Merge #2024 from branch '1058-enrichWithRvk' of github.com:hbz/lobid-…

…resources
hbz · Jun 17, 2024 · 57f318e · 57f318e
2 parents f50dfb3 + 8cca4e8
commit 57f318e
Show file tree

Hide file tree

Showing 14 changed files with 60 additions and 14 deletions.
diff --git a/src/main/java/org/lobid/resources/run/AlmaMarcXmlFix2lobidJsonEs.java b/src/main/java/org/lobid/resources/run/AlmaMarcXmlFix2lobidJsonEs.java
@@ -125,6 +125,7 @@ public void run() {
                 fixVariables.put("classification.tsv", "./maps/classification.tsv");
                 fixVariables.put("formangabe.tsv", "./maps/formangabe.tsv");
                 fixVariables.put("almaMmsId2rpbId", "../../../../../../lookup-tables/data/almaMmsId2rpbId.tsv");
+                fixVariables.put("rvk.tsv", "../../../../../../lookup-tables/data/rvk.tsv");
                 fixVariables.put("lobidOrganisationsMapping.tsv", "./maps/lobidOrganisationsMapping.tsv");
                 fixVariables.put("hbzowner2sigel.tsv", "./maps/hbzowner2sigel.tsv");
                 fixVariables.put("rpb2.ttl", "../../../../../../vocabs/rpb/rpb2.ttl");

diff --git a/...run/CulturegraphXmlFilterHbzRvkToCsv.java → ...run/CulturegraphXmlFilterHbzRvkToTsv.java b/...run/CulturegraphXmlFilterHbzRvkToCsv.java → ...run/CulturegraphXmlFilterHbzRvkToTsv.java
@@ -21,8 +21,8 @@
  * @author Pascal Christoph (dr0i)
  * @author Tobias Bülte (TobiasNx)
  **/
-public final class CulturegraphXmlFilterHbzRvkToCsv {
-	private static String OUTPUT_FILE="cg-concordance.csv";
+public final class CulturegraphXmlFilterHbzRvkToTsv {
+	private static String OUTPUT_FILE="rvk.tsv";
 
 	public static void main(String... args) {
 		String XML_INPUT_FILE = new File(args[0]).getAbsolutePath();
@@ -32,12 +32,16 @@ public static void main(String... args) {
 		final FileOpener opener = new FileOpener();
 		JsonDecoder jsonDecoder = new JsonDecoder();
 		jsonDecoder.setRecordPath("records");
+		CsvEncoder csvEncoder = new CsvEncoder();
+		csvEncoder.setSeparator("\t");
+		csvEncoder.setNoQuotes(true);
+
 		try {
 			opener.setReceiver(new XmlDecoder()).setReceiver(new MarcXmlHandler())
-					.setReceiver(new Metafix("src/main/resources/rvk/cg-to-rvk-csv.fix"))
+					.setReceiver(new Metafix("src/main/resources/rvk/cg-to-rvk-tsv.fix"))
 					.setReceiver(new JsonEncoder())
 					.setReceiver(jsonDecoder)
-					.setReceiver(new CsvEncoder())
+					.setReceiver(csvEncoder)
 					.setReceiver(new ObjectWriter<>(OUTPUT_FILE));
 		} catch (IOException e) {
 			e.printStackTrace();

diff --git a/src/main/resources/alma/fix/maps.fix b/src/main/resources/alma/fix/maps.fix
@@ -26,6 +26,9 @@ put_filemap("$[dnbSachgruppen]","dnbSachgruppen",sep_char:"\t")
 put_filemap("$[classification.tsv]","classificationCode2Label", sep_char:"\t",key_column:"0",value_column:"1",expected_columns:"-1")
 put_filemap("$[classification.tsv]","classificationCode2Uri", sep_char:"\t",key_column:"0",value_column:"2",expected_columns:"-1")
 
+# RVK via Culturegraph
+put_filemap("$[rvk.tsv]","rvk", sep_char:"\t")
+
 # RPB SKOS Maps
 put_rdfmap("$[rpb2.ttl]", "rpb2", target: "skos:prefLabel", select_language: "de") # LBZ-Notationen aka rpb2 fka RPB-Sachgruppen und Zusätze
 put_rdfmap("$[rpb-spatial.ttl]", "rpbr", target: "skos:prefLabel", select_language: "de") # RPB-Raumsystematik aka rpbr

diff --git a/src/main/resources/alma/fix/subjects.fix b/src/main/resources/alma/fix/subjects.fix
@@ -328,6 +328,18 @@ do list(path:"084??", "var":"$i")
   end
 end
 
+# RVK via Culturegraph
+
+copy_field("almaMmsId","@rvkNotations")
+lookup("@rvkNotations","rvk",delete:"true")
+split_field("@rvkNotations", ",")
+do list(path: "@rvkNotations","var":"$i")
+  copy_field("$i","subject[].$append.notation")
+  set_array("subject[].$last.type[]","Concept")
+  add_field("subject[].$last.source.label","RVK (Regensburger Verbundklassifikation)")
+  add_field("subject[].$last.source.id","https://d-nb.info/gnd/4449787-8")
+end
+
 # 689 RSWK Schlagwortfolgen fka: Schlagwortketten 1 - 10 - no info on repeatability
 
 call_macro("schlagwortfolge", field: "6890?")

diff --git a/src/main/resources/rvk/cg-to-rvk-csv.fix → src/main/resources/rvk/cg-to-rvk-tsv.fix b/src/main/resources/rvk/cg-to-rvk-csv.fix → src/main/resources/rvk/cg-to-rvk-tsv.fix
diff --git a/src/test/java/UnitTests.java b/src/test/java/UnitTests.java
@@ -13,7 +13,7 @@
 @Suite.SuiteClasses({
     TestGenerateContext.class,
     org.lobid.resources.AlmaMarc21XmlToLobidJsonMetafixTest.class,
-    org.lobid.resources.CulturegraphXmlFilterHbzRvkToCsvTest.class,
+    org.lobid.resources.CulturegraphXmlFilterHbzRvkToTsvTest.class,
     org.lobid.resources.CulturegraphXmlFilterHbzToJsonTest.class})
 
 public final class UnitTests {

diff --git a/src/test/java/org/lobid/resources/AlmaMarc21XmlToLobidJsonMetafixTest.java b/src/test/java/org/lobid/resources/AlmaMarc21XmlToLobidJsonMetafixTest.java
@@ -69,6 +69,7 @@ public void setup() {
         fixVariables.put("picaCreatorId2Isil.tsv", "src/main/resources/alma/maps/picaCreatorId2Isil.tsv");
         fixVariables.put("nwbibWikidataLabelTypeCoords.tsv", "src/main/resources/alma/maps/nwbibWikidataLabelTypeCoords.tsv");
         fixVariables.put("almaMmsId2rpbId", "src/test/resources/alma/maps/almaMmsId2rpbId.tsv");
+        fixVariables.put("rvk.tsv", "src/test/resources/cg/rvk.tsv");
         fixVariables.put("lobidOrganisationsMapping.tsv", "src/test/resources/alma/maps/lobidOrganisationsMapping.tsv");
         fixVariables.put("hbzowner2sigel.tsv", "src/main/resources/alma/maps/hbzowner2sigel.tsv");
         fixVariables.put("rpb2.ttl", "src/test/resources/alma/maps/rpb2.ttl");

diff --git a/...CulturegraphXmlFilterHbzRvkToCsvTest.java → ...CulturegraphXmlFilterHbzRvkToTsvTest.java b/...CulturegraphXmlFilterHbzRvkToCsvTest.java → ...CulturegraphXmlFilterHbzRvkToTsvTest.java
@@ -11,29 +11,29 @@
 import org.slf4j.LoggerFactory;
 
 import org.junit.Test;
-import org.lobid.resources.run.CulturegraphXmlFilterHbzRvkToCsv;
+import org.lobid.resources.run.CulturegraphXmlFilterHbzRvkToTsv;
 
 /**
  * Test of filtering resources with hbz holdings from culturegraph MARCXML,
  * tranforming into a CSV file.
  *
  * @author Pascal Christoph(dr0i)
  **/
-public final class CulturegraphXmlFilterHbzRvkToCsvTest {
+public final class CulturegraphXmlFilterHbzRvkToTsvTest {
 
 	private static final Logger LOG =
-			LoggerFactory.getLogger(CulturegraphXmlFilterHbzRvkToCsvTest.class);
+			LoggerFactory.getLogger(CulturegraphXmlFilterHbzRvkToTsvTest.class);
 
 	private static final String PATH_TO_TEST = "src/test/resources/";
 	public static final String OUTPUT_FILE =
-			PATH_TO_TEST + "cg/output.csv";
+			PATH_TO_TEST + "cg/rvk.tsv";
 
 	private static final String XML_INPUT_FILE = "cg/aggregate_20240507_example.marcxml";
 
