CODEWITHSUNDEEP
githashsha1
Ian Mak
Ian Mak

Reputation: 3

git hash-object vs shasum doesn't match when the content starts with a number

I am trying to work on a custom solution to connect to a git API, and when I do a shasum on content that starts with letters, the results match:

$ echo -e -n "blob 4\0test" | shasum
30d74d258442c7c65512eafab474568dd706c430 *-

$ echo -e -n "test" | git hash-object --stdin
30d74d258442c7c65512eafab474568dd706c430

It doesn't match when the content starts with a number though:

$ echo -e -n "blob 5\0test" | shasum
315a7861230f24fade469e87c0c548f0cc4bc8c8 *-

$ echo -e -n "0test" | git hash-object --stdin
475a3cf5e5dadf80fe51cc8748c9bfdabae29f07

I'm stumped as to why it's giving different results, anybody know why this is the case?

Upvotes: 0

Views: 108

Answers (1)

bk2204
bk2204

Reputation: 76409

When you use echo -e, you're providing an octal escape when you write \0. Adding additional digits after that causes them to be interpreted as octal digits in the octal escape.

Since POSIX doesn't define echo -e and it isn't portable, it's better to use printf, since you know that the octal escape will be interpreted consistently and can contain at most three digits:

$ printf "blob 5\0000test" | shasum
475a3cf5e5dadf80fe51cc8748c9bfdabae29f07  -
$ printf "0test" | git hash-object --stdin
475a3cf5e5dadf80fe51cc8748c9bfdabae29f07

In the first invocation, the \0000 is the octal escape \000 (a NUL byte) plus the character 0 (decimal 48).

Upvotes: 1

Related Questions