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: $_$;
ddb0e69dd7 2009-10-02 17:
ddb0e69dd7 2009-10-02 18: -- transforms domain into ordered array for indexing
67e762b39b 2009-10-14 19: CREATE or replace FUNCTION tripdomain(url text) RETURNS text[]
ddb0e69dd7 2009-10-02 20: LANGUAGE plpgsql IMMUTABLE STRICT
ddb0e69dd7 2009-10-02 21: AS $_$
ddb0e69dd7 2009-10-02 22: declare
ddb0e69dd7 2009-10-02 23: result text[];
ddb0e69dd7 2009-10-02 24: splitted text[];
ddb0e69dd7 2009-10-02 25: x integer;
ddb0e69dd7 2009-10-02 26: length integer;
ddb0e69dd7 2009-10-02 27: begin
ddb0e69dd7 2009-10-02 28: splitted := string_to_array($1, '.');
ddb0e69dd7 2009-10-02 29: length := array_length(splitted, 1);
ddb0e69dd7 2009-10-02 30: x := 1;
ddb0e69dd7 2009-10-02 31: loop
ddb0e69dd7 2009-10-02 32: exit when splitted[x] is null;
ddb0e69dd7 2009-10-02 33: result[x] := splitted[x] || ':' || length - x;
ddb0e69dd7 2009-10-02 34: x := x + 1;
ddb0e69dd7 2009-10-02 35: end loop;
ddb0e69dd7 2009-10-02 36: return result;
ddb0e69dd7 2009-10-02 37: end;$_$;
ddb0e69dd7 2009-10-02 38:
ddb0e69dd7 2009-10-02 39: -- transforms ordered array into domain
67e762b39b 2009-10-14 40: create or replace function untrip(site text[]) returns text
ddb0e69dd7 2009-10-02 41: language plpgsql immutable strict
ddb0e69dd7 2009-10-02 42: as $_$
ddb0e69dd7 2009-10-02 43: declare
ddb0e69dd7 2009-10-02 44: x integer;
ddb0e69dd7 2009-10-02 45: splitted text[];
ddb0e69dd7 2009-10-02 46: pair text[];
ddb0e69dd7 2009-10-02 47: begin
ddb0e69dd7 2009-10-02 48: x := array_length(site, 1);
ddb0e69dd7 2009-10-02 49: loop
ddb0e69dd7 2009-10-02 50: exit when site[x] is null;
ddb0e69dd7 2009-10-02 51: pair := string_to_array(site[x], ':');
ddb0e69dd7 2009-10-02 52: splitted[0 - pair[2]::integer] := pair[1];
ddb0e69dd7 2009-10-02 53: x := x - 1;
ddb0e69dd7 2009-10-02 54: end loop;
ddb0e69dd7 2009-10-02 55: return array_to_string(splitted, '.');
09a01deb52 2009-10-01 56: end;
09a01deb52 2009-10-01 57: $_$;
12c35e5674 2009-12-15 58:
12c35e5674 2009-12-15 59: -- this functions returns id of site
12c35e5674 2009-12-15 60: create or replace function get_site(my_site text[]) returns integer
12c35e5674 2009-12-15 61: language plpgsql strict
12c35e5674 2009-12-15 62: as $$
12c35e5674 2009-12-15 63: declare
12c35e5674 2009-12-15 64: site_id integer;
12c35e5674 2009-12-15 65: begin
12c35e5674 2009-12-15 66: select id_site from site where my_site = site into site_id;
12c35e5674 2009-12-15 67: if not found then
12c35e5674 2009-12-15 68: insert into site (site) values (my_site);
12c35e5674 2009-12-15 69: select id_site from site where my_site = site into site_id;
12c35e5674 2009-12-15 70: end if;
12c35e5674 2009-12-15 71: return site_id;
12c35e5674 2009-12-15 72: end;
12c35e5674 2009-12-15 73: $$;
12c35e5674 2009-12-15 74:
12c35e5674 2009-12-15 75: create or replace function get_site(domain text) returns integer
12c35e5674 2009-12-15 76: language sql immutable strict
12c35e5674 2009-12-15 77: as $$
12c35e5674 2009-12-15 78: select get_site(tripdomain($1)) as result;
12c35e5674 2009-12-15 79: $$;
12c35e5674 2009-12-15 80:
12c35e5674 2009-12-15 81: -- this function adds tag to site by site id
12c35e5674 2009-12-15 82: CREATE or replace FUNCTION mark(my_id_site integer, new_tag text) RETURNS integer
