use JGit for annotations (#3505)

Author: Vladimir Kotal

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

@@ -51,12 +51,17 @@
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
+import org.eclipse.jgit.api.BlameCommand;
 import org.eclipse.jgit.api.Git;
 import org.eclipse.jgit.api.errors.GitAPIException;
+import org.eclipse.jgit.blame.BlameResult;
+import org.eclipse.jgit.diff.RawText;
+import org.eclipse.jgit.diff.RawTextComparator;
 import org.eclipse.jgit.lib.Constants;
 import org.eclipse.jgit.lib.ObjectId;
 import org.eclipse.jgit.lib.ObjectLoader;
 import org.eclipse.jgit.lib.ObjectReader;
+import org.eclipse.jgit.lib.PersonIdent;
 import org.eclipse.jgit.lib.Ref;
 import org.eclipse.jgit.revwalk.RevCommit;
 import org.eclipse.jgit.revwalk.RevTree;
@@ -70,7 +75,6 @@
 import org.opengrok.indexer.configuration.RuntimeEnvironment;
 import org.opengrok.indexer.logger.LoggerFactory;
 import org.opengrok.indexer.util.Executor;
-import org.opengrok.indexer.util.HeadHandler;
 import org.opengrok.indexer.util.LazilyInstantiate;
 import org.opengrok.indexer.util.Version;
 
@@ -125,6 +129,8 @@ public class GitRepository extends Repository {
      */
     private static final LazilyInstantiate<Boolean> GIT_IS_WORKING = LazilyInstantiate.using(GitRepository::isGitWorking);
 
+    public static final int GIT_ABBREV_LEN = 8;
+
     public GitRepository() {
         type = "git";
         /*
@@ -227,6 +233,15 @@ Executor getRenamedFilesExecutor(final File file, String sinceRevision) throws I
         return new Executor(cmd, new File(getDirectoryName()), sinceRevision != null);
     }
 
+    /**
+     * Be careful, git uses only forward slashes in its command and output (not in file path).
+     * Using backslashes together with git show will get empty output and 0 status code.
+     * @return string with separator characters replaced with forward slash
+     */
+    private static String getGitFilePath(String filePath) {
+        return filePath.replace(File.separatorChar, '/');
+    }
+
     /**
      * Try to get file contents for given revision.
      *
@@ -241,15 +256,10 @@ private HistoryRevResult getHistoryRev(OutputStream out, String fullpath, String
         HistoryRevResult result = new HistoryRevResult();
         File directory = new File(getDirectoryName());
 
-        /*
-         * Be careful, git uses only forward slashes in its command and output (not in file path).
-         * Using backslashes together with git show will get empty output and 0 status code.
-         */
         String filename;
         result.success = false;
         try {
-            filename = Paths.get(getCanonicalDirectoryName()).relativize(
-                    Paths.get(fullpath)).toString().replace(File.separatorChar, '/');
+            filename = getGitFilePath(Paths.get(getCanonicalDirectoryName()).relativize(Paths.get(fullpath)).toString());
         } catch (IOException e) {
             LOGGER.log(Level.WARNING, String.format("Failed to relativize '%s' in for repository '%s'",
                     fullpath, directory), e);
@@ -447,37 +457,6 @@ String findOriginalName(String fullpath, String changeset)
         return originalFile;
     }
 
