09a01deb52 2009-10-01 1: CREATE PROCEDURAL LANGUAGE plpgsql;
09a01deb52 2009-10-01 2:
e0ecab03f9 2009-10-07 3: -- general array sorting functions
e0ecab03f9 2009-10-07 4: -- sorts array
67e762b39b 2009-10-14 5: CREATE or replace FUNCTION sort(original anyarray) RETURNS anyarray
e0ecab03f9 2009-10-07 6: LANGUAGE sql IMMUTABLE STRICT
e0ecab03f9 2009-10-07 7: AS $_$
e0ecab03f9 2009-10-07 8: select array_agg(item) as result from (select unnest($1) as item order by item) a;
e0ecab03f9 2009-10-07 9: $_$;
e0ecab03f9 2009-10-07 10:
e0ecab03f9 2009-10-07 11: -- sorts array and removes duplicates
67e762b39b 2009-10-14 12: CREATE or replace FUNCTION usort(original anyarray) RETURNS anyarray
e0ecab03f9 2009-10-07 13: LANGUAGE sql IMMUTABLE STRICT
e0ecab03f9 2009-10-07 14: AS $_$
e0ecab03f9 2009-10-07 15: select array_agg(item) as result from (select distinct unnest($1) as item order by item) a;
e0ecab03f9 2009-10-07 16: $_$;
67e762b39b 2009-10-14 17:
67e762b39b 2009-10-14 18: -- this functions returns id of site
67e762b39b 2009-10-14 19: create or replace function get_site(my_site text[]) returns integer
67e762b39b 2009-10-14 20: language plpgsql strict
67e762b39b 2009-10-14 21: as $$
67e762b39b 2009-10-14 22: declare
67e762b39b 2009-10-14 23: site_id integer;
67e762b39b 2009-10-14 24: begin
67e762b39b 2009-10-14 25: select id_site from site where my_site = site into site_id;
67e762b39b 2009-10-14 26: if not found then
67e762b39b 2009-10-14 27: insert into site (site) values (my_site);
67e762b39b 2009-10-14 28: select id_site from site where my_site = site into site_id;
67e762b39b 2009-10-14 29: end if;
67e762b39b 2009-10-14 30: return site_id;
67e762b39b 2009-10-14 31: end;
67e762b39b 2009-10-14 32: $$;
67e762b39b 2009-10-14 33:
67e762b39b 2009-10-14 34: create or replace function get_site(domain text) returns integer
67e762b39b 2009-10-14 35: language sql immutable strict
67e762b39b 2009-10-14 36: as $$
67e762b39b 2009-10-14 37: select get_site(tripdomain($1)) as result;
67e762b39b 2009-10-14 38: $$;
c90fda69e1 2009-10-12 39:
c90fda69e1 2009-10-12 40: -- this function adds tag to domain
c90fda69e1 2009-10-12 41: CREATE or replace FUNCTION mark(domain text, new_tag text) RETURNS void
c90fda69e1 2009-10-12 42: LANGUAGE sql immutable STRICT
c90fda69e1 2009-10-12 43: AS $$
c90fda69e1 2009-10-12 44: select mark(get_site($1), $2) as result;
e0ecab03f9 2009-10-07 45: $$;
ddb0e69dd7 2009-10-02 46:
ddb0e69dd7 2009-10-02 47: -- this function adds tag to site by site id
2326f3bb9a 2009-10-08 48: CREATE or replace FUNCTION mark(my_id_site integer, new_tag text) RETURNS void
ddb0e69dd7 2009-10-02 49: LANGUAGE plpgsql STRICT
ddb0e69dd7 2009-10-02 50: AS $$
ddb0e69dd7 2009-10-02 51: declare
ddb0e69dd7 2009-10-02 52: -- maybe check should be added to make sure supplied site id really exists
ddb0e69dd7 2009-10-02 53: my_tag text[];
2326f3bb9a 2009-10-08 54: my_tag_id integer;
ddb0e69dd7 2009-10-02 55: begin
ddb0e69dd7 2009-10-02 56: -- selecting tags site already have and adding new tag to them
ddb0e69dd7 2009-10-02 57: -- note that tags should be sorted to eliminate permutations
2326f3bb9a 2009-10-08 58: select coalesce(tag, '{}'::text[]) from urls natural left join tag where id_site = my_id_site into my_tag;
ddb0e69dd7 2009-10-02 59: if not found then
ddb0e69dd7 2009-10-02 60: -- no records found - creating new tag
2326f3bb9a 2009-10-08 61: insert into urls (id_site, id_tag) values (my_id_site, get_tag(array[new_tag]));
ddb0e69dd7 2009-10-02 62: else
ddb0e69dd7 2009-10-02 63: -- joining tags