12c35e5674 2009-12-15 83: LANGUAGE plpgsql STRICT
12c35e5674 2009-12-15 84: AS $$
12c35e5674 2009-12-15 85: declare
12c35e5674 2009-12-15 86: -- maybe check should be added to make sure supplied site id really exists
12c35e5674 2009-12-15 87: my_tag text[];
12c35e5674 2009-12-15 88: my_tag_id integer;
12c35e5674 2009-12-15 89: begin
12c35e5674 2009-12-15 90: -- selecting tags site already have and adding new tag to them
12c35e5674 2009-12-15 91: -- note that tags should be sorted to eliminate permutations
12c35e5674 2009-12-15 92: select coalesce(tag, '{}'::text[]) from urls natural left join tag where id_site = my_id_site into my_tag;
12c35e5674 2009-12-15 93: if not found then
12c35e5674 2009-12-15 94: -- no records found - creating new tag
12c35e5674 2009-12-15 95: insert into urls (id_site, id_tag) values (my_id_site, get_tag(array[new_tag]));
12c35e5674 2009-12-15 96: else
12c35e5674 2009-12-15 97: -- joining tags
12c35e5674 2009-12-15 98: select usort(my_tag || array[new_tag]) into my_tag;
12c35e5674 2009-12-15 99: -- updating existing record
12c35e5674 2009-12-15 100: update urls set id_tag = get_tag(my_tag || array[new_tag]) where id_site = my_id_site;
12c35e5674 2009-12-15 101: end if;
12c35e5674 2009-12-15 102: return my_id_site;
12c35e5674 2009-12-15 103: end;
12c35e5674 2009-12-15 104: $$;
12c35e5674 2009-12-15 105:
12c35e5674 2009-12-15 106: -- this function adds tag to domain
12c35e5674 2009-12-15 107: CREATE or replace FUNCTION mark(domain text, new_tag text) RETURNS integer
12c35e5674 2009-12-15 108: LANGUAGE sql immutable STRICT
12c35e5674 2009-12-15 109: AS $$
12c35e5674 2009-12-15 110: select mark(get_site($1), $2) as result;
12c35e5674 2009-12-15 111: $$;
12c35e5674 2009-12-15 112:
12c35e5674 2009-12-15 113: -- this function returns id of tag array
12c35e5674 2009-12-15 114: create or replace function get_tag(my_tag text[]) returns integer
12c35e5674 2009-12-15 115: language plpgsql strict
12c35e5674 2009-12-15 116: as $$
12c35e5674 2009-12-15 117: declare
12c35e5674 2009-12-15 118: tag_id integer;
12c35e5674 2009-12-15 119: begin
12c35e5674 2009-12-15 120: select id_tag from tag where usort(my_tag) = tag into tag_id;
12c35e5674 2009-12-15 121: if not found then
12c35e5674 2009-12-15 122: insert into tag (tag) values (usort(my_tag));
12c35e5674 2009-12-15 123: select id_tag from tag where usort(my_tag) = tag into tag_id;
12c35e5674 2009-12-15 124: end if;
12c35e5674 2009-12-15 125: return tag_id;
12c35e5674 2009-12-15 126: end;
12c35e5674 2009-12-15 127: $$;
4b22e25f24 2009-10-07 128:
ddb0e69dd7 2009-10-02 129: -- table to hold all rules
09a01deb52 2009-10-01 130: CREATE TABLE rules (
ddb0e69dd7 2009-10-02 131: netmask cidr NOT NULL,
ddb0e69dd7 2009-10-02 132: redirect_url text DEFAULT 'about::blank'::text NOT NULL,
ddb0e69dd7 2009-10-02 133: from_weekday smallint DEFAULT 0 NOT NULL,
ddb0e69dd7 2009-10-02 134: to_weekday smallint DEFAULT 6 NOT NULL,
ddb0e69dd7 2009-10-02 135: from_time time without time zone DEFAULT '00:00:00'::time without time zone NOT NULL,
ddb0e69dd7 2009-10-02 136: to_time time without time zone DEFAULT '23:59:59'::time without time zone NOT NULL,
ddb0e69dd7 2009-10-02 137: id_tag smallint NOT NULL
ddb0e69dd7 2009-10-02 138: );
09a01deb52 2009-10-01 139:
