Shell Programming and Scripting

I have got a large file with content as below.

LOCATION A B C 1

Line 3
Line 4



*Line 8
**Line 9


TAG END


LOCATION A B" C 3

Line 33
Line 34



Line 38

*Line 40
**Line 41


TAG END



LOCATION A' B C 4


Line 51
Line 52

*Line 54
**Line 55


TAG END



LOCATION A B C 5


Line 51
Line 52

*Line 54
**Line 55


TAG END

Now I need to format this file to the below format

LOC A B C 1

Line 3
Line 4



*Line 8
**Line 9


TAG END


LOC A B`` C 3

Line 33
Line 34



Line 38

*Line 40
**Line 41


TAG END



LOC A` B C 4


Line 51
Line 52

*Line 54
**Line 55


TAG END



LOC A B C 5


Line 51
Line 52

*Line 54
**Line 55


TAG END

Putting into words, my large file has got several tags which start with a Location name (eg, LOC A B ) and ends with TAG END ( "TAG END" word is fixed). Now the location might contain some single quote (') or double quote (") in it's value. All I have to do is to search for all the loc names for single and double quote and replace them with back quotes ( one back quote for single quote and 2 back quotes for a double quote).
Note 1: One tag ( from location name till TAG END) might contain any number of lines.
Note 2: There could be one, two, three or more number of blank lines between TAG END and next location name.
Note 3: I should not alter any single or double quote present inside the tag, but only those present in the location name.

One idea is to look for the string "TAG END" and check for the next non blank line. This must be a location name. Only execption is the first non blank line which itself is the store name. To make it uniform we can add the line "TAG END" as first line and finally remove this in output file.

Any other suggestion is welcome. Please help me in achieving this.

Thanks in advance..
Rinku

Code:

sed "/LOC/{s:':\`:g;s:\":\`\`:g;}" infile

should work, but if you get trouble with the single quote, then, escape it :

Code:

sed "/LOC/{s:\':\`:g;s:\":\`\`:g;}" infile

or

Code:

sed "/LOC/s:\':\`:g;/LOC/s:\":\`\`:g" infile

Hi

Thanks for resposonse but it does not solve the problem. As I said, Location name is not fixed and LOC A B is just an example of location name. It does not mean that it starts with LOC. Sorry for the confusion.

Thanks,
Rinku

copy/paste a representative sample of the content of your file then.
We cannot help to parse a file if we ignore its structure.

Code:

 awk -vRS="TAG END" -vFS="\n" '/LOCATION/{gsub(/\047/,"`",$1);gsub(/\042/,"``",$1);print}' ORS="TAG END" file