e0ecab03f9 2009-10-07 64: select usort(my_tag || array[new_tag]) into my_tag;
2326f3bb9a 2009-10-08 65: -- updating existing record
2326f3bb9a 2009-10-08 66: update urls set id_tag = get_tag(my_tag || array[new_tag]) where id_site = my_id_site;
ddb0e69dd7 2009-10-02 67: end if;
2326f3bb9a 2009-10-08 68: end;
2326f3bb9a 2009-10-08 69: $$;
ddb0e69dd7 2009-10-02 70:
2326f3bb9a 2009-10-08 71: -- this function returns id of tag array
2326f3bb9a 2009-10-08 72: create or replace function get_tag(my_tag text[]) returns integer
2326f3bb9a 2009-10-08 73: language plpgsql strict
2326f3bb9a 2009-10-08 74: as $$
2326f3bb9a 2009-10-08 75: declare
2326f3bb9a 2009-10-08 76: tag_id integer;
2326f3bb9a 2009-10-08 77: begin
2326f3bb9a 2009-10-08 78: select id_tag from tag where usort(my_tag) = tag into tag_id;
ddb0e69dd7 2009-10-02 79: if not found then
2326f3bb9a 2009-10-08 80: insert into tag (tag) values (usort(my_tag));
2326f3bb9a 2009-10-08 81: select id_tag from tag where usort(my_tag) = tag into tag_id;
ddb0e69dd7 2009-10-02 82: end if;
2326f3bb9a 2009-10-08 83: return tag_id;
e0ecab03f9 2009-10-07 84: end;
e0ecab03f9 2009-10-07 85: $$;
ddb0e69dd7 2009-10-02 86:
ddb0e69dd7 2009-10-02 87: -- transforms domain into ordered array for indexing
67e762b39b 2009-10-14 88: CREATE or replace FUNCTION tripdomain(url text) RETURNS text[]
ddb0e69dd7 2009-10-02 89: LANGUAGE plpgsql IMMUTABLE STRICT
ddb0e69dd7 2009-10-02 90: AS $_$
ddb0e69dd7 2009-10-02 91: declare
ddb0e69dd7 2009-10-02 92: result text[];
ddb0e69dd7 2009-10-02 93: splitted text[];
ddb0e69dd7 2009-10-02 94: x integer;
ddb0e69dd7 2009-10-02 95: length integer;
ddb0e69dd7 2009-10-02 96: begin
ddb0e69dd7 2009-10-02 97: splitted := string_to_array($1, '.');
ddb0e69dd7 2009-10-02 98: length := array_length(splitted, 1);
ddb0e69dd7 2009-10-02 99: x := 1;
ddb0e69dd7 2009-10-02 100: loop
ddb0e69dd7 2009-10-02 101: exit when splitted[x] is null;
ddb0e69dd7 2009-10-02 102: result[x] := splitted[x] || ':' || length - x;
ddb0e69dd7 2009-10-02 103: x := x + 1;
ddb0e69dd7 2009-10-02 104: end loop;
ddb0e69dd7 2009-10-02 105: return result;
ddb0e69dd7 2009-10-02 106: end;$_$;
ddb0e69dd7 2009-10-02 107:
ddb0e69dd7 2009-10-02 108: -- transforms ordered array into domain
67e762b39b 2009-10-14 109: create or replace function untrip(site text[]) returns text
ddb0e69dd7 2009-10-02 110: language plpgsql immutable strict
ddb0e69dd7 2009-10-02 111: as $_$
ddb0e69dd7 2009-10-02 112: declare
ddb0e69dd7 2009-10-02 113: x integer;
ddb0e69dd7 2009-10-02 114: splitted text[];
ddb0e69dd7 2009-10-02 115: pair text[];
ddb0e69dd7 2009-10-02 116: begin
ddb0e69dd7 2009-10-02 117: x := array_length(site, 1);
ddb0e69dd7 2009-10-02 118: loop
ddb0e69dd7 2009-10-02 119: exit when site[x] is null;
ddb0e69dd7 2009-10-02 120: pair := string_to_array(site[x], ':');
ddb0e69dd7 2009-10-02 121: splitted[0 - pair[2]::integer] := pair[1];
ddb0e69dd7 2009-10-02 122: x := x - 1;
ddb0e69dd7 2009-10-02 123: end loop;
ddb0e69dd7 2009-10-02 124: return array_to_string(splitted, '.');
09a01deb52 2009-10-01 125: end;
09a01deb52 2009-10-01 126: $_$;
09a01deb52 2009-10-01 127:
ddb0e69dd7 2009-10-02 128: -- table to hold all rules
ddb0e69dd7 2009-10-02 129: CREATE TABLE rules (
ddb0e69dd7 2009-10-02 130: netmask cidr NOT NULL,
ddb0e69dd7 2009-10-02 131: redirect_url text DEFAULT 'about::blank'::text NOT NULL,
ddb0e69dd7 2009-10-02 132: from_weekday smallint DEFAULT 0 NOT NULL,
ddb0e69dd7 2009-10-02 133: to_weekday smallint DEFAULT 6 NOT NULL,
ddb0e69dd7 2009-10-02 134: from_time time without time zone DEFAULT '00:00:00'::time without time zone NOT NULL,
