Crash Course to Hollywood Programming

Author: jPV/RNO (Proofreading by Marq)

Preface

Hollywood - the cross-platform multimedia application layer - is an easy programming/scripting language that is designed for the creation of graphical or multimedia-oriented applications, but it can also be used to create other kinds of programs, especially with its comprehensive selection of plugins. Hollywood is loosely based on the popular Lua language.

Hollywood programs are created by writing script files, which are then run from the Hollywood main program, or they can be compiled into applets or standalone executables to be used without the Hollywood environment.

Applets are small platform-independent binary files which can be run using the free Hollywood player on any supported platform. Standalone executables contain the player, a script and optional data files linked into a single executable file. Linking the whole player binary into an executable makes the minimum size of Hollywood executables around 2 megabytes on MorphOS, but then the resulting programs are completely independent and can be run out of the box without any prior installation.

What makes Hollywood special is its ability to cross-compile programs between all supported platforms. Windows, Linux, OS X, or Amiga programs can be compiled just by using MorphOS, and so on. Cross-compiled programs don't usually need any modifications and the same source script will work on all platforms.

Hollywood is an easy, safe, and very well documented language, and thus also suitable for beginners. All Hollywood programs run inside a sandbox, and you can't break or crash the underlying operating system with your programming mistakes. If there's a bug causing an error in the script execution, Hollywood just closes the program and frees the resources with no further harm.

Getting Started

Requirements

Hollywood

Hollywood is a commercial product which can be bought here. The main package is enough for Hollywood programming and compiling for all other platforms except Android (Windows-only APK Compiler is required for the Android support).

Hollywood Designer is an Amiga-only add-on to create multimedia presentations in a WYSIWYG GUI. It works on Amiga compatible platforms including MorphOS, but the resulting presentations can be compiled for any other supported platform. Anyway, it isn't required for programming with Hollywood, but your own code can be inserted into presentation scripts too.

Text Editor

Hollywood features an IDE (integrated development environment) for Windows, but on other platforms a text editor or a 3rd party IDE is needed.

Cubic IDE offers the best development environment for Hollywood programming on MorphOS. With the Cubic IDE add-on you'll get syntax highlighting, automatic casing, online help, quick access to documentation, automatic indentation, bracket matching, auto completion of keywords, compile and run options, and a lot more.

MiniHollyEdit NG could also be a good freeware option, but it's broken in its current state and needs an update first.

An IDE with Hollywood support is the most comfortable solution of course, but any other text editor will also do.

Running a Script

Hollywood scripts can be run from the Hollywood main executable (the interpreter), and while the main executable accepts tens of command line arguments, you don't usually need many of them.

When you're developing and testing your application, it's enough to just run the script without compiling it. It's a good practise to save your scripts with the .hws extension.

To run a script from the shell:

Hollywood MyProgram.hws

And without the information window:

Hollywood MyProgram.hws -quiet

Hollywood has a built-in resource monitor, which can be activated with the -resourcemonitor argument. You should check the resources your application uses and that you aren't increasing the memory usage carelessly:

Hollywood MyProgram.hws -resourcemonitor

If you're using Cubic IDE, you don't need the shell at all. You can just press the F4 key to run the script or select other running options from Cubic IDE's menus.

Compiling a Script

When an application is ready for publishing, it should be compiled into a standalone executable, which is done with the -compile argument:

Hollywood MyProgram.hws -compile MyProgram

With the above command you get an executable called "MyProgram" that will run on the same archtecture it was compiled on.

To compile an application for different platforms (OS4 in this example):

Hollywood MyProgram.hws -compile MyProgram -exetype amigaos4

Or for multiple platforms at once:

Hollywood MyProgram.hws -compile MyProgram.exe -exetype morphos|amigaos4|win32|applet

If you want to compress your own data (script and linked data files), add the -compress argument:

Hollywood MyProgram.hws -compile MyProgram.exe -exetype morphos|amigaos4|win32|applet -compress

Compressing can make executables smaller, but it also "hides" any plaintext data from the executables if you prefer that.

Compiling in Cubic IDE is simply done by pressing the F2 key or by selecting the platform for cross-compiling from the menus.

