MGRX 1.4.3

a C graphics library for DOS, Linux console & X11 and Win32

Programmer's Manual

Based on the original GRX manual written by: Csaba Biegl on August 10, 1992

Adapted for MGRX by: Mariano Alvarez Fernández on August 1, 2005

Last update: April 13, 2023


  MGRX is a 2D graphics library derived from the GRX library. GRX was originaly written by Csaba Biegl for DJ Delorie's DOS port of the GCC compiler.

  MGRX supports four platforms:

  For the Linux console and the Linux X11 versions three architectures are supported: i386 (32bit), x86_64 (64bit) and ARM (32bit, tested in a Raspberry Pi 4 using Raspbian).

  Additionally MGRX can work without graphic output using the memory driver.


  • MGRX versus GRX
  • Hello world
  • Data types, function declarations
  • Setting the graphics driver
  • Setting video modes
  • Graphics contexts
  • How to use graphics contexts
  • Example: double buffer
  • Color management
  • Portable use of a few colors
  • Graphics primitives
  • Non-clipping graphics primitives
  • Customized line drawing
  • Patterns (bitmap, pixmaps and gradients)
  • Pattern filled graphics primitives
  • Patterned line drawing
  • Multipolygons
  • About text encoding
  • Text drawing
  • Special UTF-8 text type considerations
  • Recode to/from GR_UCS2_TEXT chartype
  • Internationalization
  • Drawing in user coordinates
  • Graphics cursors
  • Writing/reading PNM graphics files
  • Writing/reading PNG graphics files
  • Writing/reading JPEG graphics files
  • Miscellaneous functions
  • Input API
  • Mouse cursor handling
  • Handling window resizing
  • Clipboard
  • The memory driver
  • References

  • MGRX versus GRX

      MGRX ( is a fork of GRX ( The main MGRX differences are:

      The MGRX goal is to be a small, easy to use and maintain, C graphics library.

    Hello world

      The next program draws a double frame around the screen and writes "Hello, MGRX world" centered, then it waits until a key is pressed.

    #include <string.h>
    #include <mgrx.h>
    #include <mgrxkeys.h>
    int main()
        char *message = "Hello, MGRX world";
        int x, y;
        GrTextOption grt;
        GrEvent ev;
        grt.txo_font = GrGetDefaultFont();
        grt.txo_fgcolor = GrWhite();
        grt.txo_bgcolor = GrBlack();
        grt.txo_direct = GR_TEXT_RIGHT;
        grt.txo_xalign = GR_ALIGN_CENTER;
        grt.txo_yalign = GR_ALIGN_CENTER;
        grt.txo_chrtype = GR_BYTE_TEXT;
        GrBox(0, 0, GrMaxX(), GrMaxY(), GrWhite());
        GrBox(4, 4, GrMaxX()-4, GrMaxY()-4, GrWhite());
        x = GrMaxX() / 2;
        y = GrMaxY() / 2;
        GrDrawString(message, strlen(message), x, y, &grt);
        return 0;

      How to compile the hello world program (assuming the MGRX library was previously installed)

    If you have enabled the experimental Linux console KMS/DRM driver add "-ldrm" to the command.

    Data types, function declarations

      All public data structures and graphics primitives meant for usage by the application program are declared/prototyped in the header files (in the 'include' sub-directory):

      User programs normally only include mgrx.h and mgrxkeys.h and optionally mgrxcolr.h

    Setting the graphics driver

      The graphics driver is normally autodetected or set by the final user using the environment variable MGRXDRV, but a program can set it by function:

    int GrSetDriver(char *drvspec);

    on succes it returns non-zero (TRUE). The drvspec string has the same format as the environment variable:

    <driver> gw <width> gh <height> nc <colors>

    available drivers are for:

    the optionals gw, gh and nc parameters set the desired default graphics mode. Normal values for 'nc' are 2, 16, 256, 64K and 16M.

      If neither the MGRXDRV variable exist, nor the driver is set by program, MGRX will try to detect the correct driver when first calling GrSetMode (see bellow)

      After initiated the current driver name can be obtained from:


      The type of the installed graphics adapter can be determined with the function:

    GrVideoAdapter GrAdapterType(void);

    it returns the type of the adapter as one of the following symbolic constants (defined in mgrx.h):

    typedef enum _GR_videoAdapters {
      GR_UNKNOWN = (-1),     /* not known (before driver set) */
      GR_VGA,                /* VGA adapter */
      GR_EGA,                /* EGA adapter */
      GR_XWIN,               /* X11 driver */
      GR_WIN32,              /* W32 driver */
      GR_LNXFB,              /* Linux console driver */
      GR_MEM                 /* memory only driver */
    } GrVideoAdapter;

      Note that the VESA driver return GR_VGA here.

    Setting video modes

      Before a program can do any graphics drawing it has to configure the graphics driver for the desired graphics mode. It is done with the GrSetMode function as follows:

    int GrSetMode(int which,...); 

      On succes it returns non-zero (TRUE). The which parameter can be one of the following constants, declared in mgrx.h:


      The GR_default_graphics inits the graphic screen with the resolution and color depth defined in the MGRXDRV environment variable (or by te GrSetDriver function or by the the driver defaults).

      The GR_default_text can be used, at the end of the program, to return the graphics screen to his original state.

      The GR_biggest_graphics inits the graphic screen with the maximum resolution and color depth supported by the graphic driver. For Win32 and X11 this goes to full screen. If you want to limit the maximum resolution to use, you can set the environment variable MGRXBIGG with "maxw maxh".

      The GR_width_height_graphics mode requires the two size arguments: int width and int height.

      The GR_width_height_color_graphics mode requires three arguments: int width, int height and GrColor colors.

      The GR_width_height_bpp_graphics mode requires three arguments: int width, int height and int bpp (bits per plane instead number of colors).

      The GR_custom_graphics and GR_custom_bpp_graphics modes require five arguments: int width, int height, GrColor colors or int bpp, int vx and int vy. Using this modes you can set a virtual screen of vx by vy size, but it only works on some videomodes with the DOS drivers.

      The GR_NC_... modes are equivalent to the GR_.. ones, but they don't clear the video memory.

      Graphics drivers can provide info of the supported graphics modes, use the next code skeleton to colect the data:

        GrFrameMode fm;
        const GrVideoMode *mp;
        for (fm =GR_firstGraphicsFrameMode; fm <= GR_lastGraphicsFrameMode; fm++) {
            mp = GrFirstVideoMode(fm);
            while (mp != NULL) {
                .. use the mp info
                mp = GrNextVideoMode(mp))

      Don't worry if you don't understand it, normal user programs don't need to know about FrameModes. The GrVideoMode structure has the following fields:

    typedef struct _GR_videoMode GrVideoMode;
    struct _GR_videoMode {
      char    present;                    /* is it really available? */
      char    bpp;                        /* log2 of # of colors */
      short   width,height;               /* video mode geometry */
      short   mode;                       /* BIOS mode number (if any) */
      int     lineoffset;                 /* scan line length */
      int     privdata;                   /* driver can use it for anything */
      struct _GR_videoModeExt *extinfo;   /* extra info (maybe shared) */

    the width, height and bpp members are the useful information if you are interested in set modes other than the GR_default_graphics.

    The current graphics mode (one of the valid mode argument values for GrSetMode) can be obtained with:

    GrGraphicsMode GrCurrentMode(void);

    while the current GrVideoMode can be determined with:

    GrVideoMode * GrCurrentVideoMode(void);

      A user-defined function can be invoked every time the video mode is changed (i.e. GrSetMode is called). This function should not take any parameters and don't return any value. It can be installed (for all subsequent GrSetMode calls) with the function:

    void GrSetModeHook(void (*hookfunc)(void));

      Important note. If you ask for a no supported video mode GrSetMode will return the more similar mode it founds, so be prepared to use different dimensions and number of colors of those you asked for.

    Graphics contexts

      The library supports a set of drawing regions called contexts (the GrContext structure). These can be in video memory or in system memory. Contexts in system memory always have the same memory organization as the video memory. When GrSetMode is called, a default context is created which maps to the whole graphics screen. Contexts are described by the GrContext data structure:

    typedef struct _GR_context GrContext;
    struct _GR_context {
      struct _GR_frame    gc_frame;       /* frame buffer info */
      struct _GR_context *gc_root;        /* context which owns frame */
      int    gc_xmax;                     /* max X coord (width  - 1) */
      int    gc_ymax;                     /* max Y coord (height - 1) */
      int    gc_xoffset;                  /* X offset from root's base */
      int    gc_yoffset;                  /* Y offset from root's base */
      int    gc_xcliplo;                  /* low X clipping limit */
      int    gc_ycliplo;                  /* low Y clipping limit */
      int    gc_xcliphi;                  /* high X clipping limit */
      int    gc_ycliphi;                  /* high Y clipping limit */
      int    gc_usrxbase;                 /* user window min X coordinate */
      int    gc_usrybase;                 /* user window min Y coordinate */
      int    gc_usrwidth;                 /* user window width  */
      int    gc_usrheight;                /* user window height */
    # define gc_baseaddr                  gc_frame.gf_baseaddr
    # define gc_selector                  gc_frame.gf_selector
    # define gc_onscreen                  gc_frame.gf_onscreen
    # define gc_memflags                  gc_frame.gf_memflags
    # define gc_lineoffset                gc_frame.gf_lineoffset
    # define gc_driver                    gc_frame.gf_driver

      The following four functions return information about the layout of and memory occupied by a graphics context of size width by height in the current graphics mode (as set up by GrSetMode):

    int GrLineOffset(int width); 
    int GrNumPlanes(void); 
    long GrPlaneSize(int w,int h); 
    long GrContextSize(int w,int h); 

      GrLineOffset always returns the offset between successive pixel rows of the context in bytes. GrNumPlanes returns the number of bitmap planes in the current graphics mode. GrContextSize calculates the total amount of memory needed by a context, while GrPlaneSize calculates the size of a bitplane in the context. The function:

    GrContext *GrCreateContext(int w,int h,char *memory[4],GrContext *where);

    can be used to create a new context in system memory. The NULL pointer is also accepted as the value of the memory and where arguments, in this case the library allocates the necessary amount of memory internally. It is a general convention in the library that functions returning pointers to any MGRX specific data structure have a last argument (most of the time named where in the prototypes) which can be used to pass the address of the data structure which should be filled with the result. If this where pointer has the value of NULL, then the library allocates space for the data structure internally.

      The memory argument is really a 4 pointer array, each pointer must point to space to handle GrPlaneSize(w,h) bytes, really only GrNumPlanes() pointers must be malloced, the rest can be NULL. Nevertheless the normal use (see below) is gc = GrCreateContext(w,h,NULL,NULL), so yo don't need to care about.

      The function:

    GrContext *GrCreateSubContext(int x1,int y1,int x2,int y2,
                                  const GrContext *parent,GrContext *where);

    creates a new sub-context which maps to a part of an existing context. The coordinate arguments (x1 through y2) are interpreted relative to the parent context's limits. Pixel addressing is zero-based even in sub-contexts, i.e. the address of the top left pixel is (0,0) even in a sub-context which has been mapped onto the interior of its parent context.

      Sub-contexts can be resized, but not their parents (i.e. anything returned by GrCreateContext or set up by GrSetMode cannot be resized -- because this could lead to irrecoverable "loss" of drawing memory. The following function can be used for this purpose:

    void GrResizeSubContext(GrContext *context,int x1,int y1,int x2,int y2); 

      The current context structure is stored in a static location in the library. (For efficiency reasons -- it is used quite frequently, and this way no pointer dereferencing is necessary.) The context stores all relevant information about the video organization, coordinate limits, etc... The current context can be set with the:

    void GrSetContext(const GrContext *context);

    function. This function will reset the current context to the full graphics screen if it is passed the NULL pointer as argument. The value of the current context can be saved into a GrContext structure pointed to by where using:

    GrContext *GrSaveContext(GrContext *where); 

    again, if where is NULL, the library allocates the space, and you must remember to destory it, but remember this is a copy, not the original context, so for typical use of this fuction (save the current context, set a new context, do some drawings anf restore the original context) is better to provide where. Example:

    void do_something()
        GrContext grc; 
        GrSetContext(another_context); some drawing;

    The next two functions:

    const GrContext *GrCurrentContext(void);
    const GrContext *GrScreenContext(void);

    return the current context and the screen context respectively. Contexts can be destroyed with:

    void GrDestroyContext(GrContext *context);

      This function will free the memory occupied by the context only if it was allocated originally by the library. The next three functions set up and query the clipping limits associated with the current context:

    void GrSetClipBox(int x1,int y1,int x2,int y2); 
    void GrGetClipBox(int *x1p,int *y1p,int *x2p,int *y2p); 
    void GrResetClipBox(void);

      GrResetClipBox sets the clipping limits to the limits of context. These are the limits set up initially when a context is created. There are three similar functions to sets/gets the clipping limits of any context:

    void  GrSetClipBoxC(GrContext *c,int x1,int y1,int x2,int y2);
    void  GrGetClipBoxC(const GrContext *c,int *x1p,int *y1p,int *x2p,int *y2p);
    void  GrResetClipBoxC(GrContext *c);

      The limits of the current context can be obtained using the following functions:

    int GrMaxX(void); 
    int GrMaxY(void); 
    int GrSizeX(void); 
    int GrSizeY(void);

      The Max functions return the biggest valid coordinate, while the Size functions return a value one higher. The size of the graphics screen (regardless of the current context) can be obtained with:

    int GrScreenX(void); 
    int GrScreenY(void); 

      If you had set a virtual screen (using a custom graphics mode), the limits of the virtual screen can be fetched with:

    int GrVirtualX(void);
    int GrVirtualY(void);

      The routine:

    int GrScreenIsVirtual(void);

    returns non zero if a virtual screen is set. The rectangle showed in the real screen can be set with:

    int GrSetViewport(int xpos,int ypos);

    and the current viewport position can be obtained by:

    int GrViewportX(void);
    int GrViewportY(void);

    How to use graphics contexts

      Here is an example of standard graphic context use:

    GrContext *grc;
    if ((grc = GrCreateContext(w, h, NULL, NULL)) == NULL) {
        ...process the error
    else {
        GrSetContext grc); /* now all drawing will be in grc! */ some drawing
        ...and probably bitblt to the screen context
        GrSetContext(NULL); /* return to the screen context! */

      But if you have a GrContext variable (not a pointer) that you want to use (probably because is static to some routines) you can do:

    static GrContext grc; /* not a pointer!! */
    if (GrCreateContext(w, h, NULL, &grc)) == NULL) {
        ...process the error
    else {
        GrSetContext(&grc); /* now all drawing will be in grc! */ some drawing
        ...and probably bitblt to the screen context
        GrSetContext(NULL); /* return to the screen context! */

      Note that GrDestoryContext knows if grc was automatically malloced or not!!

