You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

545 lines
22 KiB

API definition of libisoburn.
Copyright 2007-2008 Vreixo Formoso Lopes <>
and Thomas Schmitt <>
/** Overview
libisoburn is a frontend for libraries libburn and libisofs which enables
creation and expansion of ISO-9660 filesystems on all CD/DVD media supported
by libburn. This includes media like DVD+RW, which do not support multi-session
management on media level and even plain disk files or block devices.
The price for that is thorough specialization on data files in ISO-9660
filesystem images. So libisoburn is not suitable for audio (CD-DA) or any
other CD layout which does not entirely consist of ISO-9660 sessions.
Wrapper functions
The priciple of this frontend is that you may use any call of libisofs or
libburn unless it has a isoburn_*() wrapper listed in the following function
E.g. call isoburn_initialize() rather than iso_init(); burn_initialize()
and call isoburn_drive_scan_and_grab() rather than burn_drive_scan_and_grab().
But you may call burn_disc_get_profile() directly if you want to display
the media type.
The wrappers will transparently provide the necessary emulations which
are appropriate for particular target "drives" and media states.
To learn about them you have to read both API descriptions: the one of
the wrapper and the one of the underlying libburn or libisofs call.
Usage model
Additionally there are own libisoburn API calls which allow to implement the
following usage model (see also man xorriso for a end user's view):
There may be an input drive and an output drive. Either of them may be missing
with the consequence that no reading resp. writing is possible.
Both drive roles can be fulfilled by the same drive.
Input can be a random access readable libburn drive:
optical media, regular files, block devices.
Output can be any writeable libburn drive:
writeable optical media in burner, writeable file objects (no directories).\
libburn needs rw-permissions to drive device file resp. file object.
If the input drive provides a suitable ISO RockRidge image, then its tree
may be loaded into memory and can then be manipulated by libisofs API calls.
The loading is done by isoburn_read_image() under control of
struct isoburn_read_opts
>>> which the application obtains from libisoburn.
Writing of result images is controlled by libisofs related parameters
in struct isoburn_source_opts
>>> which the application obtains from libisoburn.
All multi-session aspects are handled by libisoburn according to these
settings. The application does not have to analyze media state and write
job parameters.
Setup for Growing or Modifying
There are two alternative API calls for performing the setup for two
alternative image generation strategies.
If input and output drive is the same, then isoburn_prepare_disc() is to
be used. It will lead to an add-on session on appendable or overwriteable
media with existing ISO image. With blank media it will produce a first
If the output drive is not the input drive, then it has to bear blank media
or overwriteable without a valid ISO image. To prepare for such an image
generation run, use isoburn_prepare_new_image(). The run will copy file data
from an eventual input drive with valid image, add any newly introduced data
from the local filesystem, and produce a first session on output media.
After either of these setups, some peripheral libburn drive parameter settings
like burn_write_opts_set_simulate(), burn_write_opts_set_multi(),
burn_drive_set_speed(), burn_write_opts_set_underrun_proof() should be made.
Do not set the write mode. It will be chosen by libisoburn so it matches job
and media state.
Writing the image
Then one may start image generation and write threads by isoburn_disc_write().
Progress may be watched at the output drive by burn_drive_get_status() and
At some time, the output drive will be BURN_DRIVE_IDLE indicating that
writing has ended.
One should inquire isoburn_drive_wrote_well() to learn about overall success.
Finally one must call isoburn_activate_session() which will finalize any
eventual multi-session emulation.
/* API functions */
/** Initialize libisoburn, libisofs and libburn.
Wrapper for : iso_init() and burn_initialize()
@param reason A character array for eventual messages (e.g. with errors)
@param flag Bitfield for control purposes (unused yet, submit 0)
@return 1 indicates success, 0 is failure
int isoburn_initialize(char msg[1024], int flag);
/** Obtain the three release version numbers of the library.
@param major The maturity version (0 for now, as we are still learning)
@param minor The development goal version.
@param micro The development step version. This has an additional meaning:
Pare numbers indicate a version with frozen API. I.e. you can
rely on the same set of features to be present in all
published releases with that major.minor.micro combination.
Odd numbers indicate that API upgrades are in progress.
I.e. new features might be already present or they might
be still missing. You need to check before compiling an
application which relies on freshly introduced features.
void isoburn_version(int *major, int *minor, int *micro);
/* >>> isoburn_is_compatible() */
/** Aquire a target drive by its filesystem path resp. libburn persistent
Wrapper for: burn_drive_scan_and_grab()
int isoburn_drive_scan_and_grab(struct burn_drive_info *drive_infos[],
char* adr, int load);
/** Aquire a drive from the burn_drive_info[] array which was obtained by
a previous call of burn_drive_scan().
