From 399a5b07ee0ce108779e9d8aaf3eefd5a838b2ef Mon Sep 17 00:00:00 2001 From: Lebbeous Fogle-Weekley Date: Tue, 24 Jan 2012 17:36:34 -0500 Subject: [PATCH] more efficient limit; don't autosuggest deleted bibs; use keyword config Signed-off-by: Lebbeous Fogle-Weekley --- .../src/perlmods/lib/OpenILS/WWW/AutoSuggest.pm | 6 +- Open-ILS/src/sql/Pg/030.schema.metabib.sql | 1 + .../sql/Pg/upgrade/YYYY.schema.bib_autosuggest.sql | 175 ++++++++++++++++++++- 3 files changed, 175 insertions(+), 7 deletions(-) diff --git a/Open-ILS/src/perlmods/lib/OpenILS/WWW/AutoSuggest.pm b/Open-ILS/src/perlmods/lib/OpenILS/WWW/AutoSuggest.pm index 950a306ef6..30082fd6c3 100644 --- a/Open-ILS/src/perlmods/lib/OpenILS/WWW/AutoSuggest.pm +++ b/Open-ILS/src/perlmods/lib/OpenILS/WWW/AutoSuggest.pm @@ -105,9 +105,9 @@ sub get_suggestions { prepare_for_tsquery($query), $search_class, $headline_opts, - $org_unit - ], - "limit" => $limit + $org_unit, + $limit + ] }); } diff --git a/Open-ILS/src/sql/Pg/030.schema.metabib.sql b/Open-ILS/src/sql/Pg/030.schema.metabib.sql index 96e3834468..d6acc28e7a 100644 --- a/Open-ILS/src/sql/Pg/030.schema.metabib.sql +++ b/Open-ILS/src/sql/Pg/030.schema.metabib.sql @@ -1040,6 +1040,7 @@ BEGIN DELETE FROM metabib.record_attr WHERE id = NEW.id; -- Kill the attrs hash, useless on deleted records DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible DELETE FROM biblio.peer_bib_copy_map WHERE peer_record = NEW.id; -- Separate any multi-homed items + DELETE FROM metabib.browse_entry_def_map WHERE source = NEW.id; -- Don't auto-suggest deleted bibs RETURN NEW; -- and we're done END IF; diff --git a/Open-ILS/src/sql/Pg/upgrade/YYYY.schema.bib_autosuggest.sql b/Open-ILS/src/sql/Pg/upgrade/YYYY.schema.bib_autosuggest.sql index aec97f59ca..091265ab9a 100644 --- a/Open-ILS/src/sql/Pg/upgrade/YYYY.schema.bib_autosuggest.sql +++ b/Open-ILS/src/sql/Pg/upgrade/YYYY.schema.bib_autosuggest.sql @@ -29,6 +29,171 @@ ALTER TABLE config.metabib_class ADD COLUMN bouyant BOOLEAN DEFAULT FALSE NOT NU ALTER TABLE config.metabib_class ADD COLUMN restrict BOOLEAN DEFAULT FALSE NOT NULL; ALTER TABLE config.metabib_field ADD COLUMN restrict BOOLEAN DEFAULT FALSE NOT NULL; +-- AFTER UPDATE OR INSERT trigger for biblio.record_entry +-- We're only touching it here to add a DELETE statement to the IF NEW.deleted +-- block. + +CREATE OR REPLACE FUNCTION biblio.indexing_ingest_or_delete () RETURNS TRIGGER AS $func$ +DECLARE + transformed_xml TEXT; + prev_xfrm TEXT; + normalizer RECORD; + xfrm config.xml_transform%ROWTYPE; + attr_value TEXT; + new_attrs HSTORE := ''::HSTORE; + attr_def config.record_attr_definition%ROWTYPE; +BEGIN + + IF NEW.deleted IS TRUE THEN -- If this bib is deleted + DELETE FROM metabib.metarecord_source_map WHERE source = NEW.id; -- Rid ourselves of the search-estimate-killing linkage + DELETE FROM metabib.record_attr WHERE id = NEW.id; -- Kill the attrs hash, useless on deleted records + DELETE FROM authority.bib_linking WHERE bib = NEW.id; -- Avoid updating fields in bibs that are no longer visible + DELETE FROM biblio.peer_bib_copy_map WHERE peer_record = NEW.id; -- Separate any multi-homed items + DELETE FROM metabib.browse_entry_def_map WHERE source = NEW.id; -- Don't auto-suggest deleted bibs + RETURN NEW; -- and we're done + END IF; + + IF TG_OP = 'UPDATE' THEN -- re-ingest? + PERFORM * FROM config.internal_flag WHERE name = 'ingest.reingest.force_on_same_marc' AND enabled; + + IF NOT FOUND AND OLD.marc = NEW.marc THEN -- don't do anything if the MARC didn't change + RETURN NEW; + END IF; + END IF; + + -- Record authority linking + PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_authority_linking' AND enabled; + IF NOT FOUND THEN + PERFORM biblio.map_authority_linking( NEW.id, NEW.marc ); + END IF; + + -- Flatten and insert the mfr data + PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_full_rec' AND enabled; + IF NOT FOUND THEN + PERFORM metabib.reingest_metabib_full_rec(NEW.id); + + -- Now we pull out attribute data, which is dependent on the mfr for all but XPath-based fields + PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_metabib_rec_descriptor' AND enabled; + IF NOT FOUND THEN + FOR attr_def IN SELECT * FROM config.record_attr_definition ORDER BY format LOOP + + IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection + SELECT ARRAY_TO_STRING(ARRAY_ACCUM(value), COALESCE(attr_def.joiner,' ')) INTO attr_value + FROM (SELECT * FROM metabib.full_rec ORDER BY tag, subfield) AS x + WHERE record = NEW.id + AND tag LIKE attr_def.tag + AND CASE + WHEN attr_def.sf_list IS NOT NULL + THEN POSITION(subfield IN attr_def.sf_list) > 0 + ELSE TRUE + END + GROUP BY tag + ORDER BY tag + LIMIT 1; + + ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field + attr_value := biblio.marc21_extract_fixed_field(NEW.id, attr_def.fixed_field); + + ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression + + SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format; + + -- See if we can skip the XSLT ... it's expensive + IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN + -- Can't skip the transform + IF xfrm.xslt <> '---' THEN + transformed_xml := oils_xslt_process(NEW.marc,xfrm.xslt); + ELSE + transformed_xml := NEW.marc; + END IF; + + prev_xfrm := xfrm.name; + END IF; + + IF xfrm.name IS NULL THEN + -- just grab the marcxml (empty) transform + SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1; + prev_xfrm := xfrm.name; + END IF; + + attr_value := oils_xpath_string(attr_def.xpath, transformed_xml, COALESCE(attr_def.joiner,' '), ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]); + + ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map + SELECT m.value INTO attr_value + FROM biblio.marc21_physical_characteristics(NEW.id) v + JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value) + WHERE v.subfield = attr_def.phys_char_sf + LIMIT 1; -- Just in case ... + + END IF; + + -- apply index normalizers to attr_value + FOR normalizer IN + SELECT n.func AS func, + n.param_count AS param_count, + m.params AS params + FROM config.index_normalizer n + JOIN config.record_attr_index_norm_map m ON (m.norm = n.id) + WHERE attr = attr_def.name + ORDER BY m.pos LOOP + EXECUTE 'SELECT ' || normalizer.func || '(' || + COALESCE( quote_literal( attr_value ), 'NULL' ) || + CASE + WHEN normalizer.param_count > 0 + THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'') + ELSE '' + END || + ')' INTO attr_value; + + END LOOP; + + -- Add the new value to the hstore + new_attrs := new_attrs || hstore( attr_def.name, attr_value ); + + END LOOP; + + IF TG_OP = 'INSERT' OR OLD.deleted THEN -- initial insert OR revivication + INSERT INTO metabib.record_attr (id, attrs) VALUES (NEW.id, new_attrs); + ELSE + UPDATE metabib.record_attr SET attrs = new_attrs WHERE id = NEW.id; + END IF; + + END IF; + END IF; + + -- Gather and insert the field entry data + PERFORM metabib.reingest_metabib_field_entries(NEW.id); + + -- Located URI magic + IF TG_OP = 'INSERT' THEN + PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled; + IF NOT FOUND THEN + PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor ); + END IF; + ELSE + PERFORM * FROM config.internal_flag WHERE name = 'ingest.disable_located_uri' AND enabled; + IF NOT FOUND THEN + PERFORM biblio.extract_located_uris( NEW.id, NEW.marc, NEW.editor ); + END IF; + END IF; + + -- (re)map metarecord-bib linking + IF TG_OP = 'INSERT' THEN -- if not deleted and performing an insert, check for the flag + PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_insert' AND enabled; + IF NOT FOUND THEN + PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint ); + END IF; + ELSE -- we're doing an update, and we're not deleted, remap + PERFORM * FROM config.internal_flag WHERE name = 'ingest.metarecord_mapping.skip_on_update' AND enabled; + IF NOT FOUND THEN + PERFORM metabib.remap_metarecord_for_bib( NEW.id, NEW.fingerprint ); + END IF; + END IF; + + RETURN NEW; +END; +$func$ LANGUAGE PLPGSQL; + CREATE OR REPLACE FUNCTION metabib.browse_normalize(facet_text TEXT, mapped_field INT) RETURNS TEXT AS $$ DECLARE normalizer RECORD; @@ -61,7 +226,7 @@ $$ LANGUAGE PLPGSQL; CREATE TRIGGER metabib_browse_entry_fti_trigger BEFORE INSERT OR UPDATE ON metabib.browse_entry - FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('default'); + FOR EACH ROW EXECUTE PROCEDURE oils_tsearch2('keyword'); DROP FUNCTION biblio.extract_metabib_field_entry(bigint, text); DROP FUNCTION biblio.extract_metabib_field_entry(bigint); @@ -353,7 +518,8 @@ CREATE OR REPLACE query_text TEXT, -- 'foo' or 'foo & ba:*',ready for to_tsquery() search_class TEXT, -- 'alias' or 'class' or 'class|field..', etc headline_opts TEXT, -- markup options for ts_headline() - visibility_org INTEGER -- null if you don't want opac visibility test + visibility_org INTEGER,-- null if you don't want opac visibility test + query_limit INTEGER -- use in LIMIT clause of interal query ) RETURNS TABLE ( value TEXT, -- plain match TEXT, -- marked up @@ -370,7 +536,7 @@ DECLARE search_class_join TEXT; r_fields RECORD; BEGIN - query := TO_TSQUERY(query_text); + query := TO_TSQUERY('keyword', query_text); IF visibility_org IS NOT NULL THEN opac_visibility_join := ' @@ -439,7 +605,8 @@ BEGIN ' || search_class_join || opac_visibility_join || ' WHERE $1 @@ mbe.index_vector ORDER BY 4 DESC, 5 DESC NULLS LAST, 6 DESC, 7 DESC, 8 DESC, 1 ASC - ' USING query, search_class, headline_opts, visibility_org; + LIMIT $5 + ' USING query, search_class, headline_opts, visibility_org, query_limit; -- sort order: -- bouyant AND chosen class = match class -- 2.11.0