From 13417cf9c6eec555b51da49742e47939d2f5715b Mon Sep 17 00:00:00 2001
From: James Moger <james.moger@gitblit.com>
Date: Fri, 19 Oct 2012 22:47:33 -0400
Subject: [PATCH] Exclude submodules from zip downloads (issue 151)
---
src/com/gitblit/LuceneExecutor.java | 297 ++++++++++++++++++++++++++++++++++++++++++-----------------
1 files changed, 211 insertions(+), 86 deletions(-)
diff --git a/src/com/gitblit/LuceneExecutor.java b/src/com/gitblit/LuceneExecutor.java
index 29a8569..42155f4 100644
--- a/src/com/gitblit/LuceneExecutor.java
+++ b/src/com/gitblit/LuceneExecutor.java
@@ -25,7 +25,6 @@
import java.text.MessageFormat;
import java.text.ParseException;
import java.util.ArrayList;
-import java.util.Arrays;
import java.util.Collections;
import java.util.Comparator;
import java.util.HashMap;
@@ -70,10 +69,12 @@
import org.apache.lucene.util.Version;
import org.eclipse.jgit.diff.DiffEntry.ChangeType;
import org.eclipse.jgit.lib.Constants;
+import org.eclipse.jgit.lib.FileMode;
import org.eclipse.jgit.lib.ObjectId;
import org.eclipse.jgit.lib.ObjectLoader;
import org.eclipse.jgit.lib.ObjectReader;
import org.eclipse.jgit.lib.Repository;
+import org.eclipse.jgit.lib.RepositoryCache.FileKey;
import org.eclipse.jgit.revwalk.RevCommit;
import org.eclipse.jgit.revwalk.RevTree;
import org.eclipse.jgit.revwalk.RevWalk;
@@ -105,7 +106,7 @@
public class LuceneExecutor implements Runnable {
- private static final int INDEX_VERSION = 2;
+ private static final int INDEX_VERSION = 5;
private static final String FIELD_OBJECT_TYPE = "type";
private static final String FIELD_ISSUE = "issue";
@@ -138,13 +139,17 @@
private final Map<String, IndexSearcher> searchers = new ConcurrentHashMap<String, IndexSearcher>();
private final Map<String, IndexWriter> writers = new ConcurrentHashMap<String, IndexWriter>();
- private final Set<String> excludedExtensions = new TreeSet<String>(Arrays.asList("7z", "arc",
- "arj", "bin", "bmp", "dll", "doc", "docx", "exe", "gif", "gz", "jar", "jpg", "lib",
- "lzh", "odg", "pdf", "ppt", "png", "so", "swf", "xcf", "xls", "xlsx", "zip"));
-
+ private final String luceneIgnoreExtensions = "7z arc arj bin bmp dll doc docx exe gif gz jar jpg lib lzh odg odf odt pdf ppt png so swf xcf xls xlsx zip";
+ private Set<String> excludedExtensions;
+
public LuceneExecutor(IStoredSettings settings, File repositoriesFolder) {
this.storedSettings = settings;
this.repositoriesFolder = repositoriesFolder;
+ String exts = luceneIgnoreExtensions;
+ if (settings != null) {
+ exts = settings.getString(Keys.web.luceneIgnoreExtensions, exts);
+ }
+ excludedExtensions = new TreeSet<String>(StringUtils.getStringsFromValue(exts));
}
/**
@@ -154,6 +159,14 @@
*/
@Override
public void run() {
+ if (!storedSettings.getBoolean(Keys.web.allowLuceneIndexing, true)) {
+ // Lucene indexing is disabled
+ return;
+ }
+ // reload the excluded extensions
+ String exts = storedSettings.getString(Keys.web.luceneIgnoreExtensions, luceneIgnoreExtensions);
+ excludedExtensions = new TreeSet<String>(StringUtils.getStringsFromValue(exts));
+
for (String repositoryName: GitBlit.self().getRepositoryList()) {
RepositoryModel model = GitBlit.self().getRepositoryModel(repositoryName);
if (model.hasCommits && !ArrayUtils.isEmpty(model.indexedBranches)) {
@@ -174,7 +187,7 @@
* @param repository
* the repository object
*/
- protected void index(RepositoryModel model, Repository repository) {
+ private void index(RepositoryModel model, Repository repository) {
try {
if (shouldReindex(repository)) {
// (re)build the entire index
@@ -273,7 +286,7 @@
close(repositoryName);
// delete the index folder
- File repositoryFolder = new File(repositoriesFolder, repositoryName);
+ File repositoryFolder = FileKey.resolve(new File(repositoriesFolder, repositoryName), FS.DETECTED);
File luceneIndex = new File(repositoryFolder, LUCENE_DIR);
if (luceneIndex.exists()) {
org.eclipse.jgit.util.FileUtils.delete(luceneIndex,
@@ -289,7 +302,6 @@
throw new RuntimeException(e);
}
}
-
/**
* Returns the author for the commit, if this information is available.