      Only if you don't want GrCreateContext use malloc at all, you must allocate the memory buffers and pass it to GrCreateContext.

      Using GrCreateSubContext is the same, except it doesn't need the buffer, because it uses the parent buffer.

    Example: double buffer

      Using contexts it is easy to implement double buffering, a technique to first draw each frame in memory and display them on screen at once, this makes changes to the screen appear much smoother:

    GrContext *grc;
    if ((grc = GrCreateContext(GrScreenX(), GrScreenY(), NULL, NULL)) == NULL) {
      ...process the error and exit
    GrSetContext( grc ); /* now all drawing will be in grc! */
    while (1) {
        ...draw the frame
        GrBitBlt(GrScreenContext(), 0, 0, grc, 0, 0, grc->gc_xmax, grc->gc_ymax, GrWRITE); process user input
        ...and break at exit

    Color management

      MGRX defines the type GrColor for color variables. GrColor it's a 32 bits integer. The 8 left bits are reserved for the write mode (see below). The 24 bits right are the color value.

      The library supports two models for color management. In the 'indirect' (or color table) model, color values are indices to a color table. The color table slots will be allocated with the highest resolution supported by the hardware (EGA: 2 bits, VGA: 6 bits) with respect to the component color intensities. In the 'direct' (or RGB) model, color values map directly into component color intensities with non-overlapping bitfields of the color index representing the component colors.

      Color table model is supported until 256 color modes. The RGB model is supported in 256 color and up color modes.

      In RGB model the color index map to component color intensities depend on the video mode set, so it can't be assumed the component color bitfields (but if you are curious check the GrColorInfo global structure in mgrx.h).

      After the first GrSetMode call two colors are always defined: black and white. The color values of these two colors are returned by the functions:

    GrColor GrBlack(void); 
    GrColor GrWhite(void);

      GrBlack() is guaranteed to be 0.

      The library supports five write modes (a write mode descibes the operation between the actual bit color and the one to be set): write, XOR, logical OR, logical AND and IMAGE. These can be selected with OR-ing the color value with one of the following constants declared in mgrx.h :

    #define GrWRITE       0UL            /* write color */
    #define GrXOR         0x01000000UL   /* to "XOR" any color to the screen */
    #define GrOR          0x02000000UL   /* to "OR" to the screen */
    #define GrAND         0x03000000UL   /* to "AND" to the screen */
    #define GrIMAGE       0x04000000UL   /* BLIT: write, except given color */

      The GrIMAGE write mode only works with the bitblt function.

      By convention, the no-op color is obtained by combining color value 0 (black) with the XOR operation. This no-op color has been defined in mgrx.h as:

    #define GrNOCOLOR     (GrXOR | 0)    /* GrNOCOLOR is used for "no" color */

      The write mode part and the color value part of a GrColor variable can be obtained OR-ing it with one of the following constants declared in mgrx.h:

    #define GrCVALUEMASK  0x00ffffffUL   /* color value mask */
    #define GrCMODEMASK   0xff000000UL   /* color operation mask */

      The number of colors in the current graphics mode is returned by the:

    GrColor GrNumColors(void);

    function, while the number of unused, available color can be obtained by calling:

    GrColor GrNumFreeColors(void); 

      Colors can be allocated with the:

    GrColor GrAllocColor(int r,int g,int b);
    GrColor GrAllocColor2(long hcolor);

    functions (component intensities can range from 0 to 255, hcolor must be in 0xRRGGBB format), or with the:

    GrColor GrAllocCell(void);

    function. In the second case the component intensities of the returned color can be set with:

    void GrSetColor(GrColor color,int r,int g,int b); 

      In the color table model all Alloc functions return GrNOCOLOR if there are no more free colors available. In the RGB model GrNumFreeColors returns 0 and GrAllocCell always returns GrNOCOLOR, as colors returned by GrAllocCell are meant to be changed -- what is not supposed to be done in RGB mode. Also note that GrAllocColor operates much more efficiently in RGB mode, and that it never returns GrNOCOLOR in this case.

      Color table entries can be freed (when not in RGB mode) by calling:

    void GrFreeColor(GrColor color);

      The component intensities of any color can be queried using one of this functions:

    void GrQueryColor(GrColor c,int *r,int *g,int *b); 
    void GrQueryColor2(GrColor c,long *hcolor); 

      Initially the color system is in color table (indirect) model if there are 256 or less colors. 256 color modes can be put into the RGB model by calling:

    void GrSetRGBcolorMode(void);

      The color system can be reset (i.e. put back into color table model if possible, all colors freed except for black and white) by calling:

    void GrResetColors(void);

      The function:

    void GrRefreshColors(void);

    reloads the currently allocated color values into the video hardware. This function is not needed in typical applications, unless the display adapter is programmed directly by the application.

      This functions:

    GrColor GrAllocColorID(int r,int g,int b);
    GrColor GrAllocColor2ID(long hcolor);
    void GrQueryColorID(GrColor c,int *r,int *g,int *b);
    void GrQueryColor2ID(GrColor c,long *hcolor);

    are inlined versions (except if you compile MGRX with GRX_SKIP_INLINES defined) to be used in the RGB model (in the color table model they call the normal routines).

    Portable use of a few colors

      People that only want to use a few colors find the MGRX color handling a bit confusing, but it gives the power to manage a lot of color deeps and two color models. Here are some guidelines for using a few colors in a portable way in MGRX programs.

    Using the include file "mgrxcolr.h" (new in version 1.1.2)

      The "mgrxcolr.h" provides defines and functions to use three color tables, a EGA one with 16 colors that you can use with 4bpp and more graphics modes; another table with the 138 WEB colors, as defined by the, that you can use with 8bpp and more graphics modes; and a last one with the 256 ANSI colors for use too with 8bpp and more graphics modes.

      For each table you have:

      There is an example of how to use each case in the "clrtable.c" test program.

    Using the GrAllocEgaColors function

      This is the classic method to use the 16 EGA colors in a portable way that also works in versions prior to 1.1.2. We need this MGRX function:

    GrColor *GrAllocEgaColors(void);

    it returns a 16 GrColor array with the 16 ega colors alloced (really it's a trivial function, read the source src/setup/colorega.c). We can use a construction like that:

      First, in your C code make a global pointer, and init it after set the graphics mode:

    GrColor *egacolors;
    int your_setup_function( ... )
        GrSetMode( ... )
        egacolors = GrAllocEgaColors();

      Next, add this to your main include file:

    extern GrColor *egacolors;
    #define BLACK        egacolors[0]
    #define BLUE         egacolors[1]
    #define GREEN        egacolors[2]
    #define CYAN         egacolors[3]
    #define RED          egacolors[4]
    #define MAGENTA      egacolors[5]
    #define BROWN        egacolors[6]
    #define LIGHTGRAY    egacolors[7]
    #define DARKGRAY     egacolors[8]
    #define LIGHTBLUE    egacolors[9]
    #define LIGHTGREEN   egacolors[10]
    #define LIGHTCYAN    egacolors[11]
    #define LIGHTRED     egacolors[12]
    #define LIGHTMAGENTA egacolors[13]
    #define YELLOW       egacolors[14]
    #define WHITE        egacolors[15]

      Now you can use the defined colors in your code (by example: GrClearScreen(YELLOW);). Note that if you are in color table model in a 16 color mode, you have exhausted the color table. Note too that this don't work to initialize static variables with a color, because egacolors is not initialized.

    Graphics primitives

      The screen, the current context or the current clip box can be cleared (i.e. set to a desired background color) by using one of the following three functions:

    void GrClearScreen(GrColor bg); 
    void GrClearContext(GrColor bg); 
    void GrClearClipBox(GrColor bg);

    Any context can be cleared using this function:

    void GrClearContextC(GrContext *ctx, GrColor bg);

      Thanks to the special GrColor definition, you can do more than simple clear with this functions, by example with:

    GrClearScreen( GrWhite()|GrXOR );

    the graphics screen is negativized, do it again and the screen is restored.

      The following line drawing graphics primitives are supported by the library:

    void GrPlot(int x,int y,GrColor c);
    void GrLine(int x1,int y1,int x2,int y2,GrColor c); 
    void GrHLine(int x1,int x2,int y,GrColor c); 
    void GrVLine(int x,int y1,int y2,GrColor c); 
    void GrBox(int x1,int y1,int x2,int y2,GrColor c); 
    void GrCircle(int xc,int yc,int r,GrColor c); 
    void GrEllipse(int xc,int yc,int xa,int ya,GrColor c); 
    void GrCircleArc(int xc,int yc,int r,int start,int end,int style,GrColor c); 
    void GrEllipseArc(int xc,int yc,int xa,int ya,
                      int start,int end,int style,GrColor c);
    void GrPolyLine(int numpts,int points[][2],GrColor c); 
    void GrPolygon(int numpts,int points[][2],GrColor c); 

      All primitives operate on the current graphics context. The last argument of these functions is always the color to use for the drawing. The HLine and VLine primitives are for drawing horizontal and vertical lines. They have been included in the library because they are more efficient than the general line drawing provided by GrLine. The ellipse primitives can only draw ellipses with their major axis parallel with either the X or Y coordinate axis. They take the half X and Y axis length in the xa and ya arguments. The arc (circle and ellipse) drawing functions take the start and end angles in tenths of degrees (i.e. meaningful range: 0 ... 3600). The angles are interpreted counter-clockwise starting from the positive X axis. The style argument can be one of this defines from mgrx.h:

    #define GR_ARC_STYLE_OPEN       0
    #define GR_ARC_STYLE_CLOSE1     1
    #define GR_ARC_STYLE_CLOSE2     2

      GR_ARC_STYLE_OPEN draws only the arc, GR_ARC_STYLE_CLOSE1 closes the arc with a line between his start and end point, GR_ARC_STYLE_CLOSE2 draws the typical cake slice. This routine:

    void GrLastArcCoords(int *xs,int *ys,int *xe,int *ye,int *xc,int *yc);

    can be used to retrieve the start, end, and center points used by the last arc drawing functions.

      The polyline and polygon primitives take the address of an n by 2 coordinate array. The X values should be stored in the elements with 0 second index, and the Y values in the elements with a second index value of 1. Coordinate arrays passed to the polygon primitive can either contain or omit the closing edge of the polygon -- the primitive will append it to the list if it is missing.

      Because calculating the arc points it's a very time consuming operation, there are two functions to pre-calculate the points, that can be used next with polyline and polygon primitives:

    int  GrGenerateEllipse(int xc,int yc,int xa,int ya,
                           int points[GR_MAX_ELLIPSE_POINTS][2]);
    int  GrGenerateEllipseArc(int xc,int yc,int xa,int ya,int start,int end,
                              int points[GR_MAX_ELLIPSE_POINTS][2]);

      The following filled primitives are available:

    void GrFilledBox(int x1,int y1,int x2,int y2,GrColor c); 
    void GrFramedBox(int x1,int y1,int x2,int y2,int wdt,const GrFBoxColors *c);
    void GrFilledCircle(int xc,int yc,int r,GrColor c); 
    void GrFilledEllipse(int xc,int yc,int xa,int ya,GrColor c); 
    void GrFilledCircleArc(int xc,int yc,int r,
                           int start,int end,int style,GrColor c);
    void GrFilledEllipseArc(int xc,int yc,int xa,int ya,
                            int start,int end,int style,GrColor c);
    void GrFilledPolygon(int numpts,int points[][2],GrColor c); 
    void GrFilledConvexPolygon(int numpts,int points[][2],GrColor c); 

      Similarly to the line drawing, all of the above primitives operate on the current graphics context. The GrFramedBox primitive can be used to draw motif-like shaded boxes and "ordinary" framed boxes as well. The x1 through y2 coordinates specify the interior of the box, the border is outside this area, wdt pixels wide. The primitive uses five different colors for the interior and four borders of the box which are specified in the GrFBoxColors structure:

    typedef struct { 
      GrColor fbx_intcolor; 
      GrColor fbx_topcolor; 
      GrColor fbx_rightcolor; 
      GrColor fbx_bottomcolor; 
      GrColor fbx_leftcolor; 
    } GrFBoxColors;

      The GrFilledConvexPolygon primitive can be used to fill convex polygons. It can also be used to fill some concave polygons whose boundaries do not intersect any horizontal scan line more than twice. All other concave polygons have to be filled with the (somewhat less efficient) GrFilledPolygon primitive. This primitive can also be used to fill several disjoint nonoverlapping polygons in a single operation.