Wrapper for: burn_drive_grab()
int isoburn_drive_grab(struct burn_drive *drive, int load);
/** Inquire the media status. Expect the whole spectrum of libburn BURN_DISC_*
with multi-session media. Emulated states with random access media are
Wrapper for: burn_disc_get_status()
enum burn_disc_status isoburn_disc_get_status(struct burn_drive *drive);
/** Tells whether the media can be treated by isoburn_disc_erase().
Wrapper for: burn_disc_erasable()
int isoburn_disc_erasable(struct burn_drive *d);
/** Mark the media as blank. With multi-session media this will call
burn_disc_erase(). With random access media, an eventual ISO-9660
filesystem will get invalidated by altering its start blocks on media.
In case of success, the media is in status BURN_DISC_BLANK afterwards.
Wrapper for: burn_disc_erase()
void isoburn_disc_erase(struct burn_drive *drive, int fast);
/* >>> this goes to isoburn.h */
* Options for image reading.
struct isoburn_read_opts {
unsigned int norock:1; /*< Do not read Rock Ridge extensions */
unsigned int nojoliet:1; /*< Do not read Joliet extensions */
unsigned int noiso1999:1; /*< Do not read ISO 9660:1999 enhanced tree */
unsigned int preferjoliet:1;
/*< When both Joliet and RR extensions are present, the RR
* tree is used. If you prefer using Joliet, set this to 1. */
uid_t uid; /**< Default uid when no RR */
gid_t gid; /**< Default uid when no RR */
mode_t mode; /**< Default mode when no RR (only permissions) */
* Input charset for RR file names. NULL to use default locale charset.
char *input_charset;
/* modified by the function isoburn_read_image */
unsigned int hasRR:1; /*< It will be set to 1 if RR extensions are present,
to 0 if not. */
unsigned int hasJoliet:1; /*< It will be set to 1 if Joliet extensions are
present, to 0 if not. */
* It will be set to 1 if the image is an ISO 9660:1999, i.e. it has
* a version 2 Enhanced Volume Descriptor.
unsigned int hasIso1999:1;
/** It will be set to 1 if El-Torito boot record is present, to 0 if not.*/
unsigned int hasElTorito:1;
uint32_t size; /**< Will be filled with the size (in 2048 byte block) of
* the image, as reported in the PVM. */
unsigned int pretend_blank:1; /* always create empty image */
/* >>> this goes to isoburn.h */
* Options for image generation by libisofs and image transport to libburn.
struct isoburn_source_opts {
/* Options for image generation */
int level; /**< ISO level to write at. */
/** Which extensions to support. */
unsigned int rockridge :1;
unsigned int joliet :1;
unsigned int iso1999 :1;
/* relaxed constraints */
* Relaxed constraints. Setting any of these to 1 break the specifications,
* but it is supposed to work on most moderns systems. Use with caution.
* Omit the version number (";1") at the end of the ISO-9660 identifiers.
* Version numbers are usually not used.
unsigned int omit_version_numbers :1;
* Allow ISO-9660 directory hierarchy to be deeper than 8 levels.
unsigned int allow_deep_paths :1;
* Allow path in the ISO-9660 tree to have more than 255 characters.
unsigned int allow_longer_paths :1;
* Allow a single file or directory hierarchy to have up to 37 characters.
* This is larger than the 31 characters allowed by ISO level 2, and the
* extra space is taken from the version number, so this also forces
* omit_version_numbers.
unsigned int max_37_char_filenames :1;
* ISO-9660 forces filenames to have a ".", that separates file name from
* extension. libisofs adds it if original filename doesn't has one. Set
* this to 1 to prevent this behavior
unsigned int no_force_dots :1;
* Allow lowercase characters in ISO-9660 filenames. By default, only
* uppercase characters, numbers and a few other characters are allowed.
unsigned int allow_lowercase :1;
* Allow all ASCII characters to be appear on an ISO-9660 filename. Note
* that "/" and "\0" characters are never allowed, even in RR names.
unsigned int allow_full_ascii :1;
* Allow paths in the Joliet tree to have more than 240 characters.
unsigned int joliet_longer_paths :1;
unsigned int sort_files:1;
/**< If files should be sorted based on their weight. */
* The following options set the default values for files and directory
* permissions, gid and uid. All these take one of three values: 0, 1 or 2.
* If 0, the corresponding attribute will be kept as set in the IsoNode.
* Unless you have changed it, it corresponds to the value on disc, so it
* is suitable for backup purposes. If set to 1, the corresponding attrib.
