| 1 |
root |
1.1 |
#!/usr/bin/perl |
| 2 |
|
|
|
| 3 |
|
|
use Carp; |
| 4 |
|
|
|
| 5 |
|
|
use PApp::SQL; |
| 6 |
root |
1.2 |
use BerkeleyDB; |
| 7 |
root |
1.1 |
|
| 8 |
root |
1.4 |
$DEST = "data/tsv"; |
| 9 |
|
|
|
| 10 |
root |
1.1 |
# 2031616: 0:435514 1:44163 2:107528 3:127671 4:8311 5:2028604 6:117468 7:8210 8:11754 9:24860 14:4344 |
| 11 |
|
|
# 4128768: 0:650846 1:92556 2:183027 3:170894 4:8605 5:4122242 6:147178 7:8238 8:11802 9:31453 14:6460 |
| 12 |
|
|
|
| 13 |
root |
1.4 |
my %cvtid = ( |
| 14 |
|
|
0 => "name", |
| 15 |
|
|
1 => "vorname", |
| 16 |
|
|
2 => "zusatz1", |
| 17 |
|
|
3 => "zusatz2", |
| 18 |
|
|
4 => "zusatz3", |
| 19 |
|
|
5 => "vorwahl", |
| 20 |
root |
1.1 |
# nummer |
| 21 |
root |
1.4 |
7 => "strasse", |
| 22 |
|
|
8 => "haus", |
| 23 |
|
|
9 => "plz", |
| 24 |
|
|
10 => "ort", |
| 25 |
|
|
#[11 => "ort", # Ortsteil |
| 26 |
|
|
12 => "branche", |
| 27 |
|
|
13 => "typ", |
| 28 |
root |
1.1 |
); |
| 29 |
|
|
|
| 30 |
root |
1.4 |
my @cvtid = keys %cvtid; |
| 31 |
|
|
|
| 32 |
root |
1.1 |
local $PApp::SQL::DBH = PApp::SQL::connect_cached "dinfo::", "DBI:mysql:dinfo"; |
| 33 |
|
|
|
| 34 |
|
|
sql_exec "truncate $_" |
| 35 |
|
|
for qw(branche haus name ort plz row strasse typ vorname vorwahl zusatz1 zusatz2 zusatz3); |
| 36 |
|
|
|
| 37 |
root |
1.2 |
# special treatment for name, due to it's size :( |
| 38 |
root |
1.4 |
tie %cache2, |
| 39 |
root |
1.3 |
'BerkeleyDB::Hash', |
| 40 |
root |
1.2 |
-Filename => "/tmp/dinfo-import1-cache-$$", |
| 41 |
root |
1.3 |
-Flags => DB_CREATE|DB_EXCL, |
| 42 |
root |
1.6 |
-Cachesize => 300*1024*1024; |
| 43 |
root |
1.3 |
unlink "/tmp/dinfo-import1-cache-$$"; |
| 44 |
root |
1.1 |
|
| 45 |
root |
1.4 |
open OUT, ">", "$DEST/row.txt"; |
| 46 |
|
|
select OUT; |
| 47 |
|
|
|
| 48 |
root |
1.2 |
while (<STDIN>) { |
| 49 |
root |
1.1 |
chomp; |
| 50 |
|
|
my @data = split /\t/; |
| 51 |
|
|
|
| 52 |
|
|
$data[3] =~ s/^, //; |
| 53 |
|
|
$data[7] =~ s/\.$// && $data[6] =~ s/ Geb$/Geb./; |
| 54 |
|
|
|
| 55 |
|
|
$data[6] =~ /^((?:[a-zA-Z][-.\/a-zA-Z0-9]*)?) (?:\s*\+)? \s* ([0-9\ ]+)$/x or do { |
| 56 |
|
|
warn "ERR: unparseable telnr. '$data[6]'"; |
| 57 |
|
|
next; |
| 58 |
|
|
}; |
| 59 |
|
|
|
| 60 |
|
|
$data[6] = $2; |
| 61 |
|
|
$data[13] = $1; |
| 62 |
|
|
|
| 63 |
|
|
($data[5] eq substr $data[6], 0, length $data[5]) |
| 64 |
|
|
and substr $data[6], 0, length $data[5], ""; |
| 65 |
|
|
|
| 66 |
|
|
if (length $data[6] > 12) { |
| 67 |
|
|
warn "ERR: number too long '$data[6]'"; |
| 68 |
|
|
next; |
| 69 |
|
|
} |
| 70 |
|
|
|
| 71 |
|
|
for (@cvtid) { |
| 72 |
root |
1.4 |
$data[$_] = ($cache{"$_,$data[$_]"} ||= do { |
| 73 |
|
|
my $data = $data[$_]; |
| 74 |
|
|
my $fh = $fh{$_} ||= do { |
| 75 |
|
|
open my $fh, ">", "$DEST/$cvtid{$_}.txt" |
| 76 |
|
|
or die "$DEST/$cvtid{$_}.txt: $!"; |
| 77 |
|
|
$fh; |
| 78 |
|
|
}; |
| 79 |
|
|
my $id = $data eq "" ? 1 : 1 + ++$seq[$_]; |
| 80 |
|
|
$data =~ s/([\\\x0a\x09])/\\$1/g; $data =~ s/\x00/\\0/g; |
| 81 |
|
|
|
| 82 |
|
|
print $fh "$id\t$data\n"; |
| 83 |
|
|
|
| 84 |
|
|
$id; |
| 85 |
|
|
}); |
| 86 |
root |
1.1 |
} |
| 87 |
|
|
|
| 88 |
|
|
$data[13] < 16 or die "ERR: too many typ's"; |
| 89 |
|
|
|
| 90 |
|
|
$data[6] =~ s/ //g; |
| 91 |
|
|
$data[6] = pack "H*", (substr "$data[6]000000000", 0, 12) . sprintf "%x%x", 12 - length $data[6], $data[13]; |
| 92 |
|
|
#warn unpack "H*", $data[6]; |
| 93 |
root |
1.4 |
$data[6] =~ s/([\\\x0a\x09])/\\$1/g; $data[6] =~ s/\x00/\\0/g; |
| 94 |
root |
1.1 |
|
| 95 |
|
|
#name vorname zusatz1 zusatz2 zusatz3 vorwahl nummer strasse haus plz ort branche |
| 96 |
|
|
|
| 97 |
|
|
print +(join "\t", |
| 98 |
|
|
$data[0], $data[1], $data[2], $data[3], $data[4], |
| 99 |
|
|
$data[5], $data[6], |
| 100 |
root |
1.2 |
$data[7], $data[8], $data[9], $data[10], $data[12], |
| 101 |
root |
1.1 |
), "\n"; |
| 102 |
|
|
|
| 103 |
|
|
$count++ & 4095 or warn time . " $count"; |
| 104 |
root |
1.4 |
|
| 105 |
|
|
if ($count == 20_000_000) { |
| 106 |
root |
1.7 |
while (%cache) { |
| 107 |
|
|
warn "copying cache to db...\n"; |
| 108 |
|
|
while (my ($k, $v) = each %cache) { |
| 109 |
|
|
$cache2{$k} = delete $cache{$k}; |
| 110 |
|
|
} |
| 111 |
|
|
} |
| 112 |
root |
1.5 |
warn "nuking mem cache...\n"; |
| 113 |
root |
1.4 |
undef %cache; |
| 114 |
root |
1.5 |
warn "assigning db cache...\n"; |
| 115 |
root |
1.4 |
*cache = \%cache2; |
| 116 |
|
|
} |
| 117 |
root |
1.1 |
} |
| 118 |
|
|
|
