Awesome

libiff

libiff is a portable, extensible parser library implemented in ANSI C, for EA-IFF 85: Electronic Arts' Interchange File Format (IFF). It offers the following features:

• Reading IFF files
• Programmatically creating IFF files
• Retrieving IFF file contents
• Writing IFF files
• IFF conformance checking

Furthermore, this library has an extensible parser interface, which makes it possible to parse chunk types defined in application formats.

What is the Interchange File Format (IFF)?

IFF is a generic container format designed to make the transfer of data easier between applications produced by different vendors. The main purpose is to divide files into chunks identified by a type ID consisting of 4 characters, followed by a chunk size and the specified number of bytes. The IFF standard also defines a number of predefined group chunks, which can be used to group data chunks together and to share a set of common properties over a number of sub chunks.

The IFF container format is frequently used by applications on the Commodore Amiga for storing images, music and audio samples. The actual IFF specification can be found in: doc/IFF.asc, which is included in this package.

Some application file formats using the IFF container are:

• ILBM, used for storing images. Quite often ILBM files are wrongly named as IFF files because ILBM is the most common application format
• ANIM, an extension to ILBM supporting animations which are delta compressed
• 8SVX, used for storing 8-bit sound samples
• SMUS, used for storing musical scores

This library does not implement support for these subformats, but instead the parser provides an extension interface, which can be used to handle chunks defined in these application formats. For example, the libilbm package, provides additional functions to parse ILBM image chunks by extending this IFF parser. You can easily extend the IFF parser yourself to define your own custom file format on top of IFF.

Installation on Unix-like systems

Compilation and installation of this library on Unix-like systems is straight forward, by using the standard GNU autotools build instructions:

$ ./configure
$ make
$ make install

More details about the installation process can be found in the INSTALL file included in this package.

Building with Visual C++

This package can also be built with Visual C++ for Windows platforms. First, you must copy src/libiff/ifftypes.h.in to src/libiff/ifftypes.h and edit the the latter file.

Change the line:

#define IFF_BIG_ENDIAN @IFF_BIG_ENDIAN@

into

#define IFF_BIG_ENDIAN 0

Then you can open the solution file: src/libiff.sln in Visual Studio to edit or build it. Alternatively, you can use MSBuild to compile it:

$ MSBuild libiff.sln

The output is produced in the Debug/ directory.

Portability

Because this package is implemented in ANSI C (with the small exception that the command line utilities use getopt() ), it should be pretty easy to port this package to new platforms. So far it's tested on the following platforms:

• Linux (i686-linux, x86_64-linux) using GCC
• Cygwin (i686-cygwin, x86_64-cygwin) using GCC
• AmigaOS (m68k-amigaos) using EGCS through Geek Gadgets
• Windows (i686-windows) using Visual C++ 2013

License

This library is available under the MIT license

Using the IFF library API

Using the IFF API to create, read, write and check IFF files is pretty straight forward. Full API documentation can be found in the doc/apidox directory of this package.

Reading IFF files

To read IFF files, include the iff.h header file and call the IFF_read() function:

#include <libiff/iff.h>

int main(int argc, char *argv[])
{
    /* Read an IFF file */
    IFF_Chunk *chunk = IFF_read("input.IFF", NULL);

    if(chunk != NULL)
    {
        /* Use the chunk instance for some purpose here */

        return 0;
    }
    else
        return 1; /* The chunk cannot be read for some reason */
}

Programmatically creating IFF files

An IFF file can be created by composing various IFF struct instances together. The following example creates a concatenation chunk containing a 'TEST' form. The 'TEST' form contains a 'HELO' chunk containing "abcd", and a 'BYE ' chunk containing "1234":

#include <stdlib.h>
#include <libiff/iff.h>
#include <libiff/rawchunk.h>
#include <libiff/form.h>
#include <libiff/cat.h>

