]> git.evergreen-ils.org Git - working/Evergreen.git/blob - Open-ILS/src/sql/Pg/upgrade/YYYY.functions.url_verify.sql
Link checker: middle layer work, actual parallelized machinery to check links
[working/Evergreen.git] / Open-ILS / src / sql / Pg / upgrade / YYYY.functions.url_verify.sql
1 BEGIN;
2
3 CREATE OR REPLACE FUNCTION url_verify.parse_url (url_in TEXT) RETURNS url_verify.url AS $$
4
5 use Rose::URI;
6
7 my $url_in = shift;
8 my $url = Rose::URI->new($url_in);
9
10 my %parts = map { $_ => $url->$_ } qw/scheme username password host port path query fragment/;
11
12 $parts{full_url} = $url_in;
13 ($parts{domain} = $parts{host}) =~ s/^[^.]+\.//;
14 ($parts{tld} = $parts{domain}) =~ s/(?:[^.]+\.)+//;
15 ($parts{page} = $parts{path}) =~ s#(?:[^/]*/)+##;
16
17 return \%parts;
18
19 $$ LANGUAGE PLPERLU;
20
21 CREATE OR REPLACE FUNCTION url_verify.ingest_url () RETURNS TRIGGER AS $$
22 DECLARE
23     tmp_row url_verify.url%ROWTYPE;
24 BEGIN
25     SELECT * INTO tmp_row FROM url_verify.parse_url(NEW.full_url);
26
27     NEW.scheme          := tmp_row.scheme;
28     NEW.username        := tmp_row.username;
29     NEW.password        := tmp_row.password;
30     NEW.host            := tmp_row.host;
31     NEW.domain          := tmp_row.domain;
32     NEW.tld             := tmp_row.tld;
33     NEW.port            := tmp_row.port;
34     NEW.path            := tmp_row.path;
35     NEW.page            := tmp_row.page;
36     NEW.query           := tmp_row.query;
37     NEW.fragment        := tmp_row.fragment;
38
39     RETURN NEW;
40 END;
41 $$ LANGUAGE PLPGSQL;
42
43 CREATE TRIGGER ingest_url_tgr
44     BEFORE INSERT ON url_verify.url
45     FOR EACH ROW EXECUTE PROCEDURE url_verify.ingest_url(); 
46
47
48 COMMIT;
49