If condition and for loop within sed statement

Shell Programming and Scripting
1

Hi,

I tried to go through a lot of online material but could not find concrete solution.
My issues is like this :

I've got a input file like this :

<a>
        <startDate>19700101000000</startDate>
        <endDate>20300101000000</endDate>
</a>
        <startDate>19700101000000</startDate>
        <endDate>20300101000000</endDate>
<b>
        <startDate>19700101000000</startDate>
        <endDate>20300101000000</endDate>
</b>
        <startDate>19700101000000</startDate>
        <endDate>20300101000000</endDate>
<c>
        <startDate>19700101000000</startDate>
        <endDate>20300101000000</endDate>
</c>
<d>
        <startDate>19700101000000</startDate>
        <endDate>20300101000000</endDate>
</d>

Now I want to write a single sed command in such a way that I would give $i as input to sed statement to replace date value of ith occurance.
e.g. if i = 2 , then value of 19700101000000 within <stardate></startdate> within <b></b> is changed to 20200101000000. Similarly if i =1, then startdate within a,if i=3 then from within <c> is changed to 20200101000000.
I tried doing :

sed '/startDate/{s/19700101000000/20200101000000/;};H' "FILENAME.xml"

how to iterate it to loop through a to d and change values as mentioned above.

Please assist.

2

How does your actual file look like? Does it have a,b,c,d? And why a single sed?

--ahamed

3

No Actual file does not have tags like a,b,c...
Actual file has <PRE1>,<PRE2>,<PRE3>......
Also here i mean by single sed command is a one liner command. We can append as many -e,-n but want to complete it as a single liner.

4

Will the elements within the <PREn> tags remain the same? i.e. 2?
Will awk do?

--ahamed

5

No they will also not remain same. But will definitely contain <startdate> and <enddate> tags.
I know awk would be much easier to do, but can't we do it with sed ?

6

what do you mean not in sequential order?

--ahamed

7

Please see my updated reply. :slight_smile:

8

So, you are not sure about the no of elements. You want to change the date for only certain ones you select by giving some number i.e. 1 for a etc like you mentioned in your previous post. But how will you know 1 is for this specific element?

--ahamed

9

We need not worry about specific element. As if input number is 1, it will search for first occurance of <startdate> and replace date within that tag. Similarly if input number is 2, it will search for 2nd occurance of <startdate> and replace date within second <startdate> tag.

10

Try this

awk '/startDate/{if(++x==i){sub(/>.*<\//,">"new"</")}}1' i=3 new=20200101000000 infile

--ahamed

11

Can you give me an idea on how we can achieve through sed ?

12

I am not sure if its really possible via sed. May be you should wait for an expert.

--ahamed

13

alright.
awk statement works like a charm.
Thanks a bunch for spending time and giving a faster solution.
Thanks again. :slight_smile: :b: