| 1 |
7844
|
aaronmk
|
CREATE OR REPLACE FUNCTION score_ok(score double precision)
|
| 2 |
|
|
RETURNS boolean AS
|
| 3 |
|
|
$BODY$
|
| 4 |
|
|
SELECT $1 >= 0.8
|
| 5 |
|
|
$BODY$
|
| 6 |
|
|
LANGUAGE sql VOLATILE
|
| 7 |
|
|
COST 100;
|
| 8 |
|
|
|
| 9 |
9756
|
aaronmk
|
/* IMPORTANT: when changing this table's schema, you must regenerate data.sql:
|
| 10 |
|
|
$ <this_file>/../test_taxonomic_names/test_scrub
|
| 11 |
|
|
*/
|
| 12 |
5183
|
aaronmk
|
CREATE TABLE tnrs
|
| 13 |
5110
|
aaronmk
|
(
|
| 14 |
9905
|
aaronmk
|
"Time_submitted" timestamp with time zone NOT NULL DEFAULT now(),
|
| 15 |
9908
|
aaronmk
|
"Name_number" integer NOT NULL,
|
| 16 |
5110
|
aaronmk
|
"Name_submitted" text NOT NULL,
|
| 17 |
9913
|
aaronmk
|
"Overall_score" double precision,
|
| 18 |
5110
|
aaronmk
|
"Name_matched" text,
|
| 19 |
|
|
"Name_matched_rank" text,
|
| 20 |
9913
|
aaronmk
|
"Name_score" double precision,
|
| 21 |
5110
|
aaronmk
|
"Name_matched_author" text,
|
| 22 |
|
|
"Name_matched_url" text,
|
| 23 |
|
|
"Author_matched" text,
|
| 24 |
9913
|
aaronmk
|
"Author_score" double precision,
|
| 25 |
5110
|
aaronmk
|
"Family_matched" text,
|
| 26 |
9913
|
aaronmk
|
"Family_score" double precision,
|
| 27 |
9529
|
aaronmk
|
"Name_matched_accepted_family" text,
|
| 28 |
5110
|
aaronmk
|
"Genus_matched" text,
|
| 29 |
9913
|
aaronmk
|
"Genus_score" double precision,
|
| 30 |
5110
|
aaronmk
|
"Specific_epithet_matched" text,
|
| 31 |
9913
|
aaronmk
|
"Specific_epithet_score" double precision,
|
| 32 |
5110
|
aaronmk
|
"Infraspecific_rank" text,
|
| 33 |
|
|
"Infraspecific_epithet_matched" text,
|
| 34 |
9913
|
aaronmk
|
"Infraspecific_epithet_score" double precision,
|
| 35 |
5110
|
aaronmk
|
"Infraspecific_rank_2" text,
|
| 36 |
|
|
"Infraspecific_epithet_2_matched" text,
|
| 37 |
9913
|
aaronmk
|
"Infraspecific_epithet_2_score" double precision,
|
| 38 |
5110
|
aaronmk
|
"Annotations" text,
|
| 39 |
|
|
"Unmatched_terms" text,
|
| 40 |
|
|
"Taxonomic_status" text,
|
| 41 |
|
|
"Accepted_name" text,
|
| 42 |
|
|
"Accepted_name_author" text,
|
| 43 |
|
|
"Accepted_name_rank" text,
|
| 44 |
|
|
"Accepted_name_url" text,
|
| 45 |
9762
|
aaronmk
|
"Accepted_name_species" text,
|
| 46 |
|
|
"Accepted_name_family" text,
|
| 47 |
5110
|
aaronmk
|
"Selected" text,
|
| 48 |
|
|
"Source" text,
|
| 49 |
|
|
"Warnings" text,
|
| 50 |
|
|
"Accepted_name_lsid" text,
|
| 51 |
7133
|
aaronmk
|
"Accepted_scientific_name" text,
|
| 52 |
7293
|
aaronmk