First Test

Let's type this into a text editor and save it as a file called RAM:firsttest.hws:

@BGPIC 1, "SYS:MorphOS/Prefs/Wallpapers/1024x768/Arida.png"
TextOut(100,10,"Hello World!")
WaitLeftMouse() 
; Program exits after the last line in a script, but you can also terminate it with the End() command

The first line is a preprocessor command which loads the given image file as a Hollywood background picture. Objects in Hollywood are handled by their ID numbers, and in this case we load the image as a bgpic number 1. That bgpic is also loaded initially when the application opens its window.

All preprocessor commands are prefixed with the @ character and processed before the actual script execution. Data files loaded with the preprocessor commands are also linked into the executable when compiling a program. If you don't want to link the files, you can load them separately later in the script.

The second line prints the "Hello World!" text at x=100, y=10 (in pixels) over the background image.

The third line waits for the user to press the left mouse button. Without this line, the program would just end and close its window, and we wouldn't see much.

The last line is a comment line, which is just a note inside the code. Comment lines can be inserted between /* and */, or the rest of the line can be commented out with the ; character.

Let's run the script to see the results. Hit the F4 key in Cubic IDE or type this in the shell:

Hollywood RAM:firsttest.hws -quiet

Hollywood opens a new window and takes its dimensions from the background picture. If we wouldn't preload any background picture, then the display would be opened in the default 640x480 resolution, or you could define its dimensions with the @DISPLAY preprocessor command.

Next let's compile the program into a standalone executable. Hit the F2 key in Cubic IDE or use the shell (you don't have to define the paths if you've already changed directory to RAM: in the shell):

Hollywood RAM:firsttest.hws -compile RAM:firsttest

Now we have a new executable file that also contains the preprocessed/linked background graphics in it. The program works even if you don't have the Arida.png file in the system.

Hollywood Programming Example

The previous example could be extended to have more commands to be executed in batch, which could be fine for presentations or slideshows, but let's take a more practical approach and create an interactive script whose structure looks like a real program.

@APPTITLE "Hollywood Example"
@APPVERSION "$VER: Hollywood Example 1.0 (26.10.17)"
@APPDESCRIPTION "A Hollywood example program."

There are several preprocessor commands for adding informative strings to an application. These three are especially useful for Amiga and MorphOS programs, because Hollywood programs act like commodities and you'll see this information in, for example, the Exchange program. The @APPVERSION format should respect the standard version string rules, which means that the date should be given in the DD.MM.YY format etc.

@DISPLAY {Title="Hollywood Example", ScreenTitle="Hollywood Example", Width=400, Height=300, Sizeable=True, Layers=True, ScaleMode=#SCALEMODE_LAYER, SmoothScale=True}

This time we don't use the @BGPIC preprocessor, but define the display size manually and add some extra attributes too.

Title defines what will be shown in the window title bar, and ScreenTitle what is shown in the screen title bar when our program window is active.

The Width and Height attributes define the actual drawing area (display) for our application. The application window itself will be a bit larger with the standard OS window borders added around a Hollywood display. It is also possible to open borderless and fullscreen windows without the standard window borders.

The Sizeable attribute defines if the application window is resizeable by the user.

Layers=True enables Hollywood's layer system. It depends on the program you're developing whether you want to enable layers or not. With layers enabled all drawing operations get their own layer that can be moved, removed, and otherwise processed. When they're disabled, all graphics are just painted over each other on the display and it's harder to change or undo things. But if you're just drawing fullscreen graphics to the display or something like that, it's better to keep layers disabled.

ScaleMode sets if display contents are scaled automatically when the display size changes. Many times it's better to keep this disabled and recalculate your own graphics after a resize event to keep the aspect ratio intact and to avoid slowdowns, but let's enable this now to get more functionality into our application without extra lines of code.

SmoothScale makes the scaling output prettier at the expense of speed.

@BRUSH 1, "MOSSYS:Data/Jalapeno/ReadMode.png", {LoadAlpha=True}

Brushes are the most common way to load, process, and display graphics in Hollywood. This preprocessor line loads a PNG image as a brush with ID number 1. All brush handling operations are done by referring to the ID number from now on. The last argument is an optional table containing any extra options for loading. We are using graphics with an alpha channel (transparency) now, so we tell Hollywood to load alpha channel values too.

