Erm, it's matching against the 8 hex digits of the CRC-32. How is that affected by the file size?

If the file size is 10000000 bytes or higher, your original regexp will not match against the 8 digits of the CRC-32 as you say. Instead, it will match against the 8 digits of the file size, because they come earlier. That's why I used the date as the anchor.