MediaWiki Hook for Installing Extension - mediawiki

I did some research on MediaWiki hooks.
To the best of my knowledge, the only hook for adding/updating database tables is https://www.mediawiki.org/wiki/Manual:Hooks/LoadExtensionSchemaUpdates.
However, I need a hook that fires on installing a new extension. How do I achieve that? I'd like to execute a CREATE TABLE statement only once - when the extension is installed. Installing an extension doesn't require a MediaWiki update, which is why the above hook does not suit my needs.
EDIT
To clarify: I'm developing an extension that requires access to a custom table in the database. That's why I need to execute the CREATE TABLE statement whenever this extension is installed.

First: As you have noticed, there is no such hook. You will need to do the check for installed extensions by a cron job (if it's your server), or on, say, each 100th request to the wiki, using the job queue (if you are doing this in an extension).
From there, you have a few options, depending on if you need to catch every single extension, or just most of them:
Check for registered extensions, with something like: $registry = ExtensionRegistry::getInstance();$extensions = $registry->loaded();This will work extensions using the new style. For older versions, check the global variable $wgExtensionCredits for registered extensions. Note that nothing prevents an extensions from running without registering. In fact, there are a lot of extensions like that.
Parse LocalSettings.php, and check for old and new style extension loading with a regex. Very dirty, but it's actually the way e.g. the maintainance script getConfiguration.php does it. You need to check for lines like this:wfLoadExtension( 'FooBar' );and this:wfLoadExtensions( ['Foo', 'Bar'] );and this:require_once "$IP/extensions/Foo/Bar.php"; Note however that it is possible to use other directories for extensions, and that modern skins in fact behave like extensions too.
If you need to track a certain family of extensions, and you can make sure they all use Composer, the parse $IP/composer.json for installed extensions.

I did some more research (https://www.mediawiki.org/wiki/Manual:Update.php#Details) and found out that the LoadExtensionSchemaUpdates hook is exactly what I am looking for.
# Schema updates for update.php
$wgHooks['LoadExtensionSchemaUpdates'][] = 'fnMyHook';
function fnMyHook( DatabaseUpdater $updater ) {
$updater->addExtensionTable( 'tablename',
__DIR__ . '/table.sql' );
return true;
}
When running $ php update.php for the first time after installing the extension, the script executes the CREATE TABLE statement. Then, for consecutive executions, the update script just notes that the table already exists.

Related

Pass file name to build script in PhpStorm

I have the next configuration for a build system in PhpStorm:
And it works ok, but I have a problem... my build script needs to receive the name of the file I'm running it on, so, if I build a PHP file, it will run phpcs on it, but if I'm building a CSS or JS file, it will run gulp... with Sublime Text that is possible, is it possible with PhpStorm?
There are no macros support for Run/Debug Configurations -- they are made so they do not depend on a context (currently opened file in editor). In other words -- they are pretty static and all file names/paths are basically hard coded.
For what you are describing (build script).. you need to use External Tools functionality (which can have all of that and made specifically for such tasks). Once created, you can assign custom shortcut to any External Tools entry (check Settings/Preferences | Keymap for that) so it's more convenient to use it.
If you want such script to be called on every file save automatically -- then use File Watchers -- pretty much external tools that will be called for you automatically (once per each file modified).
Since you are doing this for a build script -- maybe you should try to use dedicated (and therefore more appropriate in general) tools? For example: Gulp / Grunt .. or even Phing.
Create external tool:
https://www.jetbrains.com/phpstorm/help/external-tools.html
You can assign hotkey to execute your build command.

How do I assign Sublime key mappings for commands in the sidebar context menu?

When browsing files in the Sublime sidebar, I would like to quickly access the commands available in the context menu via shortcuts. E.g. Delete file, rename file/folder, new file/folder.
(FYI: super+N is not an ideal solution for creating new files - it isn't context aware and will usually choose an inappropriate default location).
You can enable command logging by inserting the following into the console sublime.log_commands(True). This will give you the commands and arguments being executed. You can then create a key binding with the appropriate command. My guess is the commands will use some sort of path, so you may need to write a small plugin to inject the correct paths for the various commands.
For new file creation specifically, you may want to take a look at AdvancedNewFile. Disclaimer - I'm the current maintainer of the plugin. I've tried to make it a more flexible than it originally was, with regards to specifying where to create the file. Though if you do decide to use it and have feature request, feel free to create an issue about it.