      The function:

    void GrFloodFill(int x, int y, GrColor border, GrColor c);

    flood-fills the area bounded by the color border using x, y like the starting point.

      Floodspill is a color replacer, replacing color A with color B. This is quite useful for highlighting a selected item in a list, or changing a selected color(s) in a multi colored area.

    void GrFloodSpill(int x1, int y1, int x2, int y2,
                     GrColor old_c, GrColor new_c)

    replaces old color with new color in the rectangle bounded by x1, y1, x2, y2.

    void GrFloodSpillC(GrContext *ctx, int x1, int y1, int x2, int y2,
                      GrColor old_c, GrColor new_c)

    as above but in the specified context.

    void GrFloodSpill2(int x1, int y1, int x2, int y2,
                      GrColor old_c1, GrColor new_c1,
                      GrColor old_c2, GrColor new_c2)

    replaces 2 colors, a one stop shop for highlighting a selection in a list.

    void GrFloodSpillC2(GrContext *ctx, int x1, int y1, int x2, int y2,
                      GrColor old_c1, GrColor new_c1,
                      GrColor old_c2, GrColor new_c2)

    as above but in the specified context.

      The current color value of any pixel in the current context can be obtained with:

    GrColor GrPixel(int x,int y);


    GrColor GrPixelC(GrContext *c,int x,int y);

    do the same for any context.

      Rectangular areas can be transferred within a context or between contexts by calling:

    void GrBitBlt(GrContext *dest,int x,int y,GrContext *source,
                  int x1,int y1,int x2,int y2,GrColor op);

    x, y is the position in the destination context, and x1, y1, x2, y2 the area from the source context to be transfered. The op argument should be one of supported color write modes (GrWRITE, GrXOR, GrOR, GrAND, GrIMAGE), it will control how the pixels from the source context are combined with the pixels in the destination context (the GrIMAGE op must be ored with the color value to be handled as transparent). If either the source or the destination context argument is the NULL pointer then the current context is used for that argument.

      GrBitblt works pixel by pixel, there is another function that can stretch the source to the destination:

    void GrStretchBlt(GrContext *dst,int dx1,int dy1,int dx2,int dy2,
                      GrContext *src,int x1,int y1,int x2,int y2,GrColor oper);

    here dx1, dy1, dx2, dy2 define the destination area to stretch the source area.

      A efficient form to get/put pixels from/to a context can be achieved using the next functions:

    const GrColor *GrGetScanline(int x1,int x2,int yy);
    const GrColor *GrGetScanlineC(GrContext *ctx,int x1,int x2,int yy);
    void GrPutScanline(int x1,int x2,int yy,const GrColor *c, GrColor op);

      The Get functions return a pointer to a static GrColor pixel array (or NULL if they fail) with the color values of a row (yy) segment (x1 to x2). GrGetScanline uses the current context. GrGestScanlineC uses the context ctx (that can be NULL to refer to the current context). Note that the output is only valid until the next MGRX call.

      GrPutScanline puts the GrColor pixel array c on the yy row segmet defined by x1 to x2 in the current context using the op operation. op can be any of GrWRITE, GrXOR, GrOR, GrAND or GrIMAGE. Data in c must fit GrCVALUEMASK otherwise the results are implementation dependend. So you can't supply operation code with the pixel data!.

    Non-clipping graphics primitives

      There is a non-clipping version of some of the elementary primitives. These are somewhat more efficient than the regular versions. These are to be used only in situations when it is absolutely certain that no drawing will be performed beyond the boundaries of the current context. Otherwise the program will almost certainly crash! The reason for including these functions is that they are somewhat more efficient than the regular, clipping versions. ALSO NOTE: These function do not check for conflicts with the mouse cursor. (See the explanation about the mouse cursor handling later in this document.) The list of the supported non-clipping primitives:

    void GrPlotNC(int x,int y,GrColor c); 
    void GrLineNC(int x1,int y1,int x2,int y2,GrColor c); 
    void GrHLineNC(int x1,int x2,int y,GrColor c); 
    void GrVLineNC(int x,int y1,int y2,GrColor c); 
    void GrBoxNC(int x1,int y1,int x2,int y2,GrColor c); 
    void GrFilledBoxNC(int x1,int y1,int x2,int y2,GrColor c); 
    void GrFramedBoxNC(int x1,int y1,int x2,int y2,int wdt,const GrFBoxColors *c);
    void GrBitBltNC(GrContext *dst,int x,int y,GrContext *src,
                    int x1,int y1,int x2,int y2,GrColor op);
    GrColor GrPixelNC(int x,int y);
    GrColor GrPixelCNC(GrContext *c,int x,int y);

    Customized line drawing

      The basic line drawing graphics primitives described previously always draw continuous lines which are one pixel wide. There is another group of line drawing functions which can be used to draw wide and/or patterned lines. These functions have similar parameter passing conventions as the basic ones with one difference: instead of the color value a pointer to a structure of type GrLineOption has to be passed to them. The definition of the GrLineOption structure:

    typedef struct {
      GrColor lno_color;             /* color used to draw line */
      int     lno_width;             /* width of the line */
      int     lno_pattlen;           /* length of the dash pattern */
      unsigned char *lno_dashpat;    /* draw/nodraw pattern */
    } GrLineOption;

      The lno_pattlen structure element should be equal to the number of alternating draw -- no draw section length values in the array pointed to by the lno_dashpat element. The dash pattern array is assumed to begin with a drawn section. If the pattern length is equal to zero a continuous line is drawn.

      Example, a white line 3 bits wide (thick) and pattern 6 bits draw, 4 bits nodraw:

    GrLineOption mylineop;
    mylineop.lno_color = GrWhite();
    mylineop.lno_width = 3;
    mylineop.lno_pattlen = 2;
    mylineop.lno_dashpat = "\x06\x04";

      The available custom line drawing primitives:

    void GrCustomLine(int x1,int y1,int x2,int y2,const GrLineOption *o);
    void GrCustomBox(int x1,int y1,int x2,int y2,const GrLineOption *o);
    void GrCustomCircle(int xc,int yc,int r,const GrLineOption *o);
    void GrCustomEllipse(int xc,int yc,int xa,int ya,const GrLineOption *o);
    void GrCustomCircleArc(int xc,int yc,int r,
                           int start,int end,int style,const GrLineOption *o);
    void GrCustomEllipseArc(int xc,int yc,int xa,int ya,
                            int start,int end,int style,const GrLineOption *o);
    void GrCustomPolyLine(int numpts,int points[][2],const GrLineOption *o);
    void GrCustomPolygon(int numpts,int points[][2],const GrLineOption *o);

    Patterns (bitmaps, pixmaps and gradients)

      The library also supports pattern filled and paterned versions of the filled and custom primitives described above. These functions have similar parameter passing conventions as the basic ones with one difference: instead of the color value a pointer to an union of type 'GrPattern' has to be passed to them. The GrPattern union can contain a bitmap, a pixmap or a gradient fill pattern. The first integer slot in the union determines which type it is. Bitmap fill patterns are rectangular arrays of bits, each set bit representing the foreground color of the fill operation, and each zero bit representing the background. Both the foreground and background colors can be combined with any of the supported logical operations. Bitmap fill patterns have one restriction: their width must be eight pixels. Pixmap fill patterns are very similar to contexts. Gradients are smooth color changes, can be linear or radial. Gradients can be used only in rgb color modes. The relevant structure declarations (from mgrx.h):

    #define GR_PTYPE_BITMAP   0
    #define GR_PTYPE_PIXMAP   1
    #define GR_PTYPE_GRADIENT 2
     * BITMAP: a mode independent way to specify a fill pattern of two
     *   colors. It is always 8 pixels wide (1 byte per scan line), its
     *   height is user-defined. SET THE TYPE FLAG TO GR_PTYPE_BITMAP!!!
    typedef struct _GR_bitmap {
        int     bmp_ptype;              /* type flag for pattern union */
        int     bmp_height;             /* bitmap height */
        char   *bmp_data;               /* pointer to the bit pattern */
        GrColor bmp_fgcolor;            /* foreground color for fill */
        GrColor bmp_bgcolor;            /* background color for fill */
        int     bmp_memflags;           /* set if dynamically allocated */
    } GrBitmap;
     * PIXMAP: a fill pattern stored in a layout identical to the video RAM
     *   for filling using 'bitblt'-s. It is mode dependent, typically one
     *   of the library functions is used to build it. SET THE TYPE FLAG TO
     *   GR_PTYPE_PIXMAP!!!
    typedef struct _GR_pixmap {
        int     pxp_ptype;              /* type flag for pattern union */
        int     pxp_width;              /* pixmap width (in pixels)  */
        int     pxp_height;             /* pixmap height (in pixels) */
        GrColor pxp_oper;               /* bitblt mode (SET, OR, XOR, AND, IMAGE) */
        struct _GR_frame pxp_source;    /* source context for fill */
    } GrPixmap;
     * GRADIENT: a fill pattern that change color in the direction defined by a
     *   vector (linar gradient) or from the distance to a point (radial gradient).
     *   Only can be used in rgb mode. SET THE TYPE FLAG TO GR_PTYPE_GRADIENT!!!
    typedef struct {
        int xi, yi, xf, yf;             // vector that define the gradient
        float dist_if;                  // distance from i to f (calculated)
    } GrGrdLinearData;
    typedef struct {
        int xc, yc, r;                  // center and radius that define the gradient
    } GrGrdRadialData;
    typedef struct {
        int dist;                       // normalized distance (0 to 255)
        GrColor c;                      // color
    } GrGrdStop;
    #define GR_LINEAR_GRADIENT 0
    #define GR_RADIAL_GRADIENT 1
    typedef struct {
        int     grd_ptype;              // type flag for pattern union
        int     grd_mode;               // gradient mode (linear or radial)
        GrColor grd_oper;               // bitblt mode (SET, OR, XOR, AND, IMAGE)
        int     grd_memflags;           // set if dynamically allocated
        union {
            GrGrdLinearData grd_ld;     // linear data
            GrGrdRadialData grd_rd;     // radial data
        int     grd_nstops;             // num of stops
        GrGrdStop grd_stop[GR_GRADIENT_MAXSTOPS];   // stop definition
        int     grd_genctbl;            // color table has been generated 1=yes 0=no
        GrColor grd_ctbl[257];          // color table ctbl[0]=num_colors=256 ever
    } GrGradient;
     * Fill pattern union -- can either be a bitmap, a pixmap or a gradient
    typedef union _GR_pattern {
        int        gp_ptype;            /* type flag */
        GrBitmap   gp_bitmap;           /* fill bitmap */
        GrPixmap   gp_pixmap;           /* fill pixmap */
        GrGradient gp_gradient;         /* fill gradient */
    } GrPattern;

      This define group (from mgrx.h) help to acces the GrPattern menbers:

    #define gp_bmp_data                 gp_bitmap.bmp_data
    #define gp_bmp_height               gp_bitmap.bmp_height
    #define gp_bmp_fgcolor              gp_bitmap.bmp_fgcolor
    #define gp_bmp_bgcolor              gp_bitmap.bmp_bgcolor
    #define gp_pxp_width                gp_pixmap.pxp_width
    #define gp_pxp_height               gp_pixmap.pxp_height
    #define gp_pxp_oper                 gp_pixmap.pxp_oper
    #define gp_pxp_source               gp_pixmap.pxp_source
    #define gp_grd_mode                 gp_gradient.grd_mode
    #define gp_grd_oper                 gp_gradient.grd_oper
    #define gp_grd_memflags             gp_gradient.grd_memflags
    #define gp_grd_ld                   gp_gradient.grd_ld
    #define gp_grd_rd                   gp_gradient.grd_rd
    #define gp_grd_nstops               gp_gradient.grd_nstops
    #define gp_grd_stop                 gp_gradient.grd_stop
    #define gp_grd_genctbl              gp_gradient.grd_genctbl
    #define gp_grd_ctbl                 gp_gradient.grd_ctbl


      Bitmap patterns can be easily built from initialized character arrays and static structures by the C compiler, thus no special support is included in the library for creating them. The only action required from the application program might be changing the foreground and background colors as needed. Here is an example:

        unsigned char bmpdata[8] = {0x00, 0x7E, 0x82, 0x82, 0x82, 0x82, 0x7E, 0x00};
        GrBitmap bm1 = {GR_PTYPE_BITMAP, 8, NULL, 0, 0, 0};
        bm1.bmp_fgcolor = GrWhite();
        bm1.bmp_bgcolor = GrBlack();
        bm1.bmp_data = (char *)bmpdata;


      Pixmap patterns are more difficult to build as they replicate the layout of the video memory which changes for different video modes. For this reason the library provides three functions to create pixmap patterns in a mode-independent way:

    GrPattern *GrBuildPixmap(const char *pixels,int w,int h,const GrColorTableP ct);
    GrPattern *GrBuildPixmapNR(const char *pixels,int w,int h,const GrColorTableP ct);
    GrPattern *GrBuildPixmapFromBits(const char *bits,int w,int h,
                                     GrColor fgc,GrColor bgc);
    GrPattern *GrBuildPixmapFromBitsNR(const char *bits,int w,int h,
                                       GrColor fgc,GrColor bgc);
    GrPattern *GrConvertToPixmap(GrContext *src);

      GrBuildPixmap build a pixmap from a two dimensional (w by h) array of characters. The elements in this array are used as indices into the color table specified with the argument ct. (This means that pixmaps created this way can use at most 256 colors.) The color table pointer:

    typedef GrColor *GrColorTableP;

    should point to an array of integers with the first element being the number of colors in the table and the color values themselves starting with the second element. NOTE: any color modifiers (GrXOR, GrOR, GrAND) OR-ed to the elements of the color table are ignored.

