Git Storage

Git storage relies extensively on the SHA-1 hash algorithm which produces a 160-bit key for each blob and each tree node.

The values are not computed on the blobs themselves, but on a composite of:

1. The utf-8 string "blob "
2. The utf-8 string for the file length
3. A null byte
4. The contents of the blob

So, these two commands produce equivalent results:

• echo -n Hello Git > text.txt && git hash-object text.txt
• printf "blob 9\0Hello Git" | sha1sum

Trees are computed from:

1. The utf-8 string "tree "
2. The utf-8 string for the content length
3. A null character
4. Zero or more content markers consisting of:
   1. The mask of the item
   2. UTF-8 space character
   3. The name of the item
   4. The 20-byte hash of the item

Commits are:

1. The utf-8 string "commit "
2. The utf-8 string for the content length
3. A null byte
4. The utf-8 string "tree "
5. The utf-8 string for the hash
6. A linefeed character
7. The utf-8 string "author "
8. 1234567890 -0800]]>
9. A linefeed character
10. The utf-8 string "committer "
11. 1234567890 -0800]]>
12. Two linefeed characters
13. The commit message
14. A linefeed character