Subversion Repositories shark

#if 0 /* in case someone actually tries to compile this */

/* example.c - an example of using libpng */

/* This is an example of how to use libpng to read and write PNG files.

 * The file libpng.txt is much more verbose then this.  If you have not

 * read it, do so first.  This was designed to be a starting point of an

 * implementation.  This is not officially part of libpng, is hereby placed

 * in the public domain, and therefore does not require a copyright notice.

 *

 * This file does not currently compile, because it is missing certain

 * parts, like allocating memory to hold an image.  You will have to

 * supply these parts to get it to compile.  For an example of a minimal

 * working PNG reader/writer, see pngtest.c, included in this distribution;

 * see also the programs in the contrib directory.

 */

#include "png.h"

 /* The png_jmpbuf() macro, used in error handling, became available in

  * libpng version 1.0.6.  If you want to be able to run your code with older

  * versions of libpng, you must define the macro yourself (but only if it

  * is not already defined by libpng!).

  */

#ifndef png_jmpbuf

#  define png_jmpbuf(png_ptr) ((png_ptr)->jmpbuf)

#endif

/* Check to see if a file is a PNG file using png_sig_cmp().  png_sig_cmp()

 * returns zero if the image is a PNG and nonzero if it isn't a PNG.

 *

 * The function check_if_png() shown here, but not used, returns nonzero (true)

 * if the file can be opened and is a PNG, 0 (false) otherwise.

 *

 * If this call is successful, and you are going to keep the file open,

 * you should call png_set_sig_bytes(png_ptr, PNG_BYTES_TO_CHECK); once

 * you have created the png_ptr, so that libpng knows your application

 * has read that many bytes from the start of the file.  Make sure you

 * don't call png_set_sig_bytes() with more than 8 bytes read or give it

 * an incorrect number of bytes read, or you will either have read too

 * many bytes (your fault), or you are telling libpng to read the wrong

 * number of magic bytes (also your fault).

 *

 * Many applications already read the first 2 or 4 bytes from the start

 * of the image to determine the file type, so it would be easiest just

 * to pass the bytes to png_sig_cmp() or even skip that if you know

 * you have a PNG file, and call png_set_sig_bytes().

 */

#define PNG_BYTES_TO_CHECK 4

int check_if_png(char *file_name, FILE **fp)

{

   char buf[PNG_BYTES_TO_CHECK];

   /* Open the prospective PNG file. */

   if ((*fp = fopen(file_name, "rb")) == NULL)

      return 0;

   /* Read in some of the signature bytes */

   if (fread(buf, 1, PNG_BYTES_TO_CHECK, *fp) != PNG_BYTES_TO_CHECK)

      return 0;

   /* Compare the first PNG_BYTES_TO_CHECK bytes of the signature.

      Return nonzero (true) if they match */

   return(!png_sig_cmp(buf, (png_size_t)0, PNG_BYTES_TO_CHECK));

}

/* Read a PNG file.  You may want to return an error code if the read

 * fails (depending upon the failure).  There are two "prototypes" given

 * here - one where we are given the filename, and we need to open the

 * file, and the other where we are given an open file (possibly with

 * some or all of the magic bytes read - see comments above).

 */

#ifdef open_file /* prototype 1 */

void read_png(char *file_name)  /* We need to open the file */

{

   png_structp png_ptr;

   png_infop info_ptr;

   unsigned int sig_read = 0;

   png_uint_32 width, height;

   int bit_depth, color_type, interlace_type;

   FILE *fp;

   if ((fp = fopen(file_name, "rb")) == NULL)

      return (ERROR);

#else no_open_file /* prototype 2 */

void read_png(FILE *fp, unsigned int sig_read)  /* file is already open */

{

   png_structp png_ptr;

   png_infop info_ptr;

   png_uint_32 width, height;

   int bit_depth, color_type, interlace_type;

#endif no_open_file /* only use one prototype! */

   /* Create and initialize the png_struct with the desired error handler

    * functions.  If you want to use the default stderr and longjump method,

    * you can supply NULL for the last three parameters.  We also supply the

    * the compiler header file version, so that we know if the application

    * was compiled with a compatible version of the library.  REQUIRED

    */

   png_ptr = png_create_read_struct(PNG_LIBPNG_VER_STRING,

      png_voidp user_error_ptr, user_error_fn, user_warning_fn);

   if (png_ptr == NULL)

   {

      fclose(fp);

      return (ERROR);

   }

   /* Allocate/initialize the memory for image information.  REQUIRED. */

   info_ptr = png_create_info_struct(png_ptr);

   if (info_ptr == NULL)

   {

      fclose(fp);

      png_destroy_read_struct(&png_ptr, png_infopp_NULL, png_infopp_NULL);

      return (ERROR);

   }

   /* Set error handling if you are using the setjmp/longjmp method (this is

    * the normal method of doing things with libpng).  REQUIRED unless you

    * set up your own error handlers in the png_create_read_struct() earlier.

