1 -- make record attributes definitions that use xpath to extract values
6 SELECT evergreen.upgrade_deps_block_check('0936', :eg_version);
8 CREATE OR REPLACE FUNCTION metabib.reingest_record_attributes (rid BIGINT, pattr_list TEXT[] DEFAULT NULL, prmarc TEXT DEFAULT NULL, rdeleted BOOL DEFAULT TRUE) RETURNS VOID AS $func$
15 xfrm config.xml_transform%ROWTYPE;
16 attr_vector INT[] := '{}'::INT[];
17 attr_vector_tmp INT[];
18 attr_list TEXT[] := pattr_list;
20 norm_attr_value TEXT[];
22 attr_def config.record_attr_definition%ROWTYPE;
23 ccvm_row config.coded_value_map%ROWTYPE;
26 IF attr_list IS NULL OR rdeleted THEN -- need to do the full dance on INSERT or undelete
27 SELECT ARRAY_AGG(name) INTO attr_list FROM config.record_attr_definition;
31 SELECT marc INTO rmarc FROM biblio.record_entry WHERE id = rid;
34 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE NOT composite AND name = ANY( attr_list ) ORDER BY format LOOP
36 attr_value := '{}'::TEXT[];
37 norm_attr_value := '{}'::TEXT[];
38 attr_vector_tmp := '{}'::INT[];
40 SELECT * INTO ccvm_row FROM config.coded_value_map c WHERE c.ctype = attr_def.name LIMIT 1;
42 -- tag+sf attrs only support SVF
43 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
44 SELECT ARRAY[ARRAY_TO_STRING(ARRAY_AGG(value), COALESCE(attr_def.joiner,' '))] INTO attr_value
45 FROM (SELECT * FROM metabib.full_rec ORDER BY tag, subfield) AS x
47 AND tag LIKE attr_def.tag
49 WHEN attr_def.sf_list IS NOT NULL
50 THEN POSITION(subfield IN attr_def.sf_list) > 0
57 ELSIF attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
58 attr_value := vandelay.marc21_extract_fixed_field_list(rmarc, attr_def.fixed_field);
60 IF NOT attr_def.multi THEN
61 attr_value := ARRAY[attr_value[1]];
64 ELSIF attr_def.xpath IS NOT NULL THEN -- and xpath expression
66 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
68 -- See if we can skip the XSLT ... it's expensive
69 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
70 -- Can't skip the transform
71 IF xfrm.xslt <> '---' THEN
72 transformed_xml := oils_xslt_process(rmarc,xfrm.xslt);
74 transformed_xml := rmarc;
77 prev_xfrm := xfrm.name;
80 IF xfrm.name IS NULL THEN
81 -- just grab the marcxml (empty) transform
82 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
83 prev_xfrm := xfrm.name;
86 FOR tmp_xml IN SELECT UNNEST(oils_xpath(attr_def.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]])) LOOP
87 tmp_val := oils_xpath_string(
90 COALESCE(attr_def.joiner,' '),
91 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
93 IF tmp_val IS NOT NULL AND BTRIM(tmp_val) <> '' THEN
94 attr_value := attr_value || tmp_val;
95 EXIT WHEN NOT attr_def.multi;
99 ELSIF attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
100 SELECT ARRAY_AGG(m.value) INTO attr_value
101 FROM vandelay.marc21_physical_characteristics(rmarc) v
102 LEFT JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
103 WHERE v.subfield = attr_def.phys_char_sf AND (m.value IS NOT NULL AND BTRIM(m.value) <> '')
104 AND ( ccvm_row.id IS NULL OR ( ccvm_row.id IS NOT NULL AND v.id IS NOT NULL) );
106 IF NOT attr_def.multi THEN
107 attr_value := ARRAY[attr_value[1]];
112 -- apply index normalizers to attr_value
113 FOR tmp_val IN SELECT value FROM UNNEST(attr_value) x(value) LOOP
115 SELECT n.func AS func,
116 n.param_count AS param_count,
118 FROM config.index_normalizer n
119 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
120 WHERE attr = attr_def.name
122 EXECUTE 'SELECT ' || normalizer.func || '(' ||
