From 3fada5ce39206178031ec55aacb5d2bbc8707ebc Mon Sep 17 00:00:00 2001
From: Ere Maijala <ere.maijala@helsinki.fi>
Date: Fri, 6 Jan 2017 22:29:11 +0200
Subject: [PATCH] Added support for extended subject heading information (#879)
- Information regarding subject type and source can now be returned and exposed through the search API.
---
config/vufind/SearchApiRecordFields.yaml | 13 +++-
module/VuFind/src/VuFind/RecordDriver/EIT.php | 12 +++-
.../VuFind/src/VuFind/RecordDriver/Primo.php | 15 +++-
.../src/VuFind/RecordDriver/SolrDefault.php | 14 +++-
.../src/VuFind/RecordDriver/SolrMarc.php | 71 ++++++++++++++++---
.../VuFind/src/VuFind/RecordDriver/Summon.php | 53 ++++++++------
.../VuFindTest/RecordDriver/SolrMarcTest.php | 27 +++++++
.../VuFindApi/Formatter/RecordFormatter.php | 15 ++++
.../Formatter/RecordFormatterTest.php | 14 ++--
themes/root/templates/searchapi/swagger.phtml | 61 ++++++++++------
10 files changed, 232 insertions(+), 63 deletions(-)
diff --git a/config/vufind/SearchApiRecordFields.yaml b/config/vufind/SearchApiRecordFields.yaml
index 5d239b76201..4f67a719b64 100644
--- a/config/vufind/SearchApiRecordFields.yaml
+++ b/config/vufind/SearchApiRecordFields.yaml
@@ -321,10 +321,19 @@ shortTitle:
subjects:
vufind.method: getAllSubjectHeadings
vufind.default: true
- description: Subject headings
+ description: >
+ Subject headings as an array from the least specific to the most specific
type: array
items:
- type: string
+ type: array
+ items:
+ type: string
+subjectsExtended:
+ vufind.method: "Formatter::getExtendedSubjectHeadings"
+ description: Subject headings with type and source information
+ type: array
+ items:
+ $ref: '#/definitions/Subject'
subTitle:
vufind.method: getSubTitle
description: Subtitle
diff --git a/module/VuFind/src/VuFind/RecordDriver/EIT.php b/module/VuFind/src/VuFind/RecordDriver/EIT.php
index 009d4d53f3b..64b471f8692 100644
--- a/module/VuFind/src/VuFind/RecordDriver/EIT.php
+++ b/module/VuFind/src/VuFind/RecordDriver/EIT.php
@@ -83,9 +83,15 @@ class EIT extends SolrDefault
* returned as an array of chunks, increasing from least specific to most
* specific.
*
+ * @param bool $extended Whether to return a keyed array with the following
+ * keys:
+ * - heading: the actual subject heading chunks
+ * - type: heading type
+ * - source: source vocabulary
+ *
* @return array
*/
- public function getAllSubjectHeadings()
+ public function getAllSubjectHeadings($extended = false)
{
$su = isset($this->controlInfo['artinfo']['su'])
? $this->controlInfo['artinfo']['su'] : [];
@@ -94,7 +100,9 @@ class EIT extends SolrDefault
// format, so we'll just send each value as a single chunk.
$retval = [];
foreach ($su as $s) {
- $retval[] = [$s];
+ $retval[] = $extended
+ ? ['heading' => [$s], 'type' => '', 'source' => '']
+ : [$s];
}
return $retval;
}
diff --git a/module/VuFind/src/VuFind/RecordDriver/Primo.php b/module/VuFind/src/VuFind/RecordDriver/Primo.php
index 3641155241c..fee58726efc 100644
--- a/module/VuFind/src/VuFind/RecordDriver/Primo.php
+++ b/module/VuFind/src/VuFind/RecordDriver/Primo.php
@@ -92,14 +92,23 @@ class Primo extends SolrDefault
* Get an array of all subject headings associated with the record
* (may be empty).
*
+ * @param bool $extended Whether to return a keyed array with the following
+ * keys:
+ * - heading: the actual subject heading chunks
+ * - type: heading type
+ * - source: source vocabulary
+ *
* @return array
*/
- public function getAllSubjectHeadings()
+ public function getAllSubjectHeadings($extended = false)
{
$base = isset($this->fields['subjects'])
? $this->fields['subjects'] : [];
- $callback = function ($str) {
- return array_map('trim', explode(' -- ', $str));
+ $callback = function ($str) use ($extended) {
+ $s = array_map('trim', explode(' -- ', $str));
+ return $extended
+ ? ['heading' => $s, 'type' => '', 'source' => '']
+ : $s;
};
return array_map($callback, $base);
}
diff --git a/module/VuFind/src/VuFind/RecordDriver/SolrDefault.php b/module/VuFind/src/VuFind/RecordDriver/SolrDefault.php
index 85ceedb4e01..e9ffad89c05 100644
--- a/module/VuFind/src/VuFind/RecordDriver/SolrDefault.php
+++ b/module/VuFind/src/VuFind/RecordDriver/SolrDefault.php
@@ -199,9 +199,15 @@ class SolrDefault extends AbstractBase
* returned as an array of chunks, increasing from least specific to most
* specific.
