1 DROP TABLE IF EXISTS migration_tools.search_format_map;
2 CREATE TABLE migration_tools.search_format_map (code TEXT, itype CHAR(1), iform CHAR(1), phy CHAR(1), phyv CHAR(1), phyp SMALLINT,
3 biblevel CHAR(1), iform_exclude CHAR(1)[], srform_exclude CHAR(1)[] );
4 INSERT INTO migration_tools.search_format_map (code, itype, iform, phy, phyv, phyp, biblevel, iform_exclude, srform_exclude) VALUES
5 -- itype iform phy phyv phyp bib itemform exclude sr format exclude
6 ('blu-ray', 'g', NULL, 'v', 's', 4, NULL, NULL, NULL)
7 ,('book', 'a', NULL, NULL, NULL, NULL, 'a', '{a,b,c,f,o,q,r,s}', NULL)
8 ,('braille', 'a', 'f', NULL, NULL, NULL, NULL, NULL, NULL)
9 ,('casaudiobook', 'i', NULL, 's', 'l', 3, NULL, NULL, NULL)
10 ,('casmusic', 'j', NULL, 's', 'l', 3, NULL, NULL, NULL)
11 ,('cdaudiobook', 'i', NULL, 's', 'f', 3, NULL, NULL, NULL)
12 ,('cdmusic', 'j', NULL, 's', 'f', 3, NULL, NULL, NULL)
13 ,('dvd', 'g', NULL, 'v', 'v', 4, NULL, NULL, NULL)
14 ,('eaudio', 'i', 'o', NULL, NULL, NULL, NULL, NULL, NULL)
15 ,('ebook', 'a', 's', NULL, NULL, NULL, 'a' , NULL, NULL)
16 ,('electronic', 's', 'o', NULL, NULL, NULL, NULL, NULL, NULL)
17 ,('equip', 'r', NULL, NULL, NULL, NULL, NULL, NULL, NULL)
18 ,('evideo', 'g', 'o', NULL, NULL, NULL, NULL, NULL, NULL)
19 ,('kit', 'o', NULL, NULL, NULL, NULL, NULL, NULL, NULL)
20 ,('lpbook', 'a', 'd', NULL, NULL, NULL, 'a' , NULL, NULL)
21 ,('map', 'e', NULL, NULL, NULL, NULL, NULL, NULL, NULL)
22 ,('microform', 'a', 'b', NULL, NULL, NULL, NULL, NULL, NULL)
23 ,('music', 'j', NULL, NULL, NULL, NULL, NULL, NULL, '{a,b,c,d,e,f}')
24 ,('phonomusic', 'j', NULL, 's', 'a', 3, NULL, NULL, NULL)
25 ,('phonospoken', 'i', NULL, 's', 'a', 3, NULL, NULL, NULL)
26 ,('picture', 'k', NULL, NULL, NULL, NULL, NULL, NULL, NULL)
27 ,('preloadedaudio', 'i', 'q', NULL, NULL, NULL, NULL, NULL, '{a,b,c,d,e,f,s}')
28 ,('score', 'c', NULL, NULL, NULL, NULL, NULL, NULL, NULL)
29 ,('serial', NULL, NULL, NULL, NULL, NULL, 's' , NULL, NULL)
30 ,('software', 'm', NULL, NULL, NULL, NULL, NULL, NULL, NULL)
31 ,('vhs', 'g', NULL, 'v', 'b', 4, NULL, NULL, NULL)
34 CREATE OR REPLACE FUNCTION migration_tools.reingest_staged_record_attributes (rid BIGINT, pattr_list TEXT[] DEFAULT NULL, prmarc TEXT DEFAULT NULL, rdeleted BOOL DEFAULT TRUE) RETURNS INTEGER[] AS $func$
41 xfrm config.xml_transform%ROWTYPE;
42 attr_vector INT[] := '{}'::INT[];
43 attr_vector_tmp INT[];
44 attr_list TEXT[] := pattr_list;
46 norm_attr_value TEXT[];
49 attr_def config.record_attr_definition%ROWTYPE;
50 ccvm_row config.coded_value_map%ROWTYPE;
54 IF attr_list IS NULL OR rdeleted THEN -- need to do the full dance on INSERT or undelete
55 SELECT ARRAY_AGG(name) INTO attr_list FROM config.record_attr_definition
58 fixed_field IS NOT NULL OR
60 phys_char_sf IS NOT NULL OR
67 SELECT marc INTO rmarc FROM biblio_record_entry_legacy WHERE id = rid;
70 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE NOT composite AND name = ANY( attr_list ) ORDER BY format LOOP
72 jump_past := FALSE; -- This gets set when we are non-multi and have found something
73 attr_value := '{}'::TEXT[];
74 norm_attr_value := '{}'::TEXT[];