      The GrBuildPixmapFromBits function builds a pixmap fill pattern from bitmap data. It is useful if the width of the bitmap pattern is not eight as such bitmap patterns can not be used to build a GrBitmap structure.

      Either GrBuildPixmapFromBits and GrBuildPixmap can return a pixmap wider than w up to a multiple of 8 bytes for faster bitblt's. If that conduct is not desirable use the NR versions of the functions.

      The GrConvertToPixmap function converts a graphics context to a pixmap fill pattern. It is useful when the pattern can be created with graphics drawing operations. NOTE: the pixmap pattern and the original context share the drawing RAM, thus if the context is redrawn the fill pattern changes as well.


      Gradients can be build by simple code or using functions from the library. First you call one of these:

    GrPattern *GrCreateLinGradient(int xi, int yi, int xf, int yf);
    GrPattern *GrCreateRadGradient(int xc, int yc, int r);

    to create a linear or a radial gradient. After that you define the gradient stop points using:

    int GrAddGradientStop(GrPattern *p, int dist, GrColor c);

    dist is a integer normalized distance, from 0 to 255, that represent the distance from (xi,yi) to (xf,yf) or (r). Here is an simple example for a radial gradient with 3 stop points:

        rg1 = GrCreateRadGradient(GrMaxX()/2, GrMaxY()/2, GrMaxX()/2);
        if (rg1 == NULL) goto error;
        GrAddGradientStop(rg1, 0, GrAllocColor(255, 0, 0));
        GrAddGradientStop(rg1, 128, GrAllocColor(0, 0, 255));
        GrAddGradientStop(rg1, 255, GrAllocColor(0, 255, 0));

      In any case, after creating the gradient you must call a function to populate the color table that defines the gradient:

    int GrGenGradientColorTbl(GrPattern *p);

      When drawing with a gradient, the lib must calculate the distance to show the correct color for every point. If you want the same gradient a lot of times is faster to generate first a pixmap from the gradient:

    GrPattern *GrBuildPixmapFromGradient(GrPattern *p, int xo, int yo, int w, int h);

    w,h will be the pixmap dimensions, and xo,yo the gradient pos that will correspond to the 0,0 pixmap point.

    Destroy patterns

      Patterns which were built by library routines can be destroyed when no longer needed (i.e. the space occupied by them can be freed) by calling:

    void GrDestroyPattern(GrPattern *p);

      NOTE: when pixmap patterns converted from contexts are destroyed, the drawing RAM is not freed. It is freed when the original context is destroyed. Fill patterns built by the application have to be destroyed by the application as well (if this is needed).

    Pattern filled graphics primitives

      The list of supported pattern filled graphics primitives is shown below. These functions are very similar to their solid filled counterparts, only their last argument is different:

    void GrPatternFilledPlot(int x,int y,GrPattern *p);
    void GrPatternFilledLine(int x1,int y1,int x2,int y2,GrPattern *p);
    void GrPatternFilledBox(int x1,int y1,int x2,int y2,GrPattern *p);
    void GrPatternFilledCircle(int xc,int yc,int r,GrPattern *p);
    void GrPatternFilledEllipse(int xc,int yc,int xa,int ya,GrPattern *p);
    void GrPatternFilledCircleArc(int xc,int yc,int r,int start,int end,
                                  int style,GrPattern *p);
    void GrPatternFilledEllipseArc(int xc,int yc,int xa,int ya,int start,int end,
                                   int style,GrPattern *p);
    void GrPatternFilledConvexPolygon(int numpts,int points[][2],GrPattern *p);
    void GrPatternFilledPolygon(int numpts,int points[][2],GrPattern *p);
    void GrPatternFloodFill(int x, int y, GrColor border, GrPattern *p);

      All the above functions align the pattern origin with the context origin. Most of the time it is what is expected, so two intersecting shapes show a continuous pattern. But some times it is needing to align to other point, next functions do that:

    void GrPatAlignFilledPlot(int xo,int yo,int x,int y,GrPattern *p);
    void GrPatAlignFilledLine(int xo,int yo,int x1,int y1,int x2,int y2,GrPattern *p);
    void GrPatAlignFilledBox(int xo,int yo,int x1,int y1,int x2,int y2,GrPattern *p);
    void GrPatAlignFilledCircle(int xo,int yo,int xc,int yc,int r,GrPattern *p);
    void GrPatAlignFilledEllipse(int xo,int yo,int xc,int yc,int xa,int ya,GrPattern *p);
    void GrPatAlignFilledCircleArc(int xo,int yo,int xc,int yc,int r,
                                   int start,int end,int style,GrPattern *p);
    void GrPatAlignFilledEllipseArc(int xo,int yo,int xc,int yc,int xa,int ya,
                                    int start,int end,int style,GrPattern *p);
    void GrPatAlignFilledConvexPolygon(int xo,int yo,int n,int pt[][2],GrPattern *p);
    void GrPatAlignFilledPolygon(int xo,int yo,int n,int pt[][2],GrPattern *p);
    void GrPatAlignFloodFill(int xo,int yo,int x, int y, GrColor border, GrPattern *p);

      There are two functions to display all or part of a pixmap in the current context, aligning to the display box:

    void GrPixmapDisplay(int x,int y, GrPixmap *p);
    void GrPixmapDisplayExt(int x1,int y1,int x2,int y2, GrPixmap *p);

      And two others to transform pixmaps:

    GrPixmap *GrPixmapInverse(GrPixmap *p,int flag);
    GrPixmap *GrPixmapStretch(GrPixmap *p,int nwidth,int nheight);

    GrPixmapInverse creates a new pixmap flipping p left-right or top-down as indicated by flag that can be:

    #define GR_PIXMAP_INVLR  0x01  /* inverse left right */
    #define GR_PIXMAP_INVTD  0x02  /* inverse top down */

    GrPixmapStretch creates a new pixmap stretching p to nwidth by nheight.

    Patterned line drawing

      The custom line drawing functions introduced above also have a version when the drawn sections can be filled with a (bitmap, pixmap or gradient) fill pattern. To achieve this these functions must be passed both a custom line drawing option (GrLineOption structure) and a fill pattern (GrPattern union). These two have been combined into the GrLinePattern structure:

    typedef struct {
      GrPattern     *lnp_pattern;    /* fill pattern */
      GrLineOption  *lnp_option;     /* width + dash pattern */
    } GrLinePattern;

      All patterned line drawing functions take a pointer to this structure as their last argument. The list of available functions:

    void GrPatternedLine(int x1,int y1,int x2,int y2,GrLinePattern *lp);
    void GrPatternedBox(int x1,int y1,int x2,int y2,GrLinePattern *lp);
    void GrPatternedCircle(int xc,int yc,int r,GrLinePattern *lp);
    void GrPatternedEllipse(int xc,int yc,int xa,int ya,GrLinePattern *lp);
    void GrPatternedCircleArc(int xc,int yc,int r,int start,int end,
                              int style,GrLinePattern *lp);
    void GrPatternedEllipseArc(int xc,int yc,int xa,int ya,int start,int end,
                               int style,GrLinePattern *lp);
    void GrPatternedPolyLine(int numpts,int points[][2],GrLinePattern *lp);
    void GrPatternedPolygon(int numpts,int points[][2],GrLinePattern *lp);

      Again the above functions align the pattern origin with the context origin. To align with another point use next functions:

    void GrPatndAlignLine(int xo,int yo,int x1,int y1,int x2,int y2,GrLinePattern *lp);
    void GrPatndAlignBox(int xo,int yo,int x1,int y1,int x2,int y2,GrLinePattern *lp);
    void GrPatndAlignCircle(int xo,int yo,int xc,int yc,int r,GrLinePattern *lp);
    void GrPatndAlignEllipse(int xo,int yo,int xc,int yc,int xa,int ya,GrLinePattern *lp);
    void GrPatndAlignCircleArc(int xo,int yo,int xc,int yc,int r,
                               int start,int end,int style,GrLinePattern *lp);
    void GrPatndAlignEllipseArc(int xo,int yo,int xc,int yc,int xa,int ya,
                                int start,int end,int style,GrLinePattern *lp);
    void GrPatndAlignPolyLine(int xo,int yo,int numpts,int points[][2],GrLinePattern *lp);
    void GrPatndAlignPolygon(int xo,int yo,int numpts,int points[][2],GrLinePattern *lp);


      New in MGRX 1.4.3 version are multipolygons, they let fill and draw a group of polygons at once, so figures with holes can be drawn and filled. A multipolygon must be defined with a GrMultiPointArray structure:

    typedef struct {
        int npoints;            // number of points
        int closed;             // closed polygon if != 0 (not used for fill)
        int (*points)[2];       // Points array
    } GrPointArray;
    typedef struct {
        int npa;                // number of point arrays
        GrPointArray p[1];      // Points arrays (not actual size)
    } GrMultiPointArray;

      You can fill a multipolygon with a solid color or a pattern using next functions:

    void GrFilledMultiPolygon(GrMultiPointArray *mpa,GrColor c);
    void GrPatternFilledMultiPolygon(GrMultiPointArray *mpa,GrPattern *p);
    void GrPatAlignFilledMultiPolygon(int xo,int yo,GrMultiPointArray *mpa,GrPattern *p);

      And draw the multipolygon borders with a solid color or a pattern, even with a specific thick like we explained in previous chapters:

    void GrMultiPolygon(GrMultiPointArray *mpa,GrColor c);
    void GrCustomMultiPolygon(GrMultiPointArray *mpa,const GrLineOption *o);
    void GrPatternedMultiPolygon(GrMultiPointArray *mpa,GrLinePattern *lp);
    void GrPatndAlignMultiPolygon(int xo,int yo,GrMultiPointArray *mpa,GrLinePattern *lp);

      Note that you can not achieve the fill effect on multipolygons using normal fill polygon functions, but the draw functions are convenient functions only, the same effect can be reached using the normal polygon (for closed ones) or polyline (for opened ones) functions iterating over each member of the GrMultiPointArray struct.

    About text encoding

      If you use only ASCII text you can skip this chapter, set the chrtype to GR_BYTE_TEXT (see below) and go ahead. But if you need to use extended chars you need to know abour character encodings. MGRX has (limited) support for extended encodings.

      The problem is that you can have different encodings for your source code, for the fonts you are using and for the machine where your program is finally running. To manage this MGRX has four concepts:

    User encoding

      The User encoding is the source code encoding, the encoding the programmer have used when he wrote the source. The supported encodings are:

    #define GRENC_CP437          0   /* standard DOS encoding */
    #define GRENC_CP850          1   /* latin1 DOS encoding */
    #define GRENC_CP1252         2   /* standard Latin Win encoding */
    #define GRENC_ISO_8859_1     3   /* standard in some Linux */
    #define GRENC_UTF_8          4   /* multibyte unicode, standard in newest Linux */
    #define GRENC_UCS_2          5   /* restricted unicode, 2 bytes, only BMP range */
    #define GRENC_CP1251         6   /* standard Cyrillic Win encoding */
    #define GRENC_CP1253         7   /* standard Greek Win encoding */
    #define GRENC_LASTENCODE     7   /* last encode, for checks */

    By default MGRX sets the user encoding to GRENC_UTF_8 in linux console and X11, to GRENC_PC437 in DJGPP and to GRENC_PC1252 in Win32 after the first graphic mode is set. But as a programmer you know better what is the user encoding really used in the source code, so set it whit the GrSetUserEncoding function. You can set it even before the first graphics mode is set, in this case, MGRX will not change it.

    int GrGetUserEncoding(void);
    int GrSetUserEncoding(int enc);

    Font encoding

      The Font encoding is obsviusly the encoding used by the font glyphs, the supported font encodings are:

    #define GR_FONTENC_UNKNOWN     0     /* unknown encoding (no recode) */
    #define GR_FONTENC_CP437       1     /* standard dos encoding */
    #define GR_FONTENC_CP850       2     /* standard dos encoding */
    #define GR_FONTENC_CP1252      3     /* standard Latin Win encoding */
    #define GR_FONTENC_ISO_8859_1  4     /* standard latin encoding */
    #define GR_FONTENC_UNICODE     5     /* direct UNICODE encoding */
    #define GR_FONTENC_MGRX512     6     /* custom MGRX 512 char encoding */
    #define GR_FONTENC_ISO_8859_5  7     /* ASCII + Cyrillic */
    #define GR_FONTENC_ISO_8859_7  8     /* ASCII + Greek */
    #define GR_FONTENC_CP437EXT    9     /* CP437 + ISO-8859-1 + CP1252 */
    #define GR_FONTENC_CP1251     10     /* standard Cyrillic Win encoding */
    #define GR_FONTENC_CP1253     11     /* standard Greek Win encoding */
    #define GR_FONTENC_LASTENC    11     /* last encoding, for checks */

    Only the fonts provided whit MGRX in his own format (.fnt) have the font encoding set in their heading data, but after loading a font, you can set their encoding with the GrFontSetEncoding function.

    void GrFontSetEncoding(GrFont *font,int fontencoding);

    The MGRX512 font encoding is a specially crafted encoding used by some of the fonts provided with MGRX. It has 512 glyphs covering Latin, Cyrilic and Greek and some other miscellaneous glyphs. See the doc/mgrx512.txt file for more information.

    The CP437EXT font encoding is another MGRX custom encoding that cover CP437 + ISO8859-1 + CP1252 (by now) providing 323 glyphs. See the doc/cp437ext.txt file for more information.


      Chrtype is the specific encoding used for a text string you want to draw, it can be one of these:

    #define GR_BYTE_TEXT            0   /* 1 byte per character, unknown encoding */
    #define GR_WORD_TEXT            1   /* 2 bytes per character, unknown encoding */
    #define GR_CP437_TEXT           2   /* 1 bpc standard DOS encoding */
    #define GR_CP850_TEXT           3   /* 1 bpc latin1 DOS encoding */
    #define GR_CP1252_TEXT          4   /* 1 bpc standard Latin Win encoding */
    #define GR_ISO_8859_1_TEXT      5   /* 1 bpc latin1 standard in some Linux */
    #define GR_UTF8_TEXT            6   /* multibyte UTF-8 Unicode, restricted to 4 bytes */
    #define GR_UCS2_TEXT            7   /* 2 bpc restricted Unicode, only BMP range */
    #define GR_CP1251_TEXT          8   /* 1 bpc standard Cyrillic Win encoding */
    #define GR_CP1253_TEXT          9   /* 1 bpc standard Greek Win encoding */

    You can set the chrtype individually for every string you need to draw.

    There is a funtion to get the corresponding chrtype for the User encoding:

    char GrGetChrtypeForUserEncoding(void);

    It takes advantage that the corresponding chrtype is "Userencoding + 2" so have these into account if you want to add more encodings to MGRX.

    The MGRX text drawing funtions will try to recode from chrtype to the Font encoding before draw text.

    The GR_BYTE_TEXT and GR_WORD_TEXT are there for historical reasons and don't have a corresponding User encoding. Because they are reallly unknown encodings you can use them to draw font glyphs because they will not be recoded.

    KbSys encoding

      The KbSys encoding is the encoding used by the computer, where the program is executed, to get input from the keyboard. It can be one of the defined "user encoding".

    Every input driver try to guess the KbSys encoding used for the running program, if not sure it uses the MGRXKBSYSENCODING environment variable that can be set by users to one of these strings: "CP437", "CP850", "CP1251", "CP1252", "CP1253", "ISO_8859_1", "UTF_8" or "UCS_2". The KbSys encoding can be get using this function:

    int GrGetKbSysEncoding(void);

    If the KbSys encoding is different from the User encoding, the input driver will recode key events from KbSys encoding to User encoding. So using by example GrSetUserEncoding(GRENC_UTF_8); you can write a portable program that gets UTF-8 encoded keys in every platform.

    Text drawing

      The library supports loadable fonts. When in memory they are bit-mapped (i.e. not scalable!) fonts. A driver design allow MGRX to load different font formats, the last MGRX release come with drivers to load the MGRX own font format, the BGI Borland format, the FNA ascii font format, the PSF formats 1 & 2, the windows resource font format and (only for the X11 version) the X11 raster font format.

      The MGRX distribution come with a font collection in the MGRX own format. Some of these fonts were converted from VGA fonts, these fonts have all 256 characters from the CP437 or CP850 codepage. Some fonts were converted from fonts in the MIT X11 distribution, these are ASCII or ISO-8859-1 coded. A group of fonts were converted from the Terminus font and are MGRX512 encoded. There are too two big unicode fonts. The following fonts are included:

    Font file name       Family  Description
    pc<W>x<H>[t].fnt     pc      pc font, fixed, CP437
    px<W>x<H>.fnt        pcext   pc font, fixed, CP437EXT
    pc850-<H>[t].fnt     pc      pc font, fixed, CP850
    xm<W>x<H>[b][i].fnt  X_misc  X11, fixed, miscellaneous group, ASCII
    char<H>[b][i].fnt    char    X11, proportional, charter family, ISO-8859-1
    cour<H>[b][i].fnt    cour    X11, fixed, courier, ISO-8859-1
    helv<H>[b][i].fnt    helve   X11, proportional, helvetica, ISO-8859-1
    lucb<H>[b][i].fnt    lucb    X11, proportional, lucida bright, ISO-8859-1
    lucs<H>[b][i].fnt    lucs    X11, proportional, lucida sans serif , ISO-8859-1
    luct<H>[b][i].fnt    luct    X11, fixed, lucida typewriter, ISO-8859-1
    ncen<H>[b][i].fnt    ncen    X11, proportional, new century schoolbook, ISO-8859-1
    symb<H>.fnt          symbol  X11, proportional, greek letters, symbols 
    tms<H>[b][i].fnt     times   X11, proportional, times, ISO-8859-1
    tmgrx<H>[b].fnt      tmgrx   Ter, fixed, MGRX512
    unifont.fnt          unifont Converted from Unifont, propotional, Unicode
    unimk20.fnt          unimk   X11, propotional, Unicode

      In the font names <W> means the font width, <H> the font height. Many font families have bold and/or italic variants. The files containing these fonts contain a 'b' and/or 'i' character in their name just before the extension. Some of the pc fonts come in thin formats also, these are denoted by a 't' in their file names.

      A full font list is in the fonts/_fonts.dir file. Please read the copying.grx file to know about font licenses.

      The GrFont structure hold a font in memory. A number of 'pc' fonts are built-in to the library and don't need to be loaded:

    extern  GrFont          GrFont_PC6x8;
    extern  GrFont          GrFont_PC8x8;
    extern  GrFont          GrFont_PC8x14;
    extern  GrFont          GrFont_PC8x16;
    extern  GrFont          GrFont_PX8x18;
    extern  GrFont          GrFont_PX11x22;
    extern  GrFont          GrFont_PX14x28;

    The first four are CP437 coded and the last three CP437EXT.

      Other fonts must be loaded with the GrLoadFont function. If the font file name starts with any path separator character or character sequence ('<drive letter>:', '/' or '\') then it is loaded from the specified directory, otherwise the library try load the font first from the current directory and next from the default font path. The font path can be set up with the GrSetFontPath function. If the font path is not set then the value of the 'GRXFONT' environment variable is used as the font path. GrLoadFont will try every font driver, adding the driver extension if necesary. Font loading routines return NULL if the font was not found. When not needed any more, fonts can be unloaded (i.e. the storage occupied by them freed) by calling GrUnloadFont.

      The prototype declarations for these functions:

    GrFont *GrLoadFont(char *name);
    void GrUnloadFont(GrFont *font);
    void GrSetFontPath(char *path_list);

      Using these functions:

    GrFont *GrLoadConvertedFont(char *name,int cvt,int w,int h,
                                int minch,int maxch);
    GrFont *GrBuildConvertedFont(const GrFont *from,int cvt,int w,int h,
                                 int minch,int maxch);

    a new font can be generated from a file font or a font in memory, the 'cvt' argument direct the conversion or-ing the desired operations from these defines:

     * Font conversion flags for 'GrLoadConvertedFont'. OR them as desired.
    #define GR_FONTCVT_NONE         0     /* no conversion */
    #define GR_FONTCVT_SKIPCHARS    1     /* load only selected characters */
    #define GR_FONTCVT_RESIZE       2     /* resize the font */
    #define GR_FONTCVT_ITALICIZE    4     /* tilt font for "italic" look */
    #define GR_FONTCVT_BOLDIFY      8     /* make a "bold"(er) font  */
    #define GR_FONTCVT_FIXIFY       16    /* convert prop. font to fixed wdt */
    #define GR_FONTCVT_PROPORTION   32    /* convert fixed font to prop. wdt */

      GR_FONTCVT_SKIPCHARS needs 'minch' and 'maxch' arguments.

      GR_FONTCVT_RESIZE needs 'w' and 'h' arguments.

      The function:

    GrFont *GrLoadFontFile(char *name, char *driver);

    will try to load the named font file using the font driver specified bypassing the normal lookup.

      The function:

    int GrDumpGrxFont(const GrFont *f, char *fileName);

    writes a font to a file int the MGRX own format. It returns 0 on error or 1 on succes.

      The function:

    int GrDumpFnaFont(const GrFont *f, char *fileName);

    writes a font to an ascii font file, so it can be quickly edited with a text editor. For a description of the ascii font format, see the doc/fna.txt file. It returns 0 on error or 1 on succes.

      The function:

    int GrDumpFont(const GrFont *f,char *CsymbolName,char *fileName);

    writes a font to a C source code file, so it can be compiled and linked with a user program. GrDumpFont would not normally be used in a released program because its purpose is to produce source code. When the source code is compiled and linked into a program distributing the font file with the program is not necessary, avoiding the possibility of the font file being deleted or corrupted.

      You can use the premade src/utilprog/fnt2c.c program (see the source, it's so simple) to dump a selected font to source code, by example:

    "fnt2c helv15 myhelv15.c myhelv15"

      Next, if this line is included in your main include file:

    extern GrFont myhelv15

    and "myhelv15.c" compiled and linked with your project, you can use 'myhelv15' in every place a GrFont is required.

      This simple function:

    void GrTextXY(int x,int y,char *text,GrColor fg,GrColor bg);

    draw text in the current context in the standard direction, using the Default Font (set initially to GrFont_PC8x14 but can be changed), using the chrtype corresponding to the user encoding, with x, y like the upper left corner and the foreground and background colors given (note that bg equal to GrNOCOLOR make the background transparent).

    the Default Font can be get or set with these functions:

    GrFont *GrGetDefaultFont();
    void GrSetDefaultFont(GrFont *font);

      For other functions the GrTextOption structure specifies how to draw a character string:

    typedef struct _GR_textOption {  /* text drawing option structure */
            GrFont *txo_font;        /* font to be used */
            GrColor txo_fgcolor;     /* foreground color */
            GrColor txo_bgcolor;     /* background color */
            char    txo_chrtype;     /* character type (see above) */
            char    txo_direct;      /* direction (see above) */
            char    txo_xalign;      /* X alignment (see above) */
            char    txo_yalign;      /* Y alignment (see above) */
    } GrTextOption;

      The text can be rotated in increments of 90 degrees (txo_direct), alignments can be set in both directions (txo_xalign and txo_yalign), and separate fore and background colors can be specified. The accepted text direction values:

    #define GR_TEXT_RIGHT           0       /* normal */
    #define GR_TEXT_DOWN            1       /* downward */
    #define GR_TEXT_LEFT            2       /* upside down, right to left */
    #define GR_TEXT_UP              3       /* upward */
    #define GR_TEXT_DEFAULT         GR_TEXT_RIGHT

      The accepted horizontal and vertical alignment option values:

    #define GR_ALIGN_LEFT           0       /* X only */
    #define GR_ALIGN_TOP            0       /* Y only */
    #define GR_ALIGN_CENTER         1       /* X, Y   */
    #define GR_ALIGN_RIGHT          2       /* X only */
    #define GR_ALIGN_BOTTOM         2       /* Y only */
    #define GR_ALIGN_BASELINE       3       /* Y only */

      Text strings chrtype must be one of the types defined in the previous section. With GR_BYTE_TEXT and GR_WORD_TEXT there will be no recoding, so is up to you to select a font that match the text encoding you are using.

      Remember that GR_WORD_TEXT and GR_UCS2_TEXT are two bytes per character. GR_UTF8_TEXT is the standard multibyte UTF-8 encoding. All other chrtypes are one byte per character.

      Text strings can also be drawn underlined. This is controlled by OR-ing the constant GR_UNDERLINE_TEXT to the foreground color value:

    #define GR_UNDERLINE_TEXT       (GrXOR << 4)

      After the application initializes a text option structure with the desired values it can call one of the following two text drawing functions:

    void GrDrawChar(long chr,int x,int y,const GrTextOption *opt);
    void GrDrawString(void *text,int length,int x,int y,const GrTextOption *opt);

      You must set the length parameter according to the indicated chrtype, it must be number of words for GR_WORD_TEXT and GR_UCS2_TEXT, the real number of characters for GR_UTF8_TEXT, or the number of bytes for the rest of chrtypes. If the string is NULL terminated you can set length to 0, and the function will calculate the lenght according to the chrtype.

      NOTE: text drawing is fastest when it is drawn in the 'normal' direction, and the character does not have to be clipped. It this case the library can use the appropriate low-level video RAM access routine, while in any other case the text is drawn pixel-by-pixel by the higher-level code.

      There are pattern filed versions too:

    void GrPatternDrawChar(long chr,int x,int y,const GrTextOption *opt,GrPattern *p);
    void GrPatternDrawString(void *text,int length,int x,int y,
                             const GrTextOption *opt,GrPattern *p);
    void GrPatternDrawCharExt(long chr,int x,int y,const GrTextOption *opt,GrPattern *p);
    void GrPatternDrawStringExt(void *text,int length,int x,int y,
                                const GrTextOption *opt,GrPattern *p);

    the first two align the pattern origin with the context origin. The last two align with the text box.