*
+ * @param bool $extended Whether to return a keyed array with the following
+ * keys:
+ * - heading: the actual subject heading chunks
+ * - type: heading type
+ * - source: source vocabulary
+ *
* @return array
*/
- public function getAllSubjectHeadings()
+ public function getAllSubjectHeadings($extended = false)
{
$headings = [];
foreach (['topic', 'geographic', 'genre', 'era'] as $field) {
@@ -213,8 +219,10 @@ class SolrDefault extends AbstractBase
// The Solr index doesn't currently store subject headings in a broken-down
// format, so we'll just send each value as a single chunk. Other record
// drivers (i.e. MARC) can offer this data in a more granular format.
- $callback = function ($i) {
- return [$i];
+ $callback = function ($i) use ($extended) {
+ return $extended
+ ? ['heading' => [$i], 'type' => '', 'source' => '']
+ : [$i];
};
return array_map($callback, array_unique($headings));
}
diff --git a/module/VuFind/src/VuFind/RecordDriver/SolrMarc.php b/module/VuFind/src/VuFind/RecordDriver/SolrMarc.php
index d16f7440294..1e9d6995179 100644
--- a/module/VuFind/src/VuFind/RecordDriver/SolrMarc.php
+++ b/module/VuFind/src/VuFind/RecordDriver/SolrMarc.php
@@ -53,6 +53,42 @@ class SolrMarc extends SolrDefault
*/
protected $lazyMarcRecord = null;
+ /**
+ * Fields that may contain subject headings, and their descriptions
+ *
+ * @var array
+ */
+ protected $subjectFields = [
+ '600' => 'personal name',
+ '610' => 'corporate name',
+ '611' => 'meeting name',
+ '630' => 'uniform title',
+ '648' => 'chronological',
+ '650' => 'topic',
+ '651' => 'geographic',
+ '653' => '',
+ '655' => 'genre/form',
+ '656' => 'occupation'
+ ];
+
+ /**
+ * Mappings from subject source indicators (2nd indicator of subject fields in
+ * MARC 21) to the their codes.
+ *
+ * @var array
+ * @link https://www.loc.gov/marc/bibliographic/bd6xx.html Subject field docs
+ * @link https://www.loc.gov/standards/sourcelist/subject.html Code list
+ */
+ protected $subjectSources = [
+ '0' => 'lcsh',
+ '1' => 'lcshac',
+ '2' => 'mesh',
+ '3' => 'nal',
+ '4' => 'unknown',
+ '5' => 'cash',
+ '6' => 'rvm'
+ ];
+
/**
* Get access restriction notes for the record.
*
@@ -68,20 +104,21 @@ class SolrMarc extends SolrDefault
* returned as an array of chunks, increasing from least specific to most
* specific.
*
+ * @param bool $extended Whether to return a keyed array with the following
+ * keys:
+ * - heading: the actual subject heading chunks
+ * - type: heading type
+ * - source: source vocabulary
+ *
* @return array
*/
- public function getAllSubjectHeadings()
+ public function getAllSubjectHeadings($extended = false)
{
- // These are the fields that may contain subject headings:
- $fields = [
- '600', '610', '611', '630', '648', '650', '651', '653', '655', '656'
- ];
-
// This is all the collected data:
$retval = [];
// Try each MARC field one at a time:
- foreach ($fields as $field) {
+ foreach ($this->subjectFields as $field => $fieldType) {
// Do we have any results for the current field? If not, try the next.
$results = $this->getMarcRecord()->getFields($field);
if (!$results) {
@@ -105,7 +142,25 @@ class SolrMarc extends SolrDefault
}
// If we found at least one chunk, add a heading to our result:
if (!empty($current)) {
- $retval[] = $current;
+ if ($extended) {
+ $sourceIndicator = $result->getIndicator(2);
+ $source = '';
+ if (isset($this->subjectSources[$sourceIndicator])) {
+ $source = $this->subjectSources[$sourceIndicator];
+ } else {
+ $source = $result->getSubfield('2');
+ if ($source) {
+ $source = $source->getData();
+ }
+ }
+ $retval[] = [
+ 'heading' => $current,
+ 'type' => $fieldType,
+ 'source' => $source ?: ''
+ ];
+ } else {
+ $retval[] = $current;
+ }
}
}
}
diff --git a/module/VuFind/src/VuFind/RecordDriver/Summon.php b/module/VuFind/src/VuFind/RecordDriver/Summon.php
index e229322991f..79b49428cca 100644
--- a/module/VuFind/src/VuFind/RecordDriver/Summon.php
+++ b/module/VuFind/src/VuFind/RecordDriver/Summon.php
@@ -38,6 +38,18 @@ namespace VuFind\RecordDriver;
*/
class Summon extends SolrDefault
{
+ /**
+ * Fields that may contain subject headings, and their descriptions
+ *
+ * @var array
+ */
+ protected $subjectFields = [
+ 'SubjectTerms' => 'topic',
+ 'TemporalSubjectTerms' => 'chronological',
+ 'GeographicLocations' => 'geographic',
+ 'Keywords' => 'keyword',
+ ];
+
/**
* Date converter
*
@@ -50,32 +62,31 @@ class Summon extends SolrDefault
* returned as an array of chunks, increasing from least specific to most
* specific.
