PackDir

From Just Solve the File Format Problem
(Difference between revisions)
Jump to: navigation, search
(Timestamp formula given was not completely correct, and not needed in the age of 64-bit ints.)
Line 6: Line 6:
  
 
The LZW algorithm is the variant that is used in the [[ZOO]] archiver.
 
The LZW algorithm is the variant that is used in the [[ZOO]] archiver.
 +
 +
Its [[RISC OS filetype]] is 68E.
  
 
== File format ==
 
== File format ==
Line 12: Line 14:
 
32bit ints are in LSB format.  Strings are NULL-terminated ASCII.
 
32bit ints are in LSB format.  Strings are NULL-terminated ASCII.
  
File/directory dates and times are stored in a RISC OS format: The low 8 bits of the "load" address, combined with the "exec" address, are a 40-bit integer representing the number of 0.01-second intervals since the beginning of 1900.
+
File/directory dates and times are stored in a RISC OS format: The low 8 bits of the "load" address, combined with the "exec" address, are a 40-bit integer representing the number of 0.01-second intervals since the beginning of the year 1900.
  
 
File format:
 
File format:
Line 37: Line 39:
 
     32bit_int: Compression_type: -1 == Not compressed, otherwise compressed with LZW.
 
     32bit_int: Compression_type: -1 == Not compressed, otherwise compressed with LZW.
 
     N bytes: archived file data.  N == compressed_file_length if compression_type == -1, otherwise original_file_length
 
     N bytes: archived file data.  N == compressed_file_length if compression_type == -1, otherwise original_file_length
 +
 +
== Identification ==
 +
Files begin with "{{magic|PACK}}", followed {{magic|0x00}}. Note that this is not enough to distinguish them from [[Git]] pack format.
  
 
== Software ==
 
== Software ==

Revision as of 15:29, 10 February 2019

File Format
Name PackDir
Ontology

PackDir is an archiving tool for RISC OS. It can compress file using an LZW algorithm, or store uncompressed files.

The LZW algorithm is the variant that is used in the ZOO archiver.

Its RISC OS filetype is 68E.

Contents

File format

I have reverse engineered the file format.

32bit ints are in LSB format. Strings are NULL-terminated ASCII.

File/directory dates and times are stored in a RISC OS format: The low 8 bits of the "load" address, combined with the "exec" address, are a 40-bit integer representing the number of 0.01-second intervals since the beginning of the year 1900.

File format:

   Header
   object_entry_1
   object_entry_2
   ...
   object_entry_N

Header:

   String: "PACK\0"
   32bit_int: LZW maxbits - 12

object_entry:

   string: object_name (including trailing NULL)
   32bit_int: load (RISC OS load address)
   32bit_int: exec (RISC OS exec address)
   32bit_int: original_file_length if this object is a file, number_of_child_entries otherwise
   32bit_int: attributes (RISC OS read/write permissions)
   32bit_int: entry_type: 1 == directory, 2 == file.  NB: This field is missing for the first entry in the file - the first entry is always a directory

If the entry is a file, then these fields follow:

   32bit_int: Compressed_file_length
   32bit_int: Compression_type: -1 == Not compressed, otherwise compressed with LZW.
   N bytes: archived file data.  N == compressed_file_length if compression_type == -1, otherwise original_file_length

Identification

Files begin with "PACK", followed 0x00. Note that this is not enough to distinguish them from Git pack format.

Software

Example files

Personal tools
Namespaces

Variants
Actions
Navigation
Toolbox