Here. We need this.
i dont get it.
everyone was hating on the Windows command line :).
what should we talk about in here?
what should we talk about in here?
ksh ftw!
> ksh ftw!
!!
!!
ksh rules! zsh is not bad either, mostly for the nice tab-completion stuff.
side note: i've been bored and have this Thinkpad X41T that I always have OpenBSD on but never use full time (use a MBP normally). Been doing an experiment to see if I could actually use it as a dev environment and desktop full time and recording all the things I did to get it there... should be up soon. figured you bsd nerds might be interested.
ksh is a great openbsd native shell.
what do you mean lucas? all shells are native to every os they run on.
oh an 'set -o vi' is the best thing ever invented.
works in bash, ksh, zsh and more im sure
oh an 'set -o vi' is the best thing ever invented.
works in bash, ksh, zsh and more im sure
the ksh on openbsd is different from the ksh included with other oses.
thats absolutely true: http://forums.bsdnexus.com/viewtopic.php?id=1265
vimperator for firefox... AWESOME
I remember wishing that existed like two years ago. I am glad it does now.
side note. flash is workin in openbsd these days nicely. it used to suck ass but now its good. i only use flash mainly for finding new music and occasional youtube, but its really nice to have the option.
ZSH :)
question, maple: how do you have it working? what browser and what flash port/package? flash is one thing that kind of kills the idea of a bsd desktop for me because I remember how much of a pain it was a couple of years ago.
asemi,
thats been the case for me too. i dont really like flash but it i use it too much (youtube, alot of music sites, etc).
anyway there are two ways to do it. opera+flash or firefox+flash.
opera+flash both run under linux emulation. for this just run:
# cd /usr/ports/www/opera-flashplugin
# make && make install
that will build opera, flash and setup the fedora-base.
but what i've been running is firefox3+flash7. with this setup firefox is a native openbsd build but the flash7 is running under linux emul.
im running a snapshot of openbsd 4.5 (4.4 has firefox 2.x) on this machine.
sorry for the long link -- its cached as the site seems to be down:
http://74.125.77.132/search?q=cache :tAT7yu7vA94J:openbsd.stanleylieber.com/txt/openbsd_firefox_flash_player_7.txt+openbsd+firefox+flash+player+7&hl=en&ct=clnk&cd=8&gl=us&client=firefox-a
thats been the case for me too. i dont really like flash but it i use it too much (youtube, alot of music sites, etc).
anyway there are two ways to do it. opera+flash or firefox+flash.
opera+flash both run under linux emulation. for this just run:
# cd /usr/ports/www/opera-flashplugin
# make && make install
that will build opera, flash and setup the fedora-base.
but what i've been running is firefox3+flash7. with this setup firefox is a native openbsd build but the flash7 is running under linux emul.
im running a snapshot of openbsd 4.5 (4.4 has firefox 2.x) on this machine.
sorry for the long link -- its cached as the site seems to be down:
http://74.125.77.132/search?q=cache :tAT7yu7vA94J:openbsd.stanleylieber.com/txt/openbsd_firefox_flash_player_7.txt+openbsd+firefox+flash+player+7&hl=en&ct=clnk&cd=8&gl=us&client=firefox-a
well F that link too -- here http://www.dugoo.net/wordpress/?p=217
I'm trying this now!
yep, totally didn't work. I gave up when the nspluginwrapper package wouldn't build.
Here is a neat trick I learned today. On compromised servers you will usually see a file that has a name that you can't really edit with the shell(\ ? & etc.). Here is something easy I picked up:
# To rename shitty named files, grab the inode number with the first command. Then make sure to use the 2nd command to rename it. The {} is the file name found by find. The \ prevents expansion and the ; terminates the statement:
ls -hali
find ./ -inum INODE -exec COMMAND {} \;
# To rename shitty named files, grab the inode number with the first command. Then make sure to use the 2nd command to rename it. The {} is the file name found by find. The \ prevents expansion and the ; terminates the statement:
ls -hali
find ./ -inum INODE -exec COMMAND {} \;
cool!