ddb0e69dd7 2009-10-02 135: to_time time without time zone DEFAULT '23:59:59'::time without time zone NOT NULL,
ddb0e69dd7 2009-10-02 136: id_tag smallint NOT NULL
ddb0e69dd7 2009-10-02 137: );
ddb0e69dd7 2009-10-02 138:
ddb0e69dd7 2009-10-02 139: ALTER TABLE ONLY rules
ddb0e69dd7 2009-10-02 140: ADD CONSTRAINT rules_pkey PRIMARY KEY (netmask);
ddb0e69dd7 2009-10-02 141:
ddb0e69dd7 2009-10-02 142: -- table to hold site arrays
ddb0e69dd7 2009-10-02 143: CREATE TABLE site (
ddb0e69dd7 2009-10-02 144: id_site serial,
4b22e25f24 2009-10-07 145: site text[] NOT NULL
ddb0e69dd7 2009-10-02 146: );
ddb0e69dd7 2009-10-02 147:
ddb0e69dd7 2009-10-02 148: ALTER TABLE ONLY site
33e72616c9 2009-10-09 149: ADD CONSTRAINT site_pkey PRIMARY KEY (id_site);
ddb0e69dd7 2009-10-02 150:
33e72616c9 2009-10-09 151: CREATE UNIQUE INDEX site_u ON site (usort(site));
ddb0e69dd7 2009-10-02 152:
e0ecab03f9 2009-10-07 153: CREATE INDEX site_g ON site USING gin (site);
ddb0e69dd7 2009-10-02 154:
ddb0e69dd7 2009-10-02 155: -- table to hold tag combinations
ddb0e69dd7 2009-10-02 156: CREATE TABLE tag (
ddb0e69dd7 2009-10-02 157: id_tag serial,
ddb0e69dd7 2009-10-02 158: tag text[] NOT NULL
ddb0e69dd7 2009-10-02 159: );
ddb0e69dd7 2009-10-02 160:
ddb0e69dd7 2009-10-02 161: ALTER TABLE ONLY tag
33e72616c9 2009-10-09 162: ADD CONSTRAINT tag_pkey PRIMARY KEY (id_tag);
e0ecab03f9 2009-10-07 163:
33e72616c9 2009-10-09 164: CREATE UNIQUE INDEX tag_u ON tag (usort(tag));
09a01deb52 2009-10-01 165:
09a01deb52 2009-10-01 166: CREATE INDEX tag_g ON tag USING gin (tag);
09a01deb52 2009-10-01 167:
ddb0e69dd7 2009-10-02 168: -- table to hold tag - site links
ddb0e69dd7 2009-10-02 169: CREATE TABLE urls (
ddb0e69dd7 2009-10-02 170: date_added timestamp without time zone DEFAULT ('now'::text)::timestamp(0) without time zone NOT NULL,
ddb0e69dd7 2009-10-02 171: id_site smallint NOT NULL,
7d9c268669 2009-10-02 172: id_tag smallint NOT NULL,
7d9c268669 2009-10-02 173: regex text
ddb0e69dd7 2009-10-02 174: );
ddb0e69dd7 2009-10-02 175:
ddb0e69dd7 2009-10-02 176: ALTER TABLE ONLY urls
ddb0e69dd7 2009-10-02 177: ADD CONSTRAINT urls_pkey PRIMARY KEY (date_added);
09a01deb52 2009-10-01 178:
09a01deb52 2009-10-01 179: CREATE UNIQUE INDEX urls_id_site ON urls USING btree (id_site);
4b22e25f24 2009-10-07 180:
4b22e25f24 2009-10-07 181: CREATE UNIQUE INDEX urls_id_tag ON urls USING btree (id_tag);
7d9c268669 2009-10-02 182:
ddb0e69dd7 2009-10-02 183: -- rule to join all tables into one to simplify access
ddb0e69dd7 2009-10-02 184: -- automaticall uses current day and time data
ddb0e69dd7 2009-10-02 185: CREATE VIEW site_rule AS
7d9c268669 2009-10-02 186: SELECT a.redirect_url, a.netmask, b.site, b.regexp
ddb0e69dd7 2009-10-02 187: FROM ((
ddb0e69dd7 2009-10-02 188: SELECT rules.redirect_url, tag.tag AS rule_tag, rules.netmask
7d9c268669 2009-10-02 189: FROM rules NATURAL JOIN tag
7d9c268669 2009-10-02 190: WHERE ('now'::text)::time without time zone >= rules.from_time
7d9c268669 2009-10-02 191: AND ('now'::text)::time without time zone <= rules.to_time
7d9c268669 2009-10-02 192: AND date_part('dow'::text, now()) >= (rules.from_weekday)::double precision
7d9c268669 2009-10-02 193: AND date_part('dow'::text, now()) <= (rules.to_weekday)::double precision
ddb0e69dd7 2009-10-02 194: ) a JOIN (
7d9c268669 2009-10-02 195: SELECT site.site, tag.tag AS url_tag, regexp
7d9c268669 2009-10-02 196: FROM urls NATURAL JOIN tag NATURAL JOIN site
7d9c268669 2009-10-02 197: ) b ON (b.url_tag && a.rule_tag));