|
"Max_score" double precision,
|
| 53 |
9972
|
aaronmk
|
"Is_homonym" boolean,
|
| 54 |
9763
|
aaronmk
|
"Is_plant" boolean,
|
| 55 |
9907
|
aaronmk
|
CONSTRAINT tnrs_pkey PRIMARY KEY ("Time_submitted" , "Name_number" ),
|
| 56 |
9906
|
aaronmk
|
CONSTRAINT "tnrs_Name_submitted_key" UNIQUE ("Name_submitted" )
|
| 57 |
5110
|
aaronmk
|
)
|
| 58 |
|
|
WITH (
|
| 59 |
|
|
OIDS=FALSE
|
| 60 |
|
|
);
|
| 61 |
7132
|
aaronmk
|
|
| 62 |
7844
|
aaronmk
|
CREATE UNIQUE INDEX tnrs_score_ok
|
| 63 |
|
|
ON tnrs
|
| 64 |
|
|
USING btree
|
| 65 |
|
|
("Name_submitted" )
|
| 66 |
|
|
WHERE score_ok("Max_score");
|
| 67 |
|
|
|
| 68 |
9765
|
aaronmk
|
/* IMPORTANT: when changing this function, you must regenerate the derived cols:
|
| 69 |
9974
|
aaronmk
|
UPDATE "TNRS".tnrs SET "Name_submitted" = "Name_submitted"
|
| 70 |
9767
|
aaronmk
|
runtime: 17 min ("4992166 rows affected, 1019907 ms execution time")
|
| 71 |
9765
|
aaronmk
|
*/
|
| 72 |
9512
|
aaronmk
|
CREATE OR REPLACE FUNCTION tnrs_populate_fields()
|
| 73 |
7134
|
aaronmk
|
RETURNS trigger AS
|
| 74 |
|
|
$BODY$
|
| 75 |
9763
|
aaronmk
|
DECLARE
|
| 76 |
|
|
"Specific_epithet_is_plant" boolean :=
|
| 77 |
|
|
(CASE
|
| 78 |
|
|
WHEN new."Infraspecific_epithet_matched" IS NOT NULL
|
| 79 |
|
|
OR new."Infraspecific_epithet_2_matched" IS NOT NULL
|
| 80 |
9914
|
aaronmk
|
OR new."Specific_epithet_score" >= 0.9 -- fuzzy match
|
| 81 |
9763
|
aaronmk
|
THEN true
|
| 82 |
|
|
ELSE NULL -- ambiguous
|
| 83 |
|
|
END);
|
| 84 |
9972
|
aaronmk
|
family_is_homonym boolean = EXISTS(SELECT * FROM "IRMNG".family_homonym_epithet WHERE "taxonNameOrEpithet" = new."Family_matched");
|
| 85 |
|
|
genus_is_homonym boolean = EXISTS(SELECT * FROM "IRMNG".genus_homonym_epithet WHERE "taxonNameOrEpithet" = new."Genus_matched");
|
| 86 |
7134
|
aaronmk
|
BEGIN
|
| 87 |
7848
|
aaronmk
|
new."Accepted_scientific_name" = NULLIF(concat_ws(' '
|
| 88 |
9762
|
aaronmk
|
, NULLIF(NULLIF(new."Accepted_name_family", 'Unknown'), new."Accepted_name")
|
| 89 |
7134
|
aaronmk
|
, new."Accepted_name"
|
| 90 |
|
|
, new."Accepted_name_author"
|
| 91 |
7848
|
aaronmk
|
), '');
|
| 92 |
7293
|
aaronmk
|
new."Max_score" = GREATEST(
|
| 93 |
9914
|
aaronmk
|
new."Overall_score"
|
| 94 |
|
|
, new."Family_score"
|
| 95 |
|
|
, new."Genus_score"
|
| 96 |
|
|
, new."Specific_epithet_score"
|
| 97 |
7293
|
aaronmk
|
);
|
| 98 |
9972
|
aaronmk
|
new."Is_homonym" = (CASE
|
| 99 |
|
|
WHEN new."Author_matched" IS NOT NULL THEN false -- author disambiguates
|
| 100 |
|
|
ELSE family_is_homonym OR genus_is_homonym
|
| 101 |
|
|
END);
|
| 102 |
9763
|
aaronmk
|
new."Is_plant" = (CASE
|
| 103 |
9973
|
aaronmk
|
WHEN new."Family_score" = 1 AND NOT family_is_homonym -- exact match
|
| 104 |
|
|
THEN true
|
| 105 |
|
|
ELSE -- consider genus
|
| 106 |
9763
|
aaronmk
|
(CASE
|
| 107 |
9914
|
aaronmk
|
WHEN new."Genus_score" = 1 -- exact match
|
| 108 |
9973
|
aaronmk
|
THEN
|
| 109 |
|
|
(CASE
|
| 110 |
|
|
WHEN NOT genus_is_homonym THEN true
|
| 111 |
|
|
ELSE "Specific_epithet_is_plant"
|
| 112 |
|
|
END)
|
| 113 |
9914
|
aaronmk
|
WHEN new."Genus_score" >= 0.85 -- fuzzy match
|
| 114 |
9763
|
aaronmk
|
THEN "Specific_epithet_is_plant"
|
| 115 |
|
|
ELSE NULL -- ambiguous
|
| 116 |
|
|
END)
|
| 117 |
|
|
END);
|
| 118 |
7134
|
aaronmk
|
|
| 119 |
|
|
RETURN new;
|
| 120 |
|
|
END;
|
| 121 |
|
|
$BODY$
|
| 122 |
|
|
LANGUAGE plpgsql VOLATILE
|
| 123 |
|
|
COST 100;
|
| 124 |
|
|
|
| 125 |
9512
|
aaronmk
|
CREATE TRIGGER tnrs_populate_fields
|
| 126 |
7134
|
aaronmk
|
BEFORE INSERT OR UPDATE
|
| 127 |
|
|
ON tnrs
|
| 128 |
|
|
FOR EACH ROW
|
| 129 |
9512
|
aaronmk
|
EXECUTE PROCEDURE tnrs_populate_fields();
|
| 130 |
7251
|
aaronmk
|
|
| 131 |
|
|
|
| 132 |
9759
|
aaronmk
|
CREATE OR REPLACE VIEW "MatchedTaxon" AS
|
| 133 |
7823
|
aaronmk
|
SELECT
|
| 134 |
|
|
"Time_submitted" AS "*Name_matched.Time_submitted"
|
| 135 |
7830
|
aaronmk
|
, "Name_submitted" AS "concatenatedScientificName"
|
| 136 |
7829
|
aaronmk
|
, "Name_matched" AS "matchedTaxonName"
|
| 137 |
7823
|
aaronmk
|
, "Name_matched_rank" AS "matchedTaxonRank"
|
| 138 |
|
|
, "Name_score" AS "*Name_matched.Name_score"
|
| 139 |
|
|
, "Name_matched_author" AS "matchedScientificNameAuthorship"
|
| 140 |
|
|
, "Name_matched_url" AS "matchedScientificNameID"
|
| 141 |
|
|
, "Author_score" AS "*Name_matched.Author_score"
|
| 142 |
|
|
, "Family_score" AS "matchedFamilyConfidence_fraction"
|
| 143 |
9762
|
aaronmk
|
, COALESCE("Name_matched_accepted_family", "Accepted_name_family") AS "matchedFamily"
|
| 144 |
7831
|
aaronmk
|
, "Genus_matched" AS "matchedGenus"
|
| 145 |
7823
|
aaronmk
|
, "Genus_score" AS "matchedGenusConfidence_fraction"
|
| 146 |
|
|
, "Specific_epithet_matched" AS "matchedSpecificEpithet"
|
| 147 |
|
|
, "Specific_epithet_score" AS "matchedSpeciesConfidence_fraction"
|
| 148 |
|
|
, "Infraspecific_epithet_matched" AS "matchedInfraspecificEpithet"
|
| 149 |
|
|
