in unix system, replace string in file, with variouse position

I work on uni开发者_Go百科x server.

I have many csv files containing, among other info, date fields.

I have to replace some of this date fields with another value, for example 20110915 to 20110815. Their position is variable from a file to another.

The problem is that the substitution is position field specific. For example, if my file has row like this:

blablabla;12;0.2121;20110915;20110915;19951231;popopo;other text;321;20101010

I have to replace only first date fields and not other, transforming row in:

blablabla;12;0.2121;20110815;20110915;19951231;popopo;other text;321;20101010

Is there a way to restrict the replace in file, using some constraints?

Thanks

You can try awk:

awk  'BEGIN {FS=";";OFS=";"} {if($4=="20110915")$4="20110815"; print}' input.csv

How it works:

FS and OFS define the input and output field separators. It compares the fourth field ($4) against 20110915. If it matches, it is changed to 20110815. The line is then printed.

Here is an alternative using gsub in awk:

awk 'BEGIN {FS=";";OFS=";"} {gsub(/20110915/,20110815,$4); print}' input.csv

Here is a method, if you have to substitute in a range of fields/columns (e.g. 4-4):

awk 'BEGIN {FS=";";OFS=";"} {for(i=4;i<=4;i++){gsub(/20110915/,20110815,$i)}; print}' input.csv