      The size of a font, a character or a text string can be obtained by calling one of the following functions. These functions also take into consideration the text direction specified in the text option structure passed to them.

    int  GrFontCharPresent(const GrFont *font,int chr);
    int  GrFontCharWidth(const GrFont *font,int chr);
    int  GrFontCharHeight(const GrFont *font,int chr);
    int  GrFontCharBmpRowSize(const GrFont *font,int chr);
    int  GrFontCharBitmapSize(const GrFont *font,int chr);
    int  GrFontStringWidth(const GrFont *font,void *text,int len,int chrtype);
    int  GrFontStringHeight(const GrFont *font,void *text,int len,int chrtype);
    int  GrCharWidth(long chr,const GrTextOption *opt);
    int  GrCharHeight(long chr,const GrTextOption *opt);
    void GrCharSize(long chr,const GrTextOption *opt,int *w,int *h);
    int  GrStringWidth(void *text,int length,const GrTextOption *opt);
    int  GrStringHeight(void *text,int length,const GrTextOption *opt);
    void GrStringSize(void *text,int length,const GrTextOption *opt,int *w,int *h);

      The first five funtions don't recode, so the chr parameter is really the font glyph index. All other fucntions recode if needed. If the string is NULL terminated you can set length to 0, and let the function calculate the lenght according to the chrtype.

    Special UTF-8 text type considerations

    To deal with GR_UTF8_TEXT, MGRX provides some special funtions:

    int GrUTF8StrLen(unsigned char *s);

    returns the number of UTF-8 characters (not bytes) in s.

    int GrStrLen(const void *text, int chrtype);

    is a more general function that returns the number of characters according to the chrtype provided.

    long GrUCS2ToUTF8(unsigned short ch);

    converts a UCS-2 word character to a UTF-8 string packed in a long for convenience (some functions use it, see bellow).

    long GrNextUTF8Char(unsigned char *s, int *nb);

    returns the first UTF-8 character packed in a long, it sets nb to the number of bytes consumed, so you can advance the string pointer to the next character.

    unsigned short GrUTF8ToUCS2(unsigned char *s);

    converts the first UTF-8 character in s to a a UCS-2 word character.

    unsigned short *GrUTF8StrToUCS2Str(unsigned char *s, int *ulen, int maxlen);
    unsigned char *GrTextRecodeToUTF8(const void *text,int length,int chrtype);
    void *GrTextRecodeFromUTF8(const unsigned char *text,int length,int chrtype);

    GrUTF8StrToUCS2Strconverts a UTF-8 string to a UCS-2 string and set the ulen variable with the UTF-8 string length. The maxlen parameter indicates the maximum len you want, a 0 indicates no maxlen. GrTextRecodeFromUTF8 is a more generic funtion to recode from UTF-8 to any chrtype, and GrTextRecodeToUTF8 do the opposite. All function allocate the necesary memory, so you must check the returned pointer for NULL, and the memory must be freed when unneded.

    For font related functions that expect a string and a length you must pass the UTF-8 string and the UTF-8 length (or less) in characters (not byres), example:

    opt.txo_chrtype = GR_UTF8_TEXT;
    GrDrawString(sutf8, GrUTF8StrLen(sutf8), x, y, opt);
    or if sutf8 is NULL terminated you can let the function do the calculation:
    opt.txo_chrtype = GR_UTF8_TEXT;
    GrDrawString(sutf8, 0, x, y, opt);

    For font related functions that expect a unique character packed in a long and a chrtype, you need to pack the UTF-8 char in a long, example:

    long *c;
    char cutf8[4];
    put in cutf8 the character; // 1 to four bytes, fill with 0 the unused bytes!!
    c = (long *)&cutf8;
    opt.txo_chrtype = GR_UTF8_TEXT;
    void GrDrawChar(*c, x, y, opt);

    or e.g. using the GrNextUTF8Char function:

    long c;
    c = GrNextUTF8Char(sutf8, &nb);
    opt.txo_chrtype = GR_UTF8_TEXT;
    void GrDrawChar(c, x, y, opt);

    For font functions that expect a unique character (really a glyph number) without chrtype, you must pass the UCS-2 equivalent character, example:


    Recode to/from GR_UCS2_TEXT chartype

      GR_UCS2_TEXT chartype is a very convenient chartype to work internally for a program, because it can accommodate every character from other chartypes and any char occupies exactly two bytes (an unsigned short). MGRX provides funtions to recode to/from any chartype (including the GR_UTF8_TEXT chartype) from/to GR_UCS2_TEXT.

    unsigned short GrCharRecodeToUCS2(long chr,int chrtype);

    converts the chr character of type chrtype to a a UCS-2 character. Remember for UTF-8 you have to provide a char[4] packed in a long.

    long GrCharRecodeFromUCS2(long chr,int chrtype);

    do the opposite, converts a chr UCS-2 character to a character of type chrtype. Again, if UTF-8 is the chartype required, it returns a char[4] packed in a long.

    unsigned short *GrTextRecodeToUCS2(const void *text,int length,int chrtype);
    void *GrTextRecodeFromUCS2(const unsigned short *text,int length,int chrtype);

    GrTextRecodeToUCS2 converts a string of any chrtype to a UCS-2 string and GrTextRecodeFromUCS2 do the opposite. Both functions allocate the necesary memory, so you must check the returned pointer for NULL, and the memory must be freed when unneded.


      To help internationalize programs, MGRX includes a catalog implementation. It doesn't depend of any other MGRX function and it is Chrtype agnostic (this is why (void *) is used to pass strings).

    These are the functions:

    int GrI18nInit(int nlg, int nstr, void *defstr);
    void GrI18nEnd(void);
    int GrI18nSetLabel(int lid, void *label);
    void * GrI18nGetLabel(int lid);
    int GrI18nSetLang(int lid);
    void GrI18nAddStrings(int lid, int fsid, int nums, void **str);
    void * GrI18nGetString(int sid);

    For an example of use read the test program "test/i18ntest.c"

    Drawing in user coordinates

      There is a second set of the graphics primitives which operates in user coordinates. Every context has a user to screen coordinate mapping associated with it. An application specifies the user window by calling the GrSetUserWindow function.

    void GrSetUserWindow(int x1,int y1,int x2,int y2);

      A call to this function in fact specifies the virtual coordinate limits which will be mapped onto the current context regardless of the size of the context. For example, the call:


    tells the library that the program will perform its drawing operations in a coordinate system X:0...11999 (width = 12000) and Y:0...8999 (height = 9000). This coordinate range will be mapped onto the total area of the current context. The virtual coordinate system can also be shifted. For example:


      The user coordinates can even be used to turn the usual left-handed coordinate system (0:0 corresponds to the upper left corner) to a right handed one (0:0 corresponds to the bottom left corner) by calling:


      The library also provides three utility functions for the query of the current user coordinate limits and for converting user coordinates to screen coordinates and vice versa.

    void GrGetUserWindow(int *x1,int *y1,int *x2,int *y2);
    void GrGetScreenCoord(int *x,int *y);
    void GrGetUserCoord(int *x,int *y);

      If an application wants to take advantage of the user to screen coordinate mapping it has to use the user coordinate version of the graphics primitives. These have exactly the same parameter passing conventions as their screen coordinate counterparts. NOTE: the user coordinate system is not initialized by the library! The application has to set up its coordinate mapping before calling any of the use coordinate drawing functions -- otherwise the program will almost certainly exit (in a quite ungraceful fashion) with a 'division by zero' error. The list of supported user coordinate drawing functions:

    void GrUsrPlot(int x,int y,GrColor c);
    void GrUsrLine(int x1,int y1,int x2,int y2,GrColor c);
    void GrUsrHLine(int x1,int x2,int y,GrColor c);
    void GrUsrVLine(int x,int y1,int y2,GrColor c);
    void GrUsrBox(int x1,int y1,int x2,int y2,GrColor c);
    void GrUsrFilledBox(int x1,int y1,int x2,int y2,GrColor c);
    void GrUsrFramedBox(int x1,int y1,int x2,int y2,int wdt,GrFBoxColors *c);
    void GrUsrCircle(int xc,int yc,int r,GrColor c);
    void GrUsrEllipse(int xc,int yc,int xa,int ya,GrColor c);
    void GrUsrCircleArc(int xc,int yc,int r,int start,int end,
                        int style,GrColor c);
    void GrUsrEllipseArc(int xc,int yc,int xa,int ya,int start,int end,
                         int style,GrColor c);
    void GrUsrFilledCircle(int xc,int yc,int r,GrColor c);
    void GrUsrFilledEllipse(int xc,int yc,int xa,int ya,GrColor c);
    void GrUsrFilledCircleArc(int xc,int yc,int r,int start,int end,
                              int style,GrColor c);
    void GrUsrFilledEllipseArc(int xc,int yc,int xa,int ya,int start,int end,
                               int style,GrColor c);
    void GrUsrPolyLine(int numpts,int points[][2],GrColor c);
    void GrUsrPolygon(int numpts,int points[][2],GrColor c);
    void GrUsrFilledConvexPolygon(int numpts,int points[][2],GrColor c);
    void GrUsrFilledPolygon(int numpts,int points[][2],GrColor c);
    void GrUsrFloodFill(int x, int y, GrColor border, GrColor c);
    GrColor GrUsrPixel(int x,int y);
    GrColor GrUsrPixelC(GrContext *c,int x,int y);
    void GrUsrCustomLine(int x1,int y1,int x2,int y2,const GrLineOption *o);
    void GrUsrCustomBox(int x1,int y1,int x2,int y2,const GrLineOption *o);
    void GrUsrCustomCircle(int xc,int yc,int r,const GrLineOption *o);
    void GrUsrCustomEllipse(int xc,int yc,int xa,int ya,const GrLineOption *o);
    void GrUsrCustomCircleArc(int xc,int yc,int r,int start,int end,
                              int style,const GrLineOption *o);
    void GrUsrCustomEllipseArc(int xc,int yc,int xa,int ya,int start,int end,
                               int style,const GrLineOption *o);
    void GrUsrCustomPolyLine(int numpts,int points[][2],const GrLineOption *o);
    void GrUsrCustomPolygon(int numpts,int points[][2],const GrLineOption *o);
    void GrUsrPatternedLine(int x1,int y1,int x2,int y2,GrLinePattern *lp);
    void GrUsrPatternedBox(int x1,int y1,int x2,int y2,GrLinePattern *lp);
    void GrUsrPatternedCircle(int xc,int yc,int r,GrLinePattern *lp);
    void GrUsrPatternedEllipse(int xc,int yc,int xa,int ya,GrLinePattern *lp);
    void GrUsrPatternedCircleArc(int xc,int yc,int r,int start,int end,
                                 int style,GrLinePattern *lp);
    void GrUsrPatternedEllipseArc(int xc,int yc,int xa,int ya,int start,int end,
                                  int style,GrLinePattern *lp);
    void GrUsrPatternedPolyLine(int numpts,int points[][2],GrLinePattern *lp);
    void GrUsrPatternedPolygon(int numpts,int points[][2],GrLinePattern *lp);
    void GrUsrPatternFilledPlot(int x,int y,GrPattern *p);
    void GrUsrPatternFilledLine(int x1,int y1,int x2,int y2,GrPattern *p);
    void GrUsrPatternFilledBox(int x1,int y1,int x2,int y2,GrPattern *p);
    void GrUsrPatternFilledCircle(int xc,int yc,int r,GrPattern *p);
    void GrUsrPatternFilledEllipse(int xc,int yc,int xa,int ya,GrPattern *p);
    void GrUsrPatternFilledCircleArc(int xc,int yc,int r,int start,int end,int style,GrPattern *p);
    void GrUsrPatternFilledEllipseArc(int xc,int yc,int xa,int ya,int start,int end,int style,GrPattern *p);
    void GrUsrPatternFilledConvexPolygon(int numpts,int points[][2],GrPattern *p);
    void GrUsrPatternFilledPolygon(int numpts,int points[][2],GrPattern *p);
    void GrUsrPatternFloodFill(int x, int y, GrColor border, GrPattern *p);
    void GrUsrDrawChar(long chr,int x,int y,const GrTextOption *opt);
    void GrUsrDrawString(char *text,int length,int x,int y,const GrTextOption *opt);
    void GrUsrTextXY(int x,int y,char *text,GrColor fg,GrColor bg);

    Graphics cursors

      The library provides support for the creation and usage of an unlimited number of graphics cursors. An application can use these cursors for any purpose. Cursors always save the area they occupy before they are drawn. When moved or erased they restore this area. As a general rule of thumb, an application should erase a cursor before making changes to an area it occupies and redraw the cursor after finishing the drawing. Cursors are created with the GrBuildCursor function:

    GrCursor *GrBuildCursor(char *pixels,int pitch,int w,int h,
                            int xo,int yo,const GrColorTableP c);

      The pixels, w (=width), h (=height) and c (= color table) arguments are similar to the arguments of the pixmap building library function GrBuildPixmap (see that paragraph for a more detailed explanation.), but with two differences. First, it is not assumed that the pixels data is w x h sized, the pitch argument set the offset between rows. Second, the pixmap data is interpreted slightly differently, any pixel with value zero is taken as a "transparent" pixel, i.e. the background will show through the cursor pattern at that pixel. A pixmap data byte with value = 1 will refer to the first color in the table, and so on.

      The xo (= X offset) and yo (= Y offset) arguments specify the position (from the top left corner of the cursor pattern) of the cursor's "hot point".