#define HELO_BYTES_SIZE 4
#define BYE_BYTES_SIZE 4

#define ID_HELO IFF_MAKEID('H', 'E', 'L', 'O')
#define ID_BYE IFF_MAKEID('B', 'Y', 'E', ' ')
#define ID_TEST IFF_MAKEID('T', 'E', 'S', 'T')

static IFF_Chunk *createHeloChunk(void)
{
    IFF_RawChunk *heloChunk = IFF_createRawChunk(HELO_BYTES_SIZE);

    heloChunk->chunkData[0] = 'a';
    heloChunk->chunkData[1] = 'b';
    heloChunk->chunkData[2] = 'c';
    heloChunk->chunkData[3] = 'd';

    return (IFF_Chunk*)heloChunk;
}

static IFF_Chunk *createByeChunk(void)
{
    IFF_RawChunk *byeChunk = IFF_createRawChunk(BYE_BYTES_SIZE);

    byeChunk->chunkData[0] = '1';
    byeChunk->chunkData[1] = '2';
    byeChunk->chunkData[2] = '3';
    byeChunk->chunkData[3] = '4';

    return (IFF_Chunk*)byeChunk;
}

static IFF_Chunk *createForm(void)
{
    IFF_Chunk *heloChunk = createHeloChunk();
    IFF_Chunk *byeChunk = createByeChunk();
    IFF_Form *form = IFF_createEmptyForm(ID_TEST);

    IFF_addToForm(heloChunk);
    IFF_addToForm(byeChunk);

    return (IFF_Chunk*)form;
}

int main(int argc, char *argv[])
{
    IFF_Chunk *form = createForm();
    IFF_CAT *cat = IFF_createCAT(ID_TEST);

    IFF_addToCAT(form);

    return 0;
}

Retrieving IFF file contents

Quite often you need to retrieve specific properties from an IFF file that are interesting to you, while you may want to ignore the pieces that are not relevant.

The scope of most application formats are restricted to a FORM with a particular form type. For example, ILBM files (which purpose is to store graphics) are essentially FORM chunks with an ILBM form type. Inside these forms various aspects of the picture are specified, such as the resolution, color values of the palette and planar graphics data. The IFF_searchForms() function recursively searches for all form chunks with the given form type inside an IFF file.

Furthermore, each form chunk contains an arbitrary number of data chunks or other group chunks. The IFF_getChunkFromForm() function can be used to retrieve a property from a form. Furthermore, if the given form is a member of a list chunk, which uses shared properties, it also recursively looks up the shared property value, if the requested property has not been defined in the form itself.

It may also be possible that there are more occurences of the same chunk inside a form. In these cases, the IFF_getChunksFromForm() can be used to retrieve all possible values, however this function does not take the shared properties of a list into account.

The following example shows how these functions can be used:

#include <libiff/chunk.h>
#include <libiff/form.h>

#define ID_ILBM IFF_MAKEID('I', 'L', 'B', 'M')
#define ID_BMHD IFF_MAKEID('B', 'M', 'H', 'D')
#define ID_CRNG IFF_MAKEID('C', 'R', 'N', 'G')

int main(int argc, char *argv[])
{
    /* Declarations */
    IFF_Chunk *chunk;
    IFF_Form **ilbmForms;
    unsigned int ilbmFormsLength, i;

    /* Create or read an IFF file here */

    /* Search for all forms having an ILBM form type */
    ilbmForms = IFF_searchForms(chunk, ID_ILBM, &ilbmFormsLength);

    /* Iterate over all ILBM forms in the given IFF file */

    for(i = 0; i < ilbmFormsLength; i++)
    {
        IFF_Form *ilbmForm = ilbmForms[i];
        unsigned int colorRangesLength;

        /* Retrieve the BMHD data property from the ILBM form */
        IFF_Chunk *bitMapHeader = IFF_getChunkFromForm(ilbmForm, ID_BMHD);

        /* Retrieve all possible CRNG properties from the ILBM form */
        IFF_Chunk **colorRanges = IFF_getChunksFromForm(ilbmForm, ID_CRNG, &colorRangesLength);
    }

    return 0;
}

Writing IFF files

A composition of chunks can be written as an IFF file by invoking the IFF_write() function:

#include <libiff/iff.h>

int main(int argc, char *argv[])
{
    IFF_Chunk *chunk;

    /* Create or read a chunk */

    if(IFF_write("output.IFF", chunk, NULL))
        return 0; /* The file has been successfully written */
    else
        return 1; /* Some error occured */
}

