SourceDocument refactor for separate raw field

src/main/java/io/anserini/collection/AclAnthology.java

@@ -105,7 +105,7 @@ public void readNext() throws NoSuchElementException {
   /**
    * A document in a JSON collection.
    */
-  public class Document implements SourceDocument {
+  public class Document extends SourceDocument {
     private String id;
     private String contents;
     private JsonNode paper;

src/main/java/io/anserini/collection/BibtexCollection.java

@@ -95,7 +95,7 @@ public void readNext() throws NoSuchElementException {
   /**
    * A document in a Bibtex collection.
    */
-  public static class Document implements SourceDocument {
+  public static class Document extends SourceDocument {
     private String id;
     private String contents;
     private String type;

src/main/java/io/anserini/collection/CarCollection.java

@@ -76,7 +76,7 @@ public void readNext() {
    * The paraID serves as the id.
    * See <a href="http://trec-car.cs.unh.edu/datareleases/">this reference</a> for details.
    */
-  public static class Document implements SourceDocument {
+  public static class Document extends SourceDocument {
     private final String paraID;
     private final String paragraph;
 

src/main/java/io/anserini/collection/ClueWeb09Collection.java

@@ -345,7 +345,7 @@ public static Document readNextWarcRecord(DataInputStream in, String version)
    * A document from the <a href="https://www.lemurproject.org/clueweb09.php/">ClueWeb09 collection</a>.
    * This class derives from tools provided by CMU for reading the ClueWeb09 collection.
    */
-  public static class Document implements SourceDocument {
+  public static class Document extends SourceDocument {
     public static final String WARC_VERSION = "WARC/0.18";
     protected final static String NEWLINE = "\n";
 

src/main/java/io/anserini/collection/CoreCollection.java

@@ -113,7 +113,7 @@ public void readNext() throws NoSuchElementException {
   /**
    * A document in a Core collection.
    */
-  public static class Document implements SourceDocument {
+  public static class Document extends SourceDocument {
     private String id;
     private String contents;
     private JsonNode jsonNode;

src/main/java/io/anserini/collection/CovidCollectionDocument.java

@@ -26,7 +26,7 @@
 import java.nio.file.Files;
 import java.nio.file.Paths;
 
-public abstract class CovidCollectionDocument implements SourceDocument {
+public abstract class CovidCollectionDocument extends SourceDocument {
   private static final Logger LOG = LogManager.getLogger(CovidCollectionDocument.class);
   protected String id;
   protected String content;
@@ -76,6 +76,7 @@ public boolean indexable() {
     return true;
   }
 
+  @Override
   public String raw() {
     return raw;
   }

src/main/java/io/anserini/collection/HtmlCollection.java

@@ -100,7 +100,7 @@ private void getNextEntry() throws IOException {
   /**
    * A generic document in {@code HtmlCollection}.
    */
-  public static class Document implements SourceDocument {
+  public static class Document extends SourceDocument {
     private String id;
     private String contents;
 

src/main/java/io/anserini/collection/JsonCollection.java

@@ -127,7 +127,7 @@ public void readNext() throws NoSuchElementException {
   /**
    * A document in a JSON collection.
    */
-  public static class Document implements MultifieldSourceDocument {
+  public static class Document extends MultifieldSourceDocument {
     private String id;
     private String contents;
     private Map<String, String> fields;

src/main/java/io/anserini/collection/MultifieldSourceDocument.java

@@ -18,11 +18,11 @@
 
 import java.util.Map;
 
