takdavid

#!/usr/bin/perl

binmode(STDIN, 'utf8');
binmode(STDOUT, 'utf8');

sub smart_lowercase($)
{
    my $word = shift;
    return $word if (length($word) > 1 and uc($word) eq $word);
    return lc $word;

takdavid

#!/usr/bin/perl

use Unicode::UCD 'charinfo';
use Unicode::Normalize 'isNonStDecomp', 'NFD', 'NFC', 'isComp2nd', 'isNonStDecomp';
use charnames ':full';
use Encode;
use HTML::Entities;

#use Getopt::Std;
#our ($opt_f, $opt_r);

takdavid

#!/usr/bin/perl

my @w;
while (<STDIN>) { chomp; while (split /[^-\w]+/) { push @w, lc $_; } }

my $c = {};
while (@w) { $c->{$_} += 1; }

my @k = reverse sort { $c->{$a} <=> $c{$b} } keys %{$c};
while (@k) { print $c->{$k}, $k, "\n"; }

takdavid

#!/usr/bin/perl

my @w;
my $sw = {};
while (<DATA>) { print $_; chomp; $sw->{$_} = 1; last; }
close DATA;

while (<STDIN>) { chomp; while (split /[^-\w]+/) { $_ = lc $_; if (!$sw->{$_}) { push @w, $_; } } }

my $c = {};

takdavid

#!/bin/bash
set -e
set -x

git clone git@github.com:holgerd77/django-dynamic-scraper.git

cd django-dynamic-scraper
virtualenv -p python2.7 env
source env/bin/activate
pip install celery==2.4.6

takdavid

import memcache
import hashlib

client = None

def getClient() :
    global client
    if not client :
        client = memcache.Client(['127.0.0.1:11211'], debug=0)
    return client

takdavid

package MARC::File::HUNMARC;

=head1 NAME

MARC::File::HUNMARC - HUNMARC-specific file handling

=cut

use strict;
use integer;