      The GrCursor data structure:

    typedef struct _GR_cursor {
      struct _GR_context work;            /* work areas (4) */
      int     xcord,ycord;                /* cursor position on screen */
      int     xsize,ysize;                /* cursor size */
      int     xoffs,yoffs;                /* LU corner to hot point offset */
      int     xwork,ywork;                /* save/work area sizes */
      int     xwpos,ywpos;                /* save/work area position on screen */
      int     displayed;                  /* set if displayed */
    } GrCursor;

    is typically not used (i.e. read or changed) by the application program, it should just pass pointers to these structures to the appropriate library functions. Other cursor manipulation functions:

    void GrDisplayCursor(GrCursor *cursor);
    void GrEraseCursor(GrCursor *cursor);
    void GrMoveCursor(GrCursor *cursor,int x,int y);
    void GrDestroyCursor(GrCursor *cursor);

    Writing/reading PNM graphics files

      MGRX includes functions to load/save a context from/to a PNM file.

      PNM is a group of simple graphics formats from the NetPbm distribution. NetPbm can convert from/to PNM lots of graphics formats, and apply some transformations to PNM files (Note. You don't need the NetPbm distribution to use the MGRX functions).

      There are six PNM formats:

      MGRX can handle the binary formats only (get the NetPbm distribution if you need to convert text to binary formats).

      To save a context in a PNM file you have three functions:

    int GrSaveContextToPbm( GrContext *grc, char *pbmfn, char *docn );
    int GrSaveContextToPgm( GrContext *grc, char *pgmfn, char *docn );
    int GrSaveContextToPpm( GrContext *grc, char *ppmfn, char *docn );
    they work both in RGB and palette modes, grc must be a pointer to the context to be saved, if it is NULL the current context is saved; p-mfn is the file name to be created and docn is an optional text comment to be written in the file, it can be NULL. The functions return 0 on succes or -1 on error.

      GrSaveContextToPbm dumps a context in a PBM file (bitmap). If the pixel color isn't Black it asumes White.

      GrSaveContextToPgm dumps a context in a PGM file (gray scale). The colors are quantized to gray scale using .299r + .587g + .114b.

      GrSaveContextToPpm dumps a context in a PPM file (real color).

      To load a PNM file in a context you must use:

    int GrLoadContextFromPnm( GrContext *grc, char *pnmfn );
    it support reading PBM, PGM and PPM binary files. grc must be a pointer to the context to be written, if it is NULL the current context is used; p-mfn is the file name to be read. If context dimensions are lesser than pnm dimensions, the function loads as much as it can. If color mode is not in RGB mode, the routine allocates as much colors as it can. The function returns 0 on succes or -1 on error.

      To query the file format, width and height of a PNM file you can use:

    int GrQueryPnm( char *pnmfn, int *width, int *height, int *maxval );
    pnmfn is the name of pnm file; width returns the pnm width; height returns the pnm height; maxval returns the max color component value. The function returns 1 to 6 on success (the PNM format) or -1 on error.

      The two next functions:

    int GrLoadContextFromPnmBuffer( GrContext *grc, const char *pnmbuf );
    int GrQueryPnmBuffer( const char *pnmbuf, int *width, int *height, int *maxval );
    work like GrLoadContextFromPnm and GrQueryPnmBuffer, but they get his input from a buffer instead of a file. This way, pnm files can be embeded in a program (using the src/utilprog/bin2c program by example).

    Writing/reading PNG graphics files

      MGRX includes functions to load/save a context from/to a png file. But note, for this purpose it needs the libpng library, and to enable the png support before make the MGRX lib.

      Use next function to save a context in a PNG file:

    int GrSaveContextToPng( GrContext *grc, char *pngfn );
    it works both in RGB and palette modes, grc must be a pointer to the context to be saved, if it is NULL the current context is saved; pngfn is the file name to be created. The function returns 0 on succes or -1 on error.

      To load a PNG file in a context you must use:

    int GrLoadContextFromPng( GrContext *grc, char *pngfn, int use_alpha );
    grc must be a pointer to the context to be written, if it is NULL the current context is used; pngfn is the file name to be read; set use_alpha to 1 if you want to use the image alpha channel (if available). If context dimensions are lesser than png dimensions, the function loads as much as it can. If color mode is not in RGB mode, the routine allocates as much colors as it can. The function returns 0 on succes or -1 on error.

      To query the width and height of a PNG file you can use:

    int GrQueryPng( char *pngfn, int *width, int *height );
    pngfn is the name of png file; width returns the png width; height returns the png height. The function returns 0 on success or -1 on error.

      The function:

    int GrPngSupport( void );
    returns 1 if there is png support in the library, 0 otherwise. If there is not support for png, dummy functions are added to the library, returning error (-1) ever.

    Writing/reading JPEG graphics files

      MGRX includes functions to load/save a context from/to a jpeg file. But note, for this purpose it needs the libjpeg library, and to enable the jpeg support before make the MGRX lib.

      Use next function to save a context in a JPEG file:

    int GrSaveContextToJpeg( GrContext *grc, char *jpegfn, int quality );
    it works both in RGB and palette modes, grc must be a pointer to the context to be saved, if it is NULL the current context is saved; jpegfn is the file name to be created; quality is a number between 1 and 100 to drive the compression quality, use higher values for better quality (and bigger files), you can use 75 as a standard value, normally a value between 50 and 95 is good. The function returns 0 on succes or -1 on error.

      This function saves a context in a grayscale JPEG file:

    int GrSaveContextToGrayJpeg( GrContext *grc, char *jpegfn, int quality );
    parameters and return codes are like in GrSaveContextToJpeg. The colors are quantized to gray scale using .299r + .587g + .114b.

      To load a JPEG file in a context you must use:

    int GrLoadContextFromJpeg( GrContext *grc, char *jpegfn, int scale );
    grc must be a pointer to the context to be written, if it is NULL the current context is used; jpegfn is the file name to be read; set scale to 1, 2, 4 or 8 to reduce the loaded image to 1/1, 1/2, 1/4 or 1/8. If context dimensions are lesser than jpeg dimensions, the function loads as much as it can. If color mode is not in RGB mode, the routine allocates as much colors as it can. The function returns 0 on succes or -1 on error.

      To query the width and height of a JPEG file you can use:

    int GrQueryJpeg( char *jpegfn, int *width, int *height );
    jpegfn is the name of jpeg file; width returns the jpeg width; height returns the jpeg height. The function returns 0 on success or -1 on error.

      The function:

    int GrJpegSupport( void );
    returns 1 if there is jpeg support in the library, 0 otherwise. If there is not support for jpeg, dummy functions are added to the library, returning error (-1) ever.

    Miscellaneous functions

      Here we will describe some miscellaneous functions.

    unsigned GrGetLibraryVersion(void);

      GrGetLibraryVersion returns the MGRX version API, like a hexadecimal coded number. By example 0x0090 means 0.9.0 Because mgrx.h defines the MGRX_VERSION_API macro, you can check if both, the library and the include file, are in the same version using if( GrGetLibraryVersion() == MGRX_VERSION_API )

    unsigned GrGetLibrarySystem(void);

      This functions returns a unsigned integer identifing the system you are working in. mgrx.h defines some macros you can use:

    /* these are the supported configurations: */
    #define MGRX_VERSION_GCC_386_DJGPP       1       /* DJGPP v2 */
    #define MGRX_VERSION_GCC_386_LINUX       2       /* console framebuffer i386 */
    #define MGRX_VERSION_GCC_386_X11         3       /* X11 version i386 */
    #define MGRX_VERSION_GCC_386_WIN32       4       /* WIN32 using Mingw32 */
    #define MGRX_VERSION_GCC_X86_64_LINUX    5       /* console framebuffer x86_64 */
    #define MGRX_VERSION_GCC_X86_64_X11      6       /* X11 version x86_64 */
    #define MGRX_VERSION_GCC_ARM_LINUX       7       /* console framebuffer arm */
    #define MGRX_VERSION_GCC_ARM_X11         8       /* X11 version arm */
    void GrSetWindowTitle(char *title);

      GrSetWindowTitle sets the main window title in the X11 an Win32 versions. It does nothing in the DOS and Linux-Console versions.

    void GrSleep(int msec);

      This function stops the program execution for msec miliseconds.

    GrContext *GrCreateFrameContext(GrFrameMode md,int w,int h,
               char *memory[4],GrContext *where);

      This function is like GrCreateContext, except that you can specify any valid memory frame mode, not only the Screen associated frame mode. It can be used for special purposes (see GrBitBlt1bpp for an example).

    void GrBitBlt1bpp(GrContext *dst,int dx,int dy,GrContext *src,
         int x1,int y1,int x2,int y2,GrColor fg,GrColor bg);

      This special function does a bitblt from a 1bpp context (a bitmap really), using fg and bg like the color+opcode when bit=1 and bit=0 respectively. Here is an example:

      pContext = GrCreateFrameContext(GR_frameRAM1, sizex, sizey, NULL, NULL);
      /* draw something (black and white) into the bitmap */
      GrClearContext( GrBlack() );
      GrLine(0, 0, sizex-1, sizey-1, GrWhite());
      GrLine(0, sizey-1, sizex-1, 0, GrWhite());
      /* Put the bitmap into the screen */
      fcolor = GrAllocColor( 255,0,0 );
      bcolor = GrAllocColor( 0,0,255 );

    Input API

      MGRX can handle platform independent input. The input API is based on events and is totally diferent from the GRX input API.

      The input API must be inited after a graphics mode is set using this function:

    int GrEventInit(void);

    it returns true on success and init the keyboard and the mouse if one is found. The input API must be uninited before the program ends or before set another graphic mode whit this function:

    void GrEventUnInit(void);

      The basic function to get an event is:

    void GrEventRead(GrEvent * ev);

    the event structure is defined in mgrx.h

    typedef struct {
        int type;               /* event type */
        long time;              /* miliseconds */
        int kbstat;             /* kb status (count for GREV_EXPOSE) */
        union {                 /* GREV_KEY     GREV_MOUSE  GREV_MMOVE  GREV_EXPOSE  GREV_WSZCHG */
          long p1;              /* key          subevent    but status  x            0           */
          unsigned char cp1[4]; /* for easy access to multibyte key (like UTF-8)                 */
        long p2;                /* type/nbytes  x           x           y            0           */
        long p3;                /* --           y           y           width        winwidth    */
        long p4;                /* --           --          --          height       winheight   */
    } GrEvent;

    GrEventRead returns inmediatelly with type=GREV_NULL if no event is ready. The possible events are:

    #define GREV_NULL    0           /* no event */
    #define GREV_KEY     1           /* key pressed, p1=GRXkey (char or gr_keycode), p2=type or nbytes */
    #define GREV_MOUSE   2           /* mouse event, p1=subevent, p2=x, p3=y */
    #define GREV_MMOVE   3           /* mouse move event, p1=buttons status, p2=x, p3=y */
    #define GREV_PREKEY  4           /* key event before be recoded, internal event, users don't see it */
    #define GREV_EXPOSE  5           /* a window area must be redraw (generated only if user requested it) */
    #define GREV_WMEND   6           /* window manager wants ending (generated only if user requested it) */
    #define GREV_WSZCHG  7           /* window size changed (generated only if window resize supported) */
    #define GREV_CBREPLY 8           /* clipboard reply p1=1 if ready to paste, p1=0 no data in clipboard */
    #define GREV_USER    100         /* user event */

    a programmer can define user type events with values >= GREV_USER. When the event is GREV_KEY, p1 has the character or key and p2 the number of bytes used by the character or GRKEY_KEYCODE if it is a special keys. When the event is GREV_MOUSE, p1 has the subevent:

    #define GRMOUSE_LB_PRESSED  1    /* Left button depressed */
    #define GRMOUSE_MB_PRESSED  2    /* Middle button depressed */
    #define GRMOUSE_RB_PRESSED  3    /* Right button depressed */
    #define GRMOUSE_LB_RELEASED 4    /* Left button released */
    #define GRMOUSE_MB_RELEASED 5    /* Middle button released */
    #define GRMOUSE_RB_RELEASED 6    /* Rigth button released */
    #define GRMOUSE_B4_PRESSED  7    /* Button 4 depressed (scroll wheel) */
    #define GRMOUSE_B4_RELEASED 8    /* Button 4 released (scroll wheel) */
    #define GRMOUSE_B5_PRESSED  9    /* Button 5 depressed (scroll wheel) */
    #define GRMOUSE_B5_RELEASED 10   /* Button 5 released (scroll wheel) */

      By default MGRX doesn't generate GREV_MMOVE events except if the programmer requests it with the function:

    void GrEventGenMmove(int when);

    when can be one of these constants:

    #define GR_GEN_MMOVE_NEVER  0    /* Doesn't gen GREV_MMOVE */
    #define GR_GEN_MMOVE_IFBUT  1    /* Gen GREV_MMOVE if a button is pressed */
    #define GR_GEN_MMOVE_ALWAYS 2    /* Gen GREV_MMOVE always */

    for GREV_MMOVE p1 is filled with the mouse buttons status:

    #define GRMOUSE_LB_STATUS   1    /* Status bit for left button */
    #define GRMOUSE_MB_STATUS   4    /* Status bit for middle button */
    #define GRMOUSE_RB_STATUS   2    /* Status bit for right button */

      For GREV_KEY, GREV_MOUSE and GREV_MMOVE kbstat is filled wih the keyboard status word, or-ing it with the next defines it can be known the status of some special keys:

    #define GRKBS_RIGHTSHIFT    0x01   /* Keybd states: right shift key depressed */
    #define GRKBS_LEFTSHIFT     0x02   /* left shift key depressed */
    #define GRKBS_CTRL          0x04   /* CTRL depressed */
    #define GRKBS_ALT           0x08   /* ALT depressed */
    #define GRKBS_SCROLLOCK     0x10   /* SCROLL LOCK active */
    #define GRKBS_NUMLOCK       0x20   /* NUM LOCK active */
    #define GRKBS_CAPSLOCK      0x40   /* CAPS LOCK active */
    #define GRKBS_INSERT        0x80   /* INSERT state active */

      The file mgrxkeys.h defines the keys returned in p1 with the GREV_KEY. This is an extract:

    #define GrKey_Control_A            0x0001
    #define GrKey_Control_B            0x0002
    #define GrKey_Control_C            0x0003
    #define GrKey_A                    0x0041
    #define GrKey_B                    0x0042
    #define GrKey_C                    0x0043
    #define GrKey_F1                   0x013b
    #define GrKey_F2                   0x013c
    #define GrKey_F3                   0x013d
    #define GrKey_Alt_F1               0x0168
    #define GrKey_Alt_F2               0x0169
    #define GrKey_Alt_F3               0x016a

    but you can be confident that the standard ASCII is right maped. Remenber, if the KbSys encoding is different from the user encoding, the input driver will recode key events to user encoding.