, "Infraspecific_epithet_score" AS "*Name_matched.Infraspecific_epithet_score"
|
| 150 |
|
|
, "Annotations" AS "identificationQualifier"
|
| 151 |
|
|
, "Unmatched_terms" AS "morphospeciesSuffix"
|
| 152 |
7833
|
aaronmk
|
, "Taxonomic_status" AS "taxonomicStatus"
|
| 153 |
7823
|
aaronmk
|
, "Accepted_name" AS "acceptedTaxonName"
|
| 154 |
|
|
, "Accepted_name_author" AS "acceptedScientificNameAuthorship"
|
| 155 |
|
|
, "Accepted_name_rank" AS "acceptedTaxonRank"
|
| 156 |
|
|
, "Accepted_name_url" AS "acceptedScientificNameID"
|
| 157 |
9762
|
aaronmk
|
, "Accepted_name_species" AS "*Name_matched.Accepted_name_species"
|
| 158 |
|
|
, "Accepted_name_family" AS "acceptedFamily"
|
| 159 |
7823
|
aaronmk
|
, "Selected" AS "*Name_matched.Selected"
|
| 160 |
7828
|
aaronmk
|
, "Source" AS "*Name_matched.Source"
|
| 161 |
7823
|
aaronmk
|
, "Warnings" AS "*Name_matched.Warnings"
|
| 162 |
|
|
, "Accepted_name_lsid" AS "*Name_matched.Accepted_name_lsid"
|
| 163 |
|
|
, "Accepted_scientific_name" AS "acceptedScientificName"
|
| 164 |
|
|
, "Max_score" AS "matchedTaxonConfidence_fraction"
|
| 165 |
9616
|
aaronmk
|
FROM tnrs
|
| 166 |
|
|
;
|
| 167 |
7823
|
aaronmk
|
|
| 168 |
9759
|
aaronmk
|
CREATE OR REPLACE VIEW "ValidMatchedTaxon" AS
|
| 169 |
|
|
SELECT *
|
| 170 |
|
|
FROM "MatchedTaxon"
|
| 171 |
|
|
WHERE score_ok("matchedTaxonConfidence_fraction")
|
| 172 |
|
|
;
|
| 173 |
|
|
|
| 174 |
7823
|
aaronmk
|
CREATE OR REPLACE VIEW "AcceptedTaxon" AS
|
| 175 |
|
|
SELECT
|
| 176 |
|
|
"Time_submitted" AS "*Accepted_name.Time_submitted"
|
| 177 |
|
|
, "Name_submitted" AS "acceptedScientificName"
|
| 178 |
|
|
, "Genus_matched" AS "acceptedGenus"
|
| 179 |
|
|
, "Specific_epithet_matched" AS "acceptedSpecificEpithet"
|
| 180 |
|
|
, "Infraspecific_epithet_matched" AS "acceptedInfraspecificEpithet"
|
| 181 |
|
|
, "Annotations" AS "*Accepted_name.Annotations"
|
| 182 |
|
|
, "Taxonomic_status" AS "acceptedTaxonomicStatus"
|
| 183 |
|
|
, "Selected" AS "*Accepted_name.Selected"
|
| 184 |
|
|
, "Source" AS "*Accepted_name.Source"
|
| 185 |
|
|
, "Warnings" AS "*Accepted_name.Warnings"
|
| 186 |
|
|
, "Accepted_name_lsid" AS "*Accepted_name.Accepted_name_lsid"
|
| 187 |
9760
|
aaronmk
|
FROM tnrs
|
| 188 |
|
|
;
|
| 189 |
7823
|
aaronmk
|
|
| 190 |
|
|
CREATE OR REPLACE VIEW "ScrubbedTaxon" AS
|
| 191 |
|
|
SELECT *
|
| 192 |
9758
|
aaronmk
|
FROM "ValidMatchedTaxon"
|
| 193 |
7823
|
aaronmk
|
NATURAL LEFT JOIN "AcceptedTaxon"
|
| 194 |
|
|
;
|
