Skip to content

Commit

Permalink
Sanitise all occupations
Browse files Browse the repository at this point in the history
  • Loading branch information
@nigelhorne
nigelhorne committed Oct 11, 2023
1 parent 9c87d9b commit ed96f86
Showing 1 changed file with 122 additions and 116 deletions.
238 changes: 122 additions & 116 deletions gedcom
View file
Original file line number Diff line number Diff line change
Expand Up @@ -10442,151 +10442,157 @@ sub get_all_occupations
}

# TODO: sort out case
if(scalar(@occupations)) {
$occupations[0] =~ tr/\r//;
$occupations[0] =~ tr/\n/ /;
$occupations[0] =~ s/\.+$//;

if(($occupations[0] =~ /(.+)\sdomestic$/i) ||
($occupations[0] =~ /(.+)\sdom$/i)) {
$occupations[0] = "Domestic $1";
}

$occupations[0] =~ s/^Formerly //i;
$occupations[0] =~ s/\sretired$//i;

if($occupations[0] =~ /works? on (.+)/i) {
$occupations[0] = "$1 worker";
} elsif(($occupations[0] eq 'Ag Lab') ||
($occupations[0] eq 'Ag Labourer') ||
($occupations[0] eq 'Ag Labourer Pauper') ||
($occupations[0] eq 'Ag Lab Pauper') ||
($occupations[0] eq 'Farm Labourer') ||
($occupations[0] eq 'Agricultural Farm Labourer') ||
($occupations[0] eq 'Ordinary Agricultural Labourer') ||
($occupations[0] eq 'work on farm') ||
($occupations[0] eq 'Agricultural Lab') ||
($occupations[0] eq 'Agril Laborer') ||
($occupations[0] eq 'Labourer (Ag)')) {
$occupations[0] = 'Agricultural Labourer';
} elsif($occupations[0] eq 'Poultry Farming') {
my @rc;

while(my $occupation = shift @occupations) {
$occupation =~ tr/\r//;
$occupation =~ tr/\n/ /;
$occupation =~ s/\.+$//;

if(($occupation =~ /(.+)\sdomestic$/i) ||
($occupation =~ /(.+)\sdom$/i)) {
$occupation = "Domestic $1";
}

$occupation =~ s/^Formerly //i;
$occupation =~ s/\sretired$//i;
$occupation =~ s/Labor/Labour/i;

