Merge pull request #533 from hbz/514-import

Create import files
hbz · Jan 21, 2020 · f04dddb · f04dddb
2 parents 0784348 + ba013ad
commit f04dddb
Show file tree

Hide file tree

Showing 3 changed files with 63 additions and 36 deletions.
diff --git a/app/Import700n.java b/app/Import700n.java
@@ -1,6 +1,9 @@
 
 /* Copyright 2019 Fabian Steeg, hbz. Licensed under the GPLv2 */
 
+import static play.test.Helpers.running;
+import static play.test.Helpers.testServer;
+
 import java.io.BufferedWriter;
 import java.io.File;
 import java.io.FileOutputStream;
@@ -18,6 +21,7 @@
 
 import com.fasterxml.jackson.databind.JsonNode;
 
+import controllers.nwbib.Lobid;
 import play.libs.Json;
 
 /**
@@ -42,32 +46,55 @@ public class Import700n {
 	 * @param args Optional, the input (jsonl) and the output (txt) file names
 	 */
 	public static void main(String[] args) {
-		if (args.length == 2) {
-			dataIn = new File(args[0]);
-			dataOut = new File(args[1]);
-		}
-		try (Scanner scanner = new Scanner(dataIn, StandardCharsets.UTF_8.name());
-				BufferedWriter writer = new BufferedWriter(new OutputStreamWriter(
-						new FileOutputStream(dataOut), StandardCharsets.UTF_8))) {
-			while (scanner.hasNextLine()) {
-				JsonNode record = Json.parse(scanner.nextLine());
-				Stream<String> subjects = Streams.concat(//
-						processSpatial(record), processSubject(record));
-				String resultLine = String.format("%s\t%s", //
-						record.get("hbzId").asText(),
-						subjects.collect(Collectors.joining(", "))//
-								// https://github.com/hbz/lobid-resources/issues/1018
-								.replaceAll("spatial#N05", "spatial#N04"));
-				resultLine = resultLine//
-						.replace("spatial#N04$$0", "Westfalen$$0")
-						.replace("Siebengebirge$$0https://nwbib.de/spatial#Q4236",
-								"Siebengebirge$$0https://nwbib.de/spatial#N23");
-				System.out.println(resultLine);
-				writer.write(resultLine + "\n");
+		running(testServer(3333), () -> {
+			if (args.length == 2) {
+				dataIn = new File(args[0]);
+				dataOut = new File(args[1]);
 			}
-		} catch (IOException e) {
-			e.printStackTrace();
-		}
+			try (Scanner scanner = new Scanner(dataIn, StandardCharsets.UTF_8.name());
+					BufferedWriter writer = new BufferedWriter(new OutputStreamWriter(
+							new FileOutputStream(dataOut), StandardCharsets.UTF_8))) {
+				while (scanner.hasNextLine()) {
+					JsonNode record = Json.parse(scanner.nextLine());
+					String resultLine = processLobidResource(record);
+					// String resultLine = processNwbibSnapshot(record);
+					System.out.println(resultLine);
+					writer.write(resultLine + "\n");
+				}
+			} catch (IOException e) {
+				e.printStackTrace();
+			}
+		});
+	}
+
+	@SuppressWarnings("unused")
+	private static String processNwbibSnapshot(JsonNode record) {
+		String id = record.get("hbzId").asText();
+		String result = processLobidResource(Lobid.getResource(id));
+		// See https://github.com/hbz/nwbib/issues/516
+		result = result
+				.replace("\t",
+						"\t\"Bistum Münster$$0https://nwbib.de/spatial#Q769380\", ")
+				.replaceAll(", $", "");
+		return result;
+	}
+
+	private static String processLobidResource(JsonNode record) {
+		Stream<String> subjects = Streams.concat(//
+				processSpatial(record), processSubject(record));
+		String resultLine = String.format("%s\t%s", //
+				record.get("hbzId").asText(),
+				subjects.collect(Collectors.joining(", "))//
+						// https://github.com/hbz/lobid-resources/issues/1018
+						.replaceAll("spatial#N05", "spatial#N04"));
+		resultLine = resultLine//
+				.replace("spatial#N04$$0", "Westfalen$$0")
+				.replace("Siebengebirge$$0https://nwbib.de/spatial#Q4236",
+						"Siebengebirge$$0https://nwbib.de/spatial#N23")
+				.replace(//
+						"Kleinere geistliche Territorien im Rheinland$$0https://nwbib.de/spatial#N52", //
+						"Kleinere Territorien im Rheinland$$0https://nwbib.de/spatial#N54");
+		return resultLine;
 	}
 
 	private static Stream<String> processSubject(JsonNode record) {