vrs::FileFormat Namespace Reference

Namespace for key datastructures of a VRS file. More...

Classes
struct	FileHeader
	Every file starts with this header, which may grow but not shrink! More...
class	LittleEndian
	Placeholder layer for endianness support, if we ever need it. More...
struct	RecordHeader
	Every record starts with this header, and is followed by a raw data blob, which semantic is private to the data type handler. More...

Functions
RecordableTypeId	readRecordableTypeId (const FileFormat::LittleEndian< int32_t > &recordableTypeId)
bool	printVRSFileInternals (unique_ptr< FileHandler > &file)
constexpr uint32_t	fourCharCode (char a, char b, char c, char d)
bool	printVRSFileInternals (std::unique_ptr< FileHandler > &file)

Variables
constexpr uint32_t	kMagicHeader1 = fourCharCode('V', 'i', 's', 'i')
constexpr uint32_t	kMagicHeader2 = fourCharCode('o', 'n', 'R', 'e')
constexpr uint32_t	kMagicHeader3 = fourCharCode('c', 'o', 'r', 'd')
constexpr uint32_t	kOriginalFileFormatVersion = fourCharCode('V', 'R', 'S', '1')
	Original file format.
constexpr uint32_t	kFrontIndexFileFormatVersion = fourCharCode('V', 'R', 'S', '2')
	When we added support for place the index record at the beginning of the file.
constexpr uint32_t	kZstdFormatVersion = fourCharCode('V', 'R', 'S', '3')
	When we added support for ztd compression. Used only briefly.

Detailed Description

Namespace for key datastructures of a VRS file.

File format description:

Every file starts with one FileHeader structure. It is followed by an arbitrary number of records. That header gives you the size of the FileHeader and the size of the RecordHeader structures used for all the following records.

Be careful that both FileHeader & RecordHeader structures may grow in the future, so that when you read a file, you must use these given sizes to move around the file, while skipping possible parts of the headers your code doesn't know about.

Each record starts with a RecordHeader structure, which provides sizes so that you can find both the next and the previous record, and implicitly tells you the size of its raw byte blob.

Each RecordHeader a typeId, which tells us what handler knows how to manipulate this data. This typeId is extremely important and must not collide between types!

The RecordHeader gives an instance id, in case multiple streams of data of the same type are needed. Each RecordHeader is immediately followed by a blob of raw bytes, which meaning is only known by the type handler.

The RecordHeader provides a record format version, to be interpreted by the type handler, so that type handlers can handle changes in the format of their data.

The RecordHeader also provides a timestamp for the data, to be used when playing back the data, as well as a Record::Type.

Function Documentation

fourCharCode()

constexpr uint32_t vrs::FileFormat::fourCharCode ( char  a, char  b, char  c, char  d  )

constexpr

Assemble four letters into a uint32_t to make "good looking" magic numbers on disk... Because we're specific letters, we reduce chances of an accidental match.

printVRSFileInternals()

bool vrs::FileFormat::printVRSFileInternals

(

std::unique_ptr< FileHandler > &

file

)

Debug method to printout key internal details about a VRS file for debugging purposes.

Returns

True if the file looks "good", false if anything doesn't look right.