    */

   if (setjmp(png_jmpbuf(png_ptr)))

   {

      /* Free all of the memory associated with the png_ptr and info_ptr */

      png_destroy_read_struct(&png_ptr, &info_ptr, png_infopp_NULL);

      fclose(fp);

      /* If we get here, we had a problem reading the file */

      return (ERROR);

   }

   /* One of the following I/O initialization methods is REQUIRED */

#ifdef streams /* PNG file I/O method 1 */

   /* Set up the input control if you are using standard C streams */

   png_init_io(png_ptr, fp);

#else no_streams /* PNG file I/O method 2 */

   /* If you are using replacement read functions, instead of calling

    * png_init_io() here you would call:

    */

   png_set_read_fn(png_ptr, (void *)user_io_ptr, user_read_fn);

   /* where user_io_ptr is a structure you want available to the callbacks */

#endif no_streams /* Use only one I/O method! */

   /* If we have already read some of the signature */

   png_set_sig_bytes(png_ptr, sig_read);

#ifdef hilevel

   /*

    * If you have enough memory to read in the entire image at once,

    * and you need to specify only transforms that can be controlled

    * with one of the PNG_TRANSFORM_* bits (this presently excludes

    * dithering, filling, setting background, and doing gamma

    * adjustment), then you can read the entire image (including

    * pixels) into the info structure with this call:

    */

   png_read_png(png_ptr, info_ptr, png_transforms, png_voidp_NULL);

#else

   /* OK, you're doing it the hard way, with the lower-level functions */

   /* The call to png_read_info() gives us all of the information from the

    * PNG file before the first IDAT (image data chunk).  REQUIRED

    */

   png_read_info(png_ptr, info_ptr);

   png_get_IHDR(png_ptr, info_ptr, &width, &height, &bit_depth, &color_type,

       &interlace_type, int_p_NULL, int_p_NULL);

/* Set up the data transformations you want.  Note that these are all

 * optional.  Only call them if you want/need them.  Many of the

 * transformations only work on specific types of images, and many

 * are mutually exclusive.

 */

   /* tell libpng to strip 16 bit/color files down to 8 bits/color */

   png_set_strip_16(png_ptr);

   /* Strip alpha bytes from the input data without combining with the

    * background (not recommended).

    */

   png_set_strip_alpha(png_ptr);

   /* Extract multiple pixels with bit depths of 1, 2, and 4 from a single

    * byte into separate bytes (useful for paletted and grayscale images).

    */

   png_set_packing(png_ptr);

   /* Change the order of packed pixels to least significant bit first

    * (not useful if you are using png_set_packing). */

   png_set_packswap(png_ptr);

   /* Expand paletted colors into true RGB triplets */

   if (color_type == PNG_COLOR_TYPE_PALETTE)

      png_set_palette_rgb(png_ptr);

   /* Expand grayscale images to the full 8 bits from 1, 2, or 4 bits/pixel */

   if (color_type == PNG_COLOR_TYPE_GRAY && bit_depth < 8)

      png_set_gray_1_2_4_to_8(png_ptr);

   /* Expand paletted or RGB images with transparency to full alpha channels

    * so the data will be available as RGBA quartets.

    */

   if (png_get_valid(png_ptr, info_ptr, PNG_INFO_tRNS))

      png_set_tRNS_to_alpha(png_ptr);

   /* Set the background color to draw transparent and alpha images over.

    * It is possible to set the red, green, and blue components directly

    * for paletted images instead of supplying a palette index.  Note that

    * even if the PNG file supplies a background, you are not required to

    * use it - you should use the (solid) application background if it has one.