@@ -335,7 +347,7 @@
* @return tree
* @throws IOException
*/
- protected RevTree getTree(final RevWalk walk, final RevCommit commit)
+ private RevTree getTree(final RevWalk walk, final RevCommit commit)
throws IOException {
final RevTree tree = commit.getTree();
if (tree != null) {
@@ -375,7 +387,7 @@
* @param repository
* @return true of the on-disk index format is different than INDEX_VERSION
*/
- protected boolean shouldReindex(Repository repository) {
+ private boolean shouldReindex(Repository repository) {
try {
FileBasedConfig config = getConfig(repository);
config.load();
@@ -401,7 +413,8 @@
if (!deleteIndex(model.name)) {
return result;
}
- try {
+ try {
+ String [] encodings = storedSettings.getStrings(Keys.web.blobEncodings).toArray(new String[0]);
FileBasedConfig config = getConfig(repository);
Set<String> indexedCommits = new TreeSet<String>();
IndexWriter writer = getIndexWriter(model.name);
@@ -436,7 +449,7 @@
ObjectId defaultBranchId = JGitUtils.getDefaultBranch(repository);
for (RefModel branch : branches) {
if (branch.getObjectId().equals(defaultBranchId)) {
- defaultBranch = branch;
+ defaultBranch = branch;
break;
}
}
@@ -446,8 +459,22 @@
// walk through each branch
for (RefModel branch : branches) {
+ boolean indexBranch = false;
+ if (model.indexedBranches.contains(com.gitblit.Constants.DEFAULT_BRANCH)
+ && branch.equals(defaultBranch)) {
+ // indexing "default" branch
+ indexBranch = true;
+ } else if (IssueUtils.GB_ISSUES.equals(branch)) {
+ // skip the GB_ISSUES branch because it is indexed later
+ // note: this is different than updateIndex
+ indexBranch = false;
+ } else {
+ // normal explicit branch check
+ indexBranch = model.indexedBranches.contains(branch.getName());
+ }
+
// if this branch is not specifically indexed then skip
- if (!model.indexedBranches.contains(branch.getName())) {
+ if (!indexBranch) {
continue;
}
@@ -467,7 +494,10 @@
Map<String, ObjectId> paths = new TreeMap<String, ObjectId>();
while (treeWalk.next()) {
- paths.put(treeWalk.getPathString(), treeWalk.getObjectId(0));
+ // ensure path is not in a submodule
+ if (treeWalk.getFileMode(0) != FileMode.GITLINK) {
+ paths.put(treeWalk.getPathString(), treeWalk.getObjectId(0));
+ }
}
ByteArrayOutputStream os = new ByteArrayOutputStream();
@@ -530,14 +560,14 @@
// index the blob content
if (StringUtils.isEmpty(ext) || !excludedExtensions.contains(ext)) {
ObjectLoader ldr = repository.open(blobId, Constants.OBJ_BLOB);
- InputStream in = ldr.openStream();
+ InputStream in = ldr.openStream();
int n;
while ((n = in.read(tmp)) > 0) {
os.write(tmp, 0, n);
}
in.close();
byte[] content = os.toByteArray();
- String str = new String(content, Constants.CHARACTER_ENCODING);
+ String str = StringUtils.decodeString(content, encodings);
doc.add(new Field(FIELD_CONTENT, str, Store.YES, Index.ANALYZED));
os.reset();
}
@@ -616,13 +646,17 @@
String branch, RevCommit commit) {
IndexResult result = new IndexResult();
try {
+ String [] encodings = storedSettings.getStrings(Keys.web.blobEncodings).toArray(new String[0]);
List<PathChangeModel> changedPaths = JGitUtils.getFilesInCommit(repository, commit);
String revDate = DateTools.timeToString(commit.getCommitTime() * 1000L,
Resolution.MINUTE);
IndexWriter writer = getIndexWriter(repositoryName);
for (PathChangeModel path : changedPaths) {
+ if (path.isSubmodule()) {
+ continue;
+ }
// delete the indexed blob
- deleteBlob(repositoryName, branch, path.path);
