CODEWITHSUNDEEP
bashshellscriptingtcl
Bryan
Bryan

Reputation: 339

How to delete line with matching pattern from another file?

I want to delete lines in FILE1 contains pattern in FILE2.
How do I do this using shell/bash or Tcl?

For example:

FILE1: 

This is ECO_01  
This is ECO_02  
This is ECO_03  
This is ECO_04

FILE2: 

ECO_02  
ECO_04

Output: 

This is ECO_01   
This is ECO_03

Upvotes: 4

Views: 2097

Answers (4)

glenn jackman
glenn jackman

Reputation: 247210

Another Tcl solution:

set fid [open file2 r]
set patterns [lmap line [split [read -nonewline $fid] \n] {string trim $line}]
close $fid

set fid [open file1 r]
set lines [split [read -nonewline $fid] \n]
close $fid

set wanted [lsearch -inline -all -regexp -not $lines [join $patterns "|"]]
puts [join $wanted \n]

This is ECO_01  
This is ECO_03

Ref: lsearch man page

Upvotes: 0

Donal Fellows
Donal Fellows

Reputation: 137787

In Tcl, you'd load the file of patterns in and use them to then do the filtering. It's probably simplest to keep the main filtering flow going from standard input to standard output; you can redirect those from/to files easily enough. Since you seem to be wanting to use “is pattern a substring of” as a matching rule, you can do that with string first, leading to this code:

# Load in the patterns from the file named by the first argument
set f [open [lindex $argv 0]]
set patterns [split [string trimright [read $f] \n] \n]
close $f

# Factor out the actual matching
proc matches {theString} {
    global patterns
    foreach pat $patterns {
        # Change the next line to use other matching rules
        if {[string first $pat $theString] >= 0} {
            return true
        }
    }
    return false
}

# Read all input lines and print all non-matching lines
while {[gets stdin line] >= 0} {
    if {![match $line]} {
        puts $line
    }
}

I find it helpful to factor out procedures with key bits like “does this line match any of my patterns?” You'd probably call the above code a bit like this:

tclsh doFiltering.tcl patterns.txt <input.txt >output.txt

Upvotes: 2

shofee
shofee

Reputation: 2129

You just have to use sed command ( as shown below ) to delete the matching lines from FILE1.

macOS:

for i in `cat FILE2.txt`
do
sed -i '' "/$i/d" FILE1.txt
done

Linux:

for i in `cat FILE2.txt`
do
sed -i '/$i/d' FILE1.txt
done

Upvotes: 0

karakfa
karakfa

Reputation: 67567

most generic solution will be

$ grep -vf file2 file1

note that any substring match on any field will count. If you only restrict to exact match on an exact field (here assumed the last)

$ awk 'NR==FNR{a[$1]; next} !($NF in a)' file2 file1

Upvotes: 4

Related Questions