Искусство программирования на языке сценариев командной оболочки
Шрифт:
c) ((Lidx+=12)) ;; # Character device
*) ((Lidx+=11)) ;; # Anything else
esac
name=Lidx
case $ft in
– ) ((Lidx+=1)) ;; # The easy one
b) ((Lidx+=1)) ;; # Block device
c) ((Lidx+=1)) ;; # Character device
d) ((Lidx+=1)) ;; # The other easy one
l) ((Lidx+=3)) ;; # At LEAST two more fields
# A little more elegance here would handle pipes,
#+ sockets, deleted files - later.
*) until IsNumber ${LIST[$Lidx]} || ((Lidx >= Lcnt))
do
((Lidx+=1))
done
;; # Not required
esac
INDEX[${#INDEX[*]}]=$inode
INDEX[${#INDEX[*]}]=$name
INDEX[0]=${INDEX[0]}+1 # One more "line" found
# echo "Line: ${INDEX[0]} Type: $ft Links: $m Inode: \
# ${LIST[$inode]} Name: ${LIST[$name]}"
else
((Lidx+=1))
fi
done
case "$of" in
0) eval $2=\( \"\$\{INDEX\[@\]\}\" \) ;;
1) echo "${INDEX[@]}" > "$2" ;;
esac
return 0 # What could go wrong?
}
# # # # # Content Identify File # # # # #
#
# DigestFile Input-Array-Name Digest-Array-Name
# or
# DigestFile -if Input-FileName Digest-Array-Name
# # # # #
# Here document used as a comment block.
: <<DigestFilesDoc
The key (no pun intended) to a Unified Content File System (UCFS)
is to distinguish the files in the system based on their content.
Distinguishing files by their name is just, so, 20th Century.
The content is distinguished by computing a checksum of that content.
This version uses the md5sum program to generate a 128 bit checksum
representative of the file's contents.
There is a chance that two files having different content might
generate the same checksum using md5sum (or any checksum). Should
that become a problem, then the use of md5sum can be replace by a
cyrptographic signature. But until then...
The md5sum program is documented as outputting three fields (and it
does), but when read it appears as two fields (array elements). This
is caused by the lack of whitespace between the second and third field.
So this function gropes the md5sum output and returns:
[0] 32 character checksum in hexidecimal (UCFS filename)
[1] Single character: ' ' text file, '*' binary file
[2] Filesystem (20th Century Style) name
Note: That name may be the character '-' indicating STDIN read.
DigestFilesDoc
DigestFile
{
local if=0 # Default, variable name
local -a T1 T2
case "$#" in
3) case "$1" in
– if) if=1 ; shift ;;
* ) return 1 ;;
esac ;;
2) : ;; # Poor man's "continue"
*) return 1 ;;
esac
case $if in
0) eval T1=\( \"\$\{$1\[@\]\}\" \)
T2=( $(echo ${T1[@]} | md5sum -) )
;;
1) T2=( $(md5sum $1) )
;;
esac
case ${#T2[@]} in
0) return 1 ;;
1) return 1 ;;
2) case ${T2[1]:0:1} in # SanScrit-2.0.5
\*) T2[${#T2[@]}]=${T2[1]:1}
T2[1]=\*
;;
*) T2[${#T2[@]}]=${T2[1]}
T2[1]=" "
;;
esac
;;
3) : ;; # Assume it worked
*) return 1 ;;
esac
local -i len=${#T2[0]}
if [ $len -ne 32 ] ; then return 1 ; fi
eval $2=\( \"\$\{T2\[@\]\}\" \)
}
# # # # # Locate File # # # # #
#
# LocateFile [-l] FileName Location-Array-Name
# or
# LocateFile [-l] -of FileName Location-Array-FileName
# # # # #
# A file location is Filesystem-id and inode-number
# Here document used as a comment block.
: <<StatFieldsDoc
Based on stat, version 2.2
stat -t and stat -lt fields
[0] name
[1] Total size
File - number of bytes
Symbolic link - string length of pathname
[2] Number of (512 byte) blocks allocated
[3] File type and Access rights (hex)
[4] User ID of owner
[5] Group ID of owner
[6] Device number
[7] Inode number
[8] Number of hard links
[9] Device type (if inode device) Major
[10] Device type (if inode device) Minor
[11] Time of last access
May be disabled in 'mount' with noatime
atime of files changed by exec, read, pipe, utime, mknod (mmap?)
atime of directories changed by addition/deletion of files
[12] Time of last modification
mtime of files changed by write, truncate, utime, mknod
mtime of directories changed by addtition/deletion of files
[13] Time of last change
ctime reflects time of changed inode information (owner, group
permissions, link count
– *-*- Per:
Return code: 0
Size of array: 14
Contents of array
Element 0: /home/mszick
Element 1: 4096