@@ -296,22 +296,24 @@ def test_download_expression_with_null_uris(self):
296296
297297 def test_download_expression_with_invalid_uris (self , tmp_path ):
298298 """Test download expression with URIs that fail to download.
299-
300- This tests the exception handling in load_uri_bytes (commit c9d91080fb)
299+
300+ This tests the exception handling in load_uri_bytes
301301 where OSError is caught and None is returned for failed downloads.
302302 """
303303 # Create one valid file
304304 valid_file = tmp_path / "valid.txt"
305305 valid_file .write_bytes (b"valid content" )
306-
306+
307307 # Create URIs: one valid, one non-existent file, one invalid path
308308 table = pa .Table .from_arrays (
309309 [
310- pa .array ([
311- f"local://{ valid_file } " ,
312- f"local://{ tmp_path } /nonexistent.txt" , # File doesn't exist
313- "local:///this/path/does/not/exist/file.txt" , # Invalid path
314- ]),
310+ pa .array (
311+ [
312+ f"local://{ valid_file } " ,
313+ f"local://{ tmp_path } /nonexistent.txt" , # File doesn't exist
314+ "local:///this/path/does/not/exist/file.txt" , # Invalid path
315+ ]
316+ ),
315317 ],
316318 names = ["uri" ],
317319 )
@@ -322,28 +324,26 @@ def test_download_expression_with_invalid_uris(self, tmp_path):
322324 # Should not crash - failed downloads return None
323325 results = ds_with_downloads .take_all ()
324326 assert len (results ) == 3
325-
327+
326328 # First URI should succeed
327329 assert results [0 ]["bytes" ] == b"valid content"
328-
330+
329331 # Second and third URIs should fail gracefully (return None)
330332 assert results [1 ]["bytes" ] is None
331333 assert results [2 ]["bytes" ] is None
332334
333335 def test_download_expression_all_size_estimations_fail (self ):
334336 """Test download expression when all URI size estimations fail.
335-
336- This tests the divide-by-zero fix (commit 095973428f) where failed
337- size estimations append 0 instead of being skipped, and avg_nbytes_per_row == 0
338- is checked to prevent division by zero.
337+
338+ This tests the failed download does not cause division by zero error.
339339 """
340340 # Create URIs that will fail size estimation (non-existent files)
341341 # Using enough URIs to trigger size estimation sampling
342342 invalid_uris = [
343- f"local:///nonexistent/path/file_{ i } .txt"
343+ f"local:///nonexistent/path/file_{ i } .txt"
344344 for i in range (30 ) # More than INIT_SAMPLE_BATCH_SIZE (25)
345345 ]
346-
346+
347347 table = pa .Table .from_arrays (
348348 [pa .array (invalid_uris )],
349349 names = ["uri" ],
@@ -356,15 +356,17 @@ def test_download_expression_all_size_estimations_fail(self):
356356 # The PartitionActor should handle all failed size estimations gracefully
357357 # and fall back to using the number of rows in the block as partition size
358358 results = ds_with_downloads .take_all ()
359-
359+
360360 # All downloads should fail gracefully (return None)
361361 assert len (results ) == 30
362362 for result in results :
363363 assert result ["bytes" ] is None
364364
365- def test_download_expression_mixed_valid_and_invalid_size_estimation (self , tmp_path ):
365+ def test_download_expression_mixed_valid_and_invalid_size_estimation (
366+ self , tmp_path
367+ ):
366368 """Test download expression with mix of valid and invalid URIs for size estimation.
367-
369+
368370 This tests that size estimation handles partial failures correctly.
369371 """
370372 # Create some valid files
@@ -373,7 +375,7 @@ def test_download_expression_mixed_valid_and_invalid_size_estimation(self, tmp_p
373375 file_path = tmp_path / f"valid_{ i } .txt"
374376 file_path .write_bytes (b"x" * 100 ) # 100 bytes each
375377 valid_files .append (str (file_path ))
376-
378+
377379 # Mix valid and invalid URIs
378380 mixed_uris = []
379381 for i in range (30 ):
@@ -383,7 +385,7 @@ def test_download_expression_mixed_valid_and_invalid_size_estimation(self, tmp_p
383385 else :
384386 # Others are invalid
385387 mixed_uris .append (f"local:///nonexistent/file_{ i } .txt" )
386-
388+
387389 table = pa .Table .from_arrays (
388390 [pa .array (mixed_uris )],
389391 names = ["uri" ],
@@ -395,7 +397,7 @@ def test_download_expression_mixed_valid_and_invalid_size_estimation(self, tmp_p
395397 # Should not crash - should handle mixed valid/invalid gracefully
396398 results = ds_with_downloads .take_all ()
397399 assert len (results ) == 30
398-
400+
399401 # Verify valid URIs downloaded successfully
400402 for i , result in enumerate (results ):
401403 if i % 3 == 0 and i // 3 < len (valid_files ):
0 commit comments