User:MPC/SRT Subfile Format

From OpenStreetMap Wiki
Jump to navigation Jump to search

User:MPC

A reverse-engineering of Garmin's SRT Subfile format

SRT Header

Common Header

The structure of the common header is the same in all subfiles.

Byte Offset Length (bytes) Description
0x000 2 Header length. So far only '0x1D' and '0x27' has been seen.
0x002 10 Type GARMIN SRT
0x00C 1 0x01 ?
0x00D 1 0x00
0x00E 2 Creation year
0x010 1 Creation month
0x011 1 Creation day
0x012 1 Creation hour
0x013 1 Creation minute
0x014 1 Creation second

Specific SRT Header

Byte Offset Length (bytes) Description
0x015 2 ?? Only seen 0x0001 in here so far
0x017 4 Offset of 'Contents Section' (see below)? Only seen 0x0000001D in here so far
0x01B 2 Length (in bytes) of 'Contents Section'? Only seen 0x0010 in here so far

Contents Section

Byte Offset Length (bytes) Description
0x01D 4 Offset of 'Title Section' (see below)? Only seen 0x0000002D in here so far.
0x021 4 Length (in bytes) of 'Title Section'? Only seen 0x00000016 in here so far.
0x025 4 Offset of 'Character Lookup Table'? Only seen 0x00000043 in here so far.
0x029 4 Length (in bytes) of 'Character Lookup Table'? Only seen 0x00000355 in here so far.

Title Section

Byte Offset Length (bytes) Description
0x02D ?? A null-terminated string describing the SRT file's purpose. "Western European Sort" for instance.

The Character Lookup Table Header

Not sure about very much of this section yet.

Byte Offset Length (bytes) Description
0x043 2 ?? Only seen 0x0034 here so far
0x045 2 ?? Only seen 0x0007 here so far
0x047 2 ?? Only seen 0x0002 here so far
0x049 2 Codepage corresponding to Character Lookup Table? Only seen 0x04E4 here so far (codepage 1252?)
0x04B 2 ?? Only seen 0x2002 here so far
0x04D 2 ?? Only seen 0x0000 here so far
0x04F 4 Offset of character lookup table? 0x00000034 seen here so far
0x053 4 Length (in bytes) of character lookup table? 0x000002FD seen here so far (which is 765 - only enough for 255 entries)
0x057 4 Size (in bytes) of each entry in the character lookup table? 0x00000003 seen here so far
0x05B 4 Character lookup table flags? 0x00000000 seen here so far
0x05F 4 Offset of what follows the character lookup table? 0x00000331 seen here so far (which is 0x34 + 0x2FD (see above))
0x063 4 Length (in bytes) of what follows the character lookup table? 0x00000024 seen here so far
0x067 4 Size (in bytes) of each entry in what follows the character lookup table? 0x00000002 seen here so far
0x06B 4 Flags for what follows the character lookup table? 0x00000000 seen here so far
0x06F 2 ?? 0x0034 seen here so far

The Character Lookup Table

This is the guts of the file, and basically consists of 256 3-byte entries, one for each character of the codepage used by the map.

Byte Offset Length (bytes) Description
0x071 3 The entry in the character lookup table corresponding to character 0x00 of the codepage used by the maps.
0x074 3 The entry in the character lookup table corresponding to character 0x01 of the codepage used by the maps.
etc etc etc

Each entry in the table looks like this:

Byte Offset Length (bytes) Description
0x00 1 Flags giving the character's 'class'. A bitwise-OR of: 0x01=alphabetic, 0x02=numeric, 0x10=ligature, 0x20=mathematical(?).
0x01 1 The collating position for this character. Low values appear earlier in the sorted results.
0x02 1 Flags describing diacritic marks on a character. One of the following: (0x10='lowercase or digit', 0x20='uppercase') which is then bitwise-ORed with one of the following: (0x01='no diacritic', 0x02=acute, 0x03=grave, 0x04=circumflex(or caron), 0x05=ring, 0x06=umlaut, 0x07=tilde).

It seems that Garmin don't 100% stick to this suggested bit-allocation for the diacritic marks. It's possible that the value is more to allocate an order to the sorting of diacritic letters rather than identifying them uniquely.