From 2e45e5ef1961ba5904cf9f4a00d0fb1d61b294e5 Mon Sep 17 00:00:00 2001
From: Demian Katz <demian.katz@villanova.edu>
Date: Fri, 7 Jun 2013 08:13:21 -0400
Subject: [PATCH] Progress on VUFIND-701 (Dewey AlphaBrowse skips records w/
multiple call numbers) Some edge cases won't be fixed until next SolrMarc
upgrade (need List support)
---
import/index_scripts/dewey.bsh | 52 +++++++++++++++----
import/marc.properties | 3 +-
index-alphabetic-browse.bat | 2 +-
index-alphabetic-browse.sh | 2 +-
.../src/Connection/SolrTest.php | 21 ++++++++
solr/biblio/conf/schema.xml | 3 +-
tests/data/deweybrowse.mrc | 1 +
7 files changed, 70 insertions(+), 14 deletions(-)
create mode 100644 tests/data/deweybrowse.mrc
diff --git a/import/index_scripts/dewey.bsh b/import/index_scripts/dewey.bsh
index 1f0fb3df50a..cad5c431b52 100644
--- a/import/index_scripts/dewey.bsh
+++ b/import/index_scripts/dewey.bsh
@@ -15,8 +15,8 @@ import org.solrmarc.tools.CallNumUtils;
*
* @param record
* @param fieldSpec - which MARC fields / subfields need to be analyzed
- * @param precisionStr - a decimal number (represented in string format) showing the
- * desired precision of the returned number; i.e. 100 to round to nearest hundred,
+ * @param precisionStr - a decimal number (represented in string format) showing the
+ * desired precision of the returned number; i.e. 100 to round to nearest hundred,
* 10 to round to nearest ten, 0.1 to round to nearest tenth, etc.
* @return Set containing requested numeric portions of Dewey decimal call numbers
*/
@@ -24,23 +24,23 @@ public Set getDeweyNumber(Record record, String fieldSpec, String precisionStr)
// Initialize our return value:
Set result = new LinkedHashSet();
- // Precision comes in as a string, but we need to convert it to a float:
+ // Precision comes in as a string, but we need to convert it to a float:
float precision = Float.parseFloat(precisionStr);
-
+
// Loop through the specified MARC fields:
Set input = indexer.getFieldList(record, fieldSpec);
Iterator iter = input.iterator();
while (iter.hasNext()) {
// Get the current string to work on:
String current = iter.next();
-
+
if (CallNumUtils.isValidDewey(current)) {
// Convert the numeric portion of the call number into a float:
float currentVal = Float.parseFloat(CallNumUtils.getDeweyB4Cutter(current));
-
+
// Round the call number value to the specified precision:
Float finalVal = new Float(Math.floor(currentVal / precision) * precision);
-
+
// Convert the rounded value back to a string (with leading zeros) and save it:
result.add(CallNumUtils.normalizeFloat(finalVal.toString(), 3, -1));
}
@@ -71,14 +71,14 @@ public Set getDeweySearchable(Record record, String fieldSpec) {
while (iter.hasNext()) {
// Get the current string to work on:
String current = iter.next();
-
+
// Add valid strings to the set, normalizing them to be all uppercase
// and free from whitespace.
if (CallNumUtils.isValidDewey(current)) {
result.add(current.toUpperCase().replaceAll(" ", ""));
}
}
-
+
// If we found no call numbers, return null; otherwise, return our results:
if (result.isEmpty())
return null;
@@ -108,7 +108,39 @@ public String getDeweySortable(Record record, String fieldSpec) {
return CallNumUtils.getDeweyShelfKey(current);
}
}
-
+
// If we made it this far, we didn't find a valid sortable Dewey number:
return null;
+}
+
+/**
+ * Normalize Dewey numbers for AlphaBrowse sorting purposes (use all numbers!)
+ *
+ * Can return null
+ *
+ * @param record
+ * @param fieldSpec - which MARC fields / subfields need to be analyzed
+ * @return Set containing normalized Dewey numbers extracted from specified fields.