Function p_DrawGfx()
    Local brush_number = Rnd(2) + 1
    If LayerExists("gfx") Then RemoveLayerFX("gfx")
    SetDisplayAttributes({Title = "Brush " .. brush_number})
    DisplayBrushFX(brush_number, #CENTER, #CENTER)
    SetLayerName(0, "gfx")
EndFunction

This is how you create your own functions in Hollywood. Our own function names should always start with the p_ prefix to distinguish them from Hollywood's internal functions. In this case the function doesn't take any parameters and doesn't return any values, it can just be called from the program with p_DrawGfx() to run the lines inside it.

The purpose of this function is to show one of the two brushes we have in memory. The brush ID numbers are 1 and 2, and we randomly pick one of them. Because we have the layer system enabled, we get a new layer every time we draw a brush into the display. To avoid ending up with unlimited number of layers we remove an existing layer first. We also change the window title bar text to show the number of the shown brush.

The first line of the function declares a new local variable called brush_number and assigns a random value (1 or 2) to it. You should declare new variables as local whenever possible; read more about it here.

The second line checks if we already have a specific layer created, and removes it with a fancy random effect if we do. This line is just skipped when the function is called the first time, because we don't have the specific layer yet.

The third line changes the window title text to the brush number we got on the first line. Almost all display attributes can be changed on the fly, and you aren't restricted to the ones you defined with the @DISPLAY preprocessor command. Here you can also see how you can combine a string from other strings or variables with the .. operator, which is familiar from Lua.

DisplayBrushFX() shows the selected brush and creates a new layer on the layer stack. The position of the brush can be given in pixels, but also using inbuilt constants like #CENTER. FX at the end of the function name indicates that there's a transition effect displayed when the function is called. In most cases they are a bit over the top, and you probably end up using DisplayBrush() or other options to render the graphics.

Layers get ID numbers automatically and they are assigned dynamically, which means that removing a layer changes the ID numbers of the layers on top of it. To make life easier we can name the layers and handle them by their names instead of their ID numbers. SetLayerName() is one way to do that, and we use it to name the layer with the brush here. The function accepts a layer ID number and name as parameters. Our displayed brush creates the first layer on the layer stack and its ID number will thus be 1. SetLayerName(1,"gfx") would work in this case, and our layer would get the name "gfx" as intended. But to make life even easier we can give ID number 0 as the first parameter, and the name will be given to the last created layer. This way you don't have to keep count of layers at all.

Function p_Input(msg)
    DebugPrint("Input event!", msg.Action)
    Switch(msg.Action)
    Case "OnKeyDown":
        DebugPrint(msg.Key, "key was pressed.")
        Switch(msg.Key)
        Case " ":
            p_DrawGfx()
        Case "ESC":
            DebugPrint("Quitting...")
            End
        EndSwitch
    Case "OnMouseDown":
        p_DrawGfx()
    EndSwitch
EndFunction

Here's another function created by us. It is used to handle user input events. Event handling functions will get a message table as their parameter, and we'll get it as a local variable called "msg".

DebugPrint() is a useful function for checking what values we have in certain variables during the program execution. Don't hesitate to use it temporarily to learn what results you get from different functions and what values you have in variables. Here it is used to print the Action field from the message table, which tells us the name of the event that occured.

If the user presses a key, we get an OnKeyDown event, and if the user presses the left mouse button, we get an OnMouseDown event. A full list of standard events can be found here.

We handle different events with a Switch-Case statement in this example, but feel free to handle it in other ways if you please. It is also possible to have a separate function for each event.

If we take a look at the documentation, we see that the OnKeyDown event contains the exact key that was pressed in its Key field. It can be accessed with the msg.Key table item in our function.

The inner Switch-Case statement checks if the pressed key was space or escape. If it was space, our previous drawing function is called and you get new graphics on the display, whereas pressing the escape key quits the program with a message. Pressing any other keys will just print the key without doing anything else.

Pressing the left mouse button also calls our previous drawing function, and the display gets refreshed.

LoadBrush(2, "MOSSYS:Data/Jalapeno/TrackMode.png", {LoadAlpha=True})

The first image in our example program was loaded with the @BRUSH preprocessor command, but let's load the second image separately. The difference between them is that by doing the latter the image doesn't get linked into our executable file when we compile the program. The image is loaded from the filesystem every time you run the program, no matter if it's compiled or not.

p_DrawGfx()

Now that we have both images loaded as brushes, we can call the function we defined earlier, and we get some graphics shown in the application window.