+ deleteBlob(repositoryName, branch, path.name);
// re-index the blob
if (!ChangeType.DELETE.equals(path.changeType)) {
@@ -648,15 +682,26 @@
if (StringUtils.isEmpty(ext) || !excludedExtensions.contains(ext)) {
// read the blob content
String str = JGitUtils.getStringContent(repository, commit.getTree(),
- path.path);
- doc.add(new Field(FIELD_CONTENT, str, Store.YES, Index.ANALYZED));
- writer.addDocument(doc);
+ path.path, encodings);
+ if (str != null) {
+ doc.add(new Field(FIELD_CONTENT, str, Store.YES, Index.ANALYZED));
+ writer.addDocument(doc);
+ }
}
}
}
writer.commit();
-
- Document doc = createDocument(commit, null);
+
+ // get any annotated commit tags
+ List<String> commitTags = new ArrayList<String>();
+ for (RefModel ref : JGitUtils.getTags(repository, false, -1)) {
+ if (ref.isAnnotatedTag() && ref.getReferencedObjectId().equals(commit.getId())) {
+ commitTags.add(ref.displayName);
+ }
+ }
+
+ // create and write the Lucene document
+ Document doc = createDocument(commit, commitTags);
doc.add(new Field(FIELD_BRANCH, branch, Store.YES, Index.ANALYZED));
result.commitCount++;
result.success = index(repositoryName, doc);
@@ -692,8 +737,9 @@
* @param repositoryName
* @param issueId
* @throws Exception
+ * @return true, if deleted, false if no record was deleted
*/
- private void deleteIssue(String repositoryName, String issueId) throws Exception {
+ private boolean deleteIssue(String repositoryName, String issueId) throws Exception {
BooleanQuery query = new BooleanQuery();
Term objectTerm = new Term(FIELD_OBJECT_TYPE, SearchObjectType.issue.name());
query.add(new TermQuery(objectTerm), Occur.MUST);
@@ -701,8 +747,17 @@
query.add(new TermQuery(issueidTerm), Occur.MUST);
IndexWriter writer = getIndexWriter(repositoryName);
+ int numDocsBefore = writer.numDocs();
writer.deleteDocuments(query);
writer.commit();
+ int numDocsAfter = writer.numDocs();
+ if (numDocsBefore == numDocsAfter) {
+ logger.debug(MessageFormat.format("no records found to delete {0}", query.toString()));
+ return false;
+ } else {
+ logger.debug(MessageFormat.format("deleted {0} records with {1}", numDocsBefore - numDocsAfter, query.toString()));
+ return true;
+ }
}
/**
@@ -712,19 +767,29 @@
* @param branch
* @param path
* @throws Exception
+ * @return true, if deleted, false if no record was deleted
*/
- private void deleteBlob(String repositoryName, String branch, String path) throws Exception {
- BooleanQuery query = new BooleanQuery();
- Term objectTerm = new Term(FIELD_OBJECT_TYPE, SearchObjectType.blob.name());
- query.add(new TermQuery(objectTerm), Occur.MUST);
- Term branchTerm = new Term(FIELD_BRANCH, branch);
- query.add(new TermQuery(branchTerm), Occur.MUST);
- Term pathTerm = new Term(FIELD_PATH, path);
- query.add(new TermQuery(pathTerm), Occur.MUST);
+ public boolean deleteBlob(String repositoryName, String branch, String path) throws Exception {
+ String pattern = MessageFormat.format("{0}:'{'0} AND {1}:\"'{'1'}'\" AND {2}:\"'{'2'}'\"", FIELD_OBJECT_TYPE, FIELD_BRANCH, FIELD_PATH);
+ String q = MessageFormat.format(pattern, SearchObjectType.blob.name(), branch, path);
+ BooleanQuery query = new BooleanQuery();
+ StandardAnalyzer analyzer = new StandardAnalyzer(LUCENE_VERSION);
+ QueryParser qp = new QueryParser(LUCENE_VERSION, FIELD_SUMMARY, analyzer);
+ query.add(qp.parse(q), Occur.MUST);
+
IndexWriter writer = getIndexWriter(repositoryName);
- writer.deleteDocuments(query);
