Bugzilla::Hook - Extendable extension hooks for Bugzilla code
use Bugzilla::Hook;
Bugzilla::Hook::process("hookname", { arg => $value, arg2 => $value2 });
Bugzilla allows extension modules to drop in and add routines at arbitrary points in Bugzilla code. These points are referred to as hooks. When a piece of standard Bugzilla code wants to allow an extension to perform additional functions, it uses Bugzilla::Hook's "process" subroutine to invoke any extension code if installed.
The implementation of extensions is described in Bugzilla::Extension.
There is sample code for every hook in the Example extension, located in extensions/Example/Extension.pm.
When a hook named HOOK_NAME is run, Bugzilla looks through all enabled extensions for extensions that implement a subroutine named HOOK_NAME.
See Bugzilla::Extension for more details about how an extension can run code during a hook.
processInvoke any code hooks with a matching name from any installed extensions.
See Bugzilla::Extension for more information on Bugzilla's extension mechanism.
$name - The name of the hook to invoke.$args - A hashref. The named args to pass to the hook. They will be passed as arguments to the hook method in the extension.This describes what hooks exist in Bugzilla currently. They are mostly in alphabetical order, but some related hooks are near each other instead of being alphabetical.
This happens at the very beginning process of the attachment creation. You can edit the attachment content itself as well as all attributes of the attachment, before they are validated and inserted into the DB.
Params:
data - A reference pointing either to the content of the file being uploaded or pointing to the filehandle associated with the file.attributes - A hashref whose keys are the same as the input to "create" in Bugzilla::Attachment. The data in this hashref hasn't been validated yet.This allows you to add new login types to Bugzilla. (See Bugzilla::Auth::Login.)
Params:
modulesThis is a hash--a mapping from login-type "names" to the actual module on disk. The keys will be all the values that were passed to "login" in Bugzilla::Auth for the Login parameter. The values are the actual path to the module on disk. (For example, if the key is DB, the value is Bugzilla/Auth/Login/DB.pm.)
For your extension, the path will start with Bugzilla/Extension/Foo/, where "Foo" is the name of your Extension. (See the code in the example extension.)
If your login type is in the hash as a key, you should set that key to the right path to your module. That module's new method will be called, probably with empty parameters. If your login type is not in the hash, you should not set it.
You will be prevented from adding new keys to the hash, so make sure your key is in there before you modify it. (In other words, you can't add in login methods that weren't passed to "login" in Bugzilla::Auth.)
This works just like "auth_login_methods" except it's for login verification methods (See Bugzilla::Auth::Verify.) It also takes a modules parameter, just like "auth_login_methods".
This allows you to add new fields that will show up in every Bugzilla::Bug object. Note that you will also need to use the "bug_fields" hook in conjunction with this hook to make this work.
Params:
columns - An arrayref containing an array of column names. Push your column name(s) onto the array.This happens at the end of "create" in Bugzilla::Bug, after all other changes are made to the database. This occurs inside a database transaction.
Params:
bug - The created bug object.timestamp - The timestamp used for all updates in this transaction, as a SQL date string.This happens during "create" in Bugzilla::Bug, after all parameters have been validated, but before anything has been inserted into the database.
Params:
paramsA hashref. The validated parameters passed to create.
This happens at the end of "update" in Bugzilla::Bug, after all other changes are made to the database. This generally occurs inside a database transaction.
Params:
bugThe changed bug object, with all fields set to their updated values.
old_bugA bug object pulled from the database before the fields were set to their updated values (so it has the old values available for each field).
timestampThe timestamp used for all updates in this transaction, as a SQL date string.
changesThe hash of changed fields. $changes->{field} = [old, new]
Allows the addition of database fields from the bugs table to the standard list of allowable fields in a Bugzilla::Bug object, so that you can call the field as a method.
Note: You should add here the names of any fields you added in "bug_columns".
Params:
columns - A arrayref containing an array of column names. Push your column name(s) onto the array.Allows you to do custom parsing on comments before they are displayed. You do this by returning two regular expressions: one that matches the section you want to replace, and then another that says what you want to replace that match with.
The matching and replacement will be run with the /g switch on the regex.
Params:
regexesAn arrayref of hashrefs.
You should push a hashref containing two keys (match and replace) in to this array. match is the regular expression that matches the text you want to replace, replace is what you want to replace that text with. (This gets passed into a regular expression like s/$match/$replace/.)