    */

   png_color_16 my_background, *image_background;

   if (png_get_bKGD(png_ptr, info_ptr, &image_background))

      png_set_background(png_ptr, image_background,

                         PNG_BACKGROUND_GAMMA_FILE, 1, 1.0);

   else

      png_set_background(png_ptr, &my_background,

                         PNG_BACKGROUND_GAMMA_SCREEN, 0, 1.0);

   /* Some suggestions as to how to get a screen gamma value */

   /* Note that screen gamma is the display_exponent, which includes

    * the CRT_exponent and any correction for viewing conditions */

   if (/* We have a user-defined screen gamma value */)

   {

      screen_gamma = user-defined screen_gamma;

   }

   /* This is one way that applications share the same screen gamma value */

   else if ((gamma_str = getenv("SCREEN_GAMMA")) != NULL)

   {

      screen_gamma = atof(gamma_str);

   }

   /* If we don't have another value */

   else

   {

      screen_gamma = 2.2;  /* A good guess for a PC monitors in a dimly

                              lit room */

      screen_gamma = 1.7 or 1.0;  /* A good guess for Mac systems */

   }

   /* Tell libpng to handle the gamma conversion for you.  The final call

    * is a good guess for PC generated images, but it should be configurable

    * by the user at run time by the user.  It is strongly suggested that

    * your application support gamma correction.

    */

   int intent;

   if (png_get_sRGB(png_ptr, info_ptr, &intent))

      png_set_gamma(png_ptr, screen_gamma, 0.45455);

   else

   {

      double image_gamma;

      if (png_get_gAMA(png_ptr, info_ptr, &image_gamma))

         png_set_gamma(png_ptr, screen_gamma, image_gamma);

      else

         png_set_gamma(png_ptr, screen_gamma, 0.45455);

   }

   /* Dither RGB files down to 8 bit palette or reduce palettes

    * to the number of colors available on your screen.

    */

   if (color_type & PNG_COLOR_MASK_COLOR)

   {

      int num_palette;

      png_colorp palette;

      /* This reduces the image to the application supplied palette */

      if (/* we have our own palette */)

      {

         /* An array of colors to which the image should be dithered */

         png_color std_color_cube[MAX_SCREEN_COLORS];

         png_set_dither(png_ptr, std_color_cube, MAX_SCREEN_COLORS,

            MAX_SCREEN_COLORS, png_uint_16p_NULL, 0);

      }

      /* This reduces the image to the palette supplied in the file */

      else if (png_get_PLTE(png_ptr, info_ptr, &palette, &num_palette))

      {

         png_uint_16p histogram = NULL;

         png_get_hIST(png_ptr, info_ptr, &histogram);

         png_set_dither(png_ptr, palette, num_palette,

                        max_screen_colors, histogram, 0);

      }

   }

   /* invert monochrome files to have 0 as white and 1 as black */

   png_set_invert_mono(png_ptr);

   /* If you want to shift the pixel values from the range [0,255] or

    * [0,65535] to the original [0,7] or [0,31], or whatever range the

    * colors were originally in:

    */

   if (png_get_valid(png_ptr, info_ptr, PNG_INFO_sBIT))

   {

      png_color_8p sig_bit;

      png_get_sBIT(png_ptr, info_ptr, &sig_bit);

      png_set_shift(png_ptr, sig_bit);

   }

   /* flip the RGB pixels to BGR (or RGBA to BGRA) */

   if (color_type & PNG_COLOR_MASK_COLOR)

      png_set_bgr(png_ptr);

   /* swap the RGBA or GA data to ARGB or AG (or BGRA to ABGR) */

   png_set_swap_alpha(png_ptr);

   /* swap bytes of 16 bit files to least significant byte first */

   png_set_swap(png_ptr);

   /* Add filler (or alpha) byte (before/after each RGB triplet) */

   png_set_filler(png_ptr, 0xff, PNG_FILLER_AFTER);

   /* Turn on interlace handling.  REQUIRED if you are not using

    * png_read_image().  To see how to handle interlacing passes,

    * see the png_read_row() method below:

    */

   number_passes = png_set_interlace_handling(png_ptr);

   /* Optional call to gamma correct and add the background to the palette

    * and update info structure.  REQUIRED if you are expecting libpng to

    * update the palette for you (ie you selected such a transform above).

