From 626f2372bd3ee6eaf780921036e832569fff1fc2 Mon Sep 17 00:00:00 2001
From: miker <miker@dcc99617-32d9-48b4-a31d-7c20da2025e4>
Date: Thu, 25 Mar 2010 17:24:02 +0000
Subject: [PATCH] add a table to allow labelling (and eventual expansion) of
search classes; index each node separately for facets, instead of (well, in
addition to) globbing; remove extra (redundant) oils_xslt_process function
git-svn-id: svn://svn.open-ils.org/ILS/trunk@15979 dcc99617-32d9-48b4-a31d-7c20da2025e4
---
Open-ILS/src/sql/Pg/002.functions.config.sql | 38 ------
Open-ILS/src/sql/Pg/002.schema.config.sql | 10 +-
Open-ILS/src/sql/Pg/030.schema.metabib.sql | 140 ++++++++++++---------
Open-ILS/src/sql/Pg/950.data.seed-values.sql | 75 ++++++-----
...213.schema.config.metabib_field_class_stuff.sql | 121 ++++++++++++++++++
5 files changed, 251 insertions(+), 133 deletions(-)
create mode 100644 Open-ILS/src/sql/Pg/upgrade/0213.schema.config.metabib_field_class_stuff.sql
diff --git a/Open-ILS/src/sql/Pg/002.functions.config.sql b/Open-ILS/src/sql/Pg/002.functions.config.sql
index 0ac7600d75..f2d80abf4b 100644
--- a/Open-ILS/src/sql/Pg/002.functions.config.sql
+++ b/Open-ILS/src/sql/Pg/002.functions.config.sql
@@ -331,43 +331,5 @@ CREATE OR REPLACE FUNCTION is_json (TEXT) RETURNS BOOL AS $func$
return $@ ? 0 : 1;
$func$ LANGUAGE PLPERLU;
-CREATE OR REPLACE FUNCTION oils_xslt_process(TEXT, TEXT) RETURNS TEXT AS $func$
- use strict;
-
- use XML::LibXSLT;
- use XML::LibXML;
-
- my $doc = shift;
- my $xslt = shift;
-
- # The following approach uses the older XML::LibXML 1.69 / XML::LibXSLT 1.68
- # methods of parsing XML documents and stylesheets, in the hopes of broader
- # compatibility with distributions
- my $parser = $_SHARED{'_xslt_process'}{parsers}{xml} || XML::LibXML->new();
-
- # Cache the XML parser, if we do not already have one
- $_SHARED{'_xslt_process'}{parsers}{xml} = $parser
- unless ($_SHARED{'_xslt_process'}{parsers}{xml});
-
- my $xslt_parser = $_SHARED{'_xslt_process'}{parsers}{xslt} || XML::LibXSLT->new();
-
- # Cache the XSLT processor, if we do not already have one
- $_SHARED{'_xslt_process'}{parsers}{xslt} = $xslt_parser
- unless ($_SHARED{'_xslt_process'}{parsers}{xslt});
-
- my $stylesheet = $_SHARED{'_xslt_process'}{stylesheets}{$xslt} ||
- $xslt_parser->parse_stylesheet( $parser->parse_string($xslt) );
-
- $_SHARED{'_xslt_process'}{stylesheets}{$xslt} = $stylesheet
- unless ($_SHARED{'_xslt_process'}{stylesheets}{$xslt});
-
- return $stylesheet->output_string(
- $stylesheet->transform(
- $parser->parse_string($doc)
- )
- );
-
-$func$ LANGUAGE 'plperlu' STRICT IMMUTABLE;
-
COMMIT;
diff --git a/Open-ILS/src/sql/Pg/002.schema.config.sql b/Open-ILS/src/sql/Pg/002.schema.config.sql
index 665987db53..064320a6fa 100644
--- a/Open-ILS/src/sql/Pg/002.schema.config.sql
+++ b/Open-ILS/src/sql/Pg/002.schema.config.sql
@@ -59,7 +59,7 @@ CREATE TABLE config.upgrade_log (
install_date TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT NOW()
);
-INSERT INTO config.upgrade_log (version) VALUES ('0212'); -- miker
+INSERT INTO config.upgrade_log (version) VALUES ('0213'); -- miker
CREATE TABLE config.bib_source (
id SERIAL PRIMARY KEY,
@@ -189,10 +189,16 @@ INSERT INTO config.biblio_fingerprint (name, xpath, format, first_word)
TRUE
);
+CREATE TABLE config.metabib_class (
+ name TEXT PRIMARY KEY,
+ label TEXT NOT NULL UNIQUE
+);
+
CREATE TABLE config.metabib_field (
id SERIAL PRIMARY KEY,
- field_class TEXT NOT NULL CHECK (lower(field_class) IN ('title','author','subject','keyword','series')),
+ field_class TEXT NOT NULL REFERENCES config.metabib_class (name),
name TEXT NOT NULL,
+ label TEXT NOT NULL,
xpath TEXT NOT NULL,
weight INT NOT NULL DEFAULT 1,
format TEXT NOT NULL DEFAULT 'mods33',
diff --git a/Open-ILS/src/sql/Pg/030.schema.metabib.sql b/Open-ILS/src/sql/Pg/030.schema.metabib.sql
index 4e2c1bbb88..2bc89da54f 100644
--- a/Open-ILS/src/sql/Pg/030.schema.metabib.sql
+++ b/Open-ILS/src/sql/Pg/030.schema.metabib.sql
@@ -233,74 +233,96 @@ CREATE TYPE metabib.field_entry_template AS (
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[];
- raw_text TEXT;
- joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