      The GREV_EXPOSE event is only generated in X11 systems and only if the user requests it using this function:

    void GrEventGenExpose(int when);

    when can be:

    #define GR_GEN_EXPOSE_NO    0    /* Doesn't gen GREV_EXPOSE (default) */
    #define GR_GEN_EXPOSE_YES   1    /* Gen GREV_EXPOSE */

    When GREV_EXPOSE events are generated, the X11 video driver doesn't use backing store, so drawings can be faster, but the program must redraw the area exposed by itsel. Check the test program "life" for an example. For systems other than X11 the function does nothing. GREV_EXPOSE populates "count" in the "kbstat" field. If count is zero, no more Expose events follow. Simple applications that do not want to optimize redisplay can just ignore all Expose events with nonzero counts and perform full redisplays on events with zero counts.

      The GREV_WMEND event is only generated in X11 and W32 systems and only if the user requests it using this function:

    void GrEventGenWMEnd(int when);

    when can be:

    #define GR_GEN_WMEND_NO    0    /* Doesn't gen GREV_WMEND (default) */
    #define GR_GEN_WMEND_YES   1    /* Gen GREV_WMEND */

    If GREV_WMEND generation is active it will be generated when the user click the close window button, then the program must handle it, normally asking the user if he wants to exit. If generation is not active the program will be aborted.

      The GREV_WSZCHG is only generated by the X11 and W32 drivers when they are initialized with window resize support. See the specific chapter bellow.

      These are the other functions of the input API:

    void GrEventFlush(void);
    int GrEventCheck(void);
    void GrEventWait(GrEvent * ev);
    void GrEventWaitKeyOrClick(GrEvent * ev);
    int GrEventEnqueue(GrEvent * ev);
    int GrEventParEnqueue(int type, long p1, long p2, long p3, long p4);
    int GrEventEnqueueFirst(GrEvent * ev);
    int GrEventParEnqueueFirst(int type, long p1, long p2, long p3, long p4);

    GrEventFlush flush the event queue.

    GrEventCheck returns true if an event is waiting.

    GrEventWait waits until an event is ready.

    GrEventWaitKeyOrClick waits until a key or the left mouse button is pressed.

    GrEventEnqueue and GrEventParEnqueue enqueue an user event, passing the whole structure or only the type and parameters.

    GrEventEnqueueFirst and GrEventParEnqueueFirst enqueue an user event at the begining of the queue.

      Starting wiht the 1.3.6 MGRX version the input driver can compose characters from hexadecimal input. It is disabled by default, to enable it call the function:

    void GrSetComposeKey(int compkey);

    after the input driver is initiated, e.g. GrSetComposeKey(GrKey_Alt_U); After that use the compose-key and up to four hexadecimal digits for GRENC_UTF_8 and GRENC_UCS_2 User encodings, only two for the rest User encodings.

    Mouse cursor handling

      After the input API is initiated an application can test whether a mouse is available by calling the function:

    int  GrMouseDetect(void);

    which will return zero if no mouse (or mouse driver) is present, non-zero otherwise.

      The mouse can be controlled with the following functions:

    void GrMouseSetSpeed(int spmult,int spdiv);
    void GrMouseSetAccel(int thresh,int accel);
    void GrMouseSetLimits(int x1,int y1,int x2,int y2);
    void GrMouseGetLimits(int *x1,int *y1,int *x2,int *y2);
    void GrMouseWarp(int x,int y);

      For DOS and linux console the library calculates the mouse position only from the mouse mickey counters. (To avoid the limit and 'rounding to the next multiple of eight' problem with some mouse driver when it finds itself in a graphics mode unknown to it.) The parameters to the GrMouseSetSpeed function specify how coordinate changes are obtained from mickey counter changes, multipling by spmult and dividing by spdiv. In high resolution graphics modes the value of one just works fine, in low resolution modes (320x200 or similar) it is best set the spdiv to two or three. (Of course, it also depends on the sensitivity the mouse.) The GrMouseSetAccel function is used to control the ballistic effect: if a mouse coordinate changes between two samplings by more than the thresh parameter, the change is multiplied by the accel parameter. NOTE: some mouse drivers perform similar calculations before reporting the coordinates in mickeys. In this case the acceleration done by the library will be additional to the one already performed by the mouse driver.

      The limits of the mouse movement can be set (passed limits will be clipped to the screen) with GrMouseSetLimits (default is the whole screen) and the current limits can be obtained with GrMouseGetLimits. GrMouseWarp sets the mouse cursor to the specified position.

      As typical mouse drivers do not know how to draw mouse cursors in high resolution graphics modes, the mouse cursor is drawn by the library. The mouse cursor can be set with:

    void GrMouseSetCursor(GrCursor *cursor);
    void GrMouseSetInternalCursor(int type,GrColor fg,GrColor bg);

      For the type parameter in GrMouseSetInternalCursor you must use one of the available internal cursors:

    #define GR_MCUR_TYPE_ARROW  0      /* MOUSE CURSOR types: arrow cursor */
    #define GR_MCUR_TYPE_CROSS  1      /* cross cursor */
    #define GR_MCUR_TYPE_GLASS  2      /* hourglass cursor */

    the color fg will be used as the interior of it and bg will be the border. The current mouse cursor can be obtained with:

    GrCursor *GrMouseGetCursor(void);

      The mouse cursor can be displayed/erased with:

    void GrMouseDisplayCursor(void);
    void GrMouseEraseCursor(void);

      The mouse cursor can be left permanently displayed. All graphics primitives except for the few non-clipping functions check for conflicts with the mouse cursor and erase it before the drawing if necessary. Of course, it may be more efficient to erase the cursor manually before a long drawing sequence and redraw it after completion. The library provides an alternative pair of calls for this purpose which will erase the cursor only if it interferes with the drawing:

    int  GrMouseBlock(GrContext *c,int x1,int y1,int x2,int y2);
    void GrMouseUnBlock(int return_value_from_GrMouseBlock);

      GrMouseBlock should be passed the context in which the drawing will take place (the usual convention of NULL meaning the current context is supported) and the limits of the affected area. It will erase the cursor only if it interferes with the drawing. When the drawing is finished GrMouseUnBlock must be called with the argument returned by GrMouseBlock.

      The status of the mouse cursor can be obtained with calling GrMouseCursorIsDisplayed. This function will return non-zero if the cursor is displayed, zero if it is erased.

    int  GrMouseCursorIsDisplayed(void);

      The library supports (beside the simple cursor drawing) three types of "rubberband" attached to the mouse cursor. The GrMouseSetCursorMode function is used to select the cursor drawing mode.

    void GrMouseSetCursorMode(int mode,...);

      The parameter mode can have the following values:

    #define GR_M_CUR_NORMAL   0    /* MOUSE CURSOR modes: just the cursor */
    #define GR_M_CUR_RUBBER   1    /* rect. rubber band (XOR-d to the screen) */
    #define GR_M_CUR_LINE     2    /* line attached to the cursor */
    #define GR_M_CUR_BOX      3    /* rectangular box dragged by the cursor */

      GrMouseSetCursorMode takes different parameters depending on the cursor drawing mode selected. The accepted call formats are:


      The anchor parameters for the rubberband and rubberline modes specify a fixed screen location to which the other corner of the primitive is bound. The dx1 through dy2 parameters define the offsets of the corners of the dragged box from the hotpoint of the mouse cursor. The color value passed is always XOR-ed to the screen, i.e. if an application wants the rubberband to appear in a given color on a given background then it has to pass the XOR of these two colors to GrMouseSetCursorMode.

    Handling window resizing

      Starting wiht the 1.3.5 MGRX version the X11 and W32 drivers support window resize. To achieve that a program must:

      Note that if the graphics driver don't support window resize no events will be generated, but the program will run without problems. So you can code your program to support window resizes and they work with video drivers that don't support it, like the DJGPP an Linux framebuffer ones. But if you want to know if there are window resize support to run differently you can test for a flag after calling GrSetDriverExt using this:

        if (GrCurrentVideoDriver()->drvflags & GR_DRIVERF_WINDOW_RESIZE) {
            ... window resize support
        } else {
            ... no window resize support

      At any time the program can change the video mode by its own but in that case you must call GrSetMode(GR_default_text) before calling GrSetMode(GR_graphics_....), so the driver can know that the new video mode is not related with a window resize.

      In the next skeleton program you can see the process using two nested loops (there are other possible solutions):

        // global dimensions
        int w = 640;
        int h = 480;
        int bpp = 8;
        ... some intitalizations 
        GrSetDriverExt(NULL, "rszwin");
        ... more initializations
        while (1) {
            int exitloop = 0;
            int szchgbyprog = 0;
            GrEvent ev;
            GrSetMode(GR_width_height_bpp_graphics, w, h, bpp);
            ... restore the program state
            ... do initial drawings
            while (1) { // this is the main loop really
                if (((ev.type == GREV_KEY) && (ev.p1 == GrKey_Escape)) { // terminate
                    exitloop = 1;
                if (ev.type == GREV_WSZCHG) { // manage a window resize
                    w = ev.p3;
                    h = ev.p4;
           the program state
                ... manage other events
                if (...something...) { // set a window size by program
                    w = ownw;
                    h = ownh;
                    szchgbyprog = 1;
           the program state
                ... manage other events
            if (szchgbyprog) {
                szchgbyprog = 0;
            if (exitloop) break;

      You can check the test/wrsztest.c program for an example.


      Starting wiht the 1.3.6 version MGRX provides a clipboard to copy and paste text. The clipboard is internal to the program for the DOS and linuxfb versions, but the X11 and W32 drivers interact with the system clipboard. Because the X11 clipboard is asynchronous a special procedure is nedeed, but you can use it with any platform.

      MGRX maintains a copy of the last copied text, by default with a maximum of 32000 characters, but you can change this with:

    void GrSetClipBoardMaxChars(int maxchars);

    If you don't want to have a limit use GR_CB_NOLIMIT

      To put text in the clipboard use:

    int GrPutClipBoard(void *buf, int len, int chrtype);

    it returns the number of characters really copied, it can be less than the required len if the clipboard limit is reached.

      To get text from the clipboard you must first call:

    int GrIsClipBoardReadyToGet(void);

    it returns 1 if the text is ready to get, 0 if the clipboard is empty or -1 if a request to the system clipboard is issued. DOS, linuxfb and W32 platforms will never return -1. Anyway if -1 is returned you need to wait after a GREV_CBREPLY event with p1=1 is read. if 1 is returned or after the event is received yoy can call:

    int GrGetClipBoardLen(int *partloaded);

    to know the clipboard content len in characters. If the partloaded parameter is not NULL it will be true if the last copy operation was partially loaded, because the clipboard limit was reached. And now you can get the clipboard content with:

    void *GrGetClipBoard(int maxlen, int chrtype);

    you must check for NULL and freeing the returned string when unneded.

      At any time you can erase the internal copy of the last copied text with:

    void GrClearClipBoard(void);

    but note that in the X11 platform this prevent other clients to get the copied text from your program.

    The memory driver

      MGRX can be used to draw in memory using the memory driver, so a graphic screen is not nedeed. After the drawing, the result can be stored in a graphics file. This is the Hello MGRX example for the memory driver:

    #include <string.h>
    #include <mgrx.h>
    int main()
        char *message = "Hello, MGRX world";
        int x, y;
        GrTextOption grt;
        GrSetDriver("memory gw 400 gh 400 nc 256");
        grt.txo_font = &GrDefaultFont;
        grt.txo_fgcolor = GrWhite();
        grt.txo_bgcolor = GrBlack();
        grt.txo_direct = GR_TEXT_RIGHT;
        grt.txo_xalign = GR_ALIGN_CENTER;
        grt.txo_yalign = GR_ALIGN_CENTER;
        grt.txo_chrtype = GR_BYTE_TEXT;
        GrBox(0, 0, GrMaxX(), GrMaxY(), GrWhite());
        GrBox(4, 4, GrMaxX()-4, GrMaxY()-4, GrWhite());
        x = GrMaxX() / 2;
        y = GrMaxY() / 2;
        GrDrawString(message, strlen(message), x, y, &grt);
        GrSaveContextToPpm(NULL, "memtest.ppm", "GRX MemTest");
        return 0;

    References Official MGRX site GRX site (MGRX is derived from GRX) NetPbm distribution Libpng library Libjpeg library