biblio Functions

DECLARE
    bib_row     biblio.record_entry%ROWTYPE;
    cn_row      asset.call_number%ROWTYPE;
    attr_set    INT[] := '{}'::INT[];
BEGIN
    SELECT * INTO bib_row FROM biblio.record_entry WHERE id = bib_id;

    IF force_source THEN
        IF new_source IS NOT NULL THEN
            attr_set := attr_set || search.calculate_visibility_attribute(new_source, 'bib_source');
        END IF;
    ELSIF bib_row.source IS NOT NULL THEN
        attr_set := attr_set || search.calculate_visibility_attribute(bib_row.source, 'bib_source');
    END IF;

    FOR cn_row IN
        SELECT  *
          FROM  asset.call_number
          WHERE record = bib_id
                AND label = '##URI##'
                AND NOT deleted
    LOOP
        attr_set := attr_set || search.calculate_visibility_attribute(cn_row.owning_lib, 'luri_org');
    END LOOP;

    RETURN attr_set;
END;

BEGIN

    IF xml_is_well_formed(NEW.marc) THEN
        RETURN NEW;
    ELSE
        RAISE EXCEPTION 'Attempted to % MARCXML that is not well formed', TG_OP;
    END IF;

END;

DECLARE
	idx		config.biblio_fingerprint%ROWTYPE;
	xfrm		config.xml_transform%ROWTYPE;
	prev_xfrm	TEXT;
	transformed_xml	TEXT;
	xml_node	TEXT;
	xml_node_list	TEXT[];
	raw_text	TEXT;
    output_text TEXT := '';
BEGIN

    IF marc IS NULL OR marc = '' THEN
        RETURN NULL;
    END IF;

	-- Loop over the indexing entries
	FOR idx IN SELECT * FROM config.biblio_fingerprint ORDER BY format, id LOOP

		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(marc,xfrm.xslt);
			ELSE
				transformed_xml := marc;
			END IF;

			prev_xfrm := xfrm.name;
		END IF;

		raw_text := COALESCE(
            naco_normalize(
                ARRAY_TO_STRING(
                    oils_xpath(
                        '//text()',
                        (oils_xpath(
                            idx.xpath,
                            transformed_xml,
                            ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
                        ))[1]
                    ),
                    ''
                )
            ),
            ''
        );

        raw_text := REGEXP_REPLACE(raw_text, E'\\[.+?\\]', E'');
        raw_text := REGEXP_REPLACE(raw_text, E'\\mthe\\M|\\man?d?d\\M', E'', 'g'); -- arg! the pain!

        IF idx.first_word IS TRUE THEN
            raw_text := REGEXP_REPLACE(raw_text, E'^(\\w+).*?$', E'\\1');
        END IF;

		output_text := output_text || idx.name || ':' ||
					   REGEXP_REPLACE(raw_text, E'\\s+', '', 'g') || ' ';

	END LOOP;

    RETURN BTRIM(output_text);

END;

DECLARE
    uris            TEXT[];
    uri_xml         TEXT;
    uri_label       TEXT;
    uri_href        TEXT;
    uri_use         TEXT;
    uri_owner_list  TEXT[];
    uri_owner       TEXT;
    uri_owner_id    INT;
    uri_id          INT;
    uri_cn_id       INT;
    uri_map_id      INT;
    current_uri     INT;
    current_map     INT;
    uri_map_count   INT;
    current_uri_map_list    INT[];
    current_map_owner_list  INT[];

BEGIN

    uris := oils_xpath('//*[@tag="856" and (@ind1="4" or @ind1="1") and (@ind2="0" or @ind2="1")]',marcxml);
    IF ARRAY_UPPER(uris,1) > 0 THEN
        FOR i IN 1 .. ARRAY_UPPER(uris, 1) LOOP
            -- First we pull info out of the 856
            uri_xml     := uris[i];

            uri_href    := (oils_xpath('//*[@code="u"]/text()',uri_xml))[1];
            uri_label   := (oils_xpath('//*[@code="y"]/text()|//*[@code="3"]/text()',uri_xml))[1];
            uri_use     := (oils_xpath('//*[@code="z"]/text()|//*[@code="2"]/text()|//*[@code="n"]/text()',uri_xml))[1];

            IF uri_label IS NULL THEN
                uri_label := uri_href;
            END IF;
            CONTINUE WHEN uri_href IS NULL;

            -- Get the distinct list of libraries wanting to use
            SELECT  ARRAY_AGG(
                        DISTINCT REGEXP_REPLACE(
                            x,
                            $re$^.*?\((\w+)\).*$$re$,
                            E'\\1'
                        )
                    ) INTO uri_owner_list
              FROM  UNNEST(
                        oils_xpath(
                            '//*[@code="9"]/text()|//*[@code="w"]/text()|//*[@code="n"]/text()',
                            uri_xml
                        )
                    )x;

            IF ARRAY_UPPER(uri_owner_list,1) > 0 THEN

                -- look for a matching uri
                IF uri_use IS NULL THEN
                    SELECT id INTO uri_id
                        FROM asset.uri
                        WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active
                        ORDER BY id LIMIT 1;
                    IF NOT FOUND THEN -- create one
                        INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
                        SELECT id INTO uri_id
                            FROM asset.uri
                            WHERE label = uri_label AND href = uri_href AND use_restriction IS NULL AND active;
                    END IF;
                ELSE
                    SELECT id INTO uri_id
                        FROM asset.uri
                        WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active
                        ORDER BY id LIMIT 1;
                    IF NOT FOUND THEN -- create one
                        INSERT INTO asset.uri (label, href, use_restriction) VALUES (uri_label, uri_href, uri_use);
                        SELECT id INTO uri_id
                            FROM asset.uri
                            WHERE label = uri_label AND href = uri_href AND use_restriction = uri_use AND active;
                    END IF;
                END IF;

                FOR j IN 1 .. ARRAY_UPPER(uri_owner_list, 1) LOOP
                    uri_owner := uri_owner_list[j];

                    SELECT id INTO uri_owner_id FROM actor.org_unit WHERE shortname = BTRIM(REPLACE(uri_owner,chr(160),''));
                    CONTINUE WHEN NOT FOUND;

                    -- we need a call number to link through
                    SELECT id INTO uri_cn_id FROM asset.call_number WHERE owning_lib = uri_owner_id AND record = bib_id AND label = '##URI##' AND NOT deleted;
                    IF NOT FOUND THEN
                        INSERT INTO asset.call_number (owning_lib, record, create_date, edit_date, creator, editor, label)
                            VALUES (uri_owner_id, bib_id, 'now', 'now', editor_id, editor_id, '##URI##');
                        SELECT id INTO uri_cn_id FROM asset.call_number WHERE owning_lib = uri_owner_id AND record = bib_id AND label = '##URI##' AND NOT deleted;
                    END IF;

                    -- now, link them if they're not already
                    SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
                    IF NOT FOUND THEN
                        INSERT INTO asset.uri_call_number_map (call_number, uri) VALUES (uri_cn_id, uri_id);
                        SELECT id INTO uri_map_id FROM asset.uri_call_number_map WHERE call_number = uri_cn_id AND uri = uri_id;
                    END IF;

                    current_uri_map_list := current_uri_map_list || uri_map_id;
                    current_map_owner_list := current_map_owner_list || uri_cn_id;

                END LOOP;

            END IF;

        END LOOP;
    END IF;

    -- Clear any orphaned URIs, URI mappings and call
    -- numbers for this bib that weren't mapped above.
    FOR current_map IN
        SELECT  m.id
          FROM  asset.uri_call_number_map m
                LEFT JOIN asset.call_number cn ON (cn.id = m.call_number)
          WHERE cn.record = bib_id
                AND cn.label = '##URI##'
                AND (NOT (m.id = ANY (current_uri_map_list))
                     OR current_uri_map_list is NULL)
    LOOP
        SELECT uri INTO current_uri FROM asset.uri_call_number_map WHERE id = current_map;
        DELETE FROM asset.uri_call_number_map WHERE id = current_map;