- output_row metabib.field_entry_template%ROWTYPE;
+ 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[];
+ raw_text TEXT;
+ curr_text TEXT;
+ joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
+ output_row metabib.field_entry_template%ROWTYPE;
BEGIN
- -- Get the record
- SELECT INTO bib * FROM biblio.record_entry WHERE id = rid;
+ -- 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
+ -- Loop over the indexing entries
+ FOR idx IN SELECT * FROM config.metabib_field ORDER BY format LOOP
- SELECT INTO xfrm * from config.xml_transform WHERE name = idx.format;
+ 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;
+ -- 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;
+ prev_xfrm := xfrm.name;
+ END IF;
- xml_node_list := oils_xpath( idx.xpath, transformed_xml, ARRAY[ARRAY[xfrm.prefix, xfrm.namespace_uri]] );
+ 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 explode_array(xml_node_list) AS x LOOP
+ CONTINUE WHEN xml_node !~ E'^\\s*<';
+
+ curr_text := ARRAY_TO_STRING(
+ oils_xpath( '//text()',
+ 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'
+ )
+ ),
+ ' '
+ );
+
+ 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 := NULL;
- FOR xml_node IN SELECT x FROM explode_array(xml_node_list) AS x LOOP
- CONTINUE WHEN xml_node !~ E'^\\s*<';
- IF raw_text IS NOT NULL THEN
- raw_text := raw_text || joiner;
- END IF;
- raw_text := COALESCE(raw_text,'') || ARRAY_TO_STRING(
- oils_xpath( '//text()',
- 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'
- )
- ),
- ' '
- );
- END LOOP;
-
- CONTINUE WHEN raw_text IS NULL;
-
- 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'));
-
- RETURN NEXT output_row;
+ raw_text := COALESCE(raw_text,'') || curr_text;
- END LOOP;
+ -- insert raw node text for faceting
+ IF idx.facet_field THEN
+
+ output_row.field_class = idx.field_class;
+ output_row.field = idx.id;
+ output_row.source = rid;
+ output_row.value = BTRIM(REGEXP_REPLACE(curr_text, E'\\s+', ' ', 'g'));
+
+ RETURN NEXT output_row;
+ 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'));
+
+ RETURN NEXT output_row;
+ END IF;
+
+ END LOOP;
END;
$func$ LANGUAGE PLPGSQL;
diff --git a/Open-ILS/src/sql/Pg/950.data.seed-values.sql b/Open-ILS/src/sql/Pg/950.data.seed-values.sql
index d60c09dd56..3eabf7487d 100644
--- a/Open-ILS/src/sql/Pg/950.data.seed-values.sql
+++ b/Open-ILS/src/sql/Pg/950.data.seed-values.sql
@@ -11,40 +11,47 @@ INSERT INTO config.standing (id, value) VALUES (1, oils_i18n_gettext(1, 'Good',
INSERT INTO config.standing (id, value) VALUES (2, oils_i18n_gettext(2, 'Barred', 'cst', 'value'));
SELECT SETVAL('config.standing_id_seq'::TEXT, 100);
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'series', 'seriestitle', 'mods32', $$//mods32:mods/mods32:relatedItem[@type="series"]/mods32:titleInfo$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'title', 'abbreviated', 'mods32', $$//mods32:mods/mods32:titleInfo[mods32:title and (@type='abbreviated')]$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'title', 'translated', 'mods32', $$//mods32:mods/mods32:titleInfo[mods32:title and (@type='translated')]$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'title', 'alternative', 'mods32', $$//mods32:mods/mods32:titleInfo[mods32:title and (@type='alternative')]$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'title', 'uniform', 'mods32', $$//mods32:mods/mods32:titleInfo[mods32:title and (@type='uniform')]$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'title', 'proper', 'mods32', $$//mods32:mods/mods32:titleInfo[mods32:title and not (@type)]$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'author', 'corporate', 'mods32', $$//mods32:mods/mods32:name[@type='corporate']/mods32:namePart[../mods32:role/mods32:roleTerm[text()='creator']]$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'author', 'personal', 'mods32', $$//mods32:mods/mods32:name[@type='personal']/mods32:namePart[../mods32:role/mods32:roleTerm[text()='creator']]$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'author', 'conference', 'mods32', $$//mods32:mods/mods32:name[@type='conference']/mods32:namePart[../mods32:role/mods32:roleTerm[text()='creator']]$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'author', 