-public interface MultifieldSourceDocument extends SourceDocument {
+public abstract class MultifieldSourceDocument extends SourceDocument {
   /**
    * Returns a map of fields associated with this document.
    *
    * @return a map of fields associated with this document
    */
-  Map<String, String> fields();
+  public abstract Map<String, String> fields();
 }

src/main/java/io/anserini/collection/NewYorkTimesCollection.java

@@ -128,7 +128,7 @@ private void getNextEntry() throws IOException {
    * A document from the <a href="https://catalog.ldc.upenn.edu/products/LDC2008T19">New York Times
    * Annotated Corpus</a>.
    */
-  public static class Document implements SourceDocument {
+  public static class Document extends SourceDocument {
     private final RawDocument raw;
     private String id;
     private String contents;

src/main/java/io/anserini/collection/SourceDocument.java

@@ -21,26 +21,35 @@
  * Lucene {@link org.apache.lucene.document.Document}, which is the Lucene representation that
  * can be directly inserted into an index.
  */
-public interface SourceDocument {
+public abstract class SourceDocument {
   /**
    * Returns the unique identifier of the document.
    *
    * @return the unique identifier of the document
    */
-  String id();
+  public abstract String id();
+
+  /**
+   * Returns the content of the document to be searched.
+   *
+   * @return the content of the document to be searched
+   */
+  public abstract String content();
 
   /**
    * Returns the raw content of the document.
    *
    * @return the raw content of the document
    */
-  String content();
+  public String raw() {
+    return content();
+  };
 
   /**
    * Returns whether this document is meant to be indexed. Certain collections (e.g., ClueWeb)
    * contained metadata records that aren't meant to be indexed.
    *
    * @return <code>true</code> if this document is meant to be indexed
    */
-  boolean indexable();
+  public abstract boolean indexable();
 }

src/main/java/io/anserini/collection/TrecCollection.java

@@ -190,7 +190,7 @@ private void parseRecord(StringBuilder builder) {
   /**
    * A document in a classic TREC <i>ad hoc</i> document collection.
    */
-  public static class Document implements SourceDocument {
+  public static class Document extends SourceDocument {
 
     protected static final String DOCNO = "<DOCNO>";
     protected static final String TERMINATING_DOCNO = "</DOCNO>";

src/main/java/io/anserini/collection/TweetCollection.java

@@ -202,7 +202,7 @@ private boolean isFieldAvailable(Object field) {
   /**
    * A tweet (i.e., status).
    */
-  public static class Document implements SourceDocument {
+  public static class Document extends SourceDocument {
     // Required fields
     protected String screenName;
     protected int followersCount;

src/main/java/io/anserini/collection/WashingtonPostCollection.java

@@ -110,7 +110,7 @@ private void parseRecord(String record) {
   /**
    * A document from the <a href="https://trec.nist.gov/data/wapost/">TREC Washington Post Corpus</a>.
    */
-  public static class Document implements SourceDocument {
+  public static class Document extends SourceDocument {
     private static final Logger LOG = LogManager.getLogger(Document.class);
 
     // Required fields

src/main/java/io/anserini/collection/WikipediaCollection.java

@@ -90,7 +90,7 @@ public void readNext() {
   /**
    * A Wikipedia article. The article title serves as the id.
    */
-  public static class Document implements SourceDocument {
+  public static class Document extends SourceDocument {
     private final String title;
     private final String contents;
 

src/test/java/io/anserini/collection/AclAnthologyTest.java

@@ -118,6 +118,7 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
         assertEquals(expectedValue, aclDoc.id());
       } else if (expectedKey.equals("contents")) {
         assertEquals(expectedValue, aclDoc.content());
+        assertEquals(expectedValue, doc.raw());
       } else if (expectedKey.equals("authors")) {
         assertEquals(expectedValue, String.join(" ", aclDoc.authors()));
       } else if (expectedKey.equals("sigs")) {

src/test/java/io/anserini/collection/BibtexCollectionTest.java

@@ -100,6 +100,7 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
         assertEquals(expectedValue, ((BibtexCollection.Document) doc).type());
       } else if (expectedKey.equals("contents")) {
         assertEquals(expectedValue, doc.content());
+        assertEquals(expectedValue, doc.raw());
       } else {
         Value parsedValue = parsedFields.get(new Key(expectedKey));
         assertNotNull(parsedValue);

src/test/java/io/anserini/collection/ClueWeb09CollectionTest.java

@@ -62,5 +62,6 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
       assertEquals(expected.get("id"), doc.id());
     }
     assertEquals(expected.get("content"), doc.content());
+    assertEquals(expected.get("content"), doc.raw());
   }
 }

