UNIX for Dummies Questions & Answers

Could anybody tell me how I can use sed to modify lines following specific lines? the file is as following:

"TEST/SI1573.lab"
3670 8920 h#
8920 9530 hh
9530 10694 ih
.
"TEST/DR1/FAKS0/SI2203.lab"
9730 9580 h#
9580 9840 dh
9840 10652 ix
10652 11997 r
........

I want to modify the first fields to 0 of the lines which follows the lines started with "TEST. The file after operation should be:
"TEST/SI1573.lab"
0 8920 h#
8920 9530 hh
9530 10694 ih
.
"TEST/DR1/FAKS0/SI2203.lab"
0 9580 h#
9580 9840 dh
9840 10652 ix
10652 11997 r
........
could anybody tell my how to do that? Thanks in advance.

Code:

sed -e '/^"TEST/{p;N;s/.*\n[0-9]*/0/;}' filename

There are many different versions of sed, so yours might not understand exactly the same dialect as mine.

This looks for '"TEST' (with an opening double quote) at beginning of line. If found, it prints that line (p), and appends the next line to the pattern space (N). This causes the pattern space to contain two lines; the TEST line and the following line, separated by a newline. Then it replaces (s///) the first line in the pattern space, the newline, and any numbers just after the newline with a zero. At that point, we are done; whatever is left in the pattern space will be printed as usual.

sed syntax is very terse; if you don't have a specific reason to use sed for this, perhaps an equivalent awk or Perl script would be more maintainable (especially if you are not very familiar with sed).

Thanks, it works.

I wonder how to use awk to do this.

Code:

awk '/^"TEST/ { t=1 ; print; next }
t==1 { $1 = "0"; t=0; }1' filename

This causes the variable t to be set to 1 on the following line after the TEST line. If this is the case, replace the first field with a zero, and set t back to zero. The final 1 causes any line which reaches that point in the script to be printed. (It's a shorthand; the default action of awk is to print if the condition is true; and 1 as a condition is always true.)

Thanks, era.

Do you think you can recommend me some book or websit? I would like to learn more about script. I alwasy encounter some problems, and could not figure out by myself.

One of my personal favorites are Kernighan & Pike's The Unix Programming Environment but it's very old, so some of the peripheral details have changed quite a lot since then. Another perennial is Friedl's Mastering Regular Expressions, although you might struggle with it at first if you are not familiar with the tools it covers.

The O'Reilly sed & awk book is a good introduction to sed and awk, although there are a lot of on-line tutorials which are probably just as good.

The FAQ section has more links; I'm new to Unix. Which books should I read? - The UNIX Forums