IFF conformance checking

The IFF standard defines several constraints that may not be violated. For example, chunk IDs may only contain ASCII characters from a certain range in the ASCII table. Furthermore, the chunk size of a group chunk must be correct and may not be truncated. The IFF_check() function can be used to check whether a composition of chunks conform to the IFF standard:

#include <libiff/iff.h>

int main(int argc, char *argv[])
{
    IFF_Chunk *chunk;

    /* Create or read a chunk here */

    if(IFF_check(chunk, NULL))
        return 0; /* A valid IFF file */
    else
        return 1; /* Not a valid IFF file */
}

Comparing IFF chunk hierarchies

In some cases, it may also be useful to compare IFF chunk hierarchies. For example, to check whether to files (or subsets thereof) are identical. These hierarchies can be compared by using the IFF_compare() function:

#include <libiff/iff.h>

int main(int argc, char *argv[])
{
    IFF_Chunk *chunk1, *chunk2;

    /* Read or create the chunks here */

    if(IFF_compare(chunk1, chunk2, NULL))
        return 0; /* Both chunk hierarchies are equal */
    else
        return 1; /* Both chunk hierarchies are not equal */
}

Command-line utilities

Apart from an API to handle IFF files, this package also includes a number of command-line utilities to make usage of IFF files more convenient:

• iffpp can be used to pretty print an IFF file into a textual representation, so that it can be manually inspected
• iffjoin can be used to join an arbitrary number of IFF files into a new IFF file storing these in an concationation chunk

Consult the manual pages of these tools for more information.

Extending the IFF parser

This package only implements a generic IFF parser, which provides support for a number of group chunks and "raw" data chunks. Application formats typically define various other data chunks, each having their own structure. In order to support a particular application format, the IFF parser needs to be extended.

To implement a parser for an application format you need to:

• Create an application format interface
• Implement extension chunk modules

Creating an application format interface

Every function in the iff.h header requires at least one parameter named: chunkRegistry. So far, we have omitted these extension parameters by specifying a NULL pointer.

These function parameters can be used to support application format chunks by specifying an array of form types pointing to a functions which read, write, free, check and print extension chunks for a given data property.

Because it is incovenient to bother library users with specifying these paramaters over and over again, it is recommended to create a new application format interface, which hides these details. For example, in the libilbm package, you can include the ilbm.h header file, instead of the iff.h header file to conveniently access ILBM files and their application chunks.

The following example shows how to define a header file (test.h) for the interface of an imaginary TEST file format:

#ifndef __TEST_H
#define __TEST_H

#include <stdio.h>
#include <libiff/chunk.h>

#define TEST_ID_TEST IFF_MAKEID('T', 'E', 'S', 'T')

IFF_Chunk *TEST_read(const char *filename);

IFF_Bool TEST_write(const char *filename, const IFF_Chunk *chunk);

void TEST_free(IFF_Chunk *chunk);

void TEST_print(const IFF_Chunk *chunk, const unsigned int indentLevel);

IFF_Bool TEST_compare(const IFF_Chunk *chunk1, const IFF_Chunk *chunk2);

#endif

As you may notice, this header file is almost the same as the iff.h header file, except that the IFF_ prefixes are replaced by the TEST_ prefixes and the chunkRegistry parameter is removed.

The implementation of this interface (test.c) may look as follows:

#include "test.h"
#include <libiff/iff.h>
#include <libiff/id.h>
#include "hello.h"
#include "bye.h"

#define TEST_NUM_OF_FORM_CHUNK_TYPES 1
#define TEST_NUM_OF_CHUNK_TYPES 2

/*
 * Defines how the 'HELO' and 'BYE ' chunks should be managed. The chunks must
 * be alphabetically sorted so that they can be found by a binary search
 * algorithm.
 */
static IFF_ChunkType applicationChunkTypes[] = {
    {TEST_ID_BYE, &TEST_createBye, &TEST_readBye, &TEST_writeBye, &TEST_checkBye, &TEST_freeBye, &TEST_printBye, &TEST_compareBye},
    {TEST_ID_HELO, &TEST_createHello, &TEST_readHello, &TEST_writeHello, &TEST_checkHello, &TEST_freeHello, &TEST_printHello, &TEST_compareHello}
};