75 attr_vector_tmp := '{}'::INT[];
77 SELECT * INTO ccvm_row FROM config.coded_value_map c WHERE c.ctype = attr_def.name LIMIT 1;
79 IF attr_def.tag IS NOT NULL THEN -- tag (and optional subfield list) selection
80 SELECT ARRAY_AGG(value) INTO attr_value
81 FROM (SELECT * FROM metabib.full_rec ORDER BY tag, subfield) AS x
83 AND tag LIKE attr_def.tag
85 WHEN attr_def.sf_list IS NOT NULL
86 THEN POSITION(subfield IN attr_def.sf_list) > 0
92 IF NOT attr_def.multi THEN
93 attr_value := ARRAY[ARRAY_TO_STRING(attr_value, COALESCE(attr_def.joiner,' '))];
98 IF NOT jump_past AND attr_def.fixed_field IS NOT NULL THEN -- a named fixed field, see config.marc21_ff_pos_map.fixed_field
99 attr_value := attr_value || vandelay.marc21_extract_fixed_field_list(rmarc, attr_def.fixed_field);
101 IF NOT attr_def.multi THEN
102 attr_value := ARRAY[attr_value[1]];
107 IF NOT jump_past AND attr_def.xpath IS NOT NULL THEN -- and xpath expression
109 SELECT INTO xfrm * FROM config.xml_transform WHERE name = attr_def.format;
111 -- See if we can skip the XSLT ... it's expensive
112 IF prev_xfrm IS NULL OR prev_xfrm <> xfrm.name THEN
113 -- Can't skip the transform
114 IF xfrm.xslt <> '---' THEN
115 transformed_xml := oils_xslt_process(rmarc,xfrm.xslt);
117 transformed_xml := rmarc;
120 prev_xfrm := xfrm.name;
123 IF xfrm.name IS NULL THEN
124 -- just grab the marcxml (empty) transform
125 SELECT INTO xfrm * FROM config.xml_transform WHERE xslt = '---' LIMIT 1;
126 prev_xfrm := xfrm.name;
129 FOR tmp_xml IN SELECT UNNEST(oils_xpath(attr_def.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]])) LOOP
130 tmp_val := oils_xpath_string(
133 COALESCE(attr_def.joiner,' '),
134 ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]]
136 IF tmp_val IS NOT NULL AND BTRIM(tmp_val) <> '' THEN
137 attr_value := attr_value || tmp_val;
138 EXIT WHEN NOT attr_def.multi;
143 IF NOT jump_past AND attr_def.phys_char_sf IS NOT NULL THEN -- a named Physical Characteristic, see config.marc21_physical_characteristic_*_map
144 SELECT ARRAY_AGG(m.value) INTO tmp_array
145 FROM vandelay.marc21_physical_characteristics(rmarc) v
146 LEFT JOIN config.marc21_physical_characteristic_value_map m ON (m.id = v.value)
147 WHERE v.subfield = attr_def.phys_char_sf AND (m.value IS NOT NULL AND BTRIM(m.value) <> '')
148 AND ( ccvm_row.id IS NULL OR ( ccvm_row.id IS NOT NULL AND v.id IS NOT NULL) );
150 attr_value := attr_value || tmp_array;
152 IF NOT attr_def.multi THEN
153 attr_value := ARRAY[attr_value[1]];
158 -- apply index normalizers to attr_value
159 FOR tmp_val IN SELECT value FROM UNNEST(attr_value) x(value) LOOP
161 SELECT n.func AS func,
162 n.param_count AS param_count,
164 FROM config.index_normalizer n
165 JOIN config.record_attr_index_norm_map m ON (m.norm = n.id)
166 WHERE attr = attr_def.name
168 EXECUTE 'SELECT ' || normalizer.func || '(' ||
169 COALESCE( quote_literal( tmp_val ), 'NULL' ) ||
171 WHEN normalizer.param_count > 0
172 THEN ',' || REPLACE(REPLACE(BTRIM(normalizer.params,'[]'),E'\'',E'\\\''),E'"',E'\'')
178 IF tmp_val IS NOT NULL AND tmp_val <> '' THEN
179 -- note that a string that contains only blanks
