Prevent dropping of users when publishing a DACPAC using SqlPackage.exe - sql-server-data-tools

Is there any way of preventing users being dropped when publishing a DACPAC using SqlPackage.exe, other than changing the setting below, which prevents all objects from being dropped if they're not in the DACPAC.
<DropObjectsNotInSource>True</DropObjectsNotInSource>
We deploy to a number of environments, each with different users. Current workarounds are to either:
Script the users for each environment to recreate them after deploying
Use /Action:Script and manually change the deployment script.
Neither of these are ideal though...

Use SqlPackage.exe parameters (since February 2015 release: New Advanced Publish Options to Specify Object Types to Exclude or Not Drop):
Here's the actual parameters we use in our deployment:
/p:DropObjectsNotInSource=True
/p:ExcludeObjectTypes=Users;Logins;RoleMembership;Permissions
The first line cleans all, but the next line further refines what not to drop. This combination proved the most effective with us to drop all unnecessary objects, yet retain the login mappings as they were.
Detailed documentation of all the parameters and their possible values can be found from MSDN - SqlPackage.exe

I ran into the same issue and used Pre/Post deployment scripts to reinsert users, permissions, roles, etc like the suggested blog post. However this became unmaintainable in the long run (users unable to authenticate during deployment, if the deployment fails permissions are not restored, security changes require going through source control and re-deployment).
Recently, I reevaluated the problem as we were migrating our deployment platform. With the DacFx API (and bug fixes) released, I was able to extend the deployment process in SSDT by creating a DeploymentPlanModifier. They provide an example for filtering objects on creation, with simple modifications I filter any drops for permission based object types (using /p:AdditionalDeploymentContributors argument).
[ExportDeploymentPlanModifier( UserMappingFilter.PlanFiltererContributorId, "1.0.0.0" )]
public class UserMappingFilter : DeploymentPlanModifier
{
public const string PlanFiltererContributorId = "Dac.UserMappingFilter";
protected override void OnExecute( DeploymentPlanContributorContext context )
{
DeploymentStep next = context.PlanHandle.Head;
while( next != null )
{
DeploymentStep current = next;
next = current.Next;
DropElementStep dropStep = current as DropElementStep;
if( dropStep != null && ShouldFilter( dropStep ) )
{
base.Remove( context.PlanHandle, dropStep );
}
}
}
private bool ShouldFilter( DropElementStep createStep )
{
TSqlObject target = createStep.TargetElement;
if( target.ObjectType.Name == "RoleMembership" || target.ObjectType.Name == "User" || target.ObjectType.Name == "Role" )
{
return true;
}
return false;
}
}

We handle this in post-deploy scripts. It's a bit harder to set up, but once set up allows you to configure a slightly different script for each environment. We use this in conjunction with Publish Profiles with a different profile per environment. Basically, you use Powershell to generate a bunch of scripts for users and permissions, add those scripts to your project(s), and then Include the files in the project. Add what is referred to in the blog post as "SecurityAdditionsWrapper.sql" to your post-deploy script, and you should be good. Just remove the other security from your project to ensure that it's set correctly.
http://schottsql.blogspot.com/2013/05/ssdt-setting-different-permissions-per.html
There are also options in SSDT for:
"Drop Permissions not in source" - False
"Drop role members not defined in source" - False
"Ignore permissions" - True
"Ignore role membership" - True
We use those, but if you need better control over your users/permissions by environment, I'd strongly recommend checking out that blog post. (With thanks to Jamie Thomson for the original idea.)

Related

Is possible to modify arcanist/differential template?