*
+ * @param bool $extended Whether to return a keyed array with the following
+ * keys:
+ * - heading: the actual subject heading chunks
+ * - type: heading type
+ * - source: source vocabulary
+ *
* @return array
*/
- public function getAllSubjectHeadings()
+ public function getAllSubjectHeadings($extended = false)
{
$retval = [];
- $topic = isset($this->fields['SubjectTerms']) ?
- $this->fields['SubjectTerms'] : [];
- $temporal = isset($this->fields['TemporalSubjectTerms']) ?
- $this->fields['TemporalSubjectTerms'] : [];
- $geo = isset($this->fields['GeographicLocations']) ?
- $this->fields['GeographicLocations'] : [];
- $key = isset($this->fields['Keywords']) ?
- $this->fields['Keywords'] : [];
- $retval = [];
- foreach ($topic as $t) {
- $retval[] = [trim($t)];
- }
- foreach ($temporal as $t) {
- $retval[] = [trim($t)];
- }
- foreach ($geo as $g) {
- $retval[] = [trim($g)];
- }
- foreach ($key as $k) {
- $retval[] = [trim($k)];
+ foreach ($this->subjectFields as $field => $fieldType) {
+ if (!isset($this->fields[$field])) {
+ continue;
+ }
+ foreach ($this->fields[$field] as $topic) {
+ $topic = trim($topic);
+ $retval[] = $extended
+ ? [
+ 'heading' => [$topic],
+ 'type' => $fieldType,
+ 'source' => ''
+ ] : [$topic];
+ }
}
return $retval;
}
diff --git a/module/VuFind/tests/unit-tests/src/VuFindTest/RecordDriver/SolrMarcTest.php b/module/VuFind/tests/unit-tests/src/VuFindTest/RecordDriver/SolrMarcTest.php
index 24875095075..458895121e3 100644
--- a/module/VuFind/tests/unit-tests/src/VuFindTest/RecordDriver/SolrMarcTest.php
+++ b/module/VuFind/tests/unit-tests/src/VuFindTest/RecordDriver/SolrMarcTest.php
@@ -93,6 +93,33 @@ class SolrMarcTest extends \VuFindTest\Unit\TestCase
$this->assertEquals('2, pt. 1.', $series[0]['number']);
}
+ /**
+ * Test regular and extended subject heading support.
+ *
+ * @return void
+ */
+ public function testSubjectHeadings()
+ {
+ $config = new \Zend\Config\Config([]);
+ $record = new \VuFind\RecordDriver\SolrMarc($config);
+ $fixture = $this->loadRecordFixture('testbug1.json');
+ $record->setRawData($fixture['response']['docs'][0]);
+ $this->assertEquals(
+ [['Matematica', 'Periodici.']],
+ $record->getAllSubjectHeadings()
+ );
+ $this->assertEquals(
+ [
+ [
+ 'heading' => ['Matematica', 'Periodici.'],
+ 'type' => '',
+ 'source' => ''
+ ],
+ ],
+ $record->getAllSubjectHeadings(true)
+ );
+ }
+
/**
* Load a fixture file.