-    /**
-     * Get first revision of given file without following renames.
-     * @param fullpath file path to get first revision of
-     */
-    private String getFirstRevision(String fullpath) {
-        String[] argv = {
-                ensureCommand(CMD_PROPERTY_KEY, CMD_FALLBACK),
-                "rev-list",
-                "--reverse",
-                "HEAD",
-                "--",
-                fullpath
-        };
-
-        Executor executor = new Executor(Arrays.asList(argv), new File(getDirectoryName()),
-                RuntimeEnvironment.getInstance().getInteractiveCommandTimeout());
-        HeadHandler headHandler = new HeadHandler(1);
-        int status = executor.exec(false, headHandler);
-
-        String line;
-        if (headHandler.count() > 0 && (line = headHandler.get(0)) != null) {
-            return line.trim();
-        }
-
-        LOGGER.log(Level.WARNING,
-                "Failed to get first revision for: \"{0}\" Exit code: {1}",
-                new Object[]{fullpath, String.valueOf(status)});
-
-        return null;
-    }
-
     /**
      * Annotate the specified file/revision.
      *
@@ -488,61 +467,72 @@ private String getFirstRevision(String fullpath) {
      */
     @Override
     public Annotation annotate(File file, String revision) throws IOException {
-        List<String> cmd = new ArrayList<>();
-        ensureCommand(CMD_PROPERTY_KEY, CMD_FALLBACK);
-        cmd.add(RepoCommand);
-        cmd.add(BLAME);
-        cmd.add("-c"); // to get correctly formed changeset IDs
-        cmd.add(ABBREV_BLAME);
-        if (revision != null) {
-            cmd.add(revision);
-        } else {
-            // {@code git blame} follows renames by default. If renamed file handling is off, its output would
-            // contain invalid revisions. Therefore, the revision range needs to be constrained.
-            if (!isHandleRenamedFiles()) {
-                String firstRevision = getFirstRevision(file.getAbsolutePath());
-                if (firstRevision == null) {
-                    return null;
-                }
-                cmd.add(firstRevision + "..");
-            }
+        String filePath = getPathRelativeToCanonicalRepositoryRoot(file.getCanonicalPath());
+
+        if (revision == null) {
+            revision = getFirstRevision(filePath);
         }
-        cmd.add("--");
-        cmd.add(getPathRelativeToCanonicalRepositoryRoot(file.getCanonicalPath()));
+        Annotation annotation = getAnnotation(revision, filePath);
 
-        Executor executor = new Executor(cmd, new File(getDirectoryName()),
-                RuntimeEnvironment.getInstance().getInteractiveCommandTimeout());
-        GitAnnotationParser parser = new GitAnnotationParser(file.getName());
-        int status = executor.exec(true, parser);
-
-        // File might have changed its location if it was renamed.
-        // Try to lookup its original name and get the annotation again.
-        if (status != 0 && isHandleRenamedFiles()) {
-            cmd.clear();
-            ensureCommand(CMD_PROPERTY_KEY, CMD_FALLBACK);
-            cmd.add(RepoCommand);
-            cmd.add(BLAME);
-            cmd.add("-c"); // to get correctly formed changeset IDs
-            cmd.add(ABBREV_BLAME);
-            if (revision != null) {
-                cmd.add(revision);
+        if (annotation.getRevisions().isEmpty() && isHandleRenamedFiles()) {
+            // The file might have changed its location if it was renamed.
+            // Try to lookup its original name and get the annotation again.
+            String origName = findOriginalName(file.getCanonicalPath(), revision);
+            if (origName != null) {
+                annotation = getAnnotation(revision, origName);
             }
-            cmd.add("--");
-            cmd.add(findOriginalName(file.getCanonicalPath(), revision));
-            executor = new Executor(cmd, new File(getDirectoryName()),
-                    RuntimeEnvironment.getInstance().getInteractiveCommandTimeout());
-            parser = new GitAnnotationParser(file.getName());
-            status = executor.exec(true, parser);
         }
 
-        if (status != 0) {
+        return annotation;
+    }
+
+    private String getFirstRevision(String filePath) {
+        String revision = null;
+        try (org.eclipse.jgit.lib.Repository repository = getJGitRepository(getDirectoryName())) {
+            Iterable<RevCommit> commits = new Git(repository).log().
+                    addPath(getGitFilePath(filePath)).
+                    setMaxCount(1).
+                    call();
+            RevCommit commit = commits.iterator().next();
+            if (commit != null) {
+                revision = commit.getId().getName();
+            } else {
+                LOGGER.log(Level.WARNING, String.format("cannot get first revision of '%s' in repository '%s'",
+                        filePath, getDirectoryName()));
+            }
+        } catch (IOException | GitAPIException e) {
             LOGGER.log(Level.WARNING,
-                    "Failed to get annotations for: \"{0}\" Exit code: {1}",
-                    new Object[]{file.getAbsolutePath(), String.valueOf(status)});
-            return null;
+                    String.format("cannot get first revision of '%s' in repository '%s'",
+                            filePath, getDirectoryName()), e);
         }
+        return revision;
+    }
+
+    @NotNull
+    private Annotation getAnnotation(String revision, String filePath) throws IOException {
+        Annotation annotation = new Annotation(filePath);
 
