UNIX for Dummies Questions & Answers

Please be more specific on the input file structure. Are the dots to be removed always embedded in spaces (obviously not at EOL!)? Are there integers with no dots? Are there integers surrounded by dots? Could there be single dots between integers? Are the numbers to stay intact always in the same place (column/position)? And don't surprise us again with "expanded data" once a proposal has been posted!

Code:

awk '{for (i=1;i<=2;i++) gsub( /(^|[^0-9])\.([^0-9]|$)/ , "\1 \2")} 1'

I`m sorry for all the confusion, I should have given a more specific description of the data set.

It is a tab delimited flat file, the dots represent missing data. So there will be a tab before and after the dot. I have attached a portion of a sample input from the large file.

The numbers are supposed to stay in the same place (row and column) after the dots have been removed. The decimal numbers remain intact , 12.34 remains 12.34 but simply a . or a missing data has to be represented by a white space or anything else to keep the structure of the table intact.

Hi MadeInGermany,

I ran your code for the sample data set I attached, I`m getting an L like character in place of dot. Please pardon my ignorance, is there a way i could just have white space? Also I want to be able to keep the table structure intact, so that I can import it as a table into statistical software.

I have attached the output.

Thank you for helping out

FIRST: your file is a DOS file, CR LF (0x0D 0x0A) terminated. To correctly work upon it, remove the CR chars from it (copiously covered in these fora).
Then try:

Code:

awk '{for (i=1; i<=NF; i++) if ($i==".") $i=" "}1' OFS="\t" sample_out.txt

@MadeInGermany: what you propose is a sedism; it does not work in (my m-) awk, but inserts ^A and ^B into the lines.

Now I see that gsub() does not support \1 references.
Also I see one needs 3 cycles to replace a row of dots.
With perl this becomes:

Code:

perl -pe 'for ($i=1;$i<=3;$i++) { s/(^|\D)\.(\D|$)/$1 $2/g }'

There is certain a more elegant|efficient method with perl RE ...

works like a charm !! thanks