*
diff --git a/module/VuFindApi/src/VuFindApi/Formatter/RecordFormatter.php b/module/VuFindApi/src/VuFindApi/Formatter/RecordFormatter.php
index 53d88466157..063259d0028 100644
--- a/module/VuFindApi/src/VuFindApi/Formatter/RecordFormatter.php
+++ b/module/VuFindApi/src/VuFindApi/Formatter/RecordFormatter.php
@@ -85,6 +85,21 @@ class RecordFormatter extends BaseFormatter
return $result ? $result : null;
}
+ /**
+ * Get extended subject headings
+ *
+ * @param \VuFind\RecordDriver\SolrDefault $record Record driver
+ *
+ * @return array|null
+ */
+ protected function getExtendedSubjectHeadings($record)
+ {
+ $result = $record->getAllSubjectHeadings(true);
+ // Make sure that the record driver returned the additional information and
+ // return data only if it did
+ return $result && isset($result[0]['heading']) ? $result : null;
+ }
+
/**
* Get full record for a record as XML
*
diff --git a/module/VuFindApi/tests/unit-tests/src/VuFindTest/Formatter/RecordFormatterTest.php b/module/VuFindApi/tests/unit-tests/src/VuFindTest/Formatter/RecordFormatterTest.php
index 5f186bf3adc..61f14068408 100644
--- a/module/VuFindApi/tests/unit-tests/src/VuFindTest/Formatter/RecordFormatterTest.php
+++ b/module/VuFindApi/tests/unit-tests/src/VuFindTest/Formatter/RecordFormatterTest.php
@@ -63,7 +63,10 @@ class RecordFormatterTest extends \VuFindTest\Unit\TestCase
'fullRecord' => ['vufind.method' => 'Formatter::getFullRecord'],
'rawData' => ['vufind.method' => 'Formatter::getRawData'],
'buildings' => ['vufind.method' => 'getBuilding'],
- 'recordPage' => ['vufind.method' => 'Formatter::getRecordPage']
+ 'recordPage' => ['vufind.method' => 'Formatter::getRecordPage'],
+ 'subjectsExtended' => [
+ 'vufind.method' => 'Formatter::getExtendedSubjectHeadings'
+ ],
];
}
@@ -113,7 +116,8 @@ class RecordFormatterTest extends \VuFindTest\Unit\TestCase
'DedupData' => [['id' => 'bar']],
'fullrecord' => 'xyzzy',
'spelling' => 's',
- 'Building' => ['foo', new TranslatableString('bar', 'xyzzy')]
+ 'Building' => ['foo', new TranslatableString('bar', 'xyzzy')],
+ 'AllSubjectHeadings' => [['heading' => 'subject']],
]
);
return $driver;
@@ -149,7 +153,8 @@ class RecordFormatterTest extends \VuFindTest\Unit\TestCase
'fullRecord' => 'xyzzy',
'rawData' => $expectedRaw,
'buildings' => ['foo', ['value' => 'bar', 'translated' => 'xyzzy']],
- 'recordPage' => 'http://record'
+ 'recordPage' => 'http://record',
+ 'subjectsExtended' => [['heading' => 'subject']],
],
];
$this->assertEquals($expected, $results);
@@ -187,7 +192,8 @@ class RecordFormatterTest extends \VuFindTest\Unit\TestCase
'fullRecord' => [],
'rawData' => [],
'buildings' => [],
- 'recordPage' => []
+ 'recordPage' => [],
+ 'subjectsExtended' => [],
];
$this->assertEquals($expected, $results);
}
diff --git a/themes/root/templates/searchapi/swagger.phtml b/themes/root/templates/searchapi/swagger.phtml
index e7b05359ade..abc463709de 100644
--- a/themes/root/templates/searchapi/swagger.phtml
+++ b/themes/root/templates/searchapi/swagger.phtml
@@ -267,26 +267,6 @@
}
}
},
- "Record": {
- "type": "object",
- "properties": <?=json_encode($this->recordFields) ?>
- },
- "RecordLink": {
- "type": "object",
- "properties": {
- "title": {
- "description": "Link title",
- "type": "string"
- },
- "value": {
- "description": "Link value",
- "type": "string"
- },
- "link": {
- "$ref": "#/definitions/Link"
- }
- }
- },
"Link": {
"type": "object",
"properties": {
@@ -307,6 +287,26 @@
}
}
},
+ "Record": {
+ "type": "object",
+ "properties": <?=json_encode($this->recordFields) ?>
+ },
+ "RecordLink": {
+ "type": "object",
+ "properties": {
+ "title": {
+ "description": "Link title",
+ "type": "string"
+ },
+ "value": {
+ "description": "Link value",
+ "type": "string"
+ },
+ "link": {
+ "$ref": "#/definitions/Link"
+ }
+ }
+ },
"SearchResponse": {
"type": "object",
"properties": {
@@ -336,6 +336,27 @@
},
"required": ["resultCount", "status"]
},
+ "Subject": {
+ "type": "object",
+ "properties": {
+ "heading": {
+ "description": "Subject heading parts as an array from the least specific to the most specific",
+ "type": "array",
+ "items": {
+ "type": "string"
+ }
+ },
+ "type": {
+ "description": "Subject type",
+ "type": "string",
+ "enum": ["", "personal name", "corporate name", "meeting name", "uniform title", "chronological", "topic", "geographic", "genre\/form", "occupation", "keyword"]
+ },
+ "source": {
+ "description": "Subject source/thesaurus (e.g. lcsh, mesh)",
+ "type": "string"
+ }
+ }
+ },
"Url": {
"type": "object",
"properties": {
--
GitLab