Split files based on text within the file

[IMAUser]

I have an Oracle program producing a huge xml file which is in a concatenated form with every new type within the file having
<?xml version=&quot;1.0&quot; encoding=&quot;UTF-8&quot;?> as the first line in the file. A few lines after the above starter line there is another line
<index_name>US_RPI</index_name>

Now the question, what I need to be able to do is something like
-- my_awk_script.aw

OUTFILE=/tmp/my_out_file.lst
if text is /..xml.version/ --look for <?xml version this marks the begining of new type
{
if $OUTFILE exists
new_file_name=`grep index_name in $OUTFILE | get the text within > < `
mv $OUTFILE /tmp/$new_file_name
fi
-- Start the $OUTFILE afresh with new data lines
print $0 > $OUTFILE
}
else
{
-- Since this line is not a new type keep appending to the existing file
print $0 >> $OUTFILE
fi
}

I hope I was clear enough. The hash key seems to have gone bonkers on my keyboard and hence have used -- for coments.

Can I get some pointers please.

Thanks

 

[aigles]

Try and adapt :
[tt]
awk -v OUTFILE=/tmp/my_out_file.lst '

#
# If first line contains &quot;<?xml version&quot;, following lines are memorized
# until we found the &quot;<index_name&quot; line.
#

NR==1 && /<\?xml version/ {
memo = 1
}

#
# If Memorize phase, add line to Lines[]
#

memo {
Lines[++LinesCount] = $0;
}

#
# If Normal phase, print line to OUTPUT file
#

! memo {
print $0 >> OUTFILE;
}

#
# If memorize phase and line contains &quot;<index_name&quot;, get the new output file name
# and write memorized lines to it. Set normal phase (no memo)
#

memo && /<index_name>/ {
sub(/^[^>]*>/,&quot;&quot;
sub(/<.*$/,&quot;&quot;
close(OUTFILE);
OUTFILE = &quot;/tmp/&quot; $0;
for (iline=1; iline<=LinesCount; iline++)
print Lines[iline] >> OUTFILE;
memo = 0;
}
' input_file
[/tt]

Jean Pierre.

 

[IMAUser]

Thanks very much, It works great with a few changes.

But can you please explain whats happening here and what is sub () all about. I just dont get the pattern matching and the way it extracts the text from within
&lt;index_name&gt;US_RPI&lt;/index_name&gt;


sub(/^[^&gt;]*&gt;/,&quot;&quot;
sub(/&lt;.*$/,&quot;&quot;

Many Thanks

 

[aigles]

[tt]sub(/^[^&gt;]*&gt;/,&quot;&quot;[/tt]

Remove in $0 all chars from begining to the first '&gt;'
^ Begining of string
[^&gt;] All chars except '&gt;'
* Zero or more times
&gt; Char '&gt;'

[tt]sub(/&lt;.*$/,&quot;&quot;[/tt]

Remove in $0 all chars from the first '&lt;' to end.
&lt; Char '&lt;'
. Any single character
* Zero or more times
$ End of string


Jean Pierre.

 

[IMAUser]

Thanks Jean,
It makes more sense now and will be very helpful in future.

Thanks for all your help on this.