I'm trying to configure a phabricator instance, and I find that change the arcanist default template when we use arc diff can be very useful for the team.
Actually the template contains this text:
<<Replace this line with your Revision Title>>
Summary:
Test Plan:
Reviewers:
Subscribers:
# Tip: Write "Fixes T123" in your summary to automatically close the
# corresponding task when this change lands.
# NEW DIFFERENTIAL REVISION
# Describe the changes in this new revision.
#
# arc could not identify any existing revision in your working copy.
# If you intended to update an existing revision, use:
#
# $ arc diff --update <revision>
I'm googling to find any way to change this default template, but I can't find it...
There is any way to "personalize" this template?
As reported in Phabricator Task T12276 by a question from user #milianw, actually it seems there is not the ability to customize the commit message.
This is the official reason:
Please keep in mind Phabricator is an enterprise tool, and
the majority of installs (99%) are businesses who rely on the
accountability we've built into the software.
― chad, Feb 18 2017, 11:55 PM
Anyway I tried to explore the class DifferentialCommitMessageField and I've found this method producing the list of all available fields:
final public static function getAllFields() {
return id(new PhutilClassMapQuery())
->setAncestorClass(__CLASS__)
->setUniqueMethod('getCommitMessageFieldKey')
->setSortMethod('getFieldOrder')
->execute();
}
And look at all the class that are inheriting DifferentialCommitMessageField. Some of them:
DifferentialTagsCommitMessageField
DifferentialSubscribersCommitMessageField
DifferentialAuditorsCommitMessageField
DifferentialReviewedByCommitMessageField
DifferentialTestPlanCommitMessageField
DifferentialTitleCommitMessageField
DifferentialSummaryCommitMessageField
...
So maybe you can customize a field changing the related class. You can change some default values, or you can try disabling a field declaring this method in one of these classes:
/**
* This method is inherited from DifferentialCommitMessageField
*
* #override
*/
public function isFieldEnabled() {
// return true;
return false
}
In short you can try to extend Phabricator to do it. Currently this feature is not a priority for their general enterprise use cases.
Anyway, do not forgot that Phabricator is a Free/Libre and Open Source software. You have all the rights to play with the code and make some improvements. If you are really interested in this feature and you have the possibility to add this customization feature, some users may be interested in your patch, so you may also consider to propose your changes to upstream, if it works and does not introduce regressions.

List of permissions for Drupal8 routing file

I'm working on custom Drupal8 module. My module uses this routing file:
kalvis.routing.yml
kalvis.content:
path: '/kalvis/{from}/{to}'
defaults:
_controller: '\Drupal\kalvis\Controller\kalvisController::content'
_title: ''
requirements:
_permission: 'access content'
What does _permission part stand for and where can I find a list of all possible values for this parameter?(in tut's I've watched were used only access content and access administrative content but I suppose there is a lot more of them)
PS: I'm using Drupal 8 beta 10 installed on WAMP
If you want to see a list of all permission, the code below should work. work. If you are coding your own module you can define your own permissions and test if a user has a role with that permission.
function my_module_page_attachments_alter(array &$attachments) {
$perms = array_keys(\Drupal::service('user.permissions')->getPermissions());
}
To answer the question what is the _permission part of the routing structure. Here is a quote from the drupal docs about what it does.
_permission: A permission string (e.g., _permission: 'access content'). You can specify multiple permissions by separating them with ',' (comma) (e.g., _permission: 'access content,access user profiles') for AND logic or '+' (plus) for OR logic (e.g., _permission: 'access content+access user profiles' means a visitor needs either the access content permission or the access user profiles permission to view the page. Having both is fine, too.). Module-specific permission strings can be defined in my_module_name.permissions.yml. See hook_permission() replaced with permissions defined in a my_module_name.permissions.yml file for details.
source: https://www.drupal.org/docs/drupal-apis/routing-system/structure-of-routes
To put it simply this restricts access to this route by only allowing users with the specified permission(s) to access it. To use it you need to know the system name of the permission(s) you want to use to restrict access. Then you just place then as a string behind this paramerter. Like in the quote above. You can choose to use multiple permissions by separating them with , for AND logic or + for OR logic. Permissions system names are allowed to have spaces in them and frequently do.
I don't think there is any way to directly see it in ui if you are talking about the system names of the permissions. You can ofcource see all permissions on www.site.com/admin/people/permissions. If you are in a hurry and/or looking for a specific permission you can always look through the module.permissions.yml file of the module this permission is defined in.
If you do want to see all permissions you can make your own list of all the system names.
You can use the PermissionHandler service from the core module.
This does the following gets all yaml's and creates a list.
You would call this by calling Drupal::service('user.permissions')->getPermissions() (https://api.drupal.org/api/drupal/core%21modules%21user%21src%21PermissionHandler.php/function/PermissionHandler%3A%3AgetPermissions/8.2.x)
You can use or try to write similar code to the functionality of the user_role_permissions function from the user.module file in drupal core. It looks like this:
function user_role_permissions(array $roles) {
if (defined('MAINTENANCE_MODE') && MAINTENANCE_MODE == 'update') {
return _user_role_permissions_update($roles);
}
$entities = Role::loadMultiple($roles);
$role_permissions = array();
foreach ($roles as $rid) {
$role_permissions[$rid] = isset($entities[$rid]) ? $entities[$rid]
->getPermissions() : array();
}
return $role_permissions;
}
This code as you can see just loads all the role entities with loadMultiple (although technically you should use the entitytypemanager to load the entities whenever possible like $entities = \Drupal::entityTypeManager()->getStorage($entity_type)->loadMultiple([1, 2, 3]); for more information see the drupal entity api (https://www.drupal.org/docs/drupal-apis/entity-api/working-with-the-entity-api)).
After loading all the roles it makes a list of all permissions.
Source information below. This should stay up to date because drupal keeps their documentation versioned. But because comments suggested it I figured I might as well write it out to save you some clicks.
Original drupal documentation.
https://api.drupal.org/api/drupal/core!modules!user!user.module/function/user_role_permissions/8.2.x
Hope this helps! :)
You can confirm in the page '/admin/people/permissions'.
A quick and dirty way to see them is to create a View with a Page display. Then in the 'Access' section, ensure 'Permission' is selected and open up the options as if you were going to choose a different permission.
You can now inspect the HTML of the <select> element, the Ids of each option is the correct name for each permission:

Laravel/blade caching css files

I am working on Nginx server, with PHP-FPM. I installed Laravel 4.1 and bootstrap v3.1.1., and here is the problem. For the last 30 minutes, I have been trying to change a css rule that I first declared to check boostrap.
.jumbotron{
background: red;
}
The first time it worked. The jumbotron container was red. So, I removed that css value and started working, but still no matter which browse I use, the container is red. I even checked the css file through the Google Chromes inspection tool, and it is showing me that first value when jumbotron had a background:red. I deleted the css file and renamed it and add new styles, I configured chrome not to cache pages. But Still the same value. I'm convinced now, that Laravel has kept a cache of the first style declaration.
Is there any way to disable this at all?
General explanation
When you access a Laravel Blade view, it will generate it to a temporary file so it doesn't have to process the Blade syntax every time you access to a view. These files are stored in app/storage/view with a filename that is the MD5 hash of the file path.
Usually when you change a view, Laravel regenerate these files automatically at the next view access and everything goes on. This is done by comparing the modification times of the generated file and the view's source file through the filemtime() function. Probably in your case there was a problem and the temporary file wasn't regenerated. In this case, you have to delete these files, so they can be regenerated. It doesn't harm anything, because they are autogenerated from your views and can be regenerated anytime. They are only for cache purposes.
Normally, they should be refreshed automatically, but you can delete these files anytime if they get stuck and you have problems like these, but as I said these should be just rare exceptions.
Code break down
All the following codes are from laravel/framerok/src/Illuminate/View/. I added some extra comments to the originals.
Get view
Starting from Engines/CompilerEngine.php we have the main code we need to understand the mechanics.
public function get($path, array $data = array())
{
// Push the path to the stack of the last compiled templates.
$this->lastCompiled[] = $path;
// If this given view has expired, which means it has simply been edited since
// it was last compiled, we will re-compile the views so we can evaluate a
// fresh copy of the view. We'll pass the compiler the path of the view.
if ($this->compiler->isExpired($path))
{
$this->compiler->compile($path);
}
// Return the MD5 hash of the path concatenated
// to the app's view storage folder path.
$compiled = $this->compiler->getCompiledPath($path);
// Once we have the path to the compiled file, we will evaluate the paths with
// typical PHP just like any other templates. We also keep a stack of views
// which have been rendered for right exception messages to be generated.
$results = $this->evaluatePath($compiled, $data);
// Remove last compiled path.
array_pop($this->lastCompiled);
return $results;
}
Check if regeneration required
This will be done in Compilers/Compiler.php. This is an important function. Depending on the result it will be decided whether the view should be recompiled. If this returns false instead of true that can be a reason for views not being regenerated.
public function isExpired($path)
{
$compiled = $this->getCompiledPath($path);
// If the compiled file doesn't exist we will indicate that the view is expired
// so that it can be re-compiled. Else, we will verify the last modification
// of the views is less than the modification times of the compiled views.
if ( ! $this->cachePath || ! $this->files->exists($compiled))
{
return true;
}
$lastModified = $this->files->lastModified($path);
return $lastModified >= $this->files->lastModified($compiled);
}
Regenerate view
If the view is expired it will be regenerated. In Compilers\BladeCompiler.php we see that the compiler will loop through all Blade keywords and finally give back a string that contains the compiled PHP code. Then it will check if the view storage path is set and save the file there with a filename that is the MD5 hash of the view's filename.