That'd also be good for network shares to other filesystems...
posted this link in another thread, but incase someone reads this in the future
openbsd+flash = http://fakeempire.com/posts/firefox-flash-vim … r-openbsd/
openbsd+flash = http://fakeempire.com/posts/firefox-flash-vim … r-openbsd/
gnash is working pretty great for me lately. most website that use it for navigation and stuff work perfectly. youtube is hit or miss but works sometimes.
Gatz, that's cool... Sometimes what I do when I don't want to figure out all the escaping is to use some programming language to rename the file without going through the shell special character stuff. For exameple, File.mv('&?\', 'stupid_file'), in Ruby.
^ Andre, I just saw this. Wow. That is pretty nifty. I do like to use the built in tools for the OS. Those people thought of everything!
Here is one trick I feel I should share. I was on a customer's server and I noticed that there were some viruses on it. So of course, I use clamscan to get the detected files out from the log:
Oh no! There seems to be a colon at the end! I can't pipe this to xargs or write a loop, since it will complain the file is not there. What to do? Enter ... rev.
A simple cut will not work, since it works on the length of the string. So here is what you do:
The command is doing the following:
Filtering for all found output | filtering the filtered output for mail(there were some false positives) | only return the first column(I don't need found) | reverses the line | cuts the first 2 characters of the line(there is a \n so 2 are required and the - removes) | reverse the reversed output
And here are the results:
Pretty nice huh? I know it was only for five files here. But typically you have ~100-~1000 of these lines. This little tidbit will save you an enormous amount of time.
Here is one trick I feel I should share. I was on a customer's server and I noticed that there were some viruses on it. So of course, I use clamscan to get the detected files out from the log:
root@server [~]# grep FOUND clamscan.log | grep mail | awk {'print $1'}
/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257842183.H61105P30680.server.virtualassistant.org,S=42582:2,S:
/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257754383.H617595P8760.server.virtualassistant.org,S=31079:2,S:
/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257832532.H374077P7133.server.virtualassistant.org,S=42571:2,S:
/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257794501.H93549P13514.server.virtualassistant.org,S=42599:2,S:
/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257812246.H821056P24563.server.virtualassistant.org,S=42577:2,S:
Oh no! There seems to be a colon at the end! I can't pipe this to xargs or write a loop, since it will complain the file is not there. What to do? Enter ... rev.
A simple cut will not work, since it works on the length of the string. So here is what you do:
root@server [~]# grep FOUND clamscan.log | grep mail | awk {'print $1'} | rev | cut -b 2- | rev
/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257842183.H61105P30680.server.virtualassistant.org,S=42582:2,S
/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257754383.H617595P8760.server.virtualassistant.org,S=31079:2,S
/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257832532.H374077P7133.server.virtualassistant.org,S=42571:2,S
/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257794501.H93549P13514.server.virtualassistant.org,S=42599:2,S
/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257812246.H821056P24563.server.virtualassistant.org,S=42577:2,S
The command is doing the following:
Filtering for all found output | filtering the filtered output for mail(there were some false positives) | only return the first column(I don't need found) | reverses the line | cuts the first 2 characters of the line(there is a \n so 2 are required and the - removes) | reverse the reversed output
And here are the results:
root@server [~]# grep FOUND clamscan.log | grep mail | awk {'print $1'} | rev | cut -b 2- | rev | xargs rm -fv
removed `/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257842183.H61105P30680.server.virtualassistant.org,S=42582:2,S'
removed `/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257754383.H617595P8760.server.virtualassistant.org,S=31079:2,S'
removed `/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257832532.H374077P7133.server.virtualassistant.org,S=42571:2,S'
removed `/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257794501.H93549P13514.server.virtualassistant.org,S=42599:2,S'
removed `/home/sw1mm1ng/mail/swimmingpooldehumidifiers.org/craigd/cur/1257812246.H821056P24563.server.virtualassistant.org,S=42577:2,S'
Pretty nice huh? I know it was only for five files here. But typically you have ~100-~1000 of these lines. This little tidbit will save you an enormous amount of time.