180 -- is a valid value for some attributes
181 norm_attr_value := norm_attr_value || tmp_val;
185 IF attr_def.filter THEN
186 -- Create unknown uncontrolled values and find the IDs of the values
187 IF ccvm_row.id IS NULL THEN
188 FOR tmp_val IN SELECT value FROM UNNEST(norm_attr_value) x(value) LOOP
189 IF tmp_val IS NOT NULL AND BTRIM(tmp_val) <> '' THEN
190 BEGIN -- use subtransaction to isolate unique constraint violations
191 INSERT INTO metabib.uncontrolled_record_attr_value ( attr, value ) VALUES ( attr_def.name, tmp_val );
192 EXCEPTION WHEN unique_violation THEN END;
196 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 );
198 SELECT ARRAY_AGG(id) INTO attr_vector_tmp FROM config.coded_value_map WHERE ctype = attr_def.name AND code = ANY( norm_attr_value );
201 -- Add the new value to the vector
202 attr_vector := attr_vector || attr_vector_tmp;
207 IF ARRAY_LENGTH(pattr_list, 1) > 0 THEN
208 SELECT vlist INTO attr_vector_tmp FROM metabib.record_attr_vector_list WHERE source = rid;
209 SELECT attr_vector_tmp - ARRAY_AGG(id::INT) INTO attr_vector_tmp FROM metabib.full_attr_id_map WHERE attr = ANY (pattr_list);
210 attr_vector := attr_vector || attr_vector_tmp;
213 -- On to composite attributes, now that the record attrs have been pulled. Processed in name order, so later composite
214 -- attributes can depend on earlier ones.
215 PERFORM metabib.compile_composite_attr_cache_init();
216 FOR attr_def IN SELECT * FROM config.record_attr_definition WHERE composite AND name = ANY( attr_list ) ORDER BY name LOOP
218 FOR ccvm_row IN SELECT * FROM config.coded_value_map c WHERE c.ctype = attr_def.name ORDER BY value LOOP
220 tmp_val := metabib.compile_composite_attr( ccvm_row.id );
221 CONTINUE WHEN tmp_val IS NULL OR tmp_val = ''; -- nothing to do
223 IF attr_def.filter THEN
224 IF attr_vector @@ tmp_val::query_int THEN
225 attr_vector = attr_vector + intset(ccvm_row.id);
226 EXIT WHEN NOT attr_def.multi;
230 IF attr_def.sorter THEN
231 IF attr_vector @@ tmp_val THEN
232 DELETE FROM metabib.record_sorter WHERE source = rid AND attr = attr_def.name;
233 INSERT INTO metabib.record_sorter (source, attr, value) VALUES (rid, attr_def.name, ccvm_row.code);
243 $func$ LANGUAGE PLPGSQL;
246 CREATE OR REPLACE FUNCTION migration_tools.view_staged_vlist (rid BIGINT) RETURNS TABLE (r_ctype text, r_code text, r_value text) AS $func$
252 SELECT migration_tools.reingest_staged_record_attributes(rid) INTO vlist;
254 RETURN QUERY SELECT ctype, code, value FROM config.coded_value_map WHERE id IN (SELECT UNNEST(vlist));
256 $func$ LANGUAGE PLPGSQL;
258 CREATE OR REPLACE FUNCTION migration_tools.stage_vlist (rid BIGINT) RETURNS VOID AS $func$
263 SELECT migration_tools.reingest_staged_record_attributes(rid) INTO vlist;
265 SELECT ARRAY_AGG(code) FROM config.coded_value_map WHERE id IN (SELECT UNNEST(vlist))
266 AND ctype = 'search_format' INTO search;
268 UPDATE biblio_record_entry_legacy SET x_search_format = search WHERE id = rid;
270 $func$ LANGUAGE PLPGSQL;
272 CREATE OR REPLACE FUNCTION migration_tools.show_staged_vlist (rid BIGINT) RETURNS TEXT[] AS $func$
