| 1 |
#!/usr/bin/perl |
| 2 |
|
| 3 |
use Carp; |
| 4 |
|
| 5 |
use PApp::SQL; |
| 6 |
use BerkeleyDB; |
| 7 |
|
| 8 |
# 2031616: 0:435514 1:44163 2:107528 3:127671 4:8311 5:2028604 6:117468 7:8210 8:11754 9:24860 14:4344 |
| 9 |
# 4128768: 0:650846 1:92556 2:183027 3:170894 4:8605 5:4122242 6:147178 7:8238 8:11802 9:31453 14:6460 |
| 10 |
|
| 11 |
my @cvtid = ( |
| 12 |
[name => 0], |
| 13 |
[vorname => 1], |
| 14 |
[zusatz1 => 2], |
| 15 |
[zusatz2 => 3], |
| 16 |
[zusatz3 => 4], |
| 17 |
[vorwahl => 5], |
| 18 |
# nummer |
| 19 |
[strasse => 7], |
| 20 |
[haus => 8], |
| 21 |
[plz => 9], |
| 22 |
[ort => 10], |
| 23 |
#[ort => 11], # Ortsteil |
| 24 |
[branche => 12], |
| 25 |
[typ => 13], |
| 26 |
); |
| 27 |
|
| 28 |
local $PApp::SQL::DBH = PApp::SQL::connect_cached "dinfo::", "DBI:mysql:dinfo"; |
| 29 |
|
| 30 |
sql_exec "truncate $_" |
| 31 |
for qw(branche haus name ort plz row strasse typ vorname vorwahl zusatz1 zusatz2 zusatz3); |
| 32 |
|
| 33 |
sql_exec "lock tables " . join ", ", map "$_ write", |
| 34 |
qw(branche haus name ort plz strasse typ vorname vorwahl zusatz1 zusatz2 zusatz3); |
| 35 |
|
| 36 |
# special treatment for name, due to it's size :( |
| 37 |
unlink "/tmp/dinfo-import1-cache-$$"; |
| 38 |
tie %cache, |
| 39 |
'BerkeleyDB::Btree', |
| 40 |
-Filename => "/tmp/dinfo-import1-cache-$$", |
| 41 |
-Flags => DB_CREATE, |
| 42 |
-Cachesize => 380*1024*1024; |
| 43 |
|
| 44 |
while (<STDIN>) { |
| 45 |
chomp; |
| 46 |
my @data = split /\t/; |
| 47 |
|
| 48 |
$data[3] =~ s/^, //; |
| 49 |
$data[7] =~ s/\.$// && $data[6] =~ s/ Geb$/Geb./; |
| 50 |
|
| 51 |
$data[6] =~ /^((?:[a-zA-Z][-.\/a-zA-Z0-9]*)?) (?:\s*\+)? \s* ([0-9\ ]+)$/x or do { |
| 52 |
warn "ERR: unparseable telnr. '$data[6]'"; |
| 53 |
next; |
| 54 |
}; |
| 55 |
|
| 56 |
$data[6] = $2; |
| 57 |
$data[13] = $1; |
| 58 |
|
| 59 |
($data[5] eq substr $data[6], 0, length $data[5]) |
| 60 |
and substr $data[6], 0, length $data[5], ""; |
| 61 |
|
| 62 |
if (length $data[6] > 12) { |
| 63 |
warn "ERR: number too long '$data[6]'"; |
| 64 |
next; |
| 65 |
} |
| 66 |
|
| 67 |
for (@cvtid) { |
| 68 |
my ($name, $idx) = @$_; |
| 69 |
$data[$idx] = ($cache{"$idx,$data[$idx]"} |
| 70 |
||= sql_insertid sql_exec |
| 71 |
"insert into $name (id, data) values (NULL, ?)", "$data[$idx]"); |
| 72 |
} |
| 73 |
|
| 74 |
$data[13] < 16 or die "ERR: too many typ's"; |
| 75 |
|
| 76 |
$data[6] =~ s/ //g; |
| 77 |
$data[6] = pack "H*", (substr "$data[6]000000000", 0, 12) . sprintf "%x%x", 12 - length $data[6], $data[13]; |
| 78 |
#warn unpack "H*", $data[6]; |
| 79 |
$data[6] =~ s/([\\\x0a\x09])/\\$1/g; |
| 80 |
$data[6] =~ s/\x00/\\0/g; |
| 81 |
|
| 82 |
#name vorname zusatz1 zusatz2 zusatz3 vorwahl nummer strasse haus plz ort branche |
| 83 |
|
| 84 |
print +(join "\t", |
| 85 |
$data[0], $data[1], $data[2], $data[3], $data[4], |
| 86 |
$data[5], $data[6], |
| 87 |
$data[7], $data[8], $data[9], $data[10], $data[12], |
| 88 |
), "\n"; |
| 89 |
|
| 90 |
$count++ & 4095 or warn time . " $count"; |
| 91 |
} |
| 92 |
|