Gatz,
$ awk -F ':' '/mail.*FOUND$/ { print $1FS$2 }' clamscan.log | xargs rm -fv
Also, a simple cut does work:
$ grep "mail.*FOUND" clamscan.log | awk {'print $1'} | cut -d: -f-2
Yea but you guys weren't online when I needed this.
andre, arun, your options work but it's hardly as cool as tossing a couple reverses on the line :D
anyway thanks for this thread!
anyway thanks for this thread!
I like the way gatz's brain worked here. I would've never thought about it that way.
Resurrecting this because my find command was not working. There was a suspicious file in /. When I tried searching, it took an absurd amount of time. I have resolved it with the following:
# find / -xdev -inum INODE_NUM -ok mv '{}' suspicious.txt \;
-ok will prompt you. If you want to skip the prompt, use -exec.
# find / -xdev -inum INODE_NUM -ok mv '{}' suspicious.txt \;
-ok will prompt you. If you want to skip the prompt, use -exec.
Fuck this shit is killing me. I have a huge FTP site and I need to download certain files with a bash script. The FTP program supports globbing. However, I can't get the glob working. Here is the pattern I am using:
firstRegex = XNYM_CL_CL
mls $firstRegex* -
output
-------------------------------------
XNYM_CL_CL_FUT_20120426.zip
XNYM_CL_CL_FUT_20120427.zip
XNYM_CL_CL_FUT_20120428.zip
XNYM_CL_CL_FUT_20120429.zip
XNYM_CL_CL_FUT_20120430.zip
mls $firstRegex*_$year* -
output
------------------------------------
XNYM_CL_CL_FUT_20120506.zip
XNYM_CL_CL_FUT_20120507.zip
XNYM_CL_CL_FUT_20120508.zip
XNYM_CL_CL_FUT_20120509.zip
XNYM_CL_CL_FUT_20120510.zip
Why does it not pick up the 04 files?
firstRegex = XNYM_CL_CL
mls $firstRegex* -
output
-------------------------------------
XNYM_CL_CL_FUT_20120426.zip
XNYM_CL_CL_FUT_20120427.zip
XNYM_CL_CL_FUT_20120428.zip
XNYM_CL_CL_FUT_20120429.zip
XNYM_CL_CL_FUT_20120430.zip
mls $firstRegex*_$year* -
output
------------------------------------
XNYM_CL_CL_FUT_20120506.zip
XNYM_CL_CL_FUT_20120507.zip
XNYM_CL_CL_FUT_20120508.zip
XNYM_CL_CL_FUT_20120509.zip
XNYM_CL_CL_FUT_20120510.zip
Why does it not pick up the 04 files?
WTF shit just starts working?!?!
hahahaha
This was NOT fun.
I literally lost 4 hours today because I did not know the paste command. WTF ... :(
You can put FTP commands in a batch file and redirect them to FTP. Pretty sweet:
# ftp HOST < batch_file.txt
# ftp HOST < batch_file.txt
Never really was much of a shell junkie. Turns out {#..#} works great when you need to filter based on range:
Edit: Years of TTF, always search for the guide.