 	@SuppressWarnings("static-method")
 	@Test
 	public void testExtractLookupTableFromCgAsHbzRvk() {
-		CulturegraphXmlFilterHbzRvkToCsv.main(PATH_TO_TEST + XML_INPUT_FILE,
+		CulturegraphXmlFilterHbzRvkToTsv.main(PATH_TO_TEST + XML_INPUT_FILE,
 				OUTPUT_FILE);
 	}
 

diff --git a/src/test/java/org/lobid/resources/CulturegraphXmlFilterHbzToJsonTest.java b/src/test/java/org/lobid/resources/CulturegraphXmlFilterHbzToJsonTest.java
@@ -49,7 +49,7 @@ public final class CulturegraphXmlFilterHbzToJsonTest {
 
 	private static final String PATH_TO_TEST = "src/test/resources/";
 	public static final String JSON_OUTPUT_FILE =
-			PATH_TO_TEST + "cg/output-es-bulk.ndjson";
+			PATH_TO_TEST + "cg/rvk-es-bulk.ndjson";
 
 	private static final String XML_INPUT_FILE = "cg/aggregate_20240507_example.marcxml";
 	private static PluginConfigurableNode node;

diff --git a/src/test/resources/alma-fix/990367731740206441.json b/src/test/resources/alma-fix/990367731740206441.json
@@ -194,6 +194,27 @@
       "label" : "RVK (Regensburger Verbundklassifikation)",
       "id" : "https://d-nb.info/gnd/4449787-8"
     }
+  }, {
+    "notation" : "NQ 2360",
+    "type" : [ "Concept" ],
+    "source" : {
+      "label" : "RVK (Regensburger Verbundklassifikation)",
+      "id" : "https://d-nb.info/gnd/4449787-8"
+    }
+  }, {
+    "notation" : "NY 4760",
+    "type" : [ "Concept" ],
+    "source" : {
+      "label" : "RVK (Regensburger Verbundklassifikation)",
+      "id" : "https://d-nb.info/gnd/4449787-8"
+    }
+  }, {
+    "notation" : "BD 7100",
+    "type" : [ "Concept" ],
+    "source" : {
+      "label" : "RVK (Regensburger Verbundklassifikation)",
+      "id" : "https://d-nb.info/gnd/4449787-8"
+    }
   }, {
     "type" : [ "ComplexSubject" ],
     "label" : "Pandemie | Gesellschaft | Geschichte",

diff --git a/src/test/resources/cg/aggregate_20240507_example.marcxml b/src/test/resources/cg/aggregate_20240507_example.marcxml
@@ -717,7 +717,7 @@
       <marc:subfield code="8">9\p</marc:subfield>
     </marc:datafield>
     <marc:datafield ind2=" " ind1=" " tag="035">
-      <marc:subfield code="a">(DE-605)990063057720206441</marc:subfield>
+      <marc:subfield code="a">(DE-605)990367731740206441</marc:subfield>
       <marc:subfield code="8">6\p</marc:subfield>
     </marc:datafield>
     <marc:datafield ind2=" " ind1=" " tag="035">

diff --git a/src/test/resources/cg/output-es-bulk.ndjson → src/test/resources/cg/rvk-es-bulk.ndjson b/src/test/resources/cg/output-es-bulk.ndjson → src/test/resources/cg/rvk-es-bulk.ndjson
@@ -1,4 +1,4 @@
 {"index":{"_index":"cg","_type":"rvk"}}
 {"rvk":["MG 11380","MF 3390","MC 7200"],"id":"990067531130206441"}
 {"index":{"_index":"cg","_type":"rvk"}}
-{"rvk":["NQ 2360","NY 4760","BD 7100"],"id":"990019247190206441, 990063057720206441, 990063668050206441"}
+{"rvk":["NQ 2360","NY 4760","BD 7100"],"id":"990019247190206441, 990367731740206441, 990063668050206441"}
diff --git a/src/test/resources/cg/output.csv → src/test/resources/cg/rvk.csv b/src/test/resources/cg/output.csv → src/test/resources/cg/rvk.csv
@@ -1,4 +1,4 @@
 "990067531130206441","MG 11380,MF 3390,MC 7200"
 "990019247190206441","NQ 2360,NY 4760,BD 7100"
-"990063057720206441","NQ 2360,NY 4760,BD 7100"
+"990367731740206441","NQ 2360,NY 4760,BD 7100"
 "990063668050206441","NQ 2360,NY 4760,BD 7100"
diff --git a/src/test/resources/cg/rvk.tsv b/src/test/resources/cg/rvk.tsv
@@ -0,0 +1,4 @@
+990067531130206441	MG 11380,MF 3390,MC 7200
+990019247190206441	NQ 2360,NY 4760,BD 7100
+990367731740206441	NQ 2360,NY 4760,BD 7100
+990063668050206441	NQ 2360,NY 4760,BD 7100