UNIX for Dummies Questions & Answers

ok, didn't quite follow the last 'twist', but (thanks RudiC):

Code:

nawk -F, '$1="A" && !/five/ {$0=",,,"}1'  myFile

For printing all lines starting with A and marking those NOT having "five", try

Code:

awk '/^A/ {if (!/five/) $0=",,,"; print}' file
,,,
,,,
A,three,five,six
,,,

---------- Post updated at 20:08 ---------- Previous update was at 20:07 ----------

@vgersh99: that's an assignment to $1, no?

shooot, too quick - thanks Rudi:

Code:

nawk -F, '$1=="A" && !/five/ {$0=",,,"}1'  myFile

Often for such tasks 'grep -c ... yourdir/* |sed' is good, where sed can change the counts into na or whatever. Sometimes * makes the command line too long or the arg count too high, your OS may vary, but the usual fix is something that starts with 'find' configured for running grep with many file names, sometimes aided by '|xargs -n101' if 'find' lacks that option.

The disadvantage of that grep is the reading of a big file entirely, when the target line is at the front. You could use grep -l to get the set of file names with the pattern, and then use 'comm' to beat it against the set of the full list of tested files (lists must be binary sorted), then use 'sed' to deal with the hit and miss labeling.

I want an output for what ever I search for and when the search has no match, I want an output of some text of my choosing, like comma's or "No Match".

Exactly: 'grep -l' can find the list of files with matches in the list of all files, and 'comm -13' can compare that list to the list of all files, so you can find the files missing on the list and label them 'No Match'. Then you can merge the 'no match' list with the hit list, sorted if you care.

Code:

(
export LC_ALL=C
find * -type f | sort | tee /tmp/full.$$ | xargs -n101 grep -l '<pattern>' >/tmp/hits.$$
comm -13 /tmp/hits.$$ /tmp/full.$$ | sed 's/$/ No Match/' | sort -m /tmp/hits.$$ -
)

I like 101 as it captures >99% of available economy of scale.