$ du -chs {201204{23..30},201205*,201206*}
2.5G 20120423
2.4G 20120424
2.9G 20120425
2.5G 20120426
2.4G 20120427
311M 20120428
17M 20120429
2.1G 20120430
2.0G 20120501
2.5G 20120502
2.5G 20120503
2.8G 20120504
22M 20120505
17M 20120506
2.3G 20120507
2.8G 20120508
3.0G 20120509
2.8G 20120510
2.5G 20120511
105M 20120512
17M 20120513
2.5G 20120514
2.9G 20120515
3.5G 20120516
3.4G 20120517
3.4G 20120518
255M 20120519
17M 20120520
3.0G 20120521
3.2G 20120522
3.9G 20120523
3.6G 20120524
2.9G 20120525
16M 20120527
553M 20120528
3.2G 20120529
3.5G 20120530
3.6G 20120531
4.1G 20120601
191M 20120602
17M 20120603
3.4G 20120604
3.1G 20120605
3.3G 20120606
3.8G 20120607
3.3G 20120608
22M 20120609
17M 20120610
183M 20120611
102G total
Edit: Years of TTF, always search for the guide.
you couldn't remember pre tags?
oh, html brackets :).
Very very helpful:
|| in a shell script means that if the first command's return code is not 0, execute the second command.
Example:
$ command -v fo0 >/dev/null 2>&1 || echo >&2 "Command not found"
Command not found
An existing command is blank. :)
|| in a shell script means that if the first command's return code is not 0, execute the second command.
Example:
$ command -v fo0 >/dev/null 2>&1 || echo >&2 "Command not found"
Command not found
An existing command is blank. :)
|| is "or", like most programming languages.
I strongly prefer to check exit codes with $?, it may be a few more lines, but IMHO that's always better than squeezing multiple statements on a single line. It also allows you to do better extended error handling or add it later.
I also prefer to just show the error message, now you redirect everything from "command" to /dev/null and just *assume* that *any* error you encounter mean that "command" doesn't exist. That's just silly. I've had to hack many, MANY, *MANY* shell script which make these bad assumptions just to get a meaningful error message...
The *proper* procedure is to check if the command exists with which(1), execute the command and capture the output in a variable, check if $? is non-0 (usually, may be different for some tools, like grep), display error output from variable if it isn't.
I strongly prefer to check exit codes with $?, it may be a few more lines, but IMHO that's always better than squeezing multiple statements on a single line. It also allows you to do better extended error handling or add it later.
I also prefer to just show the error message, now you redirect everything from "command" to /dev/null and just *assume* that *any* error you encounter mean that "command" doesn't exist. That's just silly. I've had to hack many, MANY, *MANY* shell script which make these bad assumptions just to get a meaningful error message...
The *proper* procedure is to check if the command exists with which(1), execute the command and capture the output in a variable, check if $? is non-0 (usually, may be different for some tools, like grep), display error output from variable if it isn't.
Please don't say which is the proper way. That really isn't posix compliant.
I agree with the $? operator. That is a pretty neat trick I also came across today.
I agree with the $? operator. That is a pretty neat trick I also came across today.
I spent a looooot of time at the command line today. here's the script I'm about to set up as a cron job on my pf firewall to dynamically update my IP address: http://code.google.com/p/gandi-automatic-dns/
in short, it constructs a few XML RPCs given a method name and datatype/value pairs, pipes the XML (preceded by a POST) to openssl s_client, and talks to my domain registrar's API. I initially tried to do all of this in as few RPCs as possible because I just wanted to git r done but I finally gave up and wrote a function for them which cut the size of the script by more than half.
XML sucks.
in short, it constructs a few XML RPCs given a method name and datatype/value pairs, pipes the XML (preceded by a POST) to openssl s_client, and talks to my domain registrar's API. I initially tried to do all of this in as few RPCs as possible because I just wanted to git r done but I finally gave up and wrote a function for them which cut the size of the script by more than half.
XML sucks.
oh and handling arguments with while loops and shift was fun. hadn't done that before.
Yeah, I'm having issues myself with some shell script for dwm's status bar on FreeBSD.
It was easy enough on linux to just use /sys/blah/blah/BAT0 for battery information, but FreeBSD don't like that one bit, so currently I'm using a very, very ugly bit of code to parse acpiconf. I KNOW there's a simpler way to do it. Anyone have a suggestion? I feel like a noob right now. :(
acpiconf -i 0 dumps information about the battery, I need the 14th line (percent remaining, hence the greps), but only the percentage part (hence the awk).