'other', 'mods32', $$//mods32:mods/mods32:name[@type='personal']/mods32:namePart[not(../mods32:role)]$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'subject', 'geographic', 'mods32', $$//mods32:mods/mods32:subject/mods32:geographic$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'subject', 'name', 'mods32', $$//mods32:mods/mods32:subject/mods32:name$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'subject', 'temporal', 'mods32', $$//mods32:mods/mods32:subject/mods32:temporal$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'subject', 'topic', 'mods32', $$//mods32:mods/mods32:subject/mods32:topic$$ );
---INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
--- ( field_class, name, xpath ) VALUES ( 'subject', 'genre', 'mods32', $$//mods32:mods/mods32:genre$$ );
-INSERT INTO config.metabib_field ( field_class, name, format, xpath ) VALUES
- ( 'keyword', 'keyword', 'mods32', $$//mods32:mods/*[not(local-name()='originInfo')]$$ ); -- /* to fool vim */;
-INSERT INTO config.metabib_field (field_class, name, format, xpath ) VALUES
- ( 'subject', 'complete', 'mods32', $$//mods32:mods/mods32:subject//text()$$ );
+INSERT INTO config.metabib_class ( name, label ) VALUES ( 'keyword', oils_i18n_gettext('keyword', 'Keyword', 'cmc', 'name') );
+INSERT INTO config.metabib_class ( name, label ) VALUES ( 'title', oils_i18n_gettext('title', 'Title', 'cmc', 'name') );
+INSERT INTO config.metabib_class ( name, label ) VALUES ( 'author', oils_i18n_gettext('author', 'Author', 'cmc', 'name') );
+INSERT INTO config.metabib_class ( name, label ) VALUES ( 'subject', oils_i18n_gettext('subject', 'Subject', 'cmc', 'name') );
+INSERT INTO config.metabib_class ( name, label ) VALUES ( 'series', oils_i18n_gettext('series', 'Series', 'cmc', 'name') );
+
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (1, 'series', 'seriestitle', oils_i18n_gettext(1, 'Series Title', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:relatedItem[@type="series"]/mods32:titleInfo$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (2, 'title', 'abbreviated', oils_i18n_gettext(2, 'Abbreviated Title', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:titleInfo[mods32:title and (@type='abbreviated')]$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (3, 'title', 'translated', oils_i18n_gettext(3, 'Translated Title', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:titleInfo[mods32:title and (@type='translated')]$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (4, 'title', 'alternative', oils_i18n_gettext(4, 'Alternate Title', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:titleInfo[mods32:title and (@type='alternative')]$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (5, 'title', 'uniform', oils_i18n_gettext(5, 'Uniform Title', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:titleInfo[mods32:title and (@type='uniform')]$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (6, 'title', 'proper', oils_i18n_gettext(6, 'Title Proper', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:titleInfo[mods32:title and not (@type)]$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (7, 'author', 'corporate', oils_i18n_gettext(7, 'Corporate Author', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:name[@type='corporate']/mods32:namePart[../mods32:role/mods32:roleTerm[text()='creator']]$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (8, 'author', 'personal', oils_i18n_gettext(8, 'Personal Author', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:name[@type='personal']/mods32:namePart[../mods32:role/mods32:roleTerm[text()='creator']]$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (9, 'author', 'conference', oils_i18n_gettext(9, 'Conference Author', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:name[@type='conference']/mods32:namePart[../mods32:role/mods32:roleTerm[text()='creator']]$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (10, 'author', 'other', oils_i18n_gettext(10, 'Other Author', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:name[@type='personal']/mods32:namePart[not(../mods32:role)]$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (11, 'subject', 'geographic', oils_i18n_gettext(11, 'Geographic Subject', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:subject/mods32:geographic$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (12, 'subject', 'name', oils_i18n_gettext(12, 'Name Subject', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:subject/mods32:name$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (13, 'subject', 'temporal', oils_i18n_gettext(13, 'Temporal Subject', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:subject/mods32:temporal$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (14, 'subject', 'topic', oils_i18n_gettext(14, 'Topic Subject', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:subject/mods32:topic$$ );
+--INSERT INTO config.metabib_field ( id, field_class, name, format, xpath ) VALUES
+-- ( id, field_class, name, xpath ) VALUES ( 'subject', 'genre', 'mods32', $$//mods32:mods/mods32:genre$$ );
+INSERT INTO config.metabib_field ( id, field_class, name, label, format, xpath ) VALUES
+ (15, 'keyword', 'keyword', oils_i18n_gettext(15, 'General Keywords', 'cmf', 'label'), 'mods32', $$//mods32:mods/*[not(local-name()='originInfo')]$$ ); -- /* to fool vim */;
+INSERT INTO config.metabib_field (field_class, name, label, format, xpath ) VALUES
+ (16, 'subject', 'complete', oils_i18n_gettext(16, 'All Subjects', 'cmf', 'label'), 'mods32', $$//mods32:mods/mods32:subject//text()$$ );
+SELECT SETVAL('config.metabib_field_id_seq'::TEXT, (SELECT MAX(id) FROM config.metabib_field), TRUE);
INSERT INTO config.non_cataloged_type ( id, owning_lib, name ) VALUES ( 1, 1, oils_i18n_gettext(1, 'Paperback Book', 'cnct', 'name') );
SELECT SETVAL('config.non_cataloged_type_id_seq'::TEXT, 100);
diff --git a/Open-ILS/src/sql/Pg/upgrade/0213.schema.config.metabib_field_class_stuff.sql b/Open-ILS/src/sql/Pg/upgrade/0213.schema.config.metabib_field_class_stuff.sql
new file mode 100644
index 0000000000..637c8ce932
--- /dev/null
+++ b/Open-ILS/src/sql/Pg/upgrade/0213.schema.config.metabib_field_class_stuff.sql
@@ -0,0 +1,121 @@
+
+BEGIN;
+
+INSERT INTO config.upgrade_log (version) VALUES ('0213');
+
+CREATE TABLE config.metabib_class (
+ name TEXT PRIMARY KEY,
+ label TEXT NOT NULL UNIQUE
+);
+
+INSERT INTO config.metabib_class ( name, label ) VALUES ( 'keyword', oils_i18n_gettext('keyword', 'Keyword', 'cmc', 'label') );
+INSERT INTO config.metabib_class ( name, label ) VALUES ( 'title', oils_i18n_gettext('title', 'Title', 'cmc', 'label') );
+INSERT INTO config.metabib_class ( name, label ) VALUES ( 'author', oils_i18n_gettext('author', 'Author', 'cmc', 'label') );
+INSERT INTO config.metabib_class ( name, label ) VALUES ( 'subject', oils_i18n_gettext('subject', 'Subject', 'cmc', 'label') );
+INSERT INTO config.metabib_class ( name, label ) VALUES ( 'series', oils_i18n_gettext('series', 'Series', 'cmc', 'label') );
+
+ALTER TABLE config.metabib_field ADD COLUMN label TEXT;
+UPDATE config.metabib_field SET label = name;
+ALTER TABLE config.metabib_field ALTER COLUMN label SET NOT NULL;
+
+ALTER TABLE config.metabib_field ADD CONSTRAINT field_class_fkey FOREIGN KEY (field_class) REFERENCES config.metabib_class (name);
+
+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[];
+ raw_text TEXT;
+ curr_text TEXT;
+ joiner TEXT := default_joiner; -- XXX will index defs supply a joiner?
+ output_row metabib.field_entry_template%ROWTYPE;
+BEGIN
+
+ -- 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
+
+ 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 explode_array(xml_node_list) AS x LOOP
+ CONTINUE WHEN xml_node !~ E'^\\s*<';
+
+ curr_text := ARRAY_TO_STRING(
+ oils_xpath( '//text()',
+ 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'
+ )
+ ),
+ ' '
+ );
+
+ 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;
+
+ -- insert raw node text for faceting
+ IF idx.facet_field THEN
+
+ output_row.field_class = idx.field_class;
+ output_row.field = idx.id;
+ output_row.source = rid;
+ output_row.value = BTRIM(REGEXP_REPLACE(curr_text, E'\\s+', ' ', 'g'));
+
+ RETURN NEXT output_row;
+ 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'));
+
+ RETURN NEXT output_row;
+ END IF;
+
+ END LOOP;
+
+END;
+$func$ LANGUAGE PLPGSQL;
+
+
+COMMIT;
+
--
2.11.0