Stefan Sobek Blog » Linux

OCR under linux

admin — Mon, 08 Feb 2010 09:04:40 +0000

Imagine you have a pdf-file you want to make ocr-recognition. Take a scenario where you want to automatically let your linux pc do the job, e.g. in a folder.

I choose tesseract-ocr as ocr-programm. Easy to install and use and ok for my use. Unfortunately it takes only tif as input file type so that we have to convert the pdf to tif first.

To create a tif file with Ghostscript from pdf:

^?View Code BASH

1	gs -q -r300 -dBATCH -dNOPAUSE -sDEVICE=tiff24nc -sOutputFile=Dokument2.tif Dokument1.pdf

Now start OCR-recognition with tesseract-ocr (maybe you have to install it).

^?View Code BASH

1	tesseract Dokument2.tif doc.txt -l deu

-l deu means “Deutsch” for German language recognition.
-r300 means 300 DPI

Now it is easy to create a script which will automatically check a folder for new files and start ocr etc.

E.g. you can create a shell-script and start this shell script via cronjob. Maybe I will write an example here later.

Create a pdf from manpages

admin — Sat, 16 Jan 2010 16:01:30 +0000

If you want to printout or save info from manpages simply type:

^?View Code BASH

1	man -t man \| ps2pdf - > man.pdf

Download files with curl

admin — Sun, 20 Dec 2009 13:31:38 +0000

If you want to download a file with curl type the following code into bash:

^?View Code BASH

1	curl -O http://www.url.de/file.html

-O means to output to a file and not to stdout, normally the screen.

What if you need to resume a download of a big file?

^?View Code BASH

1	curl -C - -O http://www.url.de/file.html

With -C – you continue downloading the current file. If is important to write -C -. Don’t forget the -

You can also set a referrer and the user agent.

^?View Code BASH

1	curl -C - -O -A http://www.url.de -e "Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; de; rv:1.9.1.6) Gecko/20091201 Firefox/3.5.6" http://www.url.de/file.html

With -A you can set the referrer and -e sets the user agent.

Pretty helpful.

Linux kernel compilation syntax

admin — Mon, 14 Dec 2009 15:45:47 +0000

After unpacking the Kernel Archive to “/usr/src” use the following:

^?View Code BASH

make dep
make clean
make bzImagemake modules
make modules_install

Send an E-Mail with Telnet via SMTP

admin — Wed, 02 Dec 2009 10:52:27 +0000

How to send an email with telnet via smtp for testing purposes? Here is an example:

^?View Code BASH

telnet  smtp.me.com 25
220 smtp.web.de ESMTP Web.de V4.91#2 Sun, 24 Nov 2009  16:51:20 +0100
 
helo yourcomputername
250 smtp.web.de Hello yourcomputername [219.99.999.999]
 
mail  from: account@web.de
250 account@web.de is syntactically  correct
 
rcpt  to: empfaenger@provider.tld
250 empfaenger@provider.tld verified
 
data
354 Enter message,  ending with . on  a line by itself
To: anton@tirol.at;
From: hugo@schlaumeier.de
Subject: Jodelei 
 
Servus!
 
Greetz,
Peterli
.
250 OK  id=18Z6Uc-0001wm-00
 
Quit
221 smtp.web.de closing connection

To send mail with auth:

^?View Code BASH

telnet  smtp.me.com 25
220 smtp.web.de ESMTP Web.de V4.91#2 Sun, 24 Nov 2009  16:51:20 +0100
 
helo yourcomputername
250 smtp.web.de Hello yourcomputername [219.99.999.999]
 
auth login
334 VXNlcm5hbWU6
 
<base64 encoded username>
<repeated base64 encoded username>
<base64 encoded password>
<repeated base64 encoded password>
 
235 Authentication successful
 
mail  from: account@web.de
250 account@web.de is syntactically  correct
 
rcpt  to: empfaenger@provider.tld
250 empfaenger@provider.tld verified
 
data
354 Enter message,  ending with . on  a line by itself
 
Subject: Jodelei 
 
Servus!
 
Greetz,
Peterli
.
250 OK  id=18Z6Uc-0001wm-00
 
Quit
221 smtp.web.de closing connection

To base64 encode your username or pass, use another linux terminal and type:

^?View Code BASH

1 2	printf username \| base64 printf password \| base64

If you don’t have base64 on your system try mimencode instead.

scroll in screen

admin — Wed, 02 Dec 2009 09:02:24 +0000

To scroll in screen-mode of linux type the following:
Strg+a then AltGr+8

Now you can scroll up and down. This is called the copy-mode.

Source: http://blog.pregos.info/2009/02/26/scrollen-im-screen/

Tar with split syntax

admin — Mon, 30 Nov 2009 11:31:56 +0000

This is an example from http://www.linux-user.de/ausgabe/2001/07/030-split/split.html:

^?View Code BASH

1	tar cz /home/user \| split -b95m - archiv.tgz.split.

Put together files with:

^?View Code BASH

1	cat [dateiname].split.* \| tar x

Search and replace with linux in files

admin — Wed, 25 Nov 2009 15:39:23 +0000

To replace a string in all files inside a folder you can create a bash file:

^?View Code BASH

1	vi replace.sh

Put inside this file:

^?View Code BASH

#!/bin/bash
for file in $(ls -1);
do
sed 's/me/you/g' $file &gt; /tmp/dummy.txt;
mv /tmp/dummy.txt $file;
done

Save it and make it executeable (chmod 777 replace.sh)

In this case the string “me” will be replace with “you” in all files of the actual folder and the files will be overridden.

Another example

Your want to replace the tag with in all files inside a special folder. Further you want to backup the original files with suffix .old and the new files will be overridden.

Simply type on the console:

^?View Code BASH
1 2 3 4 5
for file in $(ls -1); do mv $file $file.old sed 's/<b>/<strong>/g' $file.old > $file; done

You can modify the script(s). Maybe by adding parameters etc.

Another option is to use sed with -i command or direct editing:

^?View Code BASH
1 2 3 4
for file in $(ls -1); do sed -i 's/text_to_replace/text_replacement/g' $file done

Screen command usage – part 1

admin — Wed, 25 Nov 2009 13:59:59 +0000

People often underestimate the usefulness of the linux-command screen.

What does screen do? Screen creates a terminal session for the actual user.

Type:

^?View Code BASH
1
screen

to create a terminal session.

Or type:

^?View Code BASH
1
screen -S <sessionname>

where should be replaced with your own name.

Create a new window:

^?View Code BASH
1
Crtl+a c

which means (Press Control and a together and after that press c)

Now we created a second window. To list all created windows in the actual session type:

^?View Code BASH
1
Crtl+a "

Germans have to press Shift+2 for “.

Now the two created windows will be shown:

Just choose one Window and press enter.

To quit your actual work an detach the session:

^?View Code BASH
1
Crtl+a d

To list all active sessions:

^?View Code BASH
1
screen -list

You should see a similar picture like the following:

Here you can see the name of the session: 1597.stefan

To restore the session simply type:

^?View Code BASH
1
screen -r stefan

Maybe you have to type the complete id with 1597.stefan if there exist another session with the same name.

Top command differences in Linux and Mac

admin — Thu, 19 Nov 2009 08:46:55 +0000

When switching from Mac to Linux and vice versa, I always have to think a bit about how to use the command top and sort by CPU.

With Linux it is

top

and then press

shift+p

Mac:

top -o cpu

I hope I can remember this