277 SELECT migration_tools.reingest_staged_record_attributes(rid) INTO vlist;
279 SELECT ARRAY_AGG(code) FROM config.coded_value_map WHERE id IN (SELECT UNNEST(vlist))
280 AND ctype = 'search_format' INTO search;
284 $func$ LANGUAGE PLPGSQL;
287 CREATE OR REPLACE FUNCTION migration_tools.postfix_vlist (rid BIGINT) RETURNS VOID AS $func$
292 SELECT migration_tools.reingest_staged_record_attributes(rid) INTO vlist;
294 SELECT ARRAY_AGG(code) FROM config.coded_value_map WHERE id IN (SELECT UNNEST(vlist))
295 AND ctype = 'search_format' INTO search;
297 UPDATE biblio_record_entry_legacy SET x_after_search_format = search WHERE id = rid;
299 $func$ LANGUAGE PLPGSQL;
302 CREATE OR REPLACE FUNCTION migration_tools.set_exp_sfs (rid BIGINT) RETURNS VOID AS $func$
308 SELECT circ_mods FROM biblio_record_entry_legacy WHERE id = rid INTO cms;
309 IF cms IS NOT NULL THEN
310 FOREACH y IN ARRAY cms LOOP
311 w := w || (SELECT sf1 FROM circ_mod_to_sf_map WHERE circ_mod = y);
312 w := w || (SELECT sf2 FROM circ_mod_to_sf_map WHERE circ_mod = y);
313 w := w || (SELECT sf3 FROM circ_mod_to_sf_map WHERE circ_mod = y);
315 UPDATE biblio_record_entry_legacy SET expected_sfs = w WHERE id = rid;
318 $func$ LANGUAGE PLPGSQL;
320 DROP FUNCTION IF EXISTS migration_tools.modify_staged_fixed_fields (BIGINT,TEXT);
321 CREATE OR REPLACE FUNCTION migration_tools.modify_staged_fixed_fields (bib_id BIGINT, xcode TEXT)
333 xiform_exclude CHAR(1)[];
334 xsrform_exclude CHAR(1)[];
336 ysrform_exclude TEXT;
338 SELECT itype, iform, phy, phyv, phyp, biblevel, iform_exclude, srform_exclude FROM migration_tools.search_format_map WHERE code = xcode
339 INTO xitype, xiform, xphy, xphyv, xphyp, xbiblevel, xiform_exclude, xsrform_exclude;
340 IF xiform_exclude IS NOT NULL THEN
341 yiform_exclude := ARRAY_TO_STRING(xiform_exclude,',');
343 yiform_exclude := '';
345 IF xsrform_exclude IS NOT NULL THEN
346 ysrform_exclude := ARRAY_TO_STRING(ysrform_exclude,',');
348 ysrform_exclude := '';
350 SELECT migration_tools.modify_fixed_fields(marc,xcode,xitype,xiform,xphy,xphyv,xphyp,xbiblevel,yiform_exclude,ysrform_exclude) FROM biblio_record_entry_legacy WHERE id = bib_id INTO r;
351 UPDATE biblio_record_entry_legacy SET marc = r WHERE id = bib_id;
356 DROP FUNCTION IF EXISTS migration_tools.modify_fixed_fields (TEXT, TEXT, CHAR(1), CHAR(1), CHAR(1), CHAR(1), SMALLINT, CHAR(1), TEXT, TEXT);
357 CREATE OR REPLACE FUNCTION migration_tools.modify_fixed_fields (TEXT, TEXT, CHAR(1), CHAR(1), CHAR(1), CHAR(1), SMALLINT, CHAR(1), TEXT, TEXT)
362 # assumption is that there should only be a single format per item
368 use MARC::File::XML (BinaryEncoding => 'utf8');
372 my ($marcxml, $code, $itype, $iform, $phy, $phyv, $phyp, $biblevel, $iform_exclude_temp, $srform_exclude_temp) = @_;
375 if ($iform_exclude_temp) { @iform_exclude = split /,/, $iform_exclude_temp; }
377 if ($srform_exclude_temp) { @srform_exclude = split /,/, $srform_exclude_temp; }
379 $marcxml =~ s/(<leader>.........)./${1}a/;
380 eval { $marc = MARC::Record->new_from_xml($marcxml, 'UTF-8'); };
382 import MARC::File::XML (BinaryEncoding => 'utf8');