-        return parser.getAnnotation();
+        try (org.eclipse.jgit.lib.Repository repository = getJGitRepository(getDirectoryName())) {
+            BlameCommand blameCommand = new Git(repository).blame().setFilePath(getGitFilePath(filePath));
+            ObjectId commitId = repository.resolve(revision);
+            blameCommand.setStartCommit(commitId);
+            blameCommand.setFollowFileRenames(isHandleRenamedFiles());
+            final BlameResult result = blameCommand.setTextComparator(RawTextComparator.WS_IGNORE_ALL).call();
+            if (result != null) {
+                final RawText rawText = result.getResultContents();
+                for (int i = 0; i < rawText.size(); i++) {
+                    final PersonIdent sourceAuthor = result.getSourceAuthor(i);
+                    final RevCommit sourceCommit = result.getSourceCommit(i);
+                    annotation.addLine(sourceCommit.getId().abbreviate(GIT_ABBREV_LEN).
+                            name(), sourceAuthor.getName(), true);
+                }
+            }
+        } catch (GitAPIException e) {
+            LOGGER.log(Level.FINER,
+                    String.format("failed to get annotation for file '%s' in repository '%s' in revision '%s'",
+                            filePath, getDirectoryName(), revision));
+        }
+        return annotation;
     }
 
     @Override

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

@@ -18,13 +18,12 @@
  */
 
 /*
- * Copyright (c) 2008, 2020, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2008, 2021, Oracle and/or its affiliates. All rights reserved.
  * Portions Copyright (c) 2017, 2019, Chris Fraire <[email protected]>.
  * Portions Copyright (c) 2019, Krystof Tulinger <[email protected]>.
  */
 package org.opengrok.indexer.history;
 
-import java.io.ByteArrayInputStream;
 import java.io.File;
 import java.io.IOException;
 import java.io.InputStream;
@@ -246,42 +245,6 @@ public void testDetermineParent() throws Exception {
         }
     }
 
-    /**
-     * Test of parseAnnotation method, of class GitRepository.
-     */
-    @Test
-    public void parseAnnotation() throws Exception {
-        String revId1 = "cd283405560689372626a69d5331c467bce71656";
-        String revId2 = "30ae764b12039348766291100308556675ca11ab";
-        String revId3 = "2394823984cde2390345435a9237bd7c25932342";
-        String author1 = "Author Name";
-        String author2 = "Author With Long Name";
-        String author3 = "Author Named Jr.";
-        String output = revId1 + " file1.ext   (" + author1 + "     2005-06-06 16:38:26 -0400 272) \n" +
-                revId2 + " file2.h (" + author2 + "     2007-09-10 23:02:45 -0400 273)   if (some code)\n" +
-                revId3 + " file2.c  (" + author3 + "      2006-09-20 21:47:42 -0700 274)           call_function(i);\n";
-        String fileName = "something.ext";
-
-        GitAnnotationParser parser = new GitAnnotationParser(fileName);
-        parser.processStream(new ByteArrayInputStream(output.getBytes()));
-        Annotation result = parser.getAnnotation();
-
-        assertNotNull(result);
-        assertEquals(3, result.size());
-        for (int i = 1; i <= 3; i++) {
-            assertTrue("isEnabled()", result.isEnabled(i));
-        }
-        assertEquals(revId1, result.getRevision(1));
-        assertEquals(revId2, result.getRevision(2));
-        assertEquals(revId3, result.getRevision(3));
-        assertEquals(author1, result.getAuthor(1));
-        assertEquals(author2, result.getAuthor(2));
-        assertEquals(author3, result.getAuthor(3));
-        assertEquals(author2.length(), result.getWidestAuthor());
-        assertEquals(revId1.length(), result.getWidestRevision());
-        assertEquals(fileName, result.getFilename());
-    }
-
     /**
      * Test of fileHasAnnotation method, of class GitRepository.
      */