    */

   png_read_update_info(png_ptr, info_ptr);

   /* Allocate the memory to hold the image using the fields of info_ptr. */

   /* The easiest way to read the image: */

   png_bytep row_pointers[height];

   for (row = 0; row < height; row++)

   {

      row_pointers[row] = png_malloc(png_ptr, png_get_rowbytes(png_ptr,

         info_ptr));

   }

   /* Now it's time to read the image.  One of these methods is REQUIRED */

#ifdef entire /* Read the entire image in one go */

   png_read_image(png_ptr, row_pointers);

#else no_entire /* Read the image one or more scanlines at a time */

   /* The other way to read images - deal with interlacing: */

   for (pass = 0; pass < number_passes; pass++)

   {

#ifdef single /* Read the image a single row at a time */

      for (y = 0; y < height; y++)

      {

         png_read_rows(png_ptr, &row_pointers[y], png_bytepp_NULL, 1);

      }

#else no_single /* Read the image several rows at a time */

      for (y = 0; y < height; y += number_of_rows)

      {

#ifdef sparkle /* Read the image using the "sparkle" effect. */

         png_read_rows(png_ptr, &row_pointers[y], png_bytepp_NULL,

            number_of_rows);

#else no_sparkle /* Read the image using the "rectangle" effect */

         png_read_rows(png_ptr, png_bytepp_NULL, &row_pointers[y],

            number_of_rows);

#endif no_sparkle /* use only one of these two methods */

      }

      /* if you want to display the image after every pass, do

         so here */

#endif no_single /* use only one of these two methods */

   }

#endif no_entire /* use only one of these two methods */

   /* read rest of file, and get additional chunks in info_ptr - REQUIRED */

   png_read_end(png_ptr, info_ptr);

#endif hilevel

   /* At this point you have read the entire image */

   /* clean up after the read, and free any memory allocated - REQUIRED */

   png_destroy_read_struct(&png_ptr, &info_ptr, png_infopp_NULL);

   /* close the file */

   fclose(fp);

   /* that's it */

   return (OK);

}

/* progressively read a file */

int

initialize_png_reader(png_structp *png_ptr, png_infop *info_ptr)

{

   /* Create and initialize the png_struct with the desired error handler

    * functions.  If you want to use the default stderr and longjump method,

    * you can supply NULL for the last three parameters.  We also check that

    * the library version is compatible in case we are using dynamically

    * linked libraries.

    */

   *png_ptr = png_create_read_struct(PNG_LIBPNG_VER_STRING,

       png_voidp user_error_ptr, user_error_fn, user_warning_fn);

   if (*png_ptr == NULL)

   {

      *info_ptr = NULL;

      return (ERROR);

   }

   *info_ptr = png_create_info_struct(png_ptr);

   if (*info_ptr == NULL)

   {

      png_destroy_read_struct(png_ptr, info_ptr, png_infopp_NULL);

      return (ERROR);

   }

   if (setjmp(png_jmpbuf((*png_ptr))))

   {

      png_destroy_read_struct(png_ptr, info_ptr, png_infopp_NULL);

      return (ERROR);

   }

   /* This one's new.  You will need to provide all three

    * function callbacks, even if you aren't using them all.

    * If you aren't using all functions, you can specify NULL

    * parameters.  Even when all three functions are NULL,

    * you need to call png_set_progressive_read_fn().

    * These functions shouldn't be dependent on global or

    * static variables if you are decoding several images

    * simultaneously.  You should store stream specific data

    * in a separate struct, given as the second parameter,

    * and retrieve the pointer from inside the callbacks using

    * the function png_get_progressive_ptr(png_ptr).

    */

   png_set_progressive_read_fn(*png_ptr, (void *)stream_data,

      info_callback, row_callback, end_callback);

   return (OK);

}

int

process_data(png_structp *png_ptr, png_infop *info_ptr,

   png_bytep buffer, png_uint_32 length)

{

   if (setjmp(png_jmpbuf((*png_ptr))))

   {

      /* Free the png_ptr and info_ptr memory on error */

      png_destroy_read_struct(png_ptr, info_ptr, png_infopp_NULL);

      return (ERROR);

   }

   /* This one's new also.  Simply give it chunks of data as

    * they arrive from the data stream (in order, of course).

    * On Segmented machines, don't give it any more than 64K.

    * The library seems to run fine with sizes of 4K, although

    * you can give it much less if necessary (I assume you can

    * give it chunks of 1 byte, but I haven't tried with less

    * than 256 bytes yet).  When this function returns, you may

    * want to display any rows that were generated in the row

    * callback, if you aren't already displaying them there.

    */

   png_process_data(*png_ptr, *info_ptr, buffer, length);

   return (OK);

}

info_callback(png_structp png_ptr, png_infop info)

{

/* do any setup here, including setting any of the transformations

 * mentioned in the Reading PNG files section.  For now, you _must_

 * call either png_start_read_image() or png_read_update_info()

 * after all the transformations are set (even if you don't set

 * any).  You may start getting rows before png_process_data()

 * returns, so this is your last chance to prepare for that.

 */

}

row_callback(png_structp png_ptr, png_bytep new_row,

   png_uint_32 row_num, int pass)

{

/*

 * This function is called for every row in the image.  If the

 * image is interlaced, and you turned on the interlace handler,

 * this function will be called for every row in every pass.

 *

 * In this function you will receive a pointer to new row data from

 * libpng called new_row that is to replace a corresponding row (of

 * the same data format) in a buffer allocated by your application.

 *

 * The new row data pointer new_row may be NULL, indicating there is

 * no new data to be replaced (in cases of interlace loading).

 *

 * If new_row is not NULL then you need to call

 * png_progressive_combine_row() to replace the corresponding row as

 * shown below:

 */

   /* Check if row_num is in bounds. */

   if((row_num >= 0) && (row_num < height))

   {

     /* Get pointer to corresponding row in our

      * PNG read buffer.

      */

     png_bytep old_row = ((png_bytep *)our_data)[row_num];

     /* If both rows are allocated then copy the new row

      * data to the corresponding row data.

      */

     if((old_row != NULL) && (new_row != NULL))

     png_progressive_combine_row(png_ptr, old_row, new_row);

   }

/*

 * The rows and passes are called in order, so you don't really

 * need the row_num and pass, but I'm supplying them because it

 * may make your life easier.

 *

 * For the non-NULL rows of interlaced images, you must call

 * png_progressive_combine_row() passing in the new row and the

 * old row, as demonstrated above.  You can call this function for

 * NULL rows (it will just return) and for non-interlaced images

 * (it just does the png_memcpy for you) if it will make the code

 * easier.  Thus, you can just do this for all cases:

 */

   png_progressive_combine_row(png_ptr, old_row, new_row);

/* where old_row is what was displayed for previous rows.  Note

 * that the first pass (pass == 0 really) will completely cover

 * the old row, so the rows do not have to be initialized.  After

 * the first pass (and only for interlaced images), you will have

 * to pass the current row as new_row, and the function will combine

 * the old row and the new row.

 */

}

end_callback(png_structp png_ptr, png_infop info)

{

/* this function is called when the whole image has been read,

 * including any chunks after the image (up to and including

 * the IEND).  You will usually have the same info chunk as you

 * had in the header, although some data may have been added

 * to the comments and time fields.

 *

 * Most people won't do much here, perhaps setting a flag that

 * marks the image as finished.

 */

}

/* write a png file */

void write_png(char *file_name /* , ... other image information ... */)

{

   FILE *fp;

   png_structp png_ptr;

   png_infop info_ptr;

   png_colorp palette;

   /* open the file */

   fp = fopen(file_name, "wb");

   if (fp == NULL)

      return (ERROR);

   /* Create and initialize the png_struct with the desired error handler

    * functions.  If you want to use the default stderr and longjump method,

    * you can supply NULL for the last three parameters.  We also check that

    * the library version is compatible with the one used at compile time,

    * in case we are using dynamically linked libraries.  REQUIRED.

    */

   png_ptr = png_create_write_struct(PNG_LIBPNG_VER_STRING,

      png_voidp user_error_ptr, user_error_fn, user_warning_fn);

   if (png_ptr == NULL)

   {

      fclose(fp);

      return (ERROR);

   }

   /* Allocate/initialize the image information data.  REQUIRED */

   info_ptr = png_create_info_struct(png_ptr);

   if (info_ptr == NULL)

   {

      fclose(fp);

      png_destroy_write_struct(&png_ptr,  png_infopp_NULL);

      return (ERROR);

   }

   /* Set error handling.  REQUIRED if you aren't supplying your own

    * error handling functions in the png_create_write_struct() call.

