</xsl:template>
</xsl:stylesheet>$$ WHERE name = 'mods32';
+CREATE OR REPLACE FUNCTION biblio.extract_metabib_field_entry ( rid BIGINT, default_joiner TEXT ) RETURNS SETOF metabib.field_entry_template AS $func$
+DECLARE
+ bib biblio.record_entry%ROWTYPE;
+ idx config.metabib_field%ROWTYPE;
+ xfrm config.xml_transform%ROWTYPE;
+ prev_xfrm TEXT;
+ transformed_xml TEXT;
+ xml_node TEXT;
+ xml_node_list TEXT[];
+ facet_text TEXT;
+ browse_text TEXT;
+ sort_value TEXT;
+ raw_text TEXT;
+ curr_text TEXT;
+ joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
+ authority_text TEXT;
+ authority_link BIGINT;
+ output_row metabib.field_entry_template%ROWTYPE;
+BEGIN
+
+ -- Start out with no field-use bools set
+ output_row.browse_field = FALSE;
+ output_row.facet_field = FALSE;
+ output_row.search_field = FALSE;
+
+ -- Get the record
+ SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
+
+ -- Loop over the indexing entries
+ FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
+
+ joiner := COALESCE(idx.joiner, default_joiner);
+
+ SELECT INTO xfrm * from config.xml_transform WHERE name = idx.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(bib.marc,xfrm.xslt);
+ ELSE
+ transformed_xml := bib.marc;
+ END IF;
+
+ prev_xfrm := xfrm.name;
+ END IF;
+
+ xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
+
+ raw_text := NULL;
+ FOR xml_node IN SELECT x FROM unnest(xml_node_list) AS x LOOP
+ CONTINUE WHEN xml_node !~ E'^\\s*<';
+
+ -- XXX much of this should be moved into oils_xpath_string...
+ curr_text := ARRAY_TO_STRING(evergreen.array_remove_item_by_value(evergreen.array_remove_item_by_value(
+ oils_xpath( '//text()',
+ REGEXP_REPLACE(
+ REGEXP_REPLACE( -- This escapes all &s not followed by "amp;". Data ise returned from oils_xpath (above) in UTF-8, not entity encoded
+ REGEXP_REPLACE( -- This escapes embeded <s
+ xml_node,
+ $re$(>[^<]+)(<)([^>]+<)$re$,
+ E'\\1<\\3',
+ 'g'
+ ),
+ '&(?!amp;)',
+ '&',
+ 'g'
+ ),
+ E'\\s+',
+ ' ',
+ 'g'
+ )
+ ), ' '), ''),
+ joiner
+ );
+
+ CONTINUE WHEN curr_text IS NULL OR curr_text = '';
+
+ IF raw_text IS NOT NULL THEN
+ raw_text := raw_text || joiner;
+ END IF;
+
+ raw_text := COALESCE(raw_text,'') || curr_text;
+
+ -- autosuggest/metabib.browse_entry
+ IF idx.browse_field THEN
+
+ IF idx.browse_xpath IS NOT NULL AND idx.browse_xpath <> '' THEN
+ browse_text := oils_xpath_string( idx.browse_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
+ ELSE
+ browse_text := curr_text;
+ END IF;
+
+ IF idx.browse_sort_xpath IS NOT NULL AND
+ idx.browse_sort_xpath <> '' THEN
+
+ sort_value := oils_xpath_string(
+ idx.browse_sort_xpath, xml_node, joiner,
+ ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
+ );
+ ELSE
+ sort_value := browse_text;
+ END IF;
+
+ output_row.field_class = idx.field_class;
+ output_row.field = idx.id;
+ output_row.source = rid;
+ output_row.value = BTRIM(REGEXP_REPLACE(browse_text, E'\\s+', ' ', 'g'));
+ output_row.sort_value :=
+ public.naco_normalize(sort_value);
+
+ output_row.authority := NULL;
+
+ IF idx.authority_xpath IS NOT NULL AND idx.authority_xpath <> '' THEN
+ authority_text := oils_xpath_string(
+ idx.authority_xpath, xml_node, joiner,