InstallEventHandler({OnKeyDown = p_Input, OnMouseDown = p_Input})

Our previously defined p_Input() function doesn't get any events unless we install event handlers. A table passed to the InstallEventHandler() function defines which events are listened to and what functions they call when triggered. Here we define both OnKeyDown and OnMouseDown events to call our own p_Input() function, and user input will work from now on. As you can see, events can call the same function or they can be made to call other functions if necessary.

Repeat
    WaitEvent
Forever 

This is the main loop of the program. The WaitEvent function is called repeatedly as long as our program is running and it only waits for any event to happen. The program just sleeps and doesn't waste any CPU time when there aren't any events.

System-Friendly GUI Example

Standard Hollywood applications are somewhat limited when implementing GUI functionality into them. System menus and all kinds of requesters can be made, but if you want buttons or other GUI elements inside the Hollywood display, you have to draw them with your own graphics and handle them in custom ways.

But luckily Hollywood has a powerful cross-platform plugin system, and there are many nice plugins available for it. Some of the most impressive ones are MUI Royale and RapaGUI plugins, which can be used to create native system-friendly GUIs for applications. Both are official plugins by the author of Hollywood, and thus well supported and documented.

MUI Royale, as its name suggests, allows you to create MUI GUIs for Amiga compatible platforms. RapaGUI, on the other hand, is a beast at its own level, because it allows you to create GUIs that work on several completely different platforms using the underlying GUI toolkit on each platform. Be it Windows, Linux, Mac, or Amiga compatible, you just write one GUI definition and it'll work on every OS without modifications. When you run a RapaGUI application, it will look and feel just like other programs on each platform, because it uses MUI on Amiga compatibles, GTK on Linux, and so on.

Both MUI Royale and RapaGUI are very similar - the definition of a GUI is created by writing an XML file, which is then included into your Hollywood program. The following example is made with MUI Royale, because we are now focusing on Hollywood programming under MorphOS, and MUI Royale offers a bit more operating system specific features than the more generic RapaGUI. In any case, when you have learned one of these, it's easy to jump to the other too.

So, let's start.

The most important thing to learn now is to look at information about applicability in the documentation. Every available object attribute has its applicability described on its documentation page. Applicability is simply marked with four letters: I, S, G, and N. If there's I, you can use the attribute at initialization, which means it can be used when writing the XML file. S (set value of a MUI object attribute) and G (get value of a MUI object attribute) can be used from the actual Hollywood script at runtime. N can be used in both to handle notifications of attributes.

Let's have a look at the XML file first. Create a text file and save it as muigui.xml.

<?xml version="1.0" encoding="iso-8859-1"?>

This is a standard XML prologue line, and the file should start with it. ISO-8859-1 might be a safer bet than UTF-8 in our case. Version means the version of the XML markup language, not the application version. Just copy and paste this line as is.

<application base="HWEXAMPLE" id="app" dropobject="disp" icon="muiexample.info">

The application object is the master object, and all other definitions should be under it in the object tree.

The base attribute (applicability IG) is important here, because it defines what name is used when saving MUI preferences and it's also used when creating an ARexx port for the application. Don't use spaces or special characters, and try to pick a unique name that won't be used by other programs.

The id attribute (I) is needed for every object you plan to access later from the Hollwood script. It's good to have it set for the application object too.

The last two attributes are pretty much optional, but let's have them as examples, because they're quite useful in general. The dropobject attribute (IS) defines which object gets notified if you drag and drop a file onto its appicon. In this case the object called "disp", which we'll define later, gets the notification.

A program's application icon can be defined with the icon attribute (I). Without a path it takes the file from the directory the program was launched from.

As you can see, all of these attributes have I in their applicability information. When you browse the documentation and look for new attributes or how to do things, applicability tells if you can use them in the XML file or during the program execution.

    <menustrip id="examplemenustrip">
        <menu title="Project">
            <item id="menu_about" notify="selected">About...</item>
            <item id="menu_about_mui" notify="selected">About MUI...</item>
            <item id="menu_about_muiroyale" notify="selected">About MUI Royale...</item>
            <item/>
            <item id="menu_quit" notify="selected" shortcut="Q">Quit</item>
        </menu>
        <menu title="Settings">
            <item id="menu_muisettings" notify="selected">MUI...</item>
        </menu> 
    </menustrip>

Every MUI application should contain a pull-down menu with items to open About MUI and About MUI Royale windows, and also an item to open the MUI preferences program. We define a menu here and assign it later to the main window of our program. Remember to define id attributes for all functional items as well as for the menustrip object itself. Let's also add our own about window as the first item in the "Project" menu and a quit option as the last item. The <item/> line adds a separator bar to the menu.

