Course-notes-md

# git under the hood

## About hashes

------

### The sha1 hash

```
❯ echo "Hello, Git" | sha1sum
c9d5d04925b93d2fb99c73ab2b5869bde7405ca4  -
```

Note that `echo` will add a line break into the string

To avoid the new line, add the `-n` switch:

```
❯ echo -n "Hello, Git" | sha1sum
0a2b198f595e55060dec9f0e196c10de86f2ca1c  -

❯ echo -n "Hello, Git!" | sha1sum
1d4d7d92f79dc328154dc91424e6e740f8f5a563  -
```

For the two examples above, note that one single character change to the string will produce a completely different hash.

Also remember that the hash for the string "Hello, Git" is 0a2b198f595e55060dec9f0e196c10de86f2ca1c.

Later we will calculate this hash again, but using `git hbas-object`, and the hashes will not match. We will understand why.

--------

### how many sha1 hashes can be generated

git uses the sha1 algorithm to generate its hash. It outputs a 160 bit, 40 hexadecimanl characters hash.

Meaning a total probabilities of 2^160 objects

```
❯ echo "2^160" | bc
1461501637330902918203684832716283019655932542976
(2^160 yelds 1.461.501.637.330.902.918.203.684.832.716.283.019.655.932.542.976 combinations)
```
------- 

### how long until two sha1 hashes collide - or probability of the same hash for 2 different strings or files

Probability of two sha1 hashes coliding is

1/(2^320) = 4.6816764e-97 or 0.[96 zeros]4%

----

## the .git folder

### the `objects` folder

Object types

- Blob
  Files with any extensions

- Tree
  Info about directories (folders)

- Commit
  Stores different verions of the project

- Annotaded tag
  Persistent text pointer to a commit

-----

## git low-level commands

- `git hash-object`

- `git cat-files`

- 

----

### `git hash-object`

```
❯ echo "Hello, Git" | git hash-object --stdin
b7aec520dec0a7516c18eb4c68b64ae1eb9b5a5e
```

b7aec520dec0a7516c18eb4c68b64ae1eb9b5a5e is the hash for string the "Hello, Git".

Nothing has changed inside .git folder, yet.

```
❯ ls -la .git/objects/
total 16
drwxrwxr-x 4 yuri yuri 4096 mar 27 14:09 .
drwxrwxr-x 7 yuri yuri 4096 mar 27 14:09 ..
drwxrwxr-x 2 yuri yuri 4096 mar 27 14:09 info
drwxrwxr-x 2 yuri yuri 4096 mar 27 14:09 pack
```
-------- 

### Calculating hash and storing objects on .git file system

Adding the `-w` option to git hash-object.

More info with `❯ man git hash-object`

```
❯ echo "Hello, Git" | git hash-object --stdin -w
b7aec520dec0a7516c18eb4c68b64ae1eb9b5a5e
```

Checking again `.git/objects`

A new folder appeared inside `.git/objects`

```
❯ ls -la .git/objects/
total 20
drwxrwxr-x 5 yuri yuri 4096 mar 27 14:07 .
drwxrwxr-x 7 yuri yuri 4096 mar 27 13:50 ..
drwxrwxr-x 2 yuri yuri 4096 mar 27 14:07 b7
drwxrwxr-x 2 yuri yuri 4096 mar 27 13:50 info
drwxrwxr-x 2 yuri yuri 4096 mar 27 13:50 pack
```

And inside the `b7` folder

```
❯ ls -la .git/objects/b7/
total 12
drwxrwxr-x 2 yuri yuri 4096 mar 27 14:09 .
drwxrwxr-x 5 yuri yuri 4096 mar 27 14:09 ..
-r--r--r-- 1 yuri yuri   27 mar 27 14:09 aec520dec0a7516c18eb4c68b64ae1eb9b5a5e
```

Note that:

The string "Hello, Git" will yeld hash

```
❯ echo "Hello, Git" | git hash-object --stdin
b7aec520dec0a7516c18eb4c68b64ae1eb9b5a5e
```