/*
 * Composes a node for the above array so that it can be added to a linked list
 */
static IFF_ChunkTypesNode applicationChunkTypesNode = {
    TEST_NUM_OF_CHUNK_TYPES, applicationChunkTypes, NULL
};

/*
 * Refers to the chunk types within the TEST form scope.
 * Also these form types must be alphabetically sorted.
 */
static IFF_FormChunkTypes formChunkTypes[] = {
    { TEST_ID_TEST, &applicationChunkTypesNode }
};

/*
 * Extends the default chunk registry with our custom form chunk types
 */
static const IFF_ChunkRegistry chunkRegistry = IFF_EXTEND_DEFAULT_REGISTRY_WITH_FORM_CHUNK_TYPES(TEST_NUM_OF_FORM_CHUNK_TYPES, formChunkTypes);

/*
 * Represents a registry that specifies how chunks should be managed
 */
static IFF_ChunkRegistry chunkRegistry = {
    TEST_NUM_OF_FORM_CHUNK_TYPES, formChunkTypes
};

/* The following functions hide the the extension parameters for this application format */
IFF_Chunk *TEST_read(const char *filename)
{
    return IFF_read(filename, &chunkRegistry);
}

IFF_Bool TEST_write(const char *filename, const IFF_Chunk *chunk)
{
    return IFF_write(filename, chunk, &chunkRegistry);
}

void TEST_free(IFF_Chunk *chunk)
{
    IFF_free(chunk, NULL, &chunkRegistry);
}

void TEST_print(const IFF_Chunk *chunk, const unsigned int indentLevel)
{
    IFF_print(chunk, indentLevel, &chunkRegistry);
}

IFF_Bool TEST_compare(const IFF_Chunk *chunk1, const IFF_Chunk *chunk2)
{
    return IFF_compare(chunk1, chunk2, &chunkRegistry);
}

In the code fragment above, an array defining extension chunkIDs and function pointers (applicationChunkTypes) specifies how to read, write, check, free and print each individual application chunk. The other array binds these extension chunks into the scope of the TEST form type. This ensures that the TEST.HELO property is parsed by our extension function and that a chunk with the same ID in a different form type is parsed ABCD.HELO differently (because they are not the same).

In the remainder of the interface, we simply call all the functions defined in iff.h with the given chunkRegistry parameter dealing with application format chunks of the TEST format.

Implementing extension chunk modules

Now that we have defined an application interface, you also need to specify how to read, write, free, check and print these application chunks. For each application chunk, you define a seperate module with an header the may look like this (this example defines hello.h to which the previous example refers):

#ifndef __TEST_HELLO_H
#define __TEST_HELLO_H

#include <stdio.h>
#include <libiff/ifftypes.h>
#include <libiff/group.h>
#include <libiff/chunk.h>
#include <libiff/id.h>

#define TEST_ID_HELO IFF_MAKE('H', 'E', 'L', 'O')
#define TEST_HELO_DEFAULT_SIZE (2 * sizeof(IFF_UByte) + sizeof(IFF_UWord))

typedef struct
{
    /* These struct members are mandatory for every chunk */
    IFF_Group *parent;

    IFF_ID chunkId;
    IFF_Long chunkSize;

    /* The remainder of the struct contains custom properties */
    IFF_UByte a;
    IFF_UByte b;
    IFF_UWord c;
}
TEST_Hello;

TEST_Hello *TEST_createHello(const IFF_ID chunkId, const IFF_Long chunkSize);

IFF_Bool TEST_readHello(FILE *file, IFF_Chunk *chunk, const IFF_ChunkRegistry *chunkRegistry, IFF_Long *bytesProcessed);

IFF_Bool TEST_writeHello(FILE *file, const IFF_Chunk *chunk, const IFF_ChunkRegistry *chunkRegistry, IFF_Long *bytesProcessed);

IFF_Bool TEST_checkHello(const IFF_Chunk *chunk, const IFF_ChunkRegistry *chunkRegistry);

void TEST_freeHello(IFF_Chunk *chunk, const IFF_ChunkRegistry *chunkRegistry);

void TEST_printHello(const IFF_Chunk *chunk, const unsigned int indentLevel, const IFF_ChunkRegistry *chunkRegistry);

IFF_Bool TEST_compareHello(const IFF_Chunk *chunk1, const IFF_Chunk *chunk2, const IFF_ChunkRegistry *chunkRegistry);

#endif

The code fragment above defines the read, write, check, free, and print functions for the 'HELO' chunk. As you may notice, every function has the TEST_ prefix, because it is part of the TEST application format. Furthermore, in each module every read, write, check, free and print function must have the same return type and function parameters.