+ ARRAY[
+ ARRAY[xfrm.prefix, xfrm.namespace_uri],
+ ARRAY['xlink','http://www.w3.org/1999/xlink']
+ ]
+ );
+
+ IF authority_text ~ '^\d+$' THEN
+ authority_link := authority_text::BIGINT;
+ PERFORM * FROM authority.record_entry WHERE id = authority_link;
+ IF FOUND THEN
+ output_row.authority := authority_link;
+ END IF;
+ END IF;
+
+ END IF;
+
+ output_row.browse_field = TRUE;
+ -- Returning browse rows with search_field = true for search+browse
+ -- configs allows us to retain granularity of being able to search
+ -- browse fields with "starts with" type operators (for example, for
+ -- titles of songs in music albums)
+ IF idx.search_field THEN
+ output_row.search_field = TRUE;
+ END IF;
+ RETURN NEXT output_row;
+ output_row.browse_field = FALSE;
+ output_row.search_field = FALSE;
+ output_row.sort_value := NULL;
+ END IF;
+
+ -- insert raw node text for faceting
+ IF idx.facet_field THEN
+
+ IF idx.facet_xpath IS NOT NULL AND idx.facet_xpath <> '' THEN
+ facet_text := oils_xpath_string( idx.facet_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
+ ELSE
+ facet_text := curr_text;
+ END IF;
+
+ output_row.field_class = idx.field_class;
+ output_row.field = -1 * idx.id;
+ output_row.source = rid;
+ output_row.value = BTRIM(REGEXP_REPLACE(facet_text, E'\\s+', ' ', 'g'));
+
+ output_row.facet_field = TRUE;
+ RETURN NEXT output_row;
+ output_row.facet_field = FALSE;
+ END IF;
+
+ END LOOP;
+
+ CONTINUE WHEN raw_text IS NULL OR raw_text = '';
+
+ -- insert combined node text for searching
+ IF idx.search_field THEN
+ output_row.field_class = idx.field_class;
+ output_row.field = idx.id;
+ output_row.source = rid;
+ output_row.value = BTRIM(REGEXP_REPLACE(raw_text, E'\\s+', ' ', 'g'));
+
+ output_row.search_field = TRUE;
+ RETURN NEXT output_row;
+ output_row.search_field = FALSE;
+ END IF;
+
+ END LOOP;
+
+END;
+
+$func$ LANGUAGE PLPGSQL;
+
+
+CREATE OR REPLACE FUNCTION metabib.reingest_metabib_field_entries( bib_id BIGINT, skip_facet BOOL DEFAULT FALSE, skip_browse BOOL DEFAULT FALSE, skip_search BOOL DEFAULT FALSE ) RETURNS VOID AS $func$
+DECLARE
+ fclass RECORD;
+ ind_data metabib.field_entry_template%ROWTYPE;
+ mbe_row metabib.browse_entry%ROWTYPE;
+ mbe_id BIGINT;
+ b_skip_facet BOOL;
+ b_skip_browse BOOL;
+ b_skip_search BOOL;
+ value_prepped TEXT;
+BEGIN
+
+ SELECT COALESCE(NULLIF(skip_facet, FALSE), EXISTS (SELECT enabled FROM config.internal_flag WHERE name = 'ingest.skip_facet_indexing' AND enabled)) INTO b_skip_facet;
+ SELECT COALESCE(NULLIF(skip_browse, FALSE), EXISTS (SELECT enabled FROM config.internal_flag WHERE name = 'ingest.skip_browse_indexing' AND enabled)) INTO b_skip_browse;
+ SELECT COALESCE(NULLIF(skip_search, FALSE), EXISTS (SELECT enabled FROM config.internal_flag WHERE name = 'ingest.skip_search_indexing' AND enabled)) INTO b_skip_search;
+
+ PERFORM * FROM config.internal_flag WHERE name = 'ingest.assume_inserts_only' AND enabled;
+ IF NOT FOUND THEN
+ IF NOT b_skip_search THEN
+ FOR fclass IN SELECT * FROM config.metabib_class LOOP
+ -- RAISE NOTICE 'Emptying out %', fclass.name;
+ EXECUTE $$DELETE FROM metabib.$$ || fclass.name || $$_field_entry WHERE source = $$ || bib_id;