Now take a look at the folder name that appeared on the `.git/objects` folder

`drwxrwxr-x 2 yuri yuri 4096 mar 27 14:07 b7`

And the new object name (file) is:

`-r--r--r-- 1 yuri yuri 27 mar 27 14:09 aec520dec0a7516c18eb4c68b64ae1eb9b5a5e`

Thus we can conclude that on the git filesystem

- the first 2 characteres from the hash is the folder name
- and the filename is the hash however without folder's name

------ 

## `git cat-files`

Options

- `-p` Object contents

Returns object contents

`❯ git cat-file -p b7aec520dec0a7516c18eb4c68b64ae1eb9b5a5e
Hello, Git`

- `-s` Object size

Returns object size in bytes

`❯ git cat-file -s b7aec520dec0a7516c18eb4c68b64ae1eb9b5a5e
11`

- `-t` Object type

Returns which type of object the hash is:
- Blob
- Tree
- Commit
- Annotaded tag

`❯ git cat-file -t b7aec520dec0a7516c18eb4c68b64ae1eb9b5a5e
blob`

---- 

### Storing files outside the repository folder on the git filesystem

```
❯ echo "this file is outside the repository folder" >  ~/Desktop/new-file.txt

❯ pwd && git hash-object ../new-file.txt -w
/home/yuri/Desktop/first-project
d83851530645f5bd528944c5c6f7c241738f389c

❯ ls -la .git/objects/
total 24
drwxrwxr-x 6 yuri yuri 4096 mar 27 16:00 .
drwxrwxr-x 7 yuri yuri 4096 mar 27 14:09 ..
drwxrwxr-x 2 yuri yuri 4096 mar 27 14:09 b7
drwxrwxr-x 2 yuri yuri 4096 mar 27 16:00 d8   << new folder
drwxrwxr-x 2 yuri yuri 4096 mar 27 14:09 info
drwxrwxr-x 2 yuri yuri 4096 mar 27 14:09 pack

❯ ls -la .git/objects/d8/
total 12
drwxrwxr-x 2 yuri yuri 4096 mar 27 16:00 .
drwxrwxr-x 6 yuri yuri 4096 mar 27 16:00 ..
-r--r--r-- 1 yuri yuri   57 mar 27 16:00 3851530645f5bd528944c5c6f7c241738f389c << new object inside new folder

❯ git cat-file -p d838
this file is outside the repository folder

❯ git cat-file -s d838
43

❯ git cat-file -t d838
blob

```

But can we simply `cat` a git object file?    

**No**

```
❯ file .git/objects/b7/aec520dec0a7516c18eb4c68b64ae1eb9b5a5e 
.git/objects/b7/aec520dec0a7516c18eb4c68b64ae1eb9b5a5e: zlib compressed data

❯ file .git/objects/d8/3851530645f5bd528944c5c6f7c241738f389c 
.git/objects/d8/3851530645f5bd528944c5c6f7c241738f389c: zlib compressed data
```

------

## git objects

Remember that git uses sha1 sum to create hashes.

But when we calculated sha1 hash for "Hello, Git"

```
❯ echo -n "Hello, Git" | sha1sum
0a2b198f595e55060dec9f0e196c10de86f2ca1c  -
```

and when we created a git object with contents "Hello, Git", we got a different hash

```
❯ echo "Hello, Git" | git hash-object --stdin
b7aec520dec0a7516c18eb4c68b64ae1eb9b5a5e
```

That is because a git object is made of

- type
  - blob
  
- size
  - 11
 
- terminator
  - \0 (null character)
 
- content 
  - "Hello, Git"

Let's calculate the hash emulating a complete git object

```
❯ echo -e "blob 11\0Hello, Git" | sha1sum 
b7aec520dec0a7516c18eb4c68b64ae1eb9b5a5e  -
```

There it is.

------
#### Remembering About hashes

git is capable of storing 1.461.501.637.330.902.918.203.684.832.716.283.019.655.932.542.976 objects on each repository.