public function compile($path)
{
$contents = $this->compileString($this->files->get($path));
if ( ! is_null($this->cachePath))
{
$this->files->put($this->getCompiledPath($path), $contents);
}
}
Evaluate
Finally in Engines/PhpEngine.php the view is evaluated. It imports the data passed to the view with extract() and include the file with the passed path in a try and catch all exceptions with handleViewException() that throws the exception again. There are some output buffering too.
Same issue here. I am using VirtualBox with Shared Folders pointing to my document root.
This pointed me in the right direction:
https://stackoverflow.com/a/26583609/1036602
Which led me to this:
http://www.danhart.co.uk/blog/vagrant-virtualbox-modified-files-not-updating-via-nginx-apache
and this:
https://forums.virtualbox.org/viewtopic.php?f=1&t=24905
If you're mounting your local dev root via vboxsf Shared Folders, set EnableSendFile Off in your apache2.conf (or sendfile off if using Nginx).
For what it's worth and because this answer came up first in my google search...
I had the same problem. The CSS and JS files wouldn't update. Deleting the cache files didn't work. The timestamps were not the problem. The only way I could update them was to change the filename, load it directly to get the 404 error, and then change the name back to the original name.
In the end the problem was not related to Laravel or the browser cache at all. The problem was due to NginX using sendfile which doesn't work with remote file systems. In my case, I was using VirtualBox for the OS and the remote file system was vboxsf through Guest Additions.
I hope this saves someone else some time.
In Laravel 5.8+ you can use so:
The version method will automatically append a unique hash to the filenames of all compiled files, allowing for more convenient cache busting:
mix.js('resources/js/app.js', 'public/js').version();
After generating the versioned file, you won't know the exact file name. So, you should use Laravel's global mix function within your views to load the appropriately hashed asset. The mix function will automatically determine the current name of the hashed file:
<script src="{{ mix('/js/app.js') }}"></script>
full document: https://laravel.com/docs/5.8/mix

Sonata Media: Change context programmatically

I'm writing a little blog app where the user can publish public and private news. Users can attach files to these news. I have two contexts for this app: public_news, with files which can be accessed by everyone; and private_news, with files which can only be accessed if the user has log on.
I want to be able to move files from the public_news context to the private_news context when the user changes a news from public to private, and vice versa.
I was hoping to do something as simple as $media->setContext('private_news');, but this won't move the physical file from one directory to the other.
What do you think about recreating this media?
$oldMedia = getYourOldMedia();
// $media = clone($oldMedia); # For me it didn't work as expected
# YMMV - I didn't spend lots wondering about that
$media = new Media();
// This will work fine with image and file provider,
// but it was not tested with other providers
$pool = $container->get('sonata.media.pool');
$provider = $pool->getProvider($oldMedia->getProviderName());
$media->setBinaryContent($provider->getReferenceFile($oldMedia));
}
$media->setProviderName($oldMedia->getProviderName());
$media->setContext('private_news');
/* copy any other data you're interested in */
$mediaManager->save($media);
$mediaManager->delete($oldMedia);
$mediaManager->delete might not delete your physical files depending on provider, you might want to create your own provider if you wish to do so.
Edit:
On further research I found out that you can manualy delete your files before deleting old media:
if ($pool->getFilesystem()->has($path)) {
$pool->getFilesystem()->delete($path);
}
But don't do that before saving your new media entity.