09a01deb52 2009-10-01 140: ALTER TABLE ONLY rules
ddb0e69dd7 2009-10-02 141: ADD CONSTRAINT rules_pkey PRIMARY KEY (netmask);
ddb0e69dd7 2009-10-02 142:
ddb0e69dd7 2009-10-02 143: -- table to hold site arrays
ddb0e69dd7 2009-10-02 144: CREATE TABLE site (
ddb0e69dd7 2009-10-02 145: id_site serial,
4b22e25f24 2009-10-07 146: site text[] NOT NULL
ddb0e69dd7 2009-10-02 147: );
09a01deb52 2009-10-01 148:
09a01deb52 2009-10-01 149: ALTER TABLE ONLY site
33e72616c9 2009-10-09 150: ADD CONSTRAINT site_pkey PRIMARY KEY (id_site);
ddb0e69dd7 2009-10-02 151:
33e72616c9 2009-10-09 152: CREATE UNIQUE INDEX site_u ON site (usort(site));
ddb0e69dd7 2009-10-02 153:
e0ecab03f9 2009-10-07 154: CREATE INDEX site_g ON site USING gin (site);
ddb0e69dd7 2009-10-02 155:
ddb0e69dd7 2009-10-02 156: -- table to hold tag combinations
ddb0e69dd7 2009-10-02 157: CREATE TABLE tag (
ddb0e69dd7 2009-10-02 158: id_tag serial,
ddb0e69dd7 2009-10-02 159: tag text[] NOT NULL
ddb0e69dd7 2009-10-02 160: );
09a01deb52 2009-10-01 161:
09a01deb52 2009-10-01 162: ALTER TABLE ONLY tag
33e72616c9 2009-10-09 163: ADD CONSTRAINT tag_pkey PRIMARY KEY (id_tag);
e0ecab03f9 2009-10-07 164:
33e72616c9 2009-10-09 165: CREATE UNIQUE INDEX tag_u ON tag (usort(tag));
09a01deb52 2009-10-01 166:
09a01deb52 2009-10-01 167: CREATE INDEX tag_g ON tag USING gin (tag);
09a01deb52 2009-10-01 168:
ddb0e69dd7 2009-10-02 169: -- table to hold tag - site links
ddb0e69dd7 2009-10-02 170: CREATE TABLE urls (
ddb0e69dd7 2009-10-02 171: date_added timestamp without time zone DEFAULT ('now'::text)::timestamp(0) without time zone NOT NULL,
ddb0e69dd7 2009-10-02 172: id_site smallint NOT NULL,
7d9c268669 2009-10-02 173: id_tag smallint NOT NULL,
12c35e5674 2009-12-15 174: regexp text
ddb0e69dd7 2009-10-02 175: );
ddb0e69dd7 2009-10-02 176:
ddb0e69dd7 2009-10-02 177: ALTER TABLE ONLY urls
ddb0e69dd7 2009-10-02 178: ADD CONSTRAINT urls_pkey PRIMARY KEY (date_added);
09a01deb52 2009-10-01 179:
09a01deb52 2009-10-01 180: CREATE UNIQUE INDEX urls_id_site ON urls USING btree (id_site);
4b22e25f24 2009-10-07 181:
4b22e25f24 2009-10-07 182: CREATE UNIQUE INDEX urls_id_tag ON urls USING btree (id_tag);
7d9c268669 2009-10-02 183:
ddb0e69dd7 2009-10-02 184: -- rule to join all tables into one to simplify access
ddb0e69dd7 2009-10-02 185: -- automaticall uses current day and time data
ddb0e69dd7 2009-10-02 186: CREATE VIEW site_rule AS
7d9c268669 2009-10-02 187: SELECT a.redirect_url, a.netmask, b.site, b.regexp
ddb0e69dd7 2009-10-02 188: FROM ((
ddb0e69dd7 2009-10-02 189: SELECT rules.redirect_url, tag.tag AS rule_tag, rules.netmask
7d9c268669 2009-10-02 190: FROM rules NATURAL JOIN tag
7d9c268669 2009-10-02 191: WHERE ('now'::text)::time without time zone >= rules.from_time
7d9c268669 2009-10-02 192: AND ('now'::text)::time without time zone <= rules.to_time
7d9c268669 2009-10-02 193: AND date_part('dow'::text, now()) >= (rules.from_weekday)::double precision
7d9c268669 2009-10-02 194: AND date_part('dow'::text, now()) <= (rules.to_weekday)::double precision
ddb0e69dd7 2009-10-02 195: ) a JOIN (
7d9c268669 2009-10-02 196: SELECT site.site, tag.tag AS url_tag, regexp
7d9c268669 2009-10-02 197: FROM urls NATURAL JOIN tag NATURAL JOIN site
7d9c268669 2009-10-02 198: ) b ON (b.url_tag && a.rule_tag));