Steps to use the git-lfs Batch API to download content from GitHub (enterprise) when the raw content is actually on git-lfs

gitlfs-github-api-content-extraction.sh

# github + git lfs api contenct extraction 
# GOAL: pull a large file that was sent to git lfs (cannot download directly from github.com)

## git-lfs API: https://github.com/git-lfs/git-lfs/tree/main/docs/api
## github API: https://docs.github.com/en/rest/reference/repos#get-repository-content

OWNER=dantonnoriega
REPO=some-repo
BRANCH=develop
# -----------------
# (1) get the raw info about the object ("Accept: application/vnd.github.v3.raw")
# - we get back a response toplined "version https://git-lfs.github.com/spec/v1"
# - this informs us that the data are in git-lfs (this makes it tricky to download)
# - note that if you try the `download_url` link from github, you'll get back nothing (because its not on github)
# -----------------
curl -s -u $GITHUB_PIE_PAT:x-oauth-basic -H "Accept: application/vnd.github.v3.raw" \
-X GET "https://github.com/api/v3/repos/$OWNER/$REPO/contents/$BRANCH/data/aggregates.csv.gz"

#> version https://git-lfs.github.com/spec/v1
#> oid sha256:29201c0ab4ba47be37ed5bf6b350XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
#> size 755

# -----------------
# (2) authenticate vs git-lfs to authorize a download 
# - "AXXXXXXXXXXXXXXXXXXXXXXXXXXXX" is a stand in
# -----------------
ssh git@github.com git-lfs-authenticate $OWNER/$REPO.git download

#> {
#>   "href": "https://github.com/_lfs/$OWNER/$REPO",
#>   "header": {
#>     "Authorization": "RemoteAuth AXXXXXXXXXXXXXXXXXXXXXXXXXXXX"
#>   },
#>   "expires_at": "2021-02-11T23:51:25Z",
#>   "expires_in": 599
#> }

# -----------------
# (2) get the raw info about the object
# -----------------
curl -s -u $GITHUB_PIE_PAT:x-oauth-basic -H "Accept: application/vnd.github.v3.raw" \
-X GET "https://github.com/api/v3/repos/$OWNER/$REPO/contents/$BRANCH/data/aggregates.csv.gz"

#> version https://git-lfs.github.com/spec/v1
#> oid sha256:29201c0ab4ba47be37ed5bf6b350XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
#> size 755

# -----------------
# (3) authenticate a download request and get token/href
# -----------------
curl -s -X POST --url "https://github.com/_lfs/$OWNER/$REPO.git/info/lfs/objects/batch" \
-H "Authorization: RemoteAuth AXXXXXXXXXXXXXXXXXXXXXXXXXXXX" \
-H "Accept: application/vnd.git-lfs+json" \
-H "Content-Type: application/vnd.git-lfs+json" \
--data '{
        "operation": "download", 
        "transfer": ["basic"], 
        "objects": [
            {
                "oid": "29201c0ab4ba47be37ed5bf6b350XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX", 
                "size": 755 
            }
        ]}'

#> {
#>   "objects": [
#>     {
#>       "oid": "29201c0ab4ba47be37ed5bf6b350XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX",
#>       "size": 755,
#>       "actions": {
#>         "download": {
#>           "href": "https://media.github.com/lfs/10943703/objects/29201c0ab4ba47be37ed5bf6b350XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX",
#>           "header": {
#>             "Authorization": "RemoteAuth AYYYYYYYYYYYYYYYYYYYYYYYYYYYY"
#>           }
#>         }
#>       }
#>     }
#>   ]
#> }

# -----------------
# (4) take new token (e.g. AYYYYYYYYYYYYYYYYYYYYYYYYYYYY) and the href and download (need --output option)
# -----------------
curl -H "Authorization: RemoteAuth AYYYYYYYYYYYYYYYYYYYYYYYYYYYY" \
--url "https://media.github.com/lfs/10943703/objects/29201c0ab4ba47be37ed5bf6b350XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX" \
--output aggregates.csv.gz

# i named the file but you can see the size is correct!
# Downloads ❯ stat -l aggregates.csv.gz
#> -rw-r--r-- 1 danton staff 755 Feb 11 15:50:49 2021 aggregates.csv.gz