+ END LOOP;
+ END IF;
+ IF NOT b_skip_facet THEN
+ DELETE FROM metabib.facet_entry WHERE source = bib_id;
+ END IF;
+ IF NOT b_skip_browse THEN
+ DELETE FROM metabib.browse_entry_def_map WHERE source = bib_id;
+ END IF;
+ END IF;
+
+ FOR ind_data IN SELECT * FROM biblio.extract_metabib_field_entry( bib_id ) LOOP
+ IF ind_data.field < 0 THEN
+ ind_data.field = -1 * ind_data.field;
+ END IF;
+
+ IF ind_data.facet_field AND NOT b_skip_facet THEN
+ INSERT INTO metabib.facet_entry (field, source, value)
+ VALUES (ind_data.field, ind_data.source, ind_data.value);
+ END IF;
+
+ IF ind_data.browse_field AND NOT b_skip_browse THEN
+ -- A caveat about this SELECT: this should take care of replacing
+ -- old mbe rows when data changes, but not if normalization (by
+ -- which I mean specifically the output of
+ -- evergreen.oils_tsearch2()) changes. It may or may not be
+ -- expensive to add a comparison of index_vector to index_vector
+ -- to the WHERE clause below.
+
+ value_prepped := metabib.browse_normalize(ind_data.value, ind_data.field);
+ SELECT INTO mbe_row * FROM metabib.browse_entry
+ WHERE value = value_prepped AND sort_value = ind_data.sort_value;
+
+ IF FOUND THEN
+ mbe_id := mbe_row.id;
+ ELSE
+ INSERT INTO metabib.browse_entry
+ ( value, sort_value ) VALUES
+ ( value_prepped, ind_data.sort_value );
+
+ mbe_id := CURRVAL('metabib.browse_entry_id_seq'::REGCLASS);
+ END IF;
+
+ INSERT INTO metabib.browse_entry_def_map (entry, def, source, authority)
+ VALUES (mbe_id, ind_data.field, ind_data.source, ind_data.authority);
+ END IF;
+
+ IF ind_data.search_field AND NOT b_skip_search THEN
+ -- Avoid inserting duplicate rows
+ EXECUTE 'SELECT 1 FROM metabib.' || ind_data.field_class ||
+ '_field_entry WHERE field = $1 AND source = $2 AND value = $3'
+ INTO mbe_id USING ind_data.field, ind_data.source, ind_data.value;
+ -- RAISE NOTICE 'Search for an already matching row returned %', mbe_id;
+ IF mbe_id IS NULL THEN
+ EXECUTE $$
+ INSERT INTO metabib.$$ || ind_data.field_class || $$_field_entry (field, source, value)
+ VALUES ($$ ||
+ quote_literal(ind_data.field) || $$, $$ ||
+ quote_literal(ind_data.source) || $$, $$ ||
+ quote_literal(ind_data.value) ||
+ $$);$$;
+ END IF;
+ END IF;
+
+ END LOOP;
+
+ IF NOT b_skip_search THEN
+ PERFORM metabib.update_combined_index_vectors(bib_id);
+ END IF;
+
+ RETURN;
+END;
+$func$ LANGUAGE PLPGSQL;
+
-- Don't use Title Proper search field as the browse field
UPDATE config.metabib_field SET browse_field = FALSE, browse_xpath = NULL, browse_sort_xpath = NULL WHERE id = 6;
(31, 'title', 'browse', oils_i18n_gettext(31, 'Title Proper (Browse)', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:titleBrowse$$, FALSE, '//@xlink:href', TRUE, $$*[local-name() != "nonSort"]$$ );
COMMIT;
+
+\qecho This is a browse-only reingest of your bib records. It may take a while.
+\qecho You may cancel now without losing the effect of the rest of the
+\qecho upgrade script, and arrange the reingest later.
+\qecho .
+SELECT metabib.reingest_metabib_field_entries(id, TRUE, FALSE, TRUE)
+ FROM biblio.record_entry;
projects / evergreen / pines.git / commitdiff