* will be changed by a default suitable value. Finally, if you set it to
* 2, the attrib. will be changed with the value specified in the options
* below. Note that for mode attributes, only the permissions are set, the
* file type remains unchanged.
unsigned int replace_dir_mode :2;
unsigned int replace_file_mode :2;
unsigned int replace_uid :2;
unsigned int replace_gid :2;
mode_t dir_mode; /** Mode to use on dirs when replace_dir_mode == 2. */
mode_t file_mode; /** Mode to use on files when replace_file_mode == 2. */
uid_t uid; /** uid to use when replace_uid == 2. */
gid_t gid; /** gid to use when replace_gid == 2. */
char *output_charset; /**< NULL to use default charset */
/* Options for image transport */
/** The number of bytes to be used for the fifo which decouples libisofs
and libburn for better throughput and for reducing the risk of
interrupting signals hitting the libburn thread which operates the
MMC drive.
The size will be rounded up to the next full 2048.
Minimum is 64kiB, maximum is 1 GiB (but that is too much anyway).
int fifo_size;
/* >>> Opaque definitions of isoburn_read_opts and isoburn_source_opts */
/* >>> Constructors, destructors, getters, setters. */
/* >>> Learn what libisofs does */
/** Get the image attached to a drive, if any.
@return A reference to attached image, or NULL if the drive has no image
attached. This reference needs to be released via iso_image_unref()
when it is not longer needed.
IsoImage *isoburn_get_attached_image(struct burn_drive *d);
/** Load the ISO filesystem directory tree from the media in the given drive.
This will give libisoburn the base on which it can let libisofs perform
image growing or image modification. The loaded volset gets attached
to the drive object and handed out to the application.
Not a wrapper, but peculiar to libisoburn.
@param d The drive which holds an existing ISO filesystem or blank media.
d is allowed to be NULL which produces an empty ISO image. In
this case one has to call before writing isoburn_attach_volset()
with the volset from this call and with the intended output
@param read_opts The read options which can be chosen by the application
@param image the image read, if the disc is blank it will have no files.
This reference needs to be released via iso_image_unref() when
it is not longer needed. The drive, if not NULL, will hold an
own reference which it will release when it gets a new volset
or when it gets released via isoburn_drive_release().
You can pass NULL if you already have a reference or you plan to
obtain it later with isoburn_get_attached_image(). Of course, if
you haven't specified a valid drive (i.e., if d == NULL), this
parameter can't be NULL.
@return <=0 error , 1 = success
int isoburn_read_image(struct burn_drive *d,
struct isoburn_read_opts *read_opts,
IsoImage **image);
/** Set the IsoImage to be used with a drive. This eventually releases
the reference to the old IsoImage attached to the drive.
Caution: Use with care. It hardly makes sense to replace an image that
reflects a valid ISO image on media.
This call is rather intended for writing a newly created and populated
image to blank media. The use case in xorriso is to let an image survive
the change or demise of the outdev target drive.
@param d The drive which shall be write target of the volset.
@param image The image that represents the image to be written.
This image pointer MUST already be a valid reference suitable
for iso_image_unref().
It may have been obtained by appropriate libisofs calls or by
isoburn_read_image() with d==NULL.
@return <=0 error , 1 = success
int isoburn_attach_image(struct burn_drive *d, IsoImage *image);
/** Obtain the start block number of the most recent session on media. In
case of random access media this will always be 0. Succesfull return is
not a guarantee that there is a ISO-9660 image at all. The call will fail,
nevertheless,if isoburn_disc_get_status() returns not BURN_DISC_APPENDABLE.
Wrapper for: burn_disc_get_msc1()
int isoburn_disc_get_msc1(struct burn_drive *d, int *start_lba);
/** Use this with trackno==0 to obtain the predicted start block number of the
new session. The interesting number is returned in parameter nwa.
Wrapper for: burn_disc_track_lba_nwa()
int isoburn_disc_track_lba_nwa(struct burn_drive *d, struct burn_write_opts *o,
int trackno, int *lba, int *nwa);
/** Obtain the size which was attributed to an emulated appendable on actually
overwriteable media. This value is supposed to be <= 2048 * nwa as of
@param drive The drive holding the media.
@param start_byte The reply value counted in bytes, not in sectors.
@param flag Unused yet. Submit 0.
@return 1=stat_byte is valid, 0=not an emulated appendable, -1=error
int isoburn_get_min_start_byte(struct burn_drive *d, off_t *start_byte,
int flag);
/** Create a disc object for writing the new session from the created or loaded
iso_volset which has been manipulated via libisofs, to the same media from
where the image was eventually loaded. This struct burn_disc is ready for
use by a subsequent call to isoburn_disc_write().