+ */
+public Set getDeweySortables(Record record, String fieldSpec) {
+ // Initialize our return value:
+ Set result = new LinkedHashSet();
+
+ // Loop through the specified MARC fields:
+ Set input = indexer.getFieldList(record, fieldSpec);
+ Iterator iter = input.iterator();
+ while (iter.hasNext()) {
+ // Get the current string to work on:
+ String current = iter.next();
+
+ // If this is a valid Dewey number, return the sortable shelf key:
+ if (CallNumUtils.isValidDewey(current)) {
+ result.add(CallNumUtils.getDeweyShelfKey(current));
+ }
+ }
+
+ // If we found no call numbers, return null; otherwise, return our results:
+ if (result.isEmpty())
+ return null;
+ return result;
}
\ No newline at end of file
diff --git a/import/marc.properties b/import/marc.properties
index 570b21c708c..4a1eaedff66 100644
--- a/import/marc.properties
+++ b/import/marc.properties
@@ -76,7 +76,8 @@ dewey-tens = custom, getDeweyNumber(082a:083a, 10), ddc22_map.properties(tens)
dewey-ones = custom, getDeweyNumber(082a:083a, 1), ddc22_map.properties(ones)
dewey-full = custom, getDeweySearchable(082a:083a)
dewey-sort = custom, getDeweySortable(082a:083a)
-dewey-raw = 082a:083a, first
+dewey-sort-browse = script(dewey.bsh), getDeweySortables(082a:083a)
+dewey-raw = 082a:083a
# Extract the numeric portion of the OCLC number using a pattern map:
oclc_num = 035a, (pattern_map.oclc_num)
diff --git a/index-alphabetic-browse.bat b/index-alphabetic-browse.bat
index 842a96245bc..2b740cc72cc 100644
--- a/index-alphabetic-browse.bat
+++ b/index-alphabetic-browse.bat
@@ -64,7 +64,7 @@ call %VUFIND_HOME%\index-alphabetic-browse.bat build_browse title title_fullStr
call %VUFIND_HOME%\index-alphabetic-browse.bat build_browse topic topic_browse
call %VUFIND_HOME%\index-alphabetic-browse.bat build_browse author author_browse
call %VUFIND_HOME%\index-alphabetic-browse.bat build_browse lcc callnumber-a 1
-call %VUFIND_HOME%\index-alphabetic-browse.bat build_browse dewey dewey-raw 1 "-Dbibleech=StoredFieldLeech -Dsortfield=dewey-sort -Dvaluefield=dewey-raw"
+call %VUFIND_HOME%\index-alphabetic-browse.bat build_browse dewey dewey-raw 1 "-Dbibleech=StoredFieldLeech -Dsortfield=dewey-sort-browse -Dvaluefield=dewey-raw"
goto end
rem Function to process a single browse index:
diff --git a/index-alphabetic-browse.sh b/index-alphabetic-browse.sh
index 1aa95d5db12..b9db27164c7 100755
--- a/index-alphabetic-browse.sh
+++ b/index-alphabetic-browse.sh
@@ -39,4 +39,4 @@ build_browse "title" "title_fullStr" 1 "-Dbibleech=StoredFieldLeech -Dsortfield=
build_browse "topic" "topic_browse"
build_browse "author" "author_browse"
build_browse "lcc" "callnumber-a" 1
-build_browse "dewey" "dewey-raw" 1 "-Dbibleech=StoredFieldLeech -Dsortfield=dewey-sort -Dvaluefield=dewey-raw"
+build_browse "dewey" "dewey-raw" 1 "-Dbibleech=StoredFieldLeech -Dsortfield=dewey-sort-browse -Dvaluefield=dewey-raw"
diff --git a/module/VuFind/tests/integration-tests/src/Connection/SolrTest.php b/module/VuFind/tests/integration-tests/src/Connection/SolrTest.php
index fd305d6ca58..67a4a9fdc29 100644
--- a/module/VuFind/tests/integration-tests/src/Connection/SolrTest.php
+++ b/module/VuFind/tests/integration-tests/src/Connection/SolrTest.php
@@ -73,4 +73,25 @@ class SolrTest extends \VuFindTest\Unit\TestCase
$this->assertTrue(empty($item['seeAlso']));
$this->assertTrue(in_array('Royal Dublin Society', $item['useInstead']));
}
+
+ /**
+ * Check that expected Dewey values are present (tests VUFIND-701).
+ *
+ * @return void
+ */
+ public function testDeweyValues()
+ {
+ $solr = $this->getServiceManager()->get('VuFind\Search\BackendManager')
+ ->get('Solr');
+ $result = $solr->alphabeticBrowse('dewey', '123.45 .I39', 0, 1);
+ $item = $result['Browse']['items'][0];
+ $this->assertEquals(1, $item['count']);
+ $this->assertEquals($item['count'], count($item['ids']));
+ $this->assertEquals('123.45 .I39', $item['heading']);
+ $result = $solr->alphabeticBrowse('dewey', '123.46 .Q39', 0, 1);
+ $item = $result['Browse']['items'][0];
+ $this->assertEquals(1, $item['count']);
+ $this->assertEquals($item['count'], count($item['ids']));
+ $this->assertEquals('123.46 .Q39', $item['heading']);
+ }
}
\ No newline at end of file
diff --git a/solr/biblio/conf/schema.xml b/solr/biblio/conf/schema.xml
index c10198be83b..ed6b0fb2d9f 100644
--- a/solr/biblio/conf/schema.xml
+++ b/solr/biblio/conf/schema.xml
@@ -154,7 +154,8 @@
<field name="dewey-ones" type="string" indexed="true" stored="true" multiValued="true"/>
<field name="dewey-full" type="string" indexed="true" stored="true" multiValued="true"/>
<field name="dewey-sort" type="string" indexed="true" stored="true" />
- <field name="dewey-raw" type="string" indexed="true" stored="true" />
+ <field name="dewey-sort-browse" type="string" indexed="true" stored="true" multiValued="true" />
+ <field name="dewey-raw" type="string" indexed="true" stored="true" multiValued="true" />
<field name="author2" type="textProper" indexed="true" stored="true" multiValued="true"/>
<field name="author2Str" type="string" indexed="true" stored="true" multiValued="true"/>
<field name="author2-role" type="string" indexed="true" stored="true" multiValued="true"/>
diff --git a/tests/data/deweybrowse.mrc b/tests/data/deweybrowse.mrc
new file mode 100644
index 00000000000..2830770ea9a
--- /dev/null
+++ b/tests/data/deweybrowse.mrc
@@ -0,0 +1 @@
+00613cam a2200229Ma 4500001001600000005001700016008004100033020001500074035002300089040002500112041001800137043001200155050002400167049000900191082001600200082001600216100003000232245002200262250002300284260004700307300002900354 testdeweybrowse 20110419140028.0 110214s1992 it a b 001 0 ita d a8820737493 a(OCoLC)ocm30585539 aRBN cRBN dOCLCG dPVU 1 aita alat hlat ae-it--- 14 aDG848.15 b.V53 1992 aPVUM a123.45 .I39 a123.46 .Q39 1 aPerson, Fake, d1668-1744. 10 aDewey browse test aFictional edition. aMorano : bCentro di Studi Vichiani, c1992. a296 p. : bill. ; c24 cm.
\ No newline at end of file
--
GitLab