* initial stuff for oauth2 login, fails on:
* login button on the signIn page to start the OAuth2 flow and a callback for each provider
Only GitHub is implemented for now
* show login button only when the OAuth2 consumer is configured (and activated)
* create macaron group for oauth2 urls
* prevent net/http in modules (other then oauth2)
* use a new data sessions oauth2 folder for storing the oauth2 session data
* add missing 2FA when this is enabled on the user
* add password option for OAuth2 user , for use with git over http and login to the GUI
* add tip for registering a GitHub OAuth application
* at startup of Gitea register all configured providers and also on adding/deleting of new providers
* custom handling of errors in oauth2 request init + show better tip
* add ExternalLoginUser model and migration script to add it to database
* link a external account to an existing account (still need to handle wrong login and signup) and remove if user is removed
* remove the linked external account from the user his settings
* if user is unknown we allow him to register a new account or link it to some existing account
* sign up with button on signin page (als change OAuth2Provider structure so we can store basic stuff about providers)
* from gorilla/sessions docs:
"Important Note: If you aren't using gorilla/mux, you need to wrap your handlers with context.ClearHandler as or else you will leak memory!"
(we're using gorilla/sessions for storing oauth2 sessions)
* use updated goth lib that now supports getting the OAuth2 user if the AccessToken is still valid instead of re-authenticating (prevent flooding the OAuth2 provider)
* Protected branches system
* Moved default branch to branches section (`:org/:reponame/settings/branches`).
* Initial support Protected Branch.
- Admin does not restrict
- Owner not to limit
- To write permission restrictions
* reformat tmpl
* finished the UI and add/delete protected branch response
* remove unused comment
* indent all the template files and remove ru translations since we use crowdin
* fix the push bug
* Markdown rendering overhaul
Cleaned up and squashed commits into single one.
Signed-off-by: Andrew Boyarshin <boyarshinand@gmail.com>
* Fix markdown API, add markdown module and API tests, improve code coverage
Signed-off-by: Andrew Boyarshin <boyarshinand@gmail.com>
This change corrects a few logging issues:
* Standardized formatting errors with '%v'.
* Standardized failure warning word usage.
* Corrected an instance of using the standard log library when
the gitea log library should be used instead.
* Moved attachaments POST url from /issues/attachments to /attachments
* Implemented attachment upload on release page
* Implemented downloading attachments on the release page
* Added zip and gzip files to default allowed attachments
* Implemented uploading attachments on edit release
* Renamed UploadIssueAttachment to UploadAttachment
* Use relative URLs
* Notifications - Mark as read/unread
* Feature of pinning a notification
* On view issue, do not mark as read a pinned notification
* Tab on user profile to show starred repos
* Make golint happy and use transactions on StarRepo function
* x -> sess
* Use sess.Close() instead of sess.Rollback()
* Add copyright
* Fix lint
* Import github.com/git-lfs/lfs-test-server as lfs module base
Imported commit is 3968aac269a77b73924649b9412ae03f7ccd3198
Removed:
Dockerfile CONTRIBUTING.md mgmt* script/ vendor/ kvlogger.go
.dockerignore .gitignore README.md
* Remove config, add JWT support from github.com/mgit-at/lfs-test-server
Imported commit f0cdcc5a01599c5a955dc1bbf683bb4acecdba83
* Add LFS settings
* Add LFS meta object model
* Add LFS routes and initialization
* Import github.com/dgrijalva/jwt-go into vendor/
* Adapt LFS module: handlers, routing, meta store
* Move LFS routes to /user/repo/info/lfs/*
* Add request header checks to LFS BatchHandler / PostHandler
* Implement LFS basic authentication
* Rework JWT secret generation / load
* Implement LFS SSH token authentication with JWT
Specification: https://github.com/github/git-lfs/tree/master/docs/api
* Integrate LFS settings into install process
* Remove LFS objects when repository is deleted
Only removes objects from content store when deleted repo is the only
referencing repository
* Make LFS module stateless
Fixes bug where LFS would not work after installation without
restarting Gitea
* Change 500 'Internal Server Error' to 400 'Bad Request'
* Change sql query to xorm call
* Remove unneeded type from LFS module
* Change internal imports to code.gitea.io/gitea/
* Add Gitea authors copyright
* Change basic auth realm to "gitea-lfs"
* Add unique indexes to LFS model
* Use xorm count function in LFS check on repository delete
* Return io.ReadCloser from content store and close after usage
* Add LFS info to runWeb()
* Export LFS content store base path
* LFS file download from UI
* Work around git-lfs client issue with unauthenticated requests
Returning a dummy Authorization header for unauthenticated requests
lets git-lfs client skip asking for auth credentials
See: https://github.com/github/git-lfs/issues/1088
* Fix unauthenticated UI downloads from public repositories
* Authentication check order, Finish LFS file view logic
* Ignore LFS hooks if installed for current OS user
Fixes Gitea UI actions for repositories tracking LFS files.
Checks for minimum needed git version by parsing the semantic version
string.
