UNIX for Dummies Questions & Answers

Dear Radoulov,
When I m doign generalization for other special character also, it giving me trouble.
When I add it for lines having < character

Code:

/^<[^\/]/{N;N;s/\n&amp;\n\([^<]*\)/\&amp;\1/;}
/^<[^\/]/{N;N;s/\n&lt;\n\([^<]*\)/\&lt;\1/;}

Its not modifying the lines having &.

but when i add only

Code:

/^<[^\/]/{N;N;s/\n&amp;\n\([^<]*\)/\&amp;\1/;}

it works. So what changes do i need it to make it work for all.

PS: I have added [^\/], because the sed script was removing the last two lines of xml file.

Could you post a sample of your input with the patterns (including the &amp/&lt) you want to substitute?

This is some sample data of my file.
<EQROW CHG="U" KEY="22" ><EFFECT EFFRG="00122323"></EFFECT>
<EIN> 14XE </EIN>
<PARTDESC>ANN EMER ELECPWR/RAT/
&amp;
/EMER GEN</PARTDESC>
<LOCATION><ZONE>211</ZONE>
</LOCATION>
<HDIAGNBR><REFEXT REFMAN="SSM" REFLOC="24-24/02" REFMODEL="A319/A320" REFSPL="2D671">SSM 24-24/02</REFEXT>
</HDIAGNBR>
</EQROW>
<EXTWROW CHG="U" KEY="23" ><EFFECT EFFRG="02399 "></EFFECT>
<ACTWIRE><WIRE><WIRECODE><WIRENBR>2121-5047</WIRENBR>
<WIREAWG>24</WIREAWG>
</WIRECODE>
<WIRERTE>1M</WIRERTE>
</WIRE>
<TERMNBR>
&amp;
C</TERMNBR>
<TERMPNR>EN3155-019F2018</TERMPNR>
</ACTWIRE>
</EXTWROW>
<EXTWROW CHG="U" KEY="2317743">
<ACTWIRE><WIRE><WIRECODE><WIRENBR>2121-5047</WIRENBR>
<WIREAWG>24</WIREAWG>
</WIRECODE>
<WIRERTE>1M</WIRERTE>
</WIRE>
<TERMNBR>
&lt;
C</TERMNBR>
<TERMPNR>EN3155-019F2018</TERMPNR>
</ACTWIRE>
</EXTWROW>

Like this in whole file those character are spread. I want to remove the etxra line associated with them, and also retaing those value(&amp;&lt; ect.).

Hm,
with awk it would be:

Code:

awk '/^&(amp|lt)/ { re = $0
	getline
	printf "%s%s%s\n", rec, re, $0
	next
	}
	{ rec = $0 }
1' filename

(use nawk or /usr/xpg4/bin/awk on Solaris)

For sed I have to think a little bit

If you get how it can be done using SED, then it will be nice to me. I m running the SED in UNIX environment. no idea about solaris.

your system not only comes with sed, it comes with awk too and many other essential tools. you should use a tool that does the job well.

I agree with ghostdog74,
it seems unlikely that you have not awk on your system.

Anyway, you can try this not elegant sed solution, _based on your sample input_:

Code:

sed  '/<\/[^/>]*>$/!{N;N;s/\n\(&[^;]*;\)\n/\1/;}' filename