After this asynchronous writing has ended and the drive is BURN_DRIVE_IDLE
again, the burn_disc object has to be disposed by burn_disc_free().
@param drive The combined source and target drive, grabbed with
isoburn_drive_scan_and_grab(). .
@param disc Returns the newly created burn_disc object.
@return <=0 error , 1 = success
int isoburn_prepare_disc(struct burn_drive *d, struct burn_disc **disc,
struct isoburn_source_opts *opts);
/** Create a disc object for producing a new image from a previous image
plus the changes made by user. The generated burn_disc is suitable
to be written to any grabbed libburn drive with blank writeable media.
But you must not use the same drive for input and output, because data
will be read from the source drive while at the same time the target
drive is already writing.
The resulting burn_disc object has to be disposed when all its writing
is done and the drive is BURN_DRIVE_IDLE again after asynchronous
@param in_drive The input drive,grabbed with isoburn_drive_scan_and_grab().
@param disc Returns the newly created burn_disc object.
@param opts Options for image generation and data transport to media.
@param out_drive The libburn drive which shall be write target.
If the drive was grabbed via libisoburn then it can later
access the libisofs source fifo via
Mere libburn drives cannot obtain this info.
In that case out_drive may be NULL, as well.
@return <=0 error , 1 = success
int isoburn_prepare_new_image(struct burn_drive *in_drive,
struct burn_disc **disc,
struct isoburn_source_opts *opts,
struct burn_drive *out_drive);
/** Start writing of the new session.
This call is asynchrounous. I.e. it returns quite soon and the progress has
to be watched by a loop with call burn_drive_get_status() until
BURN_DRIVE_IDLE is returned.
Wrapper for: burn_disc_write()
void isoburn_disc_write(struct burn_write_opts *o, struct burn_disc *disc);
/** Inquire state and fill parameters of the fifo which is attached to
the emerging track. This should be done in the pacifier loop while
isoburn_disc_write() or burn_disc_write() are active.
This works only with drives obtained by isoburn_drive_scan_and_grab()
or isoburn_drive_grab(). If isoburn_prepare_new_image() was used, then
parameter out_drive must have announced the track output drive.
Hint: If only burn_write_opts and not burn_drive is known, then the drive
can be obtained by burn_write_opts_get_drive().
@parm d The drive to which the track with the fifo gets burned.
@param size The total size of the fifo
@param free_bytes The current free capacity of the fifo
@param status_text Returns a pointer to a constant text, see below
@return <0 reply invalid, >=0 fifo status code:
bit0+1=input status, bit2=consumption status, i.e:
0="standby" : data processing not started yet
1="active" : input and consumption are active
2="ending" : input has ended without error
3="failing" : input had error and ended,
4="unused" : ( consumption has ended before processing start )
5="abandoned" : consumption has ended prematurely
6="ended" : consumption has ended without input error
7="aborted" : consumption has ended after input error
int isoburn_get_fifo_status(struct burn_drive *d, int *size, int *free_bytes,
char **status_text);
/** Inquire whether the most recent write run was successful.
Wrapper for: burn_drive_wrote_well()
int isoburn_drive_wrote_well(struct burn_drive *d);
/** Call this after isoburn_disc_write has finished and burn_drive_wrote_well()
indicates success. It will eventually complete the emulation of
multi-session functionality, if needed at all. Let libisoburn decide.
Not a wrapper, but peculiar to libisoburn.
int isoburn_activate_session(struct burn_drive *drive);
#if 0
/** Write a new session to a disc.
This is a synchronous call equivalent to isoburn_prepare_disc +
isoburn_disc_write + isoburn_activate_session
@param pacifier_func If not NULL: a function to produce appeasing messages.
See burn_abort_pacifier() in libburn.h for an example.
/* TODO implement this */
int isoburn_perform_write(struct burn_write_opts *o,
int (*pacifier_func)(void *handle, int patience,
int elapsed));
#endif /* 0 */
/** Release an aquired drive.
Wrapper for: burn_drive_release()
void isoburn_drive_release(struct burn_drive *drive, int eject);
/** Shutdown all three libraries.
Wrapper for : iso_finish() and burn_finish().
void isoburn_finish(void);
The following calls are for expert applications only.
An application should have a special reason to use them.
/** Inquire wether the media needs emulation or would be suitable for
generic multi-session via libburn.
@return 0 is generic multi-session
1 is emulated multi-session
-1 is not suitable for isoburn
int isoburn_needs_emulation(struct burn_drive *drive);