GH-43374: Fix urlretrieve reporthook to report actual bytes read (#142653)

Author: savannahostrowski

Lib/test/test_urllib.py

@@ -727,7 +727,7 @@ def hooktester(block_count, block_read_size, file_size, _report=report):
         self.assertEqual(report[0][2], 8193)
         self.assertEqual(report[0][1], 8192)
         self.assertEqual(report[1][1], 8192)
-        self.assertEqual(report[2][1], 8192)
+        self.assertEqual(report[2][1], 1)  # last block only reads 1 byte
 
 
 class urlretrieve_HttpTests(unittest.TestCase, FakeHTTPMixin):

Lib/test/test_urllibnet.py

@@ -219,12 +219,14 @@ def recording_reporthook(blocks, block_size, total_size):
         self.assertEqual(records[0][2], expected_size)
         self.assertEqual(records[-1][2], expected_size)
 
-        block_sizes = {block_size for _, block_size, _ in records}
-        self.assertEqual({records[0][1]}, block_sizes,
-                         msg="block sizes in %s must be equal" % records_repr)
-        self.assertGreaterEqual(records[-1][0]*records[0][1], expected_size,
-                                msg="number of blocks * block size must be"
-                                " >= total size in %s" % records_repr)
+        self.assertEqual(records[0][1], 8192,
+                         msg="first block size should be 8192 in %s" % records_repr)
+        for block_num, block_size, total_size in records:
+            self.assertLessEqual(block_size, 8192,
+                                 msg="block size should be <= 8192 in %s" % records_repr)
+        total_read = sum(block_size for _, block_size, _ in records[1:])
+        self.assertEqual(total_read, expected_size,
+                         msg="sum of bytes read must equal total size in %s" % records_repr)
 
 
 if __name__ == "__main__":

Lib/urllib/request.py

@@ -242,7 +242,7 @@ def urlretrieve(url, filename=None, reporthook=None, data=None):
                 tfp.write(block)
                 blocknum += 1
                 if reporthook:
-                    reporthook(blocknum, bs, size)
+                    reporthook(blocknum, len(block), size)
 
     if size >= 0 and read < size:
         raise ContentTooShortError(

Misc/NEWS.d/next/Library/2025-12-12-23-17-10.gh-issue-43374.M6jGC5.rst

@@ -0,0 +1 @@
+Fix :func:`urllib.request.urlretrieve` to pass the actual number of bytes read to the *reporthook* callback, instead of always passing the block size.