383 return 'failed to parse marcxml';
386 my $ldr = $marc->leader();
387 if ($itype) { substr($ldr,6,1) = $itype; } else { substr($ldr,6,1) = '|'; }
388 if ($biblevel) { substr($ldr,7,1) = $biblevel; } else { substr($ldr,7,1) = '|'; }
392 my $zze_str = '0000000000000000000000000000000000000000';
395 $zedzedeight = $marc->field('008');
397 $zze_str = $zedzedeight->data();
399 if (length($zze_str) < 30) {
400 my $nneight = MARC::Field->new( 918, '1', '0', 'a' => $zze_str );
401 $marc->insert_fields_ordered($nneight);
402 $zze_str = '0000000000000000000000000000000000000000';
404 if ($itype eq 'e' or $itype eq 'g' or $itype eq 'k')
405 { $formchar = substr($zze_str,29,1); }
406 else { $formchar = substr($zze_str,23,1); }
407 if (@iform_exclude and $itype) {
408 if ($itype eq 'e' or $itype eq 'g' or $itype eq 'k') { #visual materials
409 if ($formchar ~~ @iform_exclude) { substr($zze_str,29,1) = '|'; }
410 } else { if ($formchar ~~ @iform_exclude) { substr($zze_str,23,1) = '|'; } }
413 if ($itype eq 'e' or $itype eq 'g' or $itype eq 'k') { #visual materials
414 substr($zze_str,29,1) = $iform;
416 substr($zze_str,23,1) = $iform;
419 if ($itype eq 'e' or $itype eq 'g' or $itype eq 'k') { #visual materials
420 substr($zze_str,29,1) = '|';
422 substr($zze_str,23,1) = '|';
426 $new_zze = MARC::Field->new('008',$zze_str);
427 if ($zedzedeight) { $zedzedeight->replace_with($new_zze); } else
428 { $marc->insert_fields_ordered($new_zze); }
430 my @todelzzsx = $marc->field('006');
431 #save the old 006s in 916 fields
432 foreach my $sx (@todelzzsx) {
433 my $nfield = MARC::Field->new( 916, '1', '0', 'a' => $sx->data() );
434 $marc->insert_fields_ordered($nfield);
436 $marc->delete_fields(@todelzzsx);
438 my $zzsx_str = '00000000000000000';
439 if ($iform) { substr($zzsx_str,6,1) = $iform; }
440 my $zedzedsix = MARC::Field->new('006', $zzsx_str);
441 $marc->insert_fields_ordered($zedzedsix);
443 my @todelzzsv = $marc->field('007');
444 #save the old 007s in 917 fields
445 foreach my $sv (@todelzzsv) {
446 my $nfield = MARC::Field->new( 917, '1', '0', 'a' => $sv->data() );
447 $marc->insert_fields_ordered($nfield);
449 $marc->delete_fields(@todelzzsv);
451 my $nn = MARC::Field->new( 919, '1', '0', 'a' => 'record modified by automated fixed field changes' );
452 $marc->insert_fields_ordered($nn);
457 if ($phy eq 'o' or $phy eq 'q' or $phy eq 'z' or $phy eq 't') { $zzs_str = '00'; }
458 if ($phy eq 's' or $phy eq 'c') { $zzs_str = '00000000000000'; }
459 if ($phy eq 'r') { $zzs_str = '00000000000'; }
460 if ($phy eq 'm') { $zzs_str = '00000000000000000000000'; }
461 if ($phy eq 'a') { $zzs_str = '00000000'; }
462 if ($phy eq 'd') { $zzs_str = '000000'; }
463 if ($phy eq 'f') { $zzs_str = '0000000000'; }
464 if ($phy eq 'g') { $zzs_str = '000000000'; }
465 if ($phy eq 'h') { $zzs_str = '0000000000000'; }
466 if ($phy eq 'k') { $zzs_str = '000000'; }
467 if ($phy eq 'v') { $zzs_str = '000000000'; }
468 substr($zzs_str,0,1) = $phy;
469 substr($zzs_str,$phyp,1) = $phyv;
470 $zedzedseven = MARC::Field->new('007', $zzs_str);
471 $marc->insert_fields_ordered($zedzedseven);
473 return $marc->as_xml_record;