How to check publishing or rendering context with TOM.NET?

SDL Tridion's Content Manager templating API (TOM.NET) offers ways to detect publishing or rendering context.
Use Cases
Present debugging information to a specific environment (e.g. TCM Uris only on Staging)
Show different markup in Preview (e.g. show a link to the published page)
Show different author-able fields in Experience Manager or SiteEdit
I've seen and tried a few examples, but after following a chat between colleagues Stan and Eric, I want to make sure I follow for TOM.NET (6.1 / Tridion 2011).
Scenarios
Publishing to a specific Publication Target (typically "Live" and "Staging")
Content Manager Explorer (CME) Preview
Session Preview rendering for Experience Manager (XPM)
(Added) Template Builder
1. Publishing to a Target (or from a Publication)
Tridion.ContentManager.Publishing.PublishEngine.GetPublishInfo(IdentifiableObject item)
Item would be a page or component. This returns a collection of PublishInfo objects, which includes PublicationTarget to confirm where you're publishing to.
Tridion.ContentManager.Templating.PublishingContext.PublicationTarget has PublicationTarget as well.
2. CME Preview
PublicationTarget is null, which makes sense because you don't have a Publication Target . :-)
3. Session Preview
Use the RenderMode Enum under Tridion.ContentManager.Publishing, which has:
'Publish' (0)
'PreviewStatic' (1)
'PreviewDynamic' (2)
PublicationTarget wouldn't be null for Session Preview, which isn't really publishing.
4. (Added) Template Builder
?
Alexander Klock also describes some related examples which cover most of this except CME Preview.
Question(s)
Am I missing any scenarios? Publish to a specific publication target, regular preview, and XPM session preview?
How should I avoid hard-coding PublicationTargets (e.g. is it better to check the string values instead of TCM Uris)?
Update: added Template Builder to the list per Vikas's answer, how do I know I'm rendering within Template Builder?
You really need a tl;dr on this question...
Here's what I know:
Template Builder
Publication target is null, RenderMode is PreviewDynamic
CME Preview
Publication Target ID is tcm:0-0-0 (or TcmUri.UriNull), RenderMode is PreviewDynamic
Session Preview
Publication Target ID is the real target ID, RenderMode is PreviewDynamic
Publish
Publication Target ID is the real one, RenderMode is Publish
EDIT
Here's some sample code I wrote recently to determine the Current mode.
private CurrentMode GetCurrentMode()
{
RenderMode renderMode = _engine.RenderMode;
if (renderMode == RenderMode.Publish) return CurrentMode.Publish;
if (renderMode == RenderMode.PreviewDynamic)
{
if (_engine.PublishingContext.PublicationTarget == null) return CurrentMode.TemplateBuilder;
PublicationTarget target = _engine.PublishingContext.PublicationTarget;
if (target.Id.Equals(TcmUri.UriNull)) return CurrentMode.CmePreview;
return CurrentMode.SessionPreview;
}
return CurrentMode.Unknown;
}
private enum CurrentMode
{
TemplateBuilder,
CmePreview,
SessionPreview,
Publish,
Unknown
}
You presented a very good view of complete publishing/preview model. Here are my thoughts..
Are we missing any scenarios?
I think you covered everything expect the template builder case, which is similar to CME preview where we get publishing target as null but can be used to check different condition so important for debugging purpose.
How should I avoid hard-coding PublicationTargets
Yes we should never use tcm uri in any code, as you suggested we could use name and even name can be configured in respective config files for that program.
Also may not be relevant here, its always good to have separate target for Tridion UI edting other than staging. Both can be configured on same server with two deployer's. One could be staging.yoursite.com and other could be tridionui.yoursite.com
Thanks..

Resources