    */

   if (setjmp(png_jmpbuf(png_ptr)))

   {

      /* If we get here, we had a problem reading the file */

      fclose(fp);

      png_destroy_write_struct(&png_ptr, &info_ptr);

      return (ERROR);

   }

   /* One of the following I/O initialization functions is REQUIRED */

#ifdef streams /* I/O initialization method 1 */

   /* set up the output control if you are using standard C streams */

   png_init_io(png_ptr, fp);

#else no_streams /* I/O initialization method 2 */

   /* If you are using replacement read functions, instead of calling

    * png_init_io() here you would call */

   png_set_write_fn(png_ptr, (void *)user_io_ptr, user_write_fn,

      user_IO_flush_function);

   /* where user_io_ptr is a structure you want available to the callbacks */

#endif no_streams /* only use one initialization method */

#ifdef hilevel

   /* This is the easy way.  Use it if you already have all the

    * image info living info in the structure.  You could "|" many

    * PNG_TRANSFORM flags into the png_transforms integer here.

    */

   png_write_png(png_ptr, info_ptr, png_transforms, png_voidp_NULL);

#else

   /* This is the hard way */

   /* Set the image information here.  Width and height are up to 2^31,

    * bit_depth is one of 1, 2, 4, 8, or 16, but valid values also depend on

    * the color_type selected. color_type is one of PNG_COLOR_TYPE_GRAY,

    * PNG_COLOR_TYPE_GRAY_ALPHA, PNG_COLOR_TYPE_PALETTE, PNG_COLOR_TYPE_RGB,

    * or PNG_COLOR_TYPE_RGB_ALPHA.  interlace is either PNG_INTERLACE_NONE or

    * PNG_INTERLACE_ADAM7, and the compression_type and filter_type MUST

    * currently be PNG_COMPRESSION_TYPE_BASE and PNG_FILTER_TYPE_BASE. REQUIRED

    */

   png_set_IHDR(png_ptr, info_ptr, width, height, bit_depth, PNG_COLOR_TYPE_???,

      PNG_INTERLACE_????, PNG_COMPRESSION_TYPE_BASE, PNG_FILTER_TYPE_BASE);

   /* set the palette if there is one.  REQUIRED for indexed-color images */

   palette = (png_colorp)png_malloc(png_ptr, PNG_MAX_PALETTE_LENGTH

             * sizeof (png_color));

   /* ... set palette colors ... */

   png_set_PLTE(png_ptr, info_ptr, palette, PNG_MAX_PALETTE_LENGTH);

   /* You must not free palette here, because png_set_PLTE only makes a link to

      the palette that you malloced.  Wait until you are about to destroy

      the png structure. */

   /* optional significant bit chunk */

   /* if we are dealing with a grayscale image then */

   sig_bit.gray = true_bit_depth;

   /* otherwise, if we are dealing with a color image then */

   sig_bit.red = true_red_bit_depth;

   sig_bit.green = true_green_bit_depth;

   sig_bit.blue = true_blue_bit_depth;

   /* if the image has an alpha channel then */

   sig_bit.alpha = true_alpha_bit_depth;

   png_set_sBIT(png_ptr, info_ptr, sig_bit);

   /* Optional gamma chunk is strongly suggested if you have any guess

    * as to the correct gamma of the image.

    */

   png_set_gAMA(png_ptr, info_ptr, gamma);

   /* Optionally write comments into the image */

   text_ptr[0].key = "Title";

   text_ptr[0].text = "Mona Lisa";

   text_ptr[0].compression = PNG_TEXT_COMPRESSION_NONE;

   text_ptr[1].key = "Author";

   text_ptr[1].text = "Leonardo DaVinci";

   text_ptr[1].compression = PNG_TEXT_COMPRESSION_NONE;

   text_ptr[2].key = "Description";

   text_ptr[2].text = "<long text>";

   text_ptr[2].compression = PNG_TEXT_COMPRESSION_zTXt;

#ifdef PNG_iTXt_SUPPORTED

   text_ptr[0].lang = NULL;

   text_ptr[1].lang = NULL;

   text_ptr[2].lang = NULL;

#endif

   png_set_text(png_ptr, info_ptr, text_ptr, 3);

   /* other optional chunks like cHRM, bKGD, tRNS, tIME, oFFs, pHYs, */

   /* note that if sRGB is present the gAMA and cHRM chunks must be ignored

    * on read and must be written in accordance with the sRGB profile */

   /* Write the file header information.  REQUIRED */

   png_write_info(png_ptr, info_ptr);

   /* If you want, you can write the info in two steps, in case you need to

    * write your private chunk ahead of PLTE:

    *

    *   png_write_info_before_PLTE(write_ptr, write_info_ptr);

    *   write_my_chunk();

    *   png_write_info(png_ptr, info_ptr);

    *

    * However, given the level of known- and unknown-chunk support in 1.1.0

    * and up, this should no longer be necessary.

