Guarded with an --incremental flag.
Stores the time of the last update and only downloads issue and
pull request data since this time. All other data is relatively
small (likely fetched with a single request) and so is simply
re-populated from scratch as before.
This prevents storing pull requests twice since the Github API returns
pull requests also as issues. Those issues will be skipped but only if
retrieving pull requests is requested as well.
Closes#23.
Reporting an error when the user's rate limit is exceeded causes
the script to terminate after resuming execution from a rate limit
sleep. Instead of generating an explicit error we just want to
inform the user that the script is going to sleep until their rate
limit count resets.
The errors list was not being cleared out after resuming a backup
from a rate limit sleep. When the backup was resumed, the non-empty
errors list caused the backup to quit after the next `retrieve_data`
request.
This makes it easier for the user to identify which repository does not
exist or is not initialised, i.e. whether it is the main repository or
the wiki repository and which clone URL was used to check.
When running unauthenticated and Github starts rate-limiting the client,
github-backup crashes because the used auth variable in _get_response()
was not available. This change should fix it.
* Now wikis will follow the same logic as the main repo
checkout for --prefer-ssh.
* The regular repository and wiki paths both use the same
function to handle either cloning or updating a local copy
of the remote repo
* All git updates will now use “git fetch --all --tags”
to ensure that tags and branches other than master will
also be backed up
Previously git clones could fail without any indication
unless you edited the source to change `logger=None` to use
a configured logger.
Now a non-zero return code will always output a message to
stderr and will display the executed command so it can be
rerun for troubleshooting.
The previous commit used the wrong URL for a private repo. This was
masked by the lack of error loging in logging_subprocess (which will be
in a separate branch)
Back up reporitory pull requests by passing the --include-pulls
argument. Pull requests are saved to
repositories/<repository name>/pulls/<pull request number>.json. Include
the --pull-request-comments argument to add review comments to the pull
request backup and pass the --pull-request-commits argument to add
commits to the pull request backup.
Pull requests are automatically backed up when the --all argument is
uesd.
Pass the -H or --github-host argument with a GitHub Enterprise hostname
to backup from that GitHub enterprise host. If no argument is passed
then back up from github.com.
