Datasets:
HDTenEightyP
/
CompuServe

Dataset card Data Studio Files
xet
 Community
text
stringlengths
0
99.6k
offset length description
------ ------ -----------
0 1 version 1 for ARC 1.xx
2 for ARC 2.xx
1 1 storage. 0=store 1=pack
2=squeez 3=crunch
4=squash 5=1 pass crunch
2 2 checksum lo,hi
4 3 original length-(bytes) lo,mid,hi
7 2 squeezed length-(blocks) lo,hi
9 1 file type. s,p,u or r
10 1 length of filename
11 n filename
The following additional bytes occur if version is 2
or higher.
11+n 1 record length if relative file.
(254 otherwise)
11+n+1 2 date in MS-DOS format.
bits: 0-4 = day
5-8 = month
9-15= year-1980
The file data follows starting at offset 11+n+2.
If the file is stored, then a copy of the file
follows. The number of bytes are specified at offset 4 in
the header.
If the file is packed, then ARCs numbered 1.xx will
choose the byte which occurs least often in the file as the
control character. ARCs numbered 2.xx always use 254 as the
control character. The control character is stored at offset
11+n+2 in the header and is followed by the packed file. A
control sequence is <control char> <count> <character to
repeat>. Count can be between 3 and 256, with 0 being 256.
ARCs numbered less than 2.00 had a bug in the pack routines.
To compensate for this bug, a count of zero should be
replaced by a count of 255, and the file will un-pack
properly. This bug has been corrected in ARCs numbered 2.00
or higher.
ARC VERSION 2.20 PAGE - 39
If the file is squeezed, squashed or crunched, then it
must be read one bit at a time from here on in. ARC fills a
byte from right to left, so the bits will appear differently
if you look at them with a disk doctor, but this is the
order in which they are written to the file.
Squeezed file encoding table.
L0 C0 L1 C1 L2 C2 ... L255 C255
Since each code can be as long as 24 bits, each length
is 5 bits long. Thus L0 is the length of the Huffman code
for ascii 0, and C0 is the actual Huffman code. The total
length of the encoding table is then:
(5 * 256) + sum(code lengths) ... in bits
If an ascii value does not occur in the file, its code
length will be zero and the bit immediately following will
be the length for the next ascii value. This is why when you
squeeze a text file, there are alot of zeros near the
beginning of the archive entry. There are alot of unused
ascii values.
Immediately following the last BIT of the encoding
table is the actual file represented by its Huffman codes.
When unsqueezing a file, ARC reads in the lengths and
codes and inserts them into tables so that they remain
sorted on the code length. If our tree was:
e m p a
/ / / /
root - - - - r
The tables would look like this after reading in the
encoding table.
ascii length code
----- ------ ----
e 1 00000000 00000000 00000000
m 2 00000000 00000000 00000010
p 3 00000000 00000000 00000110
a 4 00000000 00000000 00001110
r 4 00000000 00000000 00001111
The word "ampere" would be stored on disk as:
1110 10 110 0 1111 0