    */

   /* Once we write out the header, the compression type on the text

    * chunks gets changed to PNG_TEXT_COMPRESSION_NONE_WR or

    * PNG_TEXT_COMPRESSION_zTXt_WR, so it doesn't get written out again

    * at the end.

    */

   /* set up the transformations you want.  Note that these are

    * all optional.  Only call them if you want them.

    */

   /* invert monochrome pixels */

   png_set_invert_mono(png_ptr);

   /* Shift the pixels up to a legal bit depth and fill in

    * as appropriate to correctly scale the image.

    */

   png_set_shift(png_ptr, &sig_bit);

   /* pack pixels into bytes */

   png_set_packing(png_ptr);

   /* swap location of alpha bytes from ARGB to RGBA */

   png_set_swap_alpha(png_ptr);

   /* Get rid of filler (OR ALPHA) bytes, pack XRGB/RGBX/ARGB/RGBA into

    * RGB (4 channels -> 3 channels). The second parameter is not used.

    */

   png_set_filler(png_ptr, 0, PNG_FILLER_BEFORE);

   /* flip BGR pixels to RGB */

   png_set_bgr(png_ptr);

   /* swap bytes of 16-bit files to most significant byte first */

   png_set_swap(png_ptr);

   /* swap bits of 1, 2, 4 bit packed pixel formats */

   png_set_packswap(png_ptr);

   /* turn on interlace handling if you are not using png_write_image() */

   if (interlacing)

      number_passes = png_set_interlace_handling(png_ptr);

   else

      number_passes = 1;

   /* The easiest way to write the image (you may have a different memory

    * layout, however, so choose what fits your needs best).  You need to

    * use the first method if you aren't handling interlacing yourself.

    */

   png_uint_32 k, height, width;

   png_byte image[height][width*bytes_per_pixel];

   png_bytep row_pointers[height];

   for (k = 0; k < height; k++)

     row_pointers[k] = image + k*width*bytes_per_pixel;

   /* One of the following output methods is REQUIRED */

#ifdef entire /* write out the entire image data in one call */

   png_write_image(png_ptr, row_pointers);

   /* the other way to write the image - deal with interlacing */

#else no_entire /* write out the image data by one or more scanlines */

   /* The number of passes is either 1 for non-interlaced images,

    * or 7 for interlaced images.

    */

   for (pass = 0; pass < number_passes; pass++)

   {

      /* Write a few rows at a time. */

      png_write_rows(png_ptr, &row_pointers[first_row], number_of_rows);

      /* If you are only writing one row at a time, this works */

      for (y = 0; y < height; y++)

      {

         png_write_rows(png_ptr, &row_pointers[y], 1);

      }

   }

#endif no_entire /* use only one output method */

   /* You can write optional chunks like tEXt, zTXt, and tIME at the end

    * as well.  Shouldn't be necessary in 1.1.0 and up as all the public

    * chunks are supported and you can use png_set_unknown_chunks() to

    * register unknown chunks into the info structure to be written out.

    */

   /* It is REQUIRED to call this to finish writing the rest of the file */

   png_write_end(png_ptr, info_ptr);

#endif hilevel

   /* If you png_malloced a palette, free it here (don't free info_ptr->palette,

      as recommended in versions 1.0.5m and earlier of this example; if

      libpng mallocs info_ptr->palette, libpng will free it).  If you

      allocated it with malloc() instead of png_malloc(), use free() instead

      of png_free(). */

   png_free(png_ptr, palette);

   palette=NULL;

   /* Similarly, if you png_malloced any data that you passed in with

      png_set_something(), such as a hist or trans array, free it here,

      when you can be sure that libpng is through with it. */

   png_free(png_ptr, trans);

   trans=NULL;

   /* clean up after the write, and free any memory allocated */

   png_destroy_write_struct(&png_ptr, &info_ptr);

   /* close the file */

   fclose(fp);

   /* that's it */

   return (OK);

}

#endif /* if 0 */