Instead of specifying a regular expression for replace you can also return a coderef (a reference to a subroutine). If you want to use backreferences (using $1, $2, etc. in your replace), you have to use this method--it won't work if you specify $1, $2 in a regular expression for replace. Your subroutine will get a hashref as its only argument. This hashref contains a single key, matches. matches is an arrayref that contains $1, $2, $3, etc. in order, up to $10. Your subroutine should return what you want to replace the full match with. (See the code example for this hook if you want to see how this actually all works in code. It's simpler than it sounds.)
You are responsible for HTML-escaping your returned data. Failing to do so could open a security hole in Bugzilla.
textA reference to the exact text that you are parsing.
Generally you should not modify this yourself. Instead you should be returning regular expressions using the regexes array.
The text has not been parsed in any way. (So, for example, it is not HTML-escaped. You get "&", not "&".)
bugThe Bugzilla::Bug object that this comment is on. Sometimes this is undef, meaning that we are parsing text that is not on a bug.
commentA Bugzilla::Comment object representing the comment you are about to parse.
Sometimes this is undef, meaning that we are parsing text that is not a bug comment (but could still be some other part of a bug, like the summary line).
This happens in "COLUMNS" in Bugzilla::Search, which determines legal bug list columns for buglist.cgi and colchange.cgi. It gives you the opportunity to add additional display columns.
Params:
columns - A hashref, where the keys are unique string identifiers for the column being defined and the values are hashrefs with the following fields:name - The name of the column in the database.title - The title of the column as displayed to users.The definition is structured as:
$columns->{$id} = { name => $name, title => $title };
This allows you to modify the list of users who are going to be receiving a particular bugmail. It also allows you to specify why they are receiving the bugmail.
Users' bugmail preferences will be applied to any users that you add to the list. (So, for example, if you add somebody as though they were a CC on the bug, and their preferences state that they don't get email when they are a CC, they won't get email.)
This hook is called before watchers or globalwatchers are added to the recipient list.
Params:
bugThe Bugzilla::Bug that bugmail is being sent about.
recipientsThis is a hashref. The keys are numeric user ids from the profiles table in the database, for each user who should be receiving this bugmail. The values are hashrefs. The keys in these hashrefs correspond to the "relationship" that the user has to the bug they're being emailed about, and the value should always be 1. The "relationships" are described by the various REL_ constants in Bugzilla::Constants.
Here's an example of adding userid 123 to the recipient list as though he were on the CC list:
$recipients->{123}->{+REL_CC} = 1
(We use + in front of REL_CC so that Perl interprets it as a constant instead of as a string.)
This happens in colchange.cgi right after the list of possible display columns have been defined and gives you the opportunity to add additional display columns to the list of selectable columns.
Params:
columns - An arrayref containing an array of column IDs. Any IDs added by this hook must have been defined in the the "buglist_columns" hook.If you want to add new panels to the Parameters administrative interface, this is where you do it.
Params:
panel_modulesA hashref, where the keys are the "name" of the panel and the value is the Perl module representing that panel. For example, if the name is Auth, the value would be Bugzilla::Config::Auth.
For your extension, the Perl module would start with Bugzilla::Extension::Foo, where "Foo" is the name of your Extension. (See the code in the example extension.)
This is how you modify already-existing panels in the Parameters administrative interface. For example, if you wanted to add a new Auth method (modifying Bugzilla::Config::Auth) this is how you'd do it.
Params:
panelsA hashref, where the keys are lower-case panel "names" (like auth, admin, etc.) and the values are hashrefs. The hashref contains a single key, params. params is an arrayref--the return value from get_param_list for that module. You can modify params and your changes will be reflected in the interface.
Adding new keys to panels will have no effect. You should use "config_add_panels" if you want to add new panels.
This happens right after an inbound email is converted into an Email::MIME object, but before we start parsing the email to extract field data. This means the email has already been decoded for you. It gives you a chance to interact with the email itself before email_in starts parsing its content.
mail - An Email::MIME object. The decoded incoming email.fields - A hashref. The hash which will contain extracted data.This happens after all the data has been extracted from the email, but before the reporter is validated, during email_in. This lets you do things after the normal parsing of the email, such as sanitizing field data, changing the user account being used to file a bug, etc.
fields - A hashref. The hash containing the extracted field data.DEPRECATED - Use "template_before_process" instead.
This happens right before the template is loaded on enter_bug.cgi.
Params:
vars - A hashref. The variables that will be passed into the template.This happens at the end of "update_flags" in Bugzilla::Flag, after all other changes are made to the database and after emails are sent. It gives you a before/after snapshot of flags so you can react to specific flag changes. This generally occurs inside a database transaction.
Note that the interface to this hook is UNSTABLE and it may change in the future.
Params:
object - The changed bug or attachment object.timestamp - The timestamp used for all updates in this transaction, as a SQL date string.old_flags - The snapshot of flag summaries from before the change.new_flags - The snapshot of flag summaries after the change. Call my ($removed, $added) = diff_arrays(old_flags, new_flags) to get the list of changed flags, and search for a specific condition like added eq 'review-'.This happens in "remove_from_db" in Bugzilla::Group, after we've confirmed that the group can be deleted, but before any rows have actually been removed from the database. This occurs inside a database transaction.
Params:
group - The Bugzilla::Group being deleted.This happens at the end of "create" in Bugzilla::Group, after all other changes are made to the database. This occurs inside a database transaction.
Params:
groupThe new Bugzilla::Group object that was just created.
This happens at the end of "update" in Bugzilla::Group, after all other changes are made to the database. This occurs inside a database transaction.
Params:
group - The changed Bugzilla::Group object, with all fields set to their updated values.changes - The hash of changed fields. $changes->{$field} = [$old, $new]Allows execution of custom code before the final checks are done in checksetup.pl.
Params:
silentA flag that indicates whether or not checksetup is running in silent mode. If this is true, messages that are always printed by checksetup.pl should be suppressed, but messages about any changes that are just being done this one time should be printed.
This happens at the very end of all the tables being updated during an installation or upgrade. If you need to modify your custom schema or add new columns to existing tables, do it here. No params are passed.
This allows you to add tables to Bugzilla. Note that we recommend that you prefix the names of your tables with some word (preferably the name of your Extension), so that they don't conflict with any future Bugzilla tables.
If you wish to add new columns to existing Bugzilla tables, do that in "install_update_db".
Params:
schema - A hashref, in the format of "ABSTRACT_SCHEMA" in Bugzilla::DB::Schema. Add new hash keys to make new table definitions. checksetup.pl will automatically add these tables to the database when run.Called right before Bugzilla::Mailer sends a message to the MTA.
Params:
email - The Email::MIME object that's about to be sent.mailer_args - An arrayref that's passed as mailer_args to "new" in Email::Send.This happens at the beginning of "create" in Bugzilla::Object.
Params:
classThe name of the class that create was called on. You can check this like if ($class->isa('Some::Class')) in your code, to perform specific tasks before create for only certain classes.
paramsA hashref. The set of named parameters passed to create.
This happens in "remove_from_db" in Bugzilla::Object, after we've confirmed that the object can be deleted, but before any rows have actually been removed from the database. This sometimes occurs inside a database transaction.
Params:
object - The Bugzilla::Object being deleted. You will probably want to check its type like $object->isa('Some::Class') before doing anything with it.Called during "set" in Bugzilla::Object, before any actual work is done. You can use this to perform actions before a value is changed for specific fields on certain types of objects.
Params:
objectThe object that set was called on. You will probably want to do something like if ($object->isa('Some::Class')) in your code to limit your changes to only certain subclasses of Bugzilla::Object.
fieldThe name of the field being updated in the object.
valueThe value being set on the object.
Called at the end of "create" in Bugzilla::Object, after all other changes are made to the database. This occurs inside a database transaction.
Params:
classThe name of the class that create was called on. You can check this like if ($class->isa('Some::Class')) in your code, to perform specific tasks for only certain classes.
objectThe created object.
Called at the end of "run_create_validators" in Bugzilla::Object. You can use this to run additional validation when creating an object.
If a subclass has overridden run_create_validators, then this usually happens before the subclass does its custom validation.
Params:
classThe name of the class that create was called on. You can check this like if ($class->isa('Some::Class')) in your code, to perform specific tasks for only certain classes.
paramsA hashref. The set of named parameters passed to create, modified and validated by the VALIDATORS specified for the object.
Called during "set" in Bugzilla::Object, after all the code of the function has completed (so the value has been validated and the field has been set to the new value). You can use this to perform actions after a value is changed for specific fields on certain types of objects.
The new value is not specifically passed to this hook because you can get it as $object->{$field}.
Params:
objectThe object that set was called on. You will probably want to do something like if ($object->isa('Some::Class')) in your code to limit your changes to only certain subclasses of Bugzilla::Object.
fieldThe name of the field that was updated in the object.
This happens at the end of "set_all" in Bugzilla::Object. This is a good place to call custom set_ functions on objects, or to make changes to an object before update() is called.
Params:
objectThe Bugzilla::Object which is being updated. You will probably want to do something like if ($object->isa('Some::Class')) in your code to limit your changes to only certain subclasses of Bugzilla::Object.
paramsA hashref. The set of named parameters passed to set_all.
Called during "update" in Bugzilla::Object, after changes are made to the database, but while still inside a transaction.
Params:
objectThe object that update was called on. You will probably want to do something like if ($object->isa('Some::Class')) in your code to limit your changes to only certain subclasses of Bugzilla::Object.
old_objectThe object as it was before it was updated.
changesThe fields that have been changed, in the same format that "update" in Bugzilla::Object returns.
This is a simple way to add your own pages to Bugzilla. This hooks page.cgi, which loads templates from template/en/default/pages. For example, page.cgi?id=fields.html loads template/en/default/pages/fields.html.tmpl.
This hook is called right before the template is loaded, so that you can pass your own variables to your own pages.
You can also use this to implement complex custom pages, by doing your own output and then calling exit at the end of the hook, thus preventing the normal page.cgi behavior from occurring.
Params:
page_idThis is the name of the page being loaded, like fields.html.
Note that if two extensions use the same name, it is uncertain which will override the others, so you should be careful with how you name your pages. Usually extensions prefix their pages with a directory named after their extension, so for an extension named "Foo", page ids usually look like foo/mypage.html.
varsThis is a hashref--put variables into here if you want them passed to your template.
DEPRECATED - Use "template_before_process" instead.
Called before displaying the confirmation message when deleting a product.
Params:
vars - The template vars hashref.This hook allows for extra sanity checks to be added, for use by sanitycheck.cgi.
Params:
status - a CODEREF that allows status messages to be displayed to the user. (sanitycheck.cgi's Status)Called right after a new product has been created, allowing additional changes to be made to the new product's attributes. This occurs inside of a database transaction, so if the hook throws an error, all previous changes will be rolled back, including the creation of the new product. (However, note that such rollbacks should not normally be used, as some databases that Bugzilla supports have very bad rollback performance. If you want to validate input and throw errors before the Product is created, use "object_end_of_create_validators" instead.)
Params:
product - The new Bugzilla::Product object that was just created.This hook allows for extra sanity check repairs to be made, for use by sanitycheck.cgi.
Params:
status - a CODEREF that allows status messages to be displayed to the user. (sanitycheck.cgi's Status)This hook allows you to modify the configuration of Bugzilla::Template objects before they are created. For example, you could add a new global template variable this way.
Params:
configA hashref--the configuration that will be passed to "new" in Template. See http://template-toolkit.org/docs/modules/Template.html#section_CONFIGURATION_SUMMARY for information on how this configuration variable is structured (or just look at the code for create in Bugzilla::Template.)
This hook is called any time Bugzilla processes a template file, including calls to $template->process, PROCESS statements in templates, and INCLUDE statements in templates. It is not called when templates process a BLOCK, only when they process a file.
This hook allows you to define additional variables that will be available to the template being processed, or to modify the variables that are currently in the template. It works exactly as though you inserted code to modify template variables at the top of a template.
You probably want to restrict this hook to operating only if a certain file is being processed (which is why you get a file argument below). Otherwise, modifying the vars argument will affect every single template in Bugzilla.
Note: This hook is not called if you are already in this hook. (That is, it won't call itself recursively.) This prevents infinite recursion in situations where this hook needs to process a template (such as if this hook throws an error).
Params:
varsThis is the entire set of variables that the current template can see. Technically, this is a Template::Stash object, but you can just use it like a hashref if you want.
fileThe name of the template file being processed. This is relative to the main template directory for the language (i.e. for template/en/default/bug/show.html.tmpl, this variable will contain bug/show.html.tmpl).
contextA Template::Context object. Usually you will not have to use this, but if you need information about the template itself (other than just its name), you can get it from here.
This hook allows you to add your own modules to the WebService. (See Bugzilla::WebService.)
Params:
dispatchA hashref where you can specify the names of your modules and which Perl module handles the functions for that module. (This is actually sent to "dispatch_with" in SOAP::Lite. You can see how that's used in xmlrpc.cgi.)
The Perl module name will most likely start with Bugzilla::Extension::Foo:: (where "Foo" is the name of your extension).
Example:
$dispatch->{'Example.Blah'} = "Bugzilla::Extension::Example::Webservice::Blah";
And then you'd have a module extensions/Example/lib/Webservice/Blah.pm, and could call methods from that module like Example.Blah.Foo() using the WebServices interface.
It's recommended that all the keys you put in dispatch start with the name of your extension, so that you don't conflict with the standard Bugzilla WebService functions (and so that you also don't conflict with other plugins).
If your webservice extension throws custom errors, you can set numeric codes for those errors here.
Extensions should use error codes above 10000, unless they are re-using an already-existing error code.
Params:
error_mapA hash that maps the names of errors (like invalid_param) to numbers. See "WS_ERROR_CODE" in Bugzilla::WebService::Constants for an example.