+ int numDocsBefore = writer.numDocs();
+ writer.deleteDocuments(query);
writer.commit();
+ int numDocsAfter = writer.numDocs();
+ if (numDocsBefore == numDocsAfter) {
+ logger.debug(MessageFormat.format("no records found to delete {0}", query.toString()));
+ return false;
+ } else {
+ logger.debug(MessageFormat.format("deleted {0} records with {1}", numDocsBefore - numDocsAfter, query.toString()));
+ return true;
+ }
}
/**
@@ -734,7 +799,7 @@
* @param repository
* @return IndexResult
*/
- protected IndexResult updateIndex(RepositoryModel model, Repository repository) {
+ private IndexResult updateIndex(RepositoryModel model, Repository repository) {
IndexResult result = new IndexResult();
try {
FileBasedConfig config = getConfig(repository);
@@ -762,20 +827,55 @@
deletedBranches.add(branch);
}
- // walk through each branches
+ // get the local branches
List<RefModel> branches = JGitUtils.getLocalBranches(repository, true, -1);
+
+ // sort them by most recently updated
+ Collections.sort(branches, new Comparator<RefModel>() {
+ @Override
+ public int compare(RefModel ref1, RefModel ref2) {
+ return ref2.getDate().compareTo(ref1.getDate());
+ }
+ });
+
+ // reorder default branch to first position
+ RefModel defaultBranch = null;
+ ObjectId defaultBranchId = JGitUtils.getDefaultBranch(repository);
+ for (RefModel branch : branches) {
+ if (branch.getObjectId().equals(defaultBranchId)) {
+ defaultBranch = branch;
+ break;
+ }
+ }
+ branches.remove(defaultBranch);
+ branches.add(0, defaultBranch);
+
+ // walk through each branches
for (RefModel branch : branches) {
String branchName = branch.getName();
- // determine if we should skip this branch
- if (!IssueUtils.GB_ISSUES.equals(branch)
- && !model.indexedBranches.contains(branch.getName())) {
+ boolean indexBranch = false;
+ if (model.indexedBranches.contains(com.gitblit.Constants.DEFAULT_BRANCH)
+ && branch.equals(defaultBranch)) {
+ // indexing "default" branch
+ indexBranch = true;
+ } else if (IssueUtils.GB_ISSUES.equals(branch)) {
+ // update issues modified on the GB_ISSUES branch
+ // note: this is different than reindex
+ indexBranch = true;
+ } else {
+ // normal explicit branch check
+ indexBranch = model.indexedBranches.contains(branch.getName());
+ }
+
+ // if this branch is not specifically indexed then skip
+ if (!indexBranch) {
continue;
}
// remove this branch from the deletedBranches set
deletedBranches.remove(branchName);
-
+
// determine last commit
String keyName = getBranchKey(branchName);
String lastCommit = config.getString(CONF_BRANCH, null, keyName);
@@ -810,7 +910,9 @@
IssueModel issue = IssueUtils.getIssue(repository, issueId);
if (issue == null) {
// issue was deleted, remove from index
- deleteIssue(model.name, issueId);
+ if (!deleteIssue(model.name, issueId)) {
+ logger.error(MessageFormat.format("Failed to delete issue {0} from Lucene index!", issueId));
+ }
} else {
// issue was updated
index(model.name, issue);
@@ -973,8 +1075,8 @@
* @throws IOException
*/
private IndexWriter getIndexWriter(String repository) throws IOException {
- IndexWriter indexWriter = writers.get(repository);
- File repositoryFolder = new File(repositoriesFolder, repository);
+ IndexWriter indexWriter = writers.get(repository);
+ File repositoryFolder = FileKey.resolve(new File(repositoriesFolder, repository), FS.DETECTED);
File indexFolder = new File(repositoryFolder, LUCENE_DIR);
