Tom's Guide | Tom's Hardware | Tom's Games
 |
 |
 |
Try to find the simplest sed & awk script to
extract from input file to produce the outputfile?Input file:
A1
A2
A3 Pattern1
A4
A5
...
B1
B2
B3 Pattern1
B4
B5
...Output file:
A2
A3 Pattern1
A4
Blank line
B2
B3 Pattern1
B4
Blank line
...
I should let raymond answer, but my guess would be, for each line that contains Pattern1, output it along with the preceding line, the following line, and a blank line for separation.
If so, what to do with two adjacent lines containing Pattern1.
0 
James interpretation is correct.
There is no repeated "Pattern 1" in the same block.
What I try is to:
pkginfo -l > a_file
Then run sed & awk to extract some info about packages that installed on a particular date.
Any suggestion?
0
On my HP-UX, I don't have pkginfo. If you post a little of the pkginfo output and a sample of the output you desire, that would be a big help.
0
James:
A partial input file:
40 executables
77 blocks used (approx)PKGINST: SUNWeuezt
NAME: English UTF-8 L10N For Desktop Power Pack Applications
CATEGORY: system
ARCH: sparc
VERSION: 1.5,REV=1999.12.03.14.40
BASEDIR: /usr
VENDOR: Sun Microsystems, Inc.
DESC: American English/UTF-8 L10N For Desktop Power Pack Applications
INSTDATE: Feb 04 2000 09:23
HOTLINE: Please contact your local service provider
STATUS: completely installed
FILES: 28 installed pathnames
16 shared pathnames
13 directories
7 executables
197 blocks used (approx)PKGINST: SUNWeugrf
NAME: X11 sun_eu_greek fonts
CATEGORY: system
ARCH: sparc
VERSION: 3.6,REV=1999.10.13.14.44
BASEDIR: /usr
VENDOR: Sun Microsystems, Inc.
DESC: X11 fonts for sun_eu_greek character set
INSTDATE: Feb 04 2000 09:23
HOTLINE: Please contact your local service provider
STATUS: completely installed
FILES: 32 installed pathnames
10 shared pathnames
12 directories
88 blocks used (approx)Desired output: (after I filtered based on
"Feb 04 2000 " )PKGINST: SUNWeuezt
NAME: English UTF-8 L10N For Desktop Power Pack Applications
CATEGORY: system
VERSION: 1.5,REV=1999.12.03.14.40
BASEDIR: /usr
DESC: American English/UTF-8 L10N For Desktop Power Pack Applications
INSTDATE: Feb 04 2000 09:23
PKGINST: SUNWeugrf
NAME: X11 sun_eu_greek fonts
CATEGORY: system
VERSION: 3.6,REV=1999.10.13.14.44
BASEDIR: /usr
DESC: X11 fonts for sun_eu_greek character set
INSTDATE: Feb 04 2000 09:23...
0
In both of the scripts below, to prevent from losing my indentation, I use leading underscores instead of leading spaces, so you would need to change those back to spaces.
This first script does not do date filtering. It prints the lines of interest for all packages:
#!/bin/sh
awk '\
/^PKGI/ {\
__print ""
__print}
/^NAME/||/^CAT/||/^VER/||/^BAS/||/^DES/||/^INST/ {\
__print}' a_fileexit 0
This second script expects 3 parameters such as:
pkgsearch Feb 04 2000
Alternately, it could be coded to receive the date as a single parameter in double-quotes, such as:
pkgsearch "Feb 04 2000"
or to avoid the quotes, it could be coded to receive a single parameter such as:
pkgsearch Feb-04-2000
and it could change the dashes to spaces.
#!/bin/sh
if [ $# -ne 3 ] ; then
___echo '\nRun script with desired date of package installation.'
___echo '\nExample: pkgsearch Feb 04 2000\n'
___exit 1
fireqdate=$*
awk -v reqdate="$reqdate" '{\
if (substr($0,1,8) == "PKGINST:")
__{h1=$0
___h2=""
___h3=""
___h4=""
___h5=""
___h6=""}
if (substr($0,1,5) == "NAME:") h2=$0
if (substr($0,1,9) == "CATEGORY:") h3=$0
if (substr($0,1,8) == "VERSION:") h4=$0
if (substr($0,1,8) == "BASEDIR:") h5=$0
if (substr($0,1,5) == "DESC:") h6=$0
if (substr($0,1,9) == "INSTDATE:")
___if ($2 " " $3 " " $4 == reqdate)
_____{print ""
______print h1
______print h2
______print h3
______print h4
______print h5
______print h6
______print}
}' a_fileexit 0
The script stores the lines of interest, and prints all those out if it gets a match on INSTDATE. When it hits a PKGINST line, it nullifies all the hold lines. If it did not do this, and some of the package displays do not have all of those lines, it could be displaying hold lines left over from a previous package. If each package display always includes all of those lines of interest, then nullifying the hold lines is not required.
There are several ways to pass a shell parameter to awk. I think the -v option is pretty common. If that does not work for you, we can use one of the other approaches to feed reqdate to awk.
0
James,
I have updated to make it work. See below.
However, I am getting a bit greedy here.
I would like to run in a single line, something like
#pkginfo -l | ./pkgsearch "Feb 14 2000"
Can you help as well?
Thanks.
#!/bin/shif [ $# -ne 3 ] ; then
echo '\nAssume pkginfo_list already exist
echo '\nExample: pkgsearch Feb 14 2000\n'
exit 1
fi
reqdate=$*/usr/xpg4/bin/awk -v reqdate="$reqdate" '{\
if (substr($0,4,8) == "PKGINST:")
{h1=$0
h2=""
h3=""
h4=""
h5=""
h6=""}
if (substr($0,7,5) == "NAME:") h2=$0
if (substr($0,3,9) == "CATEGORY:") h3=$0
if (substr($0,4,8) == "VERSION:") h4=$0
if (substr($0,4,8) == "BASEDIR:") h5=$0
if (substr($0,7,5) == "DESC:") h6=$0
if (substr($0,3,9) == "INSTDATE:")
{
if ($2 " " $3 " " $4 == reqdate)
{print ""
print h1
print h2
print h3
print h4
print h5
print h6
print}
}
}' pkginfo_listexit 0
0
Yep, that's easy, but out of time today. If no one answers before tomorrow morning, I will at that time.
0
Instead of specifying the filename of pkginfo_list, you would remove that filename and instead pipe that data into awk. So, instead of:
awk -v reqdate="$reqdate" '{\
etc etc etc
}' pkginfo_listit would be:
pkginfo -l | awk -v reqdate="$reqdate" '{\
etc etc etc
}'and I used the simple awk reference just to keep the lines shorter. Of course you would need the xpg4 awk.
0  |
 |
 |
This post is quite old and has been locked from receiving new replies. Please create a new posting instead.
| Ads by Google |