if($occupation =~ /works? on (.+)/i) {
$occupation = "$1 worker";
} elsif(($occupation eq 'Ag Lab') ||
($occupation eq 'Ag Labourer') ||
($occupation eq 'Ag Labourer Pauper') ||
($occupation eq 'Ag Lab Pauper') ||
($occupation eq 'Farm Labourer') ||
($occupation eq 'Agricultural Farm Labourer') ||
($occupation eq 'Ordinary Agricultural Labourer') ||
($occupation eq 'work on farm') ||
($occupation eq 'Agricultural Lab') ||
($occupation eq 'Agril Labourer') ||
($occupation eq 'Labourer (Ag)')) {
$occupation = 'Agricultural Labourer';
} elsif($occupation eq 'Poultry Farming') {
if($language eq 'French') {
$occupations[0] = (($person->sex() eq 'M') ? 'Agriculteur de poulet' : 'Agricultrice de poulet');
$occupation = (($person->sex() eq 'M') ? 'Agriculteur de poulet' : 'Agricultrice de poulet');
} else {
$occupations[0] = 'Poultry Farmer';
$occupation = 'Poultry Farmer';
}
} elsif($occupations[0] eq 'Platelayer Railway') {
$occupations[0] = 'Railway Platelayer';
} elsif(($occupations[0] eq 'General Servant Domestic') ||
($occupations[0] =~ /^General serv.+dom/i) ||
($occupations[0] =~ /^Domestic servant$/i)) {
} elsif($occupation eq 'Platelayer Railway') {
$occupation = 'Railway Platelayer';
} elsif(($occupation eq 'General Servant Domestic') ||
($occupation =~ /^General serv.+dom/i) ||
($occupation =~ /^Domestic servant$/i)) {
if($language eq 'French') {
$occupations[0] = 'Domestique';
$occupation = 'Domestique';
} else {
$occupations[0] = 'Domestic servant';
}
} elsif($occupations[0] =~ /(.+)\sserv$/i) {
$occupations[0] = "$1 servant";
} elsif($occupations[0] eq 'Lorry Driver Heavy Worker') {
$occupations[0] = 'Lorry Driver';
} elsif($occupations[0] =~ /^Shop Assistant (.*)/) {
$occupations[0] = "$1's Shop Assistant";
} elsif($occupations[0] =~ /(.*) Assistant$/i) {
$occupation = 'Domestic servant';
}
} elsif($occupation =~ /(.+)\sserv$/i) {
$occupation = "$1 servant";
} elsif($occupation eq 'Lorry Driver Heavy Worker') {
$occupation = 'Lorry Driver';
} elsif($occupation =~ /^Shop Assistant (.*)/) {
$occupation = "$1's Shop Assistant";
} elsif($occupation =~ /(.*) Assistant$/i) {
if($1 eq 'Bakers') {
$occupations[0] = "Baker's Assistant";
$occupation = "Baker's Assistant";
} elsif($1 eq 'Butchers') {
$occupations[0] = "Butcher's Assistant";
$occupation = "Butcher's Assistant";
} elsif($1 ne 'Shop') {
$occupations[0] = "$1's Assistant" unless($1 =~ /'s$/);
}
} elsif($occupations[0] =~ /Clerk (.*)/) {
$occupations[0] = "$1 Clerk";
} elsif($occupations[0] =~ /Salesman (.*)/i) {
$occupations[0] = "$1 Salesman";
} elsif($occupations[0] =~ /Foreman (.*)/i) {
$occupations[0] = "$1 Foreman";
$occupations[0] =~ s/^of the //i;
} elsif($occupations[0] =~ /Labourer Gas Stoker/) {
$occupations[0] = 'Gas Stoker';
} elsif(($occupations[0] eq 'Under Gardener Domestic') ||
($occupations[0] eq 'Domestic Under Gardner')) { # sic
$occupations[0] = 'Domestic Gardener';
} elsif($occupations[0] eq 'Market gardener') {
$occupations[0] = 'Market Gardener';
} elsif($occupations[0] eq "Brewer's Labourer") {
$occupations[0] = 'Brewery Labourer';
} elsif($occupations[0] eq "Labourer Builders") {
$occupations[0] = "Builder's labourer";
} elsif(($occupations[0] eq 'Gardener domestic') ||
($occupations[0] eq 'Gardner and domestic servant')) {
$occupations[0] = 'Gardener and Domestic';
} elsif($occupations[0] =~ /(.+)\sserv$/i) {
$occupations[0] = "$1 servant";
} elsif($occupations[0] =~ /^(.+) on farm/i) {
$occupations[0] = ($language eq 'French') ? "$1 en un ferme" : "$1 on a farm";
} elsif($occupations[0] eq 'Plate Glass Cutter') {
} elsif($occupations[0] =~ /police$/i) {
$occupations[0] .= ' officer';
} elsif(($occupations[0] eq 'Nurse') && ($language eq 'French')) {
$occupations[0] = ($person->sex() eq 'M') ? 'infirmier' : "infirmi\N{U+00E8}re";
} elsif($occupations[0] =~ /teaching/i) {
$occupation = "$1's Assistant" unless($1 =~ /'s$/);
}
} elsif($occupation =~ /Clerk (.*)/) {
$occupation = "$1 Clerk";
} elsif($occupation =~ /^Salesman (.*)/i) {
$occupation = "$1 Salesman";
} elsif($occupation =~ /Foreman (.*)/i) {
$occupation = "$1 Foreman";
$occupation =~ s/^of the //i;
} elsif($occupation =~ /Labourer Gas Stoker/) {
$occupation = 'Gas Stoker';
} elsif(($occupation eq 'Under Gardener Domestic') ||
($occupation eq 'Domestic Under Gardner')) { # sic
$occupation = 'Domestic Gardener';
} elsif($occupation eq 'Market gardener') {
$occupation = 'Market Gardener';
} elsif($occupation eq "Brewer's Labourer") {
$occupation = 'Brewery Labourer';
} elsif($occupation eq "Labourer Builders") {
$occupation = "Builder's labourer";
} elsif(($occupation eq 'Gardener domestic') ||
($occupation eq 'Gardner and domestic servant')) {
$occupation = 'Gardener and Domestic';
} elsif($occupation =~ /(.+)\sserv$/i) {
$occupation = "$1 servant";
} elsif($occupation =~ /^(.+) on farm/i) {
$occupation = ($language eq 'French') ? "$1 en un ferme" : "$1 on a farm";
} elsif($occupation eq 'Plate Glass Cutter') {
} elsif($occupation =~ /police$/i) {
$occupation .= ' officer';
} elsif(($occupation eq 'Nurse') && ($language eq 'French')) {
$occupation = ($person->sex() eq 'M') ? 'infirmier' : "infirmi\N{U+00E8}re";
} elsif($occupation =~ /teaching/i) {
if($language eq 'French') {
$occupations[0] = 'professeur';
$occupation = 'professeur';
} elsif($language eq 'German') {
$occupations[0] = ($person->sex() eq 'M') ? 'Lehrer' : 'Lehrerin';
$occupation = ($person->sex() eq 'M') ? 'Lehrer' : 'Lehrerin';
} else {
$occupations[0] = 'teacher';
$occupation = 'teacher';
}
} elsif(($occupations[0] !~ /gas works/i) && ($occupations[0] =~ /(.+)([a-z])s\s+([a-z]+)$/i)) {
} elsif(($occupation !~ /gas works/i) && ($occupation =~ /(.+)([a-z])s\s+([a-z]+)$/i)) {
# Don't change Bus Driver to Bu's Driver
$occupations[0] = "$1$2's $3" unless(("$1$2" eq 'Bu') || ("$1$2" eq 'Harnes'));
} elsif(($occupations[0] =~ /Manager (.*)/i) &&
($occupations[0] !~ /^Manager of /i) &&
($occupations[0] !~ /Manager & /i)) {
$occupations[0] = "$1 Manager";
$occupation = "$1$2's $3" unless(("$1$2" eq 'Bu') || ("$1$2" eq 'Harnes'));
} elsif(($occupation =~ /Manager (.*)/i) &&
($occupation !~ /^Manager of /i) &&
($occupation !~ /Manager & /i)) {
$occupation = "$1 Manager";
}