        SELECT COUNT(*) INTO uri_map_count FROM asset.uri_call_number_map WHERE uri = current_uri;
        IF uri_map_count = 0 THEN
            DELETE FROM asset.uri WHERE id = current_uri;
        END IF;
    END LOOP;

    UPDATE asset.call_number
    SET deleted = TRUE, edit_date = now(), editor = editor_id
    WHERE id IN (
        SELECT  id
          FROM  asset.call_number
          WHERE record = bib_id
                AND label = '##URI##'
                AND NOT deleted
                AND (NOT (id = ANY (current_map_owner_list))
                     OR current_map_owner_list is NULL)
    );

    RETURN;
END;

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;
    display_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;
    process_idx BOOL;
BEGIN

    -- Start out with no field-use bools set
    output_row.browse_nocase = FALSE;
    output_row.browse_field = FALSE;
    output_row.facet_field = FALSE;
    output_row.display_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 WHERE id = ANY (only_fields) ORDER BY format LOOP
        CONTINUE WHEN idx.xpath IS NULL OR idx.xpath = ''; -- pure virtual field

        process_idx := FALSE;
        IF idx.display_field AND 'display' = ANY (field_types) THEN process_idx = TRUE; END IF;
        IF idx.browse_field AND 'browse' = ANY (field_types) THEN process_idx = TRUE; END IF;
        IF idx.search_field AND 'search' = ANY (field_types) THEN process_idx = TRUE; END IF;
        IF idx.facet_field AND 'facet' = ANY (field_types) THEN process_idx = TRUE; END IF;
        CONTINUE WHEN process_idx = FALSE; -- disabled for all types

        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(array_remove(array_remove(
                oils_xpath( '//text()', -- get the content of all the nodes within the main selected node
                    REGEXP_REPLACE( xml_node, E'\\s+', ' ', 'g' ) -- Translate adjacent whitespace to a single space
                ), ' '), ''),  -- throw away morally empty (bankrupt?) strings
                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
                output_row.browse_nocase = idx.browse_nocase;

                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_nocase = FALSE;
                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;

            -- insert raw node text for display
            IF idx.display_field THEN

                IF idx.display_xpath IS NOT NULL AND idx.display_xpath <> '' THEN
                    display_text := oils_xpath_string( idx.display_xpath, xml_node, joiner, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
                ELSE
                    display_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(display_text, E'\\s+', ' ', 'g'));

                output_row.display_field = TRUE;
                RETURN NEXT output_row;
                output_row.display_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;

DECLARE
    qual        INT;
    ldr         TEXT;
    tval        TEXT;
    tval_rec    RECORD;
    bval        TEXT;
    bval_rec    RECORD;
    type_map    RECORD;
    ff_pos      RECORD;
    ff_tag_data TEXT;
BEGIN

    IF marc IS NULL OR marc = '' THEN
        RETURN NULL;
    END IF;

    -- First, the count of tags
    qual := ARRAY_UPPER(oils_xpath('//*[local-name()="datafield"]', marc), 1);

    -- now go through a bunch of pain to get the record type
    IF best_type IS NOT NULL THEN
        ldr := (oils_xpath('//*[local-name()="leader"]/text()', marc))[1];

        IF ldr IS NOT NULL THEN
            SELECT * INTO tval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'Type' LIMIT 1; -- They're all the same
            SELECT * INTO bval_rec FROM config.marc21_ff_pos_map WHERE fixed_field = 'BLvl' LIMIT 1; -- They're all the same


            tval := SUBSTRING( ldr, tval_rec.start_pos + 1, tval_rec.length );
            bval := SUBSTRING( ldr, bval_rec.start_pos + 1, bval_rec.length );

            -- RAISE NOTICE 'type %, blvl %, ldr %', tval, bval, ldr;

            SELECT * INTO type_map FROM config.marc21_rec_type_map WHERE type_val LIKE '%' || tval || '%' AND blvl_val LIKE '%' || bval || '%';

            IF type_map.code IS NOT NULL THEN
                IF best_type = type_map.code THEN
                    qual := qual + qual / 2;
                END IF;

                FOR ff_pos IN SELECT * FROM config.marc21_ff_pos_map WHERE fixed_field = 'Lang' AND rec_type = type_map.code ORDER BY tag DESC LOOP
                    ff_tag_data := SUBSTRING((oils_xpath('//*[@tag="' || ff_pos.tag || '"]/text()',marc))[1], ff_pos.start_pos + 1, ff_pos.length);
                    IF ff_tag_data = best_lang THEN
                            qual := qual + 100;
                    END IF;
                END LOOP;
            END IF;
        END IF;
    END IF;

    -- Now look for some quality metrics
    -- DCL record?
    IF ARRAY_UPPER(oils_xpath('//*[@tag="040"]/*[@code="a" and contains(.,"DLC")]', marc), 1) = 1 THEN
        qual := qual + 10;
    END IF;

    -- From OCLC?
    IF (oils_xpath('//*[@tag="003"]/text()', marc))[1] ~* E'oclo?c' THEN
        qual := qual + 10;
    END IF;

    RETURN qual;

END;

BEGIN

    -- For TG_ARGV, first param is language (like 'eng'), second is record type (like 'BKS')

    IF NEW.deleted IS TRUE THEN -- we don't much care, then, do we?
        RETURN NEW;
    END IF;

    NEW.fingerprint := biblio.extract_fingerprint(NEW.marc);
    NEW.quality := biblio.extract_quality(NEW.marc, TG_ARGV[0], TG_ARGV[1]);

    RETURN NEW;

END;

DECLARE
	bib	biblio.record_entry%ROWTYPE;
	output	metabib.full_rec%ROWTYPE;
	field	RECORD;
BEGIN
	SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;

	FOR field IN SELECT * FROM vandelay.flatten_marc( bib.marc ) LOOP
		output.record := rid;
		output.ind1 := field.ind1;
		output.ind2 := field.ind2;
		output.tag := field.tag;
		output.subfield := field.subfield;
		output.value := field.value;

		RETURN NEXT output;
	END LOOP;
END;

    DELETE FROM authority.bib_linking WHERE bib = $1;
    INSERT INTO authority.bib_linking (bib, authority)
        SELECT  y.bib,
                y.authority
          FROM (    SELECT  DISTINCT $1 AS bib,
                            BTRIM(remove_paren_substring(txt))::BIGINT AS authority
                      FROM  unnest(oils_xpath('//*[@code="0"]/text()',$2)) x(txt)
                      WHERE BTRIM(remove_paren_substring(txt)) ~ $re$^\d+$$re$
                ) y JOIN authority.record_entry r ON r.id = y.authority;
    SELECT $1;

    SELECT $1 AS record, ff_name, ff_value FROM vandelay.marc21_extract_all_fixed_fields( (SELECT marc FROM biblio.record_entry WHERE id = $1), TRUE );

    SELECT * FROM vandelay.marc21_extract_fixed_field( (SELECT marc FROM biblio.record_entry WHERE id = $1), $2, TRUE );

    SELECT * FROM vandelay.marc21_extract_fixed_field_list( (SELECT marc FROM biblio.record_entry WHERE id = $1), $2, TRUE );

    SELECT id, $1 AS record, ptype, subfield, value FROM vandelay.marc21_physical_characteristics( (SELECT marc FROM biblio.record_entry WHERE id = $1) );

	BEGIN RETURN 'AUTOGENERATED-' || nextval('biblio.autogen_tcn_value_seq'::TEXT); END;

BEGIN
    NEW.label_sortkey := REGEXP_REPLACE(
        evergreen.lpad_number_substrings(
            naco_normalize(NEW.label),
            '0',
            10
        ),
        E'\\s+',
        '',
        'g'
    );
    RETURN NEW;
END;