sanitize path

fixes #2315

Author: Vladimir Kotal

opengrok-indexer/src/main/java/org/opengrok/indexer/analysis/AnalyzerGuru.java

@@ -523,6 +523,10 @@ public void populateDocument(Document doc, File file, String path,
         FileAnalyzer fa, Writer xrefOut) throws IOException,
             InterruptedException, ForbiddenSymlinkException {
 
+        // Sanitize Windows path delimiters in order not to conflict with Lucene escape character
+        // and also so the path appears as correctly formed URI in the search results.
+        path = path.replace("\\", "/");
+
         String date = DateTools.timeToString(file.lastModified(),
                 DateTools.Resolution.MILLISECOND);
         doc.add(new Field(QueryBuilder.U, Util.path2uid(path, date),

opengrok-indexer/src/main/java/org/opengrok/indexer/history/RepositoryInfo.java

@@ -157,10 +157,8 @@ public void setDirectoryName(File dir) {
         String originalPath = dir.getPath();
         try {
             path = PathUtils.getRelativeToCanonical(originalPath, rootPath);
-            // N.b. OpenGrok has a weird convention that
-            // `directoryNameRelative' must start with a '/', as it is
-            // elsewhere directly appended to env.getSourceRootPath() and
-            // also stored as such.
+            // OpenGrok has a weird convention that directoryNameRelative must start with a path separator,
+            // as it is elsewhere directly appended to env.getSourceRootPath() and also stored as such.
             if (!path.equals(originalPath)) {
                 path = File.separator + path;
             }

opengrok-indexer/src/main/java/org/opengrok/indexer/index/IndexDatabase.java

@@ -1521,9 +1521,8 @@ public static Definitions getDefinitions(File file)
             LOGGER.log(Level.FINER, e.getMessage());
             return null;
         }
-        //sanitize windows path delimiters
-        //in order not to conflict with Lucene escape character
-        path=path.replace("\\", "/");
+        // Sanitize Windows path delimiters in order not to conflict with Lucene escape character.
+        path = path.replace("\\", "/");
 
         IndexReader ireader = getIndexReader(path);
 

opengrok-indexer/src/test/java/org/opengrok/indexer/index/IndexDatabaseTest.java

@@ -24,9 +24,13 @@
 package org.opengrok.indexer.index;
 
 import java.io.File;
+import java.io.IOException;
 import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.TreeSet;
+
+import org.apache.lucene.document.Document;
+import org.apache.lucene.search.ScoreDoc;
 import org.junit.AfterClass;
 import org.junit.Assert;
 import org.junit.BeforeClass;
@@ -44,6 +48,8 @@
 import org.opengrok.indexer.configuration.RuntimeEnvironment;
 import org.opengrok.indexer.history.HistoryGuru;
 import org.opengrok.indexer.history.RepositoryFactory;
+import org.opengrok.indexer.search.QueryBuilder;
+import org.opengrok.indexer.search.SearchEngine;
 import org.opengrok.indexer.util.TestRepository;
 
 /**
@@ -180,4 +186,24 @@ public void testCleanupAfterIndexRemoval() throws Exception {
         checkDataExistence(projectName + File.separator + fileName, false);
         Assert.assertEquals(origNumFiles - 1, idb.getNumFiles());
     }
+
+    /**
+     * This is a test of {@code populateDocument} so it should be rather in {@code AnalyzerGuruTest}
+     * however it lacks the pre-requisite indexing phase.
+     * @throws IOException
+     */
+    @Test
+    public void testIndexPath() throws IOException {
+        SearchEngine instance = new SearchEngine();
+        // Use as broad search as possible.
+        instance.setFile("c");
+        instance.search();
+        ScoreDoc[] scoredocs = instance.scoreDocs();
+        assertTrue("need some search hits to perform the check",scoredocs.length > 0);
+        for (ScoreDoc sd : scoredocs) {
+            Document doc = instance.doc(sd.doc);
+            assertFalse("PATH field should not contain backslash characters",
+                    doc.getField(QueryBuilder.PATH).stringValue().contains("\\"));
+        }
+    }
 }