If we want to get informed when the user has selected a menu item, we have to add notify tags to the items. Read more about the notification mechanism here. When we read the Menuitem class chapter from the documentation, we can find out that the Selected attribute has the applicability of N, and that means we can setup a notification for it. So, notify="selected" does the trick.

You can also define keyboard shortcuts for any of the menu items with the Shortcut attribute. In this example we can quit the program by pressing the command-q key combination.

Remember the traditional Amiga style guide too. If an item opens a new interactive window or requester, add ... at the end of it.

    <window id="win" muiid="MAIN" title="Hollywood MUI Example" screentitle="Hollywood MUI Example 1.0" notify="closerequest" menustrip="examplemenustrip" appwindow="true">

Then it's time to define our application window. The most important attribute here is muiid, because without it users can't snapshot window size or position. muiid has to be exactly four characters long, and each window in your program must have a different muiid value. It doesn't matter if you use the same ID (like "MAIN") in different programs, because they save their settings to different files.

The title attribute sets the text in the window title bar of our program, and the screentitle attribute sets the text shown in the screen title bar when our window is active.

The closerequest notify will tell us if the window's close gadget has been clicked, and we can react to it in our main Hollywood script.

The menu object we defined earlier can be specified for this window with the menustrip attribute. If you use multiple windows in an application and would like to see the same menu in all of them, move this attribute from the <window> definition to the <application> definition.

The last appwindow attribute lets us listen if an icon (a file) is dragged and dropped onto our program window. It can be left out if you don't need it.

        <vgroup>
            <hgroup>
                <button id="button1" notify="pressed" cyclechain="true" shorthelp="Displays brush 1">Brush _1</button>
                <button id="button2" notify="pressed" cyclechain="true" shorthelp="Displays brush 2">Brush _2</button>
            </hgroup>

The objects in our window can be organized in vertical or horizontal groups. Here we create a vertical group with smaller horizontal groups inside it. On these lines we define that we get two buttons next to each other horizontally.

If we read the Button class documentation, we see that we can get notified by the Pressed attribute. We could also get notified by the Selected attribute if we created a toggle button.

cyclechain and shorthelp attributes are interesting, because they aren't button specific. They are actually attributes of the Area class, which is a super class for almost everything else, and you can use its attributes with about all objects. When you are defining an object, you could also check the attributes from the Area class documentation.

The cyclechain attribute defines if you can activate the object from the keyboard by using the tab key. You should usually allow this.

The contents of the MUI bubble help text can be defined with the shorthelp attribute. A bubble help window opens if you keep the mouse pointer over an object for a while.

An underscore in the button label makes the following character a keyboard shortcut. We can press "1" or "2" from the keyboard to make a button press now.

            <hgroup>
                <button id="button_random" notify="pressed" cyclechain="false" shorthelp="Displays a random brush">_Random</button>
                <rectangle/>
                <text id="clock" preparse="\33r">00:00:00</text>
            </hgroup>

Here we define another horizontal group whose objects are shown below the previous group.

The new button here is removed from the cyclechain as an example; you can't activate it with the tab key.

<rectangle/> adds some dynamic space between two objects. This can be quite useful when designing a GUI, and especially if you have some resizing problems.

The text object just displays some text, but we plan to update the text later with the current time. Let's put something as its initial value anyway. The preparse attribute contains a format code for the text, and we use it to right align the text now.

            <hollywood id="disp" display="1" minwidth="32" minheight="32" maxwidth="16384" maxheight="16384" notify="appmessage"/>

This object isn't inside of any horizonal group and takes the whole window width itself.

The hollywood object embeds a complete Hollywood display inside the GUI. display defines the ID number of the Hollywood display you want show in the object. Defining the sizes makes it resizeable too. If you want to make MUI programs without the graphical display of Hollywood, that's fine too.

        </vgroup>
    </window>
</application> 

After finishing our definitions we have working code for the GUI part of our program!