为何两段相似的SHA512转Base64代码执行结果不同？

Hey Josh, let's dig into why these two snippets are giving mismatched results—this is a super common pitfall with hashing and encoding workflows, so you're not alone here! Here are the most likely culprits:

• Character Encoding Discrepancies
  The first and most frequent issue is how each snippet converts the input string to a byte array before hashing. Standard practice (and what online tools use) is to use UTF-8 encoding, but if one snippet uses a different encoding (like UTF-16, ISO-8859-1, or your system's default encoding), it'll generate completely different raw bytes for the same string. For example, the string "test" in UTF-8 is 4 bytes, but in UTF-16 it's 8 bytes—so the SHA512 hash will be totally unrelated.

• Extra Hex Conversion Step
  I've seen this a lot: one snippet directly converts the SHA512 byte array to Base64, while the other first converts the hash bytes to a hexadecimal string, then converts that hex string back to bytes, and finally encodes to Base64. That extra detour changes the data entirely. Let's visualize:

  • Correct flow: Input String → UTF-8 Bytes → SHA512 Bytes → Base64 String
  • Broken flow: Input String → UTF-8 Bytes → SHA512 Bytes → Hex String → UTF-8 Bytes of Hex → Base64 String
    Online tools use the first flow, so the second will never match.

• Non-Standard Base64 Variants
  Not all Base64 implementations are identical. Some snippets might use URL-safe Base64 (which replaces + with - and / with _, and omits padding =), while others use the standard RFC 4648 Base64. If your two snippets use different variants, their output strings will differ even if the underlying hash is the same.

• Unaccounted-For Whitespace
  If one snippet trims leading/trailing whitespace from the input string (with something like trim()) and the other doesn't, you're effectively hashing two different strings. Even a single space at the start or end will make the SHA512 hash unrecognizable.

How to Debug This #

To pinpoint the exact issue, run both snippets with a simple test string (like "hello world") and print out each intermediate step:

1. The raw byte array of the input string (displayed in hex)
2. The SHA512 hash byte array (displayed in hex)
3. The final Base64 string

Compare these outputs step by step—whichever step differs first is your root cause.

内容的提问来源于stack exchange，提问作者Josh