From bfd52b2fa46fcd30e813b301ecababc538ed1bb2 Mon Sep 17 00:00:00 2001
From: Claus Stadler <RavenArkadon@googlemail.com>
Date: Fri, 2 Jul 2021 21:55:11 +0200
Subject: [PATCH] Added support for Lang attribute to RdfSourceFactory.
 Upgraded source maven plugin version.

---
 pom.xml                                       |  2 +-
 .../spark/io/input/api/RdfSourceFactory.java  | 21 +++++++++---
 .../io/input/impl/RdfSourceFactoryImpl.java   | 34 +++++++------------
 3 files changed, 31 insertions(+), 26 deletions(-)

diff --git a/pom.xml b/pom.xml
index e8bd16707..4deac90a4 100644
--- a/pom.xml
+++ b/pom.xml
@@ -1722,7 +1722,7 @@
                 <plugin>
                     <groupId>org.apache.maven.plugins</groupId>
                     <artifactId>maven-source-plugin</artifactId>
-                    <version>3.0.1</version>
+                    <version>3.2.1</version>
                     <executions>
                         <execution>
                             <id>attach-sources</id>
diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSourceFactory.java b/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSourceFactory.java
index b43dfcd37..adb3c6518 100644
--- a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSourceFactory.java
+++ b/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSourceFactory.java
@@ -2,6 +2,7 @@
 
 import org.apache.hadoop.fs.FileSystem;
 import org.apache.hadoop.fs.Path;
+import org.apache.jena.riot.Lang;
 
 
 /**
@@ -16,14 +17,26 @@
 public interface RdfSourceFactory {
 
     default RdfSource get(String sourceStr) {
+        return get(sourceStr, (Lang) null);
+    }
+
+    default RdfSource get(String sourceStr, Lang lang) {
+        Path path = new Path(sourceStr);
+        return get(path, null, lang);
+    }
+
+    default RdfSource get(String sourceStr, FileSystem fileSystem) {
+        Path path = new Path(sourceStr);
+        return get(path, fileSystem, null);
+    }
+
+    default RdfSource get(Path path, FileSystem fileSystem, Lang lang) {
         try {
-            return create(sourceStr);
+            return create(path, fileSystem, lang);
         } catch (Exception e) {
             throw new RuntimeException(e);
         }
     }
 
-    RdfSource create(String sourceStr) throws Exception;
-    RdfSource create(String sourceStr, FileSystem fileSystem) throws Exception;
-    RdfSource create(Path path, FileSystem fileSystem) throws Exception;
+    RdfSource create(Path path, FileSystem fileSystem, Lang lang) throws Exception;
 }
diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceFactoryImpl.java b/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceFactoryImpl.java
index 581ca1545..4583f5b40 100644
--- a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceFactoryImpl.java
+++ b/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceFactoryImpl.java
@@ -36,33 +36,25 @@ public static RdfSourceFactory from(SparkSession sparkSession) {
         return new RdfSourceFactoryImpl(sparkSession);
     }
 
-
-    public RdfSource create(String sourceStr) throws Exception {
-        Configuration hadoopConf = sparkSession.sparkContext().hadoopConfiguration();
-        FileSystem fileSystem = FileSystem.get(hadoopConf);
-        return create(sourceStr, fileSystem);
-    }
-
-
     @Override
-    public RdfSource create(String sourceStr, FileSystem fileSystem) throws Exception {
-        Path path = new Path(sourceStr);
+    public RdfSource create(Path path, FileSystem fileSystem, Lang lang) throws Exception {
 
-        return create(path, fileSystem);
-    }
+        if (fileSystem == null) {
+            Configuration hadoopConf = sparkSession.sparkContext().hadoopConfiguration();
+            fileSystem = FileSystem.get(hadoopConf);
+        }
 
-    @Override
-    public RdfSource create(Path path, FileSystem fileSystem) throws Exception {
         Path resolvedPath = fileSystem.resolvePath(path);
 
-        EntityInfo entityInfo;
-        try (InputStream in = fileSystem.open(resolvedPath)) {
-             entityInfo = RDFDataMgrEx.probeEntityInfo(in, RDFDataMgrEx.DEFAULT_PROBE_LANGS);
-        }
+        if (lang == null) {
+            EntityInfo entityInfo;
+            try (InputStream in = fileSystem.open(resolvedPath)) {
+                entityInfo = RDFDataMgrEx.probeEntityInfo(in, RDFDataMgrEx.DEFAULT_PROBE_LANGS);
+            }
+            lang = RDFLanguages.contentTypeToLang(entityInfo.getContentType());
 
-        Lang lang = RDFLanguages.contentTypeToLang(entityInfo.getContentType());
-
-        Objects.requireNonNull(lang, "Could not obtain lang for " + entityInfo.getContentType() + " from " + path);
+            Objects.requireNonNull(lang, "Could not obtain lang for " + entityInfo.getContentType() + " from " + path);
+        }
 
         return new RdfSourceImpl(sparkSession, resolvedPath, lang);
     }