123 COALESCE( quote_literal( tmp_val ), 'NULL' ) ||
125 WHEN normalizer.param_count > 0
126 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
132 IF tmp_val IS NOT NULL AND tmp_val <> '' THEN
133 -- note that a string that contains only blanks
134 -- is a valid value for some attributes
135 norm_attr_value := norm_attr_value || tmp_val;
139 IF attr_def.filter THEN
140 -- Create unknown uncontrolled values and find the IDs of the values
141 IF ccvm_row.id IS NULL THEN
142 FOR tmp_val IN SELECT value FROM UNNEST(norm_attr_value) x(value) LOOP
143 IF tmp_val IS NOT NULL AND BTRIM(tmp_val) <> '' THEN
144 BEGIN -- use subtransaction to isolate unique constraint violations
145 INSERT INTO metabib.uncontrolled_record_attr_value ( attr, value ) VALUES ( attr_def.name, tmp_val );
146 EXCEPTION WHEN unique_violation THEN END;
150 SELECT ARRAY_AGG(id) INTO attr_vector_tmp FROM metabib.uncontrolled_record_attr_value WHERE attr = attr_def.name AND value = ANY( norm_attr_value );
152 SELECT ARRAY_AGG(id) INTO attr_vector_tmp FROM config.coded_value_map WHERE ctype = attr_def.name AND code = ANY( norm_attr_value );
155 -- Add the new value to the vector
156 attr_vector := attr_vector || attr_vector_tmp;
159 IF attr_def.sorter AND norm_attr_value[1] IS NOT NULL THEN
160 DELETE FROM metabib.record_sorter WHERE source = rid AND attr = attr_def.name;
161 INSERT INTO metabib.record_sorter (source, attr, value) VALUES (rid, attr_def.name, norm_attr_value[1]);
166 /* We may need to rewrite the vlist to contain
167 the intersection of new values for requested
168 attrs and old values for ignored attrs. To
169 do this, we take the old attr vlist and
170 subtract any values that are valid for the
171 requested attrs, and then add back the new
172 set of attr values. */
174 IF ARRAY_LENGTH(pattr_list, 1) > 0 THEN
175 SELECT vlist INTO attr_vector_tmp FROM metabib.record_attr_vector_list WHERE source = rid;
176 SELECT attr_vector_tmp - ARRAY_AGG(id::INT) INTO attr_vector_tmp FROM metabib.full_attr_id_map WHERE attr = ANY (pattr_list);
177 attr_vector := attr_vector || attr_vector_tmp;
180 -- On to composite attributes, now that the record attrs have been pulled. Processed in name order, so later composite
181 -- attributes can depend on earlier ones.
182 PERFORM metabib.compile_composite_attr_cache_init();
183 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE composite AND name = ANY( attr_list ) ORDER BY name LOOP
185 FOR ccvm_row IN SELECT * FROM config.coded_value_map c WHERE c.ctype = attr_def.name ORDER BY value LOOP
187 tmp_val := metabib.compile_composite_attr( ccvm_row.id );
188 CONTINUE WHEN tmp_val IS NULL OR tmp_val = ''; -- nothing to do
190 IF attr_def.filter THEN
191 IF attr_vector @@ tmp_val::query_int THEN
192 attr_vector = attr_vector + intset(ccvm_row.id);
193 EXIT WHEN NOT attr_def.multi;
197 IF attr_def.sorter THEN
198 IF attr_vector @@ tmp_val THEN
199 DELETE FROM metabib.record_sorter WHERE source = rid AND attr = attr_def.name;
200 INSERT INTO metabib.record_sorter (source, attr, value) VALUES (rid, attr_def.name, ccvm_row.code);
208 IF ARRAY_LENGTH(attr_vector, 1) > 0 THEN
209 IF rdeleted THEN -- initial insert OR revivication
210 DELETE FROM metabib.record_attr_vector_list WHERE source = rid;
211 INSERT INTO metabib.record_attr_vector_list (source, vlist) VALUES (rid, attr_vector);
213 UPDATE metabib.record_attr_vector_list SET vlist = attr_vector WHERE source = rid;
219 $func$ LANGUAGE PLPGSQL;