if($occupations[0]) {
if($occupation) {
if($language eq 'English') {
if($ENV{'LANG'} =~ /^en_US/) {
$occupations[0] = Lingua::EN::ABC::b2a(lc($occupations[0]));
$occupation = Lingua::EN::ABC::b2a(lc($occupation));
$occupation =~ s/labour/labor/;
$occupation =~ s/Labour/Labor/;
} elsif($ENV{'LANG'} =~ /^en_CA/) {
$occupations[0] = Lingua::EN::ABC::b2c(lc($occupations[0]));
$occupation = Lingua::EN::ABC::b2c(lc($occupation));
} else {
$occupations[0] = Lingua::EN::ABC::a2b(lc($occupations[0]));
$occupation = Lingua::EN::ABC::a2b(lc($occupation));
}
} elsif($language eq 'French') {
if($occupations[0] eq 'Postman') {
$occupations[0] = (($person->sex() eq 'M') ? 'Facteur' : 'Factrisse');
} elsif($occupations[0] =~ /(.+)\sFarmer/) {
$occupations[0] = (($person->sex() eq 'M') ? "Agriculteur $1" : "Agricultrice $1");
} elsif($occupations[0] eq 'Farmer') {
$occupations[0] = (($person->sex() eq 'M') ? 'Agriculteur' : 'Agricultrice');
} elsif($occupations[0] eq 'Teacher') {
$occupations[0] = 'Professeur';
}
$occupations[0] =~ s/retired/\N{U+00E0} la retrainte/i;
}
$occupations[0] =~ s/[\(\)]//g;
$occupations[0] =~ s/\s+Own Account//i;
if($occupations[0] =~ /works? (on|for) (.+)/i) {
$occupations[0] = "$2 worker";
} elsif($occupations[0] =~ /^Cleaner\s+(.+)/i) {
$occupations[0] = "$1 cleaner";
}
$occupations[0] = ucfirst($occupations[0]);
$occupations[0] =~ s/\s+heavy worker$//i;
$occupations[0] =~ s/\s+own business$//i;
if($occupation eq 'Postman') {
$occupation = (($person->sex() eq 'M') ? 'Facteur' : 'Factrisse');
} elsif($occupation =~ /(.+)\sFarmer/) {
$occupation = (($person->sex() eq 'M') ? "Agriculteur $1" : "Agricultrice $1");
} elsif($occupation eq 'Farmer') {
$occupation = (($person->sex() eq 'M') ? 'Agriculteur' : 'Agricultrice');
} elsif($occupation eq 'Teacher') {
$occupation = 'Professeur';
}
$occupation =~ s/retired/\N{U+00E0} la retrainte/i;
}
$occupation =~ s/[\(\)]//g;
$occupation =~ s/\s+Own Account//i;
if($occupation =~ /works? (on|for) (.+)/i) {
$occupation = "$2 worker";
} elsif($occupation =~ /^Cleaner\s+(.+)/i) {
$occupation = "$1 cleaner";
}
$occupation = ucfirst($occupation);
$occupation =~ s/\s+heavy worker$//i;
$occupation =~ s/\s+own business$//i;

push @rc, $occupation;
} else {
complain({
person => $person,
warning => 'Occupation is empty'
});
shift @occupations;
}
}

return @occupations;
return @rc;
}

# Document a complete journey when both a departure and an arrival record exist
Expand Down

0 comments on commit ed96f86

Please sign in to comment.