Unit tests for exorciseIndex()

lucene/core/src/test/org/apache/lucene/index/TestCheckIndex.java

@@ -294,4 +294,73 @@ public void testPriorBrokenCommitPoint() throws Exception {
       }
     }
   }
+
+  public void testExorciseIndex() throws IOException {
+    try (MockDirectoryWrapper dir = newMockDirectory()) {
+
+      // disable this normally useful test infra feature since this test intentionally leaves broken
+      // indices:
+      dir.setCheckIndexOnClose(false);
+      dir.listAll();
+      IndexWriterConfig iwc =
+          new IndexWriterConfig()
+              .setMergePolicy(NoMergePolicy.INSTANCE)
+              .setIndexDeletionPolicy(DeleteNothingIndexDeletionPolicy.INSTANCE);
+
+      try (IndexWriter iw = new IndexWriter(dir, iwc)) {
+        // create 3 segments by creating 3 separate commit points
+        // We will corrupt the first segment by deleting its compound file (_0.cfs)
+        // We will corrupt the second segment by deleting its segment info (_1.si) (LUCENE-7820)
+        // TODO: Should we test exorciseIndex() for other corruptions like missing
+        //  term dictionary (.tis) / frequencies (.frq) / field data (.fdt) / positions (.prx) ?
+        // exorciseIndex() should be able to exorcise all corrupt segments and rewrite a new
+        // segments file
+        Document doc = new Document();
+        doc.add(new StringField("id", "a", Field.Store.NO));
+        iw.addDocument(doc);
+        iw.commit();
+        // NOTE: we are (illegally) relying on precise file naming here -- if Codec or IW's
+        // behaviour changes, this may need fixing:
+        // We will corrupt the first segment by deleting its compound file (_0.cfs)
+        assertTrue(slowFileExists(dir, "_0.cfs"));
+        iw.addDocument(doc);
+        iw.commit();
+        // We will corrupt the first segment by deleting its segment info (_1.si)
+        assertTrue(slowFileExists(dir, "_1.si"));
+        iw.addDocument(doc);
+        iw.commit();
+        // We should be able to read the third segment once the corrupt segments are exorcised
+        assertTrue(slowFileExists(dir, "_2.si"));
+      }
+
+      try (CheckIndex checkers = new CheckIndex(dir)) {
+        CheckIndex.Status checkIndexStatus = checkers.checkIndex();
+        assertTrue(checkIndexStatus.clean);
+      }
+
+      // now corrupt the segment 0 by removing its compound file (_0.cfs)
+      dir.deleteFile("_0.cfs");
+
+      // now corrupt the segment 1 by removing its segment info file (_0.si)
+      // Only works after addressing LUCENE-7820
+      dir.deleteFile("_1.si");
+
+      // Index should not be clean after corruptions
+      try (CheckIndex checkers = new CheckIndex(dir)) {
+        CheckIndex.Status checkIndexStatus = checkers.checkIndex();
+        assertFalse(checkIndexStatus.clean);
+        checkers.exorciseIndex(checkIndexStatus);
+        // Note that exorciseIndex does not remove any of the unreferenced files after it's done;
+        // you must separately open an {@link IndexWriter}, which deletes unreferenced files when
+        // it's created.
+        assertFalse(checkIndexStatus.clean); // not clean yet until we check again!
+      }
+
+      // Checking for clean index after exorcism!
+      try (CheckIndex checkers = new CheckIndex(dir)) {
+        CheckIndex.Status checkIndexStatus = checkers.checkIndex();
+        assertTrue(checkIndexStatus.clean);
+      }
+    }
+  }
 }