Directory directory = FSDirectory.open(indexFolder);
@@ -1048,7 +1150,7 @@
qp = new QueryParser(LUCENE_VERSION, FIELD_CONTENT, analyzer);
qp.setAllowLeadingWildcard(true);
query.add(qp.parse(text), Occur.SHOULD);
-
+
IndexSearcher searcher;
if (repositories.length == 1) {
// single repository search
@@ -1064,7 +1166,10 @@
MultiSourceReader reader = new MultiSourceReader(rdrs);
searcher = new IndexSearcher(reader);
}
+
Query rewrittenQuery = searcher.rewrite(query);
+ logger.debug(rewrittenQuery.toString());
+
TopScoreDocCollector collector = TopScoreDocCollector.create(5000, true);
searcher.search(rewrittenQuery, collector);
int offset = Math.max(0, (page - 1) * pageSize);
@@ -1105,42 +1210,86 @@
*/
private String getHighlightedFragment(Analyzer analyzer, Query query,
String content, SearchResult result) throws IOException, InvalidTokenOffsetsException {
- content = content == null ? "":StringUtils.escapeForHtml(content, false);
-
+ if (content == null) {
+ content = "";
+ }
+
+ int fragmentLength = SearchObjectType.commit == result.type ? 512 : 150;
+
QueryScorer scorer = new QueryScorer(query, "content");
- Fragmenter fragmenter;
-
- // TODO improve the fragmenter - hopefully on line breaks
- if (SearchObjectType.commit == result.type) {
- fragmenter = new SimpleSpanFragmenter(scorer, 1024);
- } else {
- fragmenter = new SimpleSpanFragmenter(scorer, 150);
- }
+ Fragmenter fragmenter = new SimpleSpanFragmenter(scorer, fragmentLength);
// use an artificial delimiter for the token
- String termTag = "<!--[";
- String termTagEnd = "]-->";
+ String termTag = "!!--[";
+ String termTagEnd = "]--!!";
SimpleHTMLFormatter formatter = new SimpleHTMLFormatter(termTag, termTagEnd);
Highlighter highlighter = new Highlighter(formatter, scorer);
highlighter.setTextFragmenter(fragmenter);
-
+
String [] fragments = highlighter.getBestFragments(analyzer, "content", content, 3);
if (ArrayUtils.isEmpty(fragments)) {
if (SearchObjectType.blob == result.type) {
return "";
}
- return "<pre class=\"text\">" + content + "</pre>";
+ // clip commit message
+ String fragment = content;
+ if (fragment.length() > fragmentLength) {
+ fragment = fragment.substring(0, fragmentLength) + "...";
+ }
+ return "<pre class=\"text\">" + StringUtils.escapeForHtml(fragment, true) + "</pre>";
}
+
+ // make sure we have unique fragments
+ Set<String> uniqueFragments = new LinkedHashSet<String>();
+ for (String fragment : fragments) {
+ uniqueFragments.add(fragment);
+ }
+ fragments = uniqueFragments.toArray(new String[uniqueFragments.size()]);
+
StringBuilder sb = new StringBuilder();
for (int i = 0, len = fragments.length; i < len; i++) {
String fragment = fragments[i];
-
+ String tag = "<pre class=\"text\">";
+
// resurrect the raw fragment from removing the artificial delimiters
- String raw = fragment.replace(termTag, "").replace(termTagEnd, "");
- sb.append(getPreTag(result, raw, content));
+ String raw = fragment.replace(termTag, "").replace(termTagEnd, "");
+
+ // determine position of the raw fragment in the content
+ int pos = content.indexOf(raw);
+
+ // restore complete first line of fragment
+ int c = pos;
+ while (c > 0) {
+ c--;
+ if (content.charAt(c) == '\n') {
+ break;
+ }
+ }
+ if (c > 0) {
+ // inject leading chunk of first fragment line
+ fragment = content.substring(c + 1, pos) + fragment;
+ }
+
+ if (SearchObjectType.blob == result.type) {
+ // count lines as offset into the content for this fragment
+ int line = