Random attempt at using libdvbsak#1:
Possible API for parsing a PAT table:
struct bitstream {
union {
struct {
unsigned char *buf;
size_t len;
} mem;
} u;
enum {
BITSTREAM_MEM,
} type;
unsigned int ptr;
int error;
};
struct dvbsi_section {
int pid;
uint8_t table_id;
int section_syntax_ind;
/* if section_syntax_indicator == 1: */
int table_id_ext;
int version_nr;
int current_next_ind;
uint8_t section_nr;
uint8_t last_section_nr;
uint8_t crc32[4];
struct bitstream contents;
};
struct dvbsi_table_pat {
uint16_t transport_stream_id;
uint16_t network_pid;
int program_count;
};
struct dvbsi_table_pat_program {
uint16_t number;
uint16_t map_pid;
};
int dvbsi_decode_section_header(struct dvbsi_section* section, struct
bitstream* bs);
int dvbsi_table_decode_pat(struct dvbsi_table_pat* pat, struct bitstream* bs);
int dvbsi_table_decode_pat_program(struct dvbsi_table_pat_program* program,
struct bitstream* bs);
The return value is the number of bytes used, or a negative error code. The
"struct bitstream" pointers are updated by the _caller_ using this value.
The idea is you decode the section header first with
dvbsi_decode_section_header(), then you decode the PAT with
dvbsi_table_decode_pat() if it was one. Since the PAT can have a variable
number of programs, that only decodes the invariant parts, so you use
dvbsi_table_decode_pat_program() to decode the programs themselves.
Descriptors are done similarily - you get a bitstream structure pointing at
the start of them in an SI table structure.
You use the followings to get the ID and length of the descriptor.
int dvbsi_descriptor_id(struct bitstream* bs);
int dvbsi_descriptor_length(struct bitstream* bs);
if you like the descriptor, you decode it with the appropriate function.
int dvbsi_descriptor_decode_service(struct dvbsi_descriptor_service *srv,
struct bitstream *bs);
....
Otherwise you just skip over the descriptor data to the next one.
This involves no mallocing whatsoever - and it leaves it entirely up to the
application what to do with the extracted data. Sort of like a a SAX-style
XML parser as opposed to a DOM style if you know what I mean.
Hmm, this idea reminds me of a certain other library now I think of it...
anyway lemme know what you think.
