Create and use tsv instead of csv map #1058

hbz · Jun 6, 2024 · 7dc8230 · 7dc8230
1 parent c0e2966
commit 7dc8230
Show file tree

Hide file tree

Showing 8 changed files with 36 additions and 7 deletions.
diff --git a/src/main/java/org/lobid/resources/run/AlmaMarcXmlFix2lobidJsonEs.java b/src/main/java/org/lobid/resources/run/AlmaMarcXmlFix2lobidJsonEs.java
@@ -125,7 +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.csv", "../../../../../../lookup-tables/data/rvk.csv");
+                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/src/main/java/org/lobid/resources/run/CulturegraphXmlFilterHbzRvkToCsv.java b/src/main/java/org/lobid/resources/run/CulturegraphXmlFilterHbzRvkToCsv.java
@@ -22,7 +22,7 @@
  * @author Tobias Bülte (TobiasNx)
  **/
 public final class CulturegraphXmlFilterHbzRvkToCsv {
-	private static String OUTPUT_FILE="cg-concordance.csv";
+	private static String OUTPUT_FILE="cg-concordance.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 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
@@ -27,7 +27,7 @@ put_filemap("$[classification.tsv]","classificationCode2Label", sep_char:"\t",ke
 put_filemap("$[classification.tsv]","classificationCode2Uri", sep_char:"\t",key_column:"0",value_column:"2",expected_columns:"-1")
 
 # RVK via Culturegraph
-put_filemap("$[rvk.csv]","rvk", sep_char:",")
+put_filemap("$[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

diff --git a/src/main/resources/alma/fix/subjects.fix b/src/main/resources/alma/fix/subjects.fix
@@ -332,7 +332,7 @@ end
 
 copy_field("almaMmsId","@rvkNotations")
 lookup("@rvkNotations","rvk",delete:"true")
-split_field("@rvkNotations", ", ")
+split_field("@rvkNotations", ",")
 do list(path: "@rvkNotations","var":"$i")
   copy_field("$i","subject[].$append.notation")
   set_array("subject[].$last.type[]","Concept")

diff --git a/src/test/java/org/lobid/resources/AlmaMarc21XmlToLobidJsonMetafixTest.java b/src/test/java/org/lobid/resources/AlmaMarc21XmlToLobidJsonMetafixTest.java
@@ -69,7 +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.csv", "src/test/resources/cg/rvk.csv");
+        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/src/test/java/org/lobid/resources/CulturegraphXmlFilterHbzRvkToCsvTest.java b/src/test/java/org/lobid/resources/CulturegraphXmlFilterHbzRvkToCsvTest.java
@@ -26,7 +26,7 @@ public final class CulturegraphXmlFilterHbzRvkToCsvTest {
 
 	private static final String PATH_TO_TEST = "src/test/resources/";
 	public static final String OUTPUT_FILE =
-			PATH_TO_TEST + "cg/rvk.csv";
+			PATH_TO_TEST + "cg/rvk.tsv";
 
 	private static final String XML_INPUT_FILE = "cg/aggregate_20240507_example.marcxml";
 

diff --git a/src/test/resources/alma-fix/990367731740206441.json b/src/test/resources/alma-fix/990367731740206441.json
@@ -190,6 +190,27 @@
       "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/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