* Hide LFS metafile diff from commit view, marking as binary
* Show LFS notice if file in commit view is tracked
* Add notbefore/nbf JWT claim
* Correct lint suggestions - comments for structs and functions
- Add comments to LFS model
- Function comment for GetRandomBytesAsBase64
- LFS server function comments and lint variable suggestion
* Move secret generation code out of conditional
Ensures no LFS code may run with an empty secret
* Do not hand out JWT tokens if LFS server support is disabled
* Moved conf assets into options folder
* Dropped old bindata
* Started to integrate options bindata and accessors
* Do not enforce a builtin app.ini
* Replaced bindata calls with options
* Dropped bindata task from makefile, it's the generate task now
* Always embedd app.ini to provide sane config defaults
* Use sane defaults for the configuration
* Defined default value for SSH_KEYGEN_PATH
* Dropped "NEVER EVER MODIFY THIS FILE" header from app.ini
* Fixed new paths in latest test additions
* Drop bindata with make clean task
* Set more proper default values
Export Pusher name as GITEA_PUSHER_NAME env variable
Export also GITEA_UUID, but keep the uuid env variable for backward compatibility
export pusher name ENV variable #316
change env variable prefix to GITEA_
Signed-off-by: Antonio Facciolo <afdev82@gmail.com>
Export also GITEA_UUID #316
Keep uuid env variable for backward compatibility
* Get rid of the bin folder within the build process
Signed-off-by: Thomas Boerger <thomas@webhippie.de>
* Dropped latest make task, it is unused
Signed-off-by: Thomas Boerger <thomas@webhippie.de>
* Added tidb tag to drone config
Signed-off-by: Thomas Boerger <thomas@webhippie.de>
* Dropped the cert build tag
Signed-off-by: Thomas Boerger <thomas@webhippie.de>
* Dropped useless minwinsvc build tag
Signed-off-by: Thomas Boerger <thomas@webhippie.de>
* Dropped the useless build tags from drone config
Signed-off-by: Thomas Boerger <thomas@webhippie.de>
* Dropped unused codekit config
* Integrated dynamic and static bindata for public
* Ignore public bindata
* Add a general generate make task
* Integrated flexible public assets into web command
* Updated vendoring, added all missiong govendor deps
* Made the linter happy with the bindata and dynamic code
* Moved public bindata definition to modules directory
* Ignoring the new bindata path now
* Updated to the new public modules import path
* Updated public bindata command and drop the new prefix
- Download filename was wrong for files other than images. Example: It was `download` instead of `file.pdf`
- PDF was downloading instead of showing on browser
* Can use a predefined set of labels
* Change UI
* Fix HTML file indentation
* Avoid reading file from other directory (security issue)
* Apply a better fix
* Remove not used variable
* Merge upstream/develop
* Do modifications
* Raname
* remove binding + rename variable
* Add support for federated avatars
Fixes#3105
Removes avatar fetching duplication code
Adds an "Enable Federated Avatar" checkbox in user settings
(defaults to unchecked)
Moves avatar settings all in the same form, making
local and remote avatars mutually exclusive
Renames UploadAvatarForm to AvatarForm
as it's not anymore only for uploading
* Run gofmt on all modified files
* Move Avatar form in its own page
* Add go-libravatar dependency to vendor/ dir
Hopefully helps with accepting the contribution.
See also #3214
* Revert "Add go-libravatar dependency to vendor/ dir"
This reverts commit a8cb93ae640bbb90f7d25012fc257bda9fae9b82.
* Make federated avatar setting a global configuration
Removes the per-user setting
* Move avatar handling back to base tool, disable federated avatar in offline mode
* Format, handle error
* Properly set fallback host
* Use unsupported github.com mirror for importing go-libravatar
* Remove comment showing life exists outside of github.com
... pity, but contribution would not be accepted otherwise
* Use Combo for Get and Post methods over /avatar
* FEDERATED_AVATAR -> ENABLE_FEDERATED_AVATAR
* Fix persistance of federated avatar lookup checkbox at install time
* Federated Avatars -> Enable Federated Avatars
* Use len(string) == 0 instead of string == ""
* Move import line where it belong
See
https://github.com/Unknwon/go-code-convention/blob/master/en-US/import_packages.md
Pity the import url is still the unofficial one, but oh well...
* Save a line (and waste much more expensive time)
* Remove redundant parens
* Remove an empty line
* Remove empty lines
* Reorder lines to make diff smaller
* Remove another newline
Unknwon review got me start a fight against newlines
* Move DISABLE_GRAVATAR and ENABLE_FEDERATED_AVATAR after OFFLINE_MODE
On re-reading the diff I figured what Unknwon meant here:
https://github.com/gogits/gogs/pull/3320/files#r73741106
* Remove newlines that weren't there before my intervention
Remove the possibility of using email as user name when user actually push
through combination of email and password with HTTP.
Also refactor update action function to replcae tons of arguments with
single PushUpdateOptions struct.
And define the user who pushes code as pusher, therefore variable names shouldn't
be confusing any more.
Now, the dump cmd uses setting.CustomPath and setting.LogRootPath
instead of setting.WorkDir which was kind of broken if the gogs
binary was in a different directory than gogs data.
Additionally, the backup of setting.CustomPath directory is only done
if it exists.
Using a tmp dir makes gogs dump more robust to concurrent runs.
It also permits an easier cleaning of the tmp files (gogs-db.sql and
gog-repo.zip) by just removing the tmp dir.
As a side effect, it partially fix bugs on workdir.
Previously, 'gogs dump' created the archives in the current directory,
and tried to include these archives from the directory where the
gogs binary lies.
ex: if gogs binary is in /usr/bin/gogs, and gogs dump is run from /tmp/,
/tmp/gog-repo.zip is created, but gogs dump tried to include
/usr/bin/gogs-repo.zip.
The dump cmd did not check the return value of the z.AddFile or
z.AddDir when building the final archive.
It caused the dump command to succeed even if an error occurred.
The resulted dump archive could be corrupted/empty.
(errors could be various: removal by a concurrent process, disk full,
bugs in the dump cmd itself)
Willem van Dreumel