It was easy enough on linux to just use /sys/blah/blah/BAT0 for battery information, but FreeBSD don't like that one bit, so currently I'm using a very, very ugly bit of code to parse acpiconf. I KNOW there's a simpler way to do it. Anyone have a suggestion? I feel like a noob right now. :(
acpiconf -i 0 | grep -n : | grep '14:' | awk '{print $3}'acpiconf -i 0 dumps information about the battery, I need the 14th line (percent remaining, hence the greps), but only the percentage part (hence the awk).
post the output of acpiconf -i 0. there's no command that will return more specific information? I remember being able to use the apm command on openbsd.
but I guess if you're using acpi...
The apm command won't work on FreeBSD 64-bit.
AFAIK acpiconf is the way to go, that's also how I do it (1)
You could use cut(1) instead of awk(1) though ;)
1: https://bitbucket.org/Carpetsmoker/battray/sr … at=default
PS: How do I add links on ttf lucas? You should add a cheatsheet at the bottom ...
AFAIK acpiconf is the way to go, that's also how I do it (1)
You could use cut(1) instead of awk(1) though ;)
1: https://bitbucket.org/Carpetsmoker/battray/sr … at=default
PS: How do I add links on ttf lucas? You should add a cheatsheet at the bottom ...
you search for 'guide' and then you click on a result and then you realize that it's a thread someone created by accident when searching for a guide and then you go back and click on the real guide thread.
Here's the output:
CS: thanks!
$ acpiconf -i 0 Design capacity: 6600 mAh Last full capacity: 6511 mAh Technology: secondary (rechargeable) Design voltage: 10800 mV Capacity (warn): 512 mAh Capacity (low): 256 mAh Low/warn granularity: 66 mAh Warn/full granularity: 66 mAh Model number: 1005HA Serial number: Type: LION OEM info: ASUS State: high Remaining capacity: 100% Remaining time: unknown Present rate: 0 mA (0 mW) Present voltage: 12577 mV
CS: thanks!
will someone please tell me why this:
echos only the % sign?
sed -n 's/R.*c.*\([0-9]*%\)/\1/p'
echos only the % sign?
Your last .* is too greedy. It consumes all characters between c and %.
Adding a space will make it work:
But I think this is clearer:
However, none of those will actually work, since acpiconf uses tabs ( http://svnweb.freebsd.org/base/release/9.1.0/ … iew=markup ):
This will work:
Adding a space will make it work:
sed -n 's/R.*c.* \([0-9]*%\)/\1/p'
But I think this is clearer:
sed -n 's/Remaining capacity: \+\([0-9]\+%\)/\1/p'
However, none of those will actually work, since acpiconf uses tabs ( http://svnweb.freebsd.org/base/release/9.1.0/ … iew=markup ):
if (battio.battinfo.cap == -1)
printf("Remaining capacity:\tunknown\n");
else
printf("Remaining capacity:\t%d%%\n",
battio.battinfo.cap);This will work:
sed -n 's/Remaining capacity:\t\([0-9]\+%\)/\1/p'
ah of course. damn greedy asterisk. sed is fun. the re_format man page is also a good read.
$ acpiconf -i 0 | awk '/Remaining capacity/ { print $NF }'
... this makes much more sense.
but, sed is more fun!
It's time for bed ... To Bed! I sed!
World-class joke! have you considered a career in comedy?
The only comedy for me is this joke of a life.
life is pretty hilarious.
It's not unlike a slapstick humor where the main characters keep falling on their face no matter what they do.
trying to decipher those boxes
whoa, i put them in this post, but they disappeared. :[
whoa, i put them in this post, but they disappeared. :[
that is excellent. I need to put return codes in my prompt!
pretty cool. i don't know about your user title, though. :p
tcsh has the `printexitvalue' settings, which does the same thing. Albeit in a much more boring way.