Shell Programming and Scripting

Hi gang,

I am trying to create some batch commands for many html pages I need to re-format.

I am trying the number 2b in this example to wrap anchor tags around the number that will be referenced in the footnotes.

I am trying to use the h/H hold command, but I have never tried using it before and it is not yet working.

From the sed help file example it uses the 'g/G' to extract whatever is in the hold buffer.

So basically, each of my html pages have pages numbers of text referring to the original text book numbers and used to reference each pages footnotes. So I want sed, each time it finds a page number to hold that page number in 'h/H' and then check each line after for '2b' or any number in which the line does not begin with '<' (already formatted with HTML tags) and wrap the number in anchor tags according to the last page number held in HOLD.

Most pages have multiple page references, so every time sed encounters a new page number, it should re-write the HOLD space with new page number and begin applying that to each anchor, until of course another new page number is encountered.

I thought I had that with:

Code:

sed -i -e '/[0-9]\{1,3\}$/{h;};/^</!s/\([0-9]\{1,3\}[b]\{,1\}\)/\<a href="g\1End" name="g\1Txt"><font size=-1">[g:\1]<\/font><\/a>/g' 1.html

but for this line of text:

Code:

From old Ikshváku's  2b line he came,

I get:

Code:

From old Ikshváku's  <a href="g2bEnd" name="g2bTxt"><font size=-1">[g:2b]</font></a> line he came,

It looks like 'g/G' is not what I need to extract whatever is (hopefully) in the HOLD space. Can anyone help me out here?

Originially I would have thought the two semi-colons after 'h' inside curly brackets and then outside would puke back an error, but they do not. Having just the one (inside curly brackets), does result in sed balking.

appreciated.
tt

Hi, g/G is a command in itself, it only copies from the hold space into the pattern space. I cannot be part of a substitution command. I think this will be complicated with sed and I suspect you'll find awk a better match...

so ok that explains why I am getting the 'g' in front when I want the chapter number I was hoping was being stored in hold. so no way to extract that chapter number ans use it as part of a sub command? Why have the hold command then? It seems logical to hold data that will be used (inserted/sub'd) in somewhere no?
I suspect you're probably right, but I do not know awk and even though I have been looking for a tut recently b/c I am starting to need to do more things like this and I am sure awk will be more versatile, I just don't have the time to learn what I need.

any awk one-liners that do something similar I might be able to apply?