1

How do I remove duplicated entries within each separate fields with below sample as data.

0x,9.4,,,#0,#UNIX#unix,#cli#L#فا#0#فا#0#L#SE#Cli#SE,#فارسی#فارسی#۱#1#١#1,bsh,#V & v

expected output(either delete all duplicated ones, case-insensitive, difference in Unicode "Persian #۱/Arabic #١", order of entries and which entry (ignore case) should keep doesn't matter here): 

0x,9.4,,,#0,#unix,#cli#L#فا#0#SE,#فارسی#١#۱#1,bsh,#V & v

The pattern is in this format #x, x means anything in one or more length of characters.

Unicode table for Persian/Arabic languages alphabet/numbers differences

αғsнιη
  • 41,407

3 Answers3

1

Using a command line in a shell (just a few lines) with a proper parser :

perl -CS -Mopen=":std,IN,OUT,IO,:encoding(utf8)" -MText::CSV -lne '
    BEGIN{
        our $csv = Text::CSV->new({ sep_char => "," });
        sub uniq { my %seen;  grep !$seen{lc $_}++, @_; }
    };
    $csv->parse($_) or die "parse error";
    print join ",", map { join "#", uniq split /#/ } $csv->fields();
' file.csv

Output :

0x,9.4,,,#0,#UNIX,#cli#L#فا#0#SE,#فارسی#۱#1#١,bsh,#V & v

Note :

  • require to install Text::CSV perl module : sudo apt-get install libtext-csv-perl for debian and derivative
Gilles Quénot
  • 33,867
1

Complete Python solution:

uniq_arabic_csv.py script: 

#!/bin/python

import sys, re

csv_data = sys.stdin.read().strip()
result = []

for item in csv_data.split(','):
    matches = list(re.finditer(r'#[^#]+', item, re.U | re.I))
    if not matches or len(matches) == 1:
        result.append(item)
    else:
        s, res = set(), []
        for m in matches:
            m = m.group()
            if m.lower() not in s: res.append(m)
            s.add(m.lower())
        result.append(''.join(res))

result = ','.join(result)                    
print(result)

Usage:

s="0x,9.4,,,#0,#UNIX#unix,#cli#L#فا#0#فا#0#L#SE#Cli#SE,#فارسی#فارسی#۱#1#۱#1,bsh,#V & v"
python uniq_arabic_csv.py <<<"$s"

The output:

0x,9.4,,,#0,#UNIX,#cli#L#فا#0#SE,#فارسی#۱#1,bsh,#V & v
RomanPerekhrest
  • 30,212
0

I code below simple awk solution as well:

awk -F, '{ i=0; comma=""; while (++i<=NF) {
    if ( split(tolower($i), arr, /#/)>1 ){ for(v in arr) !unq[arr[v]]++;printf comma;
        for (x in unq) {printf (x!="")?hash""x:""; hash="#"; delete unq[x]}}
    else {printf comma""$i; comma=","}}; print ""
}' infile.csv
αғsнιη
  • 41,407