Shell Programming and Scripting

Hi ,suppose I have a file
title=dsafsadf
........
.......
year=1995
author=john smith
............
title=bbbbbb
........
.......
year=1988
author=alex

I need to replace the title line with a expression that contains variables year and author. I want to use a python readline for loop. The problem is that I couldn't use commands such as 'print data[i-3].strip()' because I don't know exactly how many lines are between the current line and title line.
Is there any way to do this?

Code:

#!/bin/ksh
mInpFile='Your_File'
mYearAuthor='/tmp/YearAuthor'
egrep 'year|author' ${mInpFile} > ${mYearAuthor}
while read mLine
do
  mFirstSix=`echo $mLine | cut -c1-6`
  if [ "${mFirstSix}" = "title=" ]; then
    mYear=`sed -n '1p' ${mYearAuthor} | cut -d'=' -f2`
    mAuthor=`sed -n '2p' ${mYearAuthor} | cut -d'=' -f2`
    mOutLine='title='${mYear}${mAuthor}
    sed '1,2d' ${mYearAuthor} > $$Temp
    mv $$Temp ${mYearAuthor}
  else
    mOutLine=${mLine}
  fi
  echo ${mOutLine}
done < ${mInpFile}
rm -f ${mYearAuthor}

fantastic!
I see how you managed to match the year/author of book1 to the title of book1 and year/author of book2 to title of book2. You would delete the first two lines of the YearAuthor after each iteration.How genious! Algorithm is not something you can learn from comptuer science class.

How about combining this, sed and cut to just awk

Code:

mYear=`awk -F"=" 'NR == 1 { print $2 }' filename`

or

Code:

mYear=`awk -F"=" ' { if ( NR == 1 ) { print $2 } else { exit } }' filename`

One minor defect with the ksh script is that ksh seem to delete leading white space in the 'while ' loop. There $mline is not exactly the same as each line of input file.
so output is missing leading white space.

First change the value of the IFS variable before the read statement and place double qoutes around the variable like this:

Code:

#!/bin/ksh
mInpFile='Your_File'
mYearAuthor='/tmp/YearAuthor'
egrep 'year|author' ${mInpFile} > ${mYearAuthor}
OLDIFS=$IFS
IFS=
while read mLine
do
  mFirstSix=`echo $mLine | cut -c1-6`
  if [ "${mFirstSix}" = "title=" ]; then
    mYear=`sed -n '1p' ${mYearAuthor} | cut -d'=' -f2`
    mAuthor=`sed -n '2p' ${mYearAuthor} | cut -d'=' -f2`
    mOutLine='title='${mYear}${mAuthor}
    sed '1,2d' ${mYearAuthor} > $$Temp
    mv $$Temp ${mYearAuthor}
  else
    mOutLine=${mLine}
  fi
  echo "${mOutLine}"
done < ${mInpFile}
rm -f ${mYearAuthor}
IFS=$OLDIFS

Regards

Matrix, thank you for commenting on my code.

It shows that you read it and tried to understand it.

Here in the site, when I write a code, I do it quickly, most replies take
only a few minutes, with the first solution that comes to my mind, as
the OP is online waiting for an answer -- just making sure it is working
using whatever the OP gave as specification.

If the same problem was given to me as an assignment in my workplace,
with several days to code it, I would take an entire different approach,
which could result in a drastically different script.

Most of the time, for more involved issues, it is not written in shell scripts,
but in a compiled language.

I also have to follow the company standards, place header comments,
have in mind easy of maintenance, etc.

As for changing this specific solution from "sed+cut" to "just awk", it is
possible to do it, as you demonstrated.

It is also possible to do with just "sed", just "perl", just "python", etc.

As the saying goes:
"There are several ways to skin a cat."

Matrix, the great majority of the OPs here are new to UNIX, thus every time
I offer a solution, I keep in mind the following:
1) To keep it easy to read.
2) To keep it simple.
3) To use plain UNIX commands, exception for basic 'sed' commands.

Languages such as 'awk', 'perl', 'python' are very powerful, they are very
useful for people who know how to use them properly, which may take
years for a broad understanding of all its commands and functionality.

Before one start learning any of these languages, he/she must have a
great experience using plain UNIX commands.

A lot of times, the OPs ask for an explanation of the code, even when it
has a single command, such as 'paste' -- most of them don't even know
about 'man' page.

Writing code using advanced UNIX languages may scare an UNIX apprentice
and it may lead him to believe, as a beginner, that he must use this language
to solve his problem.

The other issue is maintenance -- if the solution is presented in advanced
UNIX languages, the OP will have no chance to easily change it.

Thus, my approach to always use plain UNIX commands -- when not possible,
to use 'sed'.

Thanks again for proposing another possible solution for the problem.