Change zip to fail if iteratables are different lengths

Author: tomwhite

tests/test_vcf_writer.py

@@ -85,7 +85,7 @@ def test_write_vcf__filtering(tmp_path, include, exclude, expected_chrom_pos):
     assert len(variants) == len(expected_chrom_pos)
     assert v.samples == ["NA00001", "NA00002", "NA00003"]
 
-    for variant, chrom_pos in zip(variants, expected_chrom_pos, strict=False):
+    for variant, chrom_pos in zip(variants, expected_chrom_pos, strict=True):
         chrom, pos = chrom_pos
         assert variant.CHROM == chrom
         assert variant.POS == pos
@@ -141,7 +141,7 @@ def test_write_vcf__regions(tmp_path, regions, targets,
 
     assert v.samples == ["NA00001", "NA00002", "NA00003"]
 
-    for variant, chrom_pos in zip(variants, expected_chrom_pos, strict=False):
+    for variant, chrom_pos in zip(variants, expected_chrom_pos, strict=True):
         chrom, pos = chrom_pos
         assert variant.CHROM == chrom
         assert variant.POS == pos
@@ -230,7 +230,7 @@ def test_write_vcf__regions_samples_filtering(
     assert len(variants) == len(expected_chrom_pos)
     assert v.samples == ["NA00001"]
 
-    for variant, chrom_pos in zip(variants, expected_chrom_pos, strict=False):
+    for variant, chrom_pos in zip(variants, expected_chrom_pos, strict=True):
         chrom, pos = chrom_pos
         assert variant.CHROM == chrom
         assert variant.POS == pos

vcztools/query.py

@@ -99,7 +99,7 @@ def generate(root):
                     end = start + v_chunk_size
 
                     for gt_row, phase in zip(
-                        gt_zarray[start:end], phase_zarray[start:end], strict=False
+                        gt_zarray[start:end], phase_zarray[start:end], strict=True
                     ):
 
                         def stringify(gt_and_phase: tuple):
@@ -113,7 +113,7 @@ def stringify(gt_and_phase: tuple):
                             return separator.join(gt)
 
                         gt_row = gt_row.tolist()
-                        yield map(stringify, zip(gt_row, phase, strict=False))
+                        yield map(stringify, zip(gt_row, phase, strict=True))
             else:
                 # TODO: Support datasets without the phasing data
                 raise NotImplementedError
@@ -219,8 +219,8 @@ def _compose_sample_loop_generator(
 
         def generate(root):
             iterables = (generator(root) for generator in generators)
-            zipped = zip(*iterables, strict=False)
-            zipped_zipped = (zip(*element, strict=False) for element in zipped)
+            zipped = zip(*iterables, strict=True)
+            zipped_zipped = (zip(*element, strict=True) for element in zipped)
             flattened_zipped_zipped = (
                 (
                     subsubelement
@@ -309,7 +309,7 @@ def generate(root) -> str:
             filter_iterable = filter_generator(root)
 
             for results, filter_indicator in zip(
-                zip(*iterables, strict=False), filter_iterable, strict=False
+                zip(*iterables, strict=True), filter_iterable, strict=True
             ):
                 if filter_indicator:
                     results = map(str, results)

vcztools/stats.py

@@ -38,7 +38,7 @@ def stats(vcz, output):
         ).astype(np.int64)
 
         for contig, contig_length, nr in zip(
-            contigs, contig_lengths, num_records_per_contig, strict=False
+            contigs, contig_lengths, num_records_per_contig, strict=True
         ):
             if nr > 0:
                 print(f"{contig}\t{contig_length}\t{nr}", file=output)