And the implementation may look as follows:

#include "hello.h"
#include <libiff/io.h>
#include <libiff/util.h>
#include "test.h"

IFF_Chunk *TEST_createHello(const IFF_ID chunkId, const IFF_Long chunkSize)
{
    TEST_Hello *hello = (TEST_Hello*)IFF_createChunk(chunkId, chunkSize, sizeof(TEST_Hello));

    if(hello != NULL)
    {
        hello->a = '\0';
        hello->b = '\0';
        hello->c = 0;
    }

    return (IFF_Chunk*)hello;
}

IFF_Bool TEST_readHello(FILE *file, IFF_Chunk *chunk, const IFF_ChunkRegistry *chunkRegistry, IFF_Long *bytesProcessed)
{
    TEST_Hello *hello = (TEST_Hello*)chunk;
    IFF_FieldStatus status;

    if((status = IFF_readUByteField(file, &hello->a, chunk, "a", bytesProcessed)) != IFF_FIELD_MORE)
        return IFF_deriveSuccess(status);

    if((status = IFF_readUByteField(file, &hello->b, chunk, "b", bytesProcessed)) != IFF_FIELD_MORE)
        return IFF_deriveSuccess(status);

    if((status = IFF_readUWordField(file, &hello->c, chunk, "c", bytesProcessed)) != IFF_FIELD_MORE)
        return IFF_deriveSuccess(status);

    return TRUE;
}

IFF_Bool TEST_writeHello(FILE *file, const IFF_Chunk *chunk, const IFF_ChunkRegistry *chunkRegistry, IFF_Long *bytesProcessed)
{
    const TEST_Hello *hello = (TEST_Hello*)chunk;
    IFF_FieldStatus status;

    if((status = IFF_writeUByteField(file, hello->a, chunk, "a", bytesProcessed)) != IFF_FIELD_MORE)
        return IFF_deriveSuccess(status);

    if((status = IFF_writeUByteField(file, hello->b, chunk, "b", bytesProcessed)) != IFF_FIELD_MORE)
        return IFF_deriveSuccess(status);

    if((status = IFF_writeUWordField(file, hello->c, chunk, "c", bytesProcessed)) != IFF_FIELD_MORE)
        return IFF_deriveSuccess(status);

    return TRUE;
}

IFF_Bool TEST_checkHello(const IFF_Chunk *chunk, const IFF_ChunkRegistry *chunkRegistry)
{
    return TRUE;
}

void TEST_freeHello(IFF_Chunk *chunk, const IFF_ChunkRegistry *chunkRegistry)
{
}

void TEST_printHello(const IFF_Chunk *chunk, const unsigned int indentLevel, const IFF_ChunkRegistry *chunkRegistry)
{
    const TEST_Hello *hello = (const TEST_Hello*)chunk;

    IFF_printIndent(stdout, indentLevel, "a = %c;\n", hello->a);
    IFF_printIndent(stdout, indentLevel, "b = %c;\n", hello->b);
    IFF_printIndent(stdout, indentLevel, "c = %u;\n", hello->c);
}

IFF_Bool TEST_compareHello(const IFF_Chunk *chunk1, const IFF_Chunk *chunk2, const IFF_ChunkRegistry *chunkRegistry)
{
    const TEST_Hello *hello1 = (const TEST_Hello*)chunk1;
    const TEST_Hello *hello2 = (const TEST_Hello*)chunk2;

    if(hello1->a != hello2->a)
        return FALSE;

    if(hello1->b != hello2->b)
        return FALSE;

    if(hello1->c != hello2->c)
        return FALSE;

    return TRUE;
}