src/test/java/io/anserini/collection/ClueWeb12CollectionTest.java

@@ -62,5 +62,6 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
       assertEquals(expected.get("id"), doc.id());
     }
     assertEquals(expected.get("content"), doc.content());
+    assertEquals(expected.get("content"), doc.raw());
   }
 }

src/test/java/io/anserini/collection/CoreCollectionTest.java

@@ -95,6 +95,7 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
         assertEquals(expectedValue, coreDoc.id());
       } else if (expectedKey.equals("contents")) {
         assertEquals(expected.get("title") + " " + expected.get("abstract"), coreDoc.content());
+        assertEquals(expected.get("title") + " " + expected.get("abstract"), coreDoc.raw());
       } else {
         assert(coreDoc.jsonNode().has(expectedKey));
       }

src/test/java/io/anserini/collection/JsonCollectionLineObjectTest.java

@@ -62,6 +62,7 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
     assertTrue(doc.indexable());
     assertEquals(expected.get("id"), doc.id());
     assertEquals(expected.get("content"), doc.content());
+    assertEquals(expected.get("content"), doc.raw());
     assertEquals(expected.get("field1"), ((JsonCollection.Document) doc).fields().get("field1"));
     assertEquals(expected.get("field2"), ((JsonCollection.Document) doc).fields().get("field2"));
   }

src/test/java/io/anserini/collection/JsonCollectionTest.java

@@ -24,5 +24,6 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
     assertTrue(doc.indexable());
     assertEquals(expected.get("id"), doc.id());
     assertEquals(expected.get("content"), doc.content());
+    assertEquals(expected.get("content"), doc.raw());
   }
 }

src/test/java/io/anserini/collection/NewYorkTimesCollectionTest.java

@@ -53,6 +53,7 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
     assertTrue(doc.indexable());
     assertEquals(expected.get("id"), nyt.id());
     assertEquals(expected.get("content"), nyt.content());
+    assertEquals(expected.get("content"), nyt.raw());
     assertEquals(expected.get("headline"), nyt.getRawDocument().getHeadline());
     assertEquals(expected.get("abstract"), nyt.getRawDocument().getArticleAbstract());
     assertEquals(expected.get("body"), nyt.getRawDocument().getBody());

src/test/java/io/anserini/collection/TrecCollectionTest.java

@@ -59,5 +59,6 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
     assertTrue(doc.indexable());
     assertEquals(expected.get("id"), doc.id());
     assertEquals(expected.get("content"), doc.content());
+    assertEquals(expected.get("content"), doc.raw());
   }
 }

src/test/java/io/anserini/collection/TrecwebCollectionTest.java

@@ -55,5 +55,6 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
     assertTrue(doc.indexable());
     assertEquals(expected.get("id"), doc.id());
     assertEquals(expected.get("content"), doc.content());
+    assertEquals(expected.get("content"), doc.raw());
   }
 }

src/test/java/io/anserini/collection/TweetCollectionTest.java

@@ -57,6 +57,7 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
     assertTrue(doc.indexable());
     assertEquals(expected.get("id"), doc.id());
     assertEquals(expected.get("content"), doc.content());
+    assertEquals(expected.get("content"), doc.raw());
     assertEquals(expected.get("screen_name"), ((TweetCollection.Document) doc).getScreenName());
     assertEquals((long) Long.valueOf(expected.get("timestamp_ms")),
         ((TweetCollection.Document) doc).getTimestampMs().getAsLong());

src/test/java/io/anserini/collection/WikipediaCollectionTest.java

@@ -48,5 +48,6 @@ void checkDocument(SourceDocument doc, Map<String, String> expected) {
     assertTrue(doc.indexable());
     assertEquals(expected.get("id"), doc.id());
     assertEquals(expected.get("content"), doc.content());
+    assertEquals(expected.get("content"), doc.raw());
   }
 }