Protect Air application content

On Mac Os, I see that all content on my application can be readable (mxml and as files).
Indeed with right clic on application, you can see all application content and so all files.
So It's very dangerous for a company to distribute air application like that.
Is a solution exist to protect those files.
Thanks
It is not possible to protect 100% your code. After all, if the computer can run it, it can be decompiled, regardless of the language. However, you can make it more difficult.
One method is to encrypt the swf as stated in another answer. But all the "attacker" needs to do is find the key and then they can decrypt all your swfs.
Another method is to use obfuscators. Obfuscators don't depend on encryption, nor they prevent decompiling, they just make it harder to understand what gets decompiled.
For example if you had a method called saveInvoice() the obfuscator would rename it to aa1() or something like that, so it would make it diffucult to guess what that function does. It basically turns everything into spaguetti code.
You can use a decompiler to see what can be obtained from a SWF file (which is alot), and play with obfuscators to see if they meet your espectations.
An example of one is http://www.kindi.com/ which I'm not endorsing btw, it just shows up quickly on google.
Although there are loads of decompilers which can read all your code. There is one guy who came up with encryption solution it might worth a try. (It's for Desktop AIR applications)
Have a look at this post: http://forums.adobe.com/message/3510525#3510525
Quoted text (in case of page being erased)
The method I use will allow you encrpyt most of your source code using
a key that is unique to every computer. The initial download of my
software is a simple air app that does not contain the actual program.
It is more like a shell that first retreaves a list of the clients mac
addresses and the user entered activation code that is created at time
of purchase. This is sent to server and logged. The activation code
is saved to a file client side. At the server the mac address and
activation key are used to create the encryption key. The bulk of the
program code is then encrypted using that key, then divided into parts
and sent back to the client. The client puts the parts back together
and saves the encrypted file. At runtime the shell finds the mac
address list and the activation key, then using same method as server
gets the encryption key and decrypts the program file. Run simple
check to make sure it loaded. For encyption i found an aes method that
works in php and javascript.
Next I use this code to load the program
var loader = air.HTMLLoader.createRootWindow(true, options, true, windowBounds);
loader.cacheResponse=false;
loader.placeLoadStringContentInApplicationSandbox=true;
loader.loadString(page);
This method makes it very difficult to copy
to another computer although since I wrote it i know there are some
weeknesses in the security but to make it harder i obv. the shell
code. It at least keeps most from pirating. However there are issues
with this that I have found. First i was using networkInfo to get the
list of mac address but this failed in a test windows XP computer.
When the wireless was off it did not return the MAC. I was not able
to recreate this in VISTA or 7. Not sure if it could happen. Was not
tested on a mac computer. To fix this (at least for windows). I
wrote a simple bat file that gets the MAC list, then converted it to
an exe which is included. This does force you to create native
installers. call the exe with this
var nativeProcessStartupInfo = new air.NativeProcessStartupInfo();
var file = air.File.applicationDirectory.resolvePath("findmac.exe");
nativeProcessStartupInfo.executable = file;
process = new air.NativeProcess();
process.start(nativeProcessStartupInfo);
process.addEventListener(air.ProgressEvent.STANDARD_OUTPUT_DATA, onOutputData);
process.addEventListener(air.ProgressEvent.STANDARD_ERROR_DATA, onErrorData);
process.addEventListener(air.NativeProcessExitEvent.EXIT, onExit);
process.addEventListener(air.IOErrorEvent.STANDARD_OUTPUT_IO_ERROR, onIOError);
process.addEventListener(air.IOErrorEvent.STANDARD_ERROR_IO_ERROR, onIOError);
put the list together in the onOutputData event using array.push and
continue on the onExit event using the findmac.exe will return the
same info every time (that i know of) beware thought that using the
native install will break the standard application update process so
you will have to write your own. My updates are processed the same way
as above. This is contents of the .bat file to get the mac list
#Echo off
SETLOCAL SET MAC = SET Media = Connected
FOR /F "Tokens=1-2 Delims=:" %%a in ('ipconfig /all^| FIND "Physical Address"') do #echo %%b ENDLOCAL
using this method makes it simple to implement at try before you by
method. at runtime if no activation code get try me version from
server instead of full version.

When developing an R package, do I have to recompile the package every time I make a change?

I am developing a package in R
When I am debugging a particular function or set of functions, what is the best way to test the function?
Do I have to either use source('function.R') or R CMD build each time I want to check my changes?
(extra credit for associated emacs ess key-bindings)
See also http://github.com/hadley/devtools/ which provides some tools to make this task easier.
for example, after making changes to source code, you build, install, and reload a package with the function install():
library(devtools)
install("package_name")
devtools also makes it easier to:
Reload complete package:
load_all("pkg")
Create or update documentation using roxygen2
document("pkg")
run all scripts in /inst/test/:
test("pkg")
build and R CMD check:
check("pkg")
Take a look at ?insertSource, which is a new function in R 2.12.0, plus the other functions in the See Also section of that help page. Also, check out ?assignInNamespace if your package has a Namespace.
The above presumes you are talking about updating and debugging R sources, not compiled code.
I generally have used the source() route to load new versions of functions I am improving/debugging, alongside the usual R debugging tools. But I haven't got Namespaces in my packages as yet. My fingers have gotten quite used to the C-c C-l keybinding in emacs+ess for sourcing a buffer over the years.
You might want to have a look at the 'mvbutils' package. I use it to live-edit my packages all the time; I can add, remove, and edit functions and documentation while the package is loaded, and the changes are reflected both in the loaded version, in the installed version (so they're kept in the next R session), and [when I tell it] in the "source package". I only re-build via R CMD when I want to distribute a zipped version to someone else. To test code, I use the 'debug' package, which works fine on a loaded package.
I even use 'mvbutils' to live-edit 'mvbutils', which can be a bit hairy sometimes.
The 'mvbutils' documentation could really do with a full demo of this in action, but in theory the existing doco should show you how to proceed.
Can't help you with Emacs, sorry...
I had got the same issue and I solved it while using RStudio.
In the editor, I check the option "Source on save" for my R file that contains function. As I'm used to save my file each time I edit it (a good habit I think), the corresponding functions loaded in my R workspace is always up to date.

How can I get a Windows batch or Perl script to run when a file is added to a directory?

I am trying to write a script that will parse a local file and upload its contents to a MySQL database. Right now, I am thinking that a batch script that runs a Perl script would work, but am not sure if this is the best method of accomplishing this.
In addition, I would like this script to run immediately when the data file is added to a certain directory. Is this possible in Windows?
Thoughts? Feedback? I'm fairly new to Perl and Windows batch scripts, so any guidance would be appreciated.
You can use Win32::ChangeNotify. Your script will be notified when a file is added to the target directory.
Checking a folder for newly created files can be implemented using the WMI functionality. Namely, you can create a Perl script that subscribes to the __InstanceCreationEvent WMI event that traces the creation of the CIM_DirectoryContainsFile class instances. Once that kind of event is fired, you know a new file has been added to the folder and can process it as you need.
These articles provide more information on the subject and contain VBScript code samples (hope it won't be hard for you to convert them to Perl):
How Can I Automatically Run a Script Any Time a File is Added to a Folder?
WMI and File System Monitoring
The function you want is ReadDirectoryChangesW. A quick search for a perl wrapper yields this Win32::ReadDirectoryChanges module.
Your script would look something like this:
use Win32::ReadDirectoryChanges;
$rdc = new Win32::ReadDirectoryChanges(path => $path,
subtree => 1,
filter => $filter);
while(1) {
#results = $rdc->read_changes;
while (scalar #results) {
my ($action, $filename) = splice(#results, 0, 2);
... run script ...
}
}
You can easily achieve this in Perl using File::ChangeNotify. This module is to be found on CPAN: http://search.cpan.org/dist/File-ChangeNotify/lib/File/ChangeNotify.pm
You can run the code as a daemon or as a service, make it watch one or more directories and then automatically execute some code (or start up a script) if some condition matches.
Best of all, it's cross-platform, so should you want to switch to a Linux machine or a Mac, it would still work.
It wouldn't be too hard to put together a small C# application that uses the FileSystemWatcher class to detect files being added to a folder and then spawn the required script. It would certainly use less CPU / system resources / hard disk bandwidth than polling the folder at regular intervals.
You need to consider what is a sufficient heuristic for determining "modified".
In increasing order of cost and accuracy:
file size (file content can still be changed as long as size is maintained)
file timestamp (If you aren't running ntpd time is not monotonic)
file sha1sum (bulletproof but expensive)
I would run ntpd, and then loop over the timestamps, and then compare the checksum if the timestamp changes. This can cover a lot of ground in little time.
These methods are not appropriate for a computer security application, they are for file management on a sane system.