKUGDINBS4M0Et6r2NRruczmd_v1NepPi5xdTHnU,3786
+sas2parquet-0.1.5.dist-info/WHEEL,sha256=3ny-bZhpXrU6vSQ1UPG34FoxZBp3lVcvK0LkgUz6VLk,88
+sas2parquet-0.1.5.dist-info/entry_points.txt,sha256=pg57h0xD_3R9ZC_YfxLLfu_2p1JNhF8xDNS6v7kiSBY,52
+sas2parquet-0.1.5.dist-info/licenses/LICENSE,sha256=ouRycIMUGF1zCj49-ijn1wIlTNknZEoLwAHUp0ifH-g,1066
+sas2parquet-0.1.5.dist-info/RECORD,,

sas2parquet-0.1.1.dist-info/RECORD

@@ -1,8 +0,0 @@
-sas2parquet/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-sas2parquet/cli.py,sha256=Gap4-lrHE3a-tOTRljqmRP6uX-epLFHnOYN0pdffU-g,879
-sas2parquet/convert.py,sha256=36hsDLM0uQF2tuwd9U8k3U5zEZtwQcEJWta3v4SkoXc,13950
-sas2parquet-0.1.1.dist-info/METADATA,sha256=LSc1S3ehB21qDV_s9cMdga27FUd3bQ8DUAUvWnmvIfY,4293
-sas2parquet-0.1.1.dist-info/WHEEL,sha256=3ny-bZhpXrU6vSQ1UPG34FoxZBp3lVcvK0LkgUz6VLk,88
-sas2parquet-0.1.1.dist-info/entry_points.txt,sha256=pg57h0xD_3R9ZC_YfxLLfu_2p1JNhF8xDNS6v7kiSBY,52
-sas2parquet-0.1.1.dist-info/licenses/LICENSE,sha256=ouRycIMUGF1zCj49-ijn1wIlTNknZEoLwAHUp0ifH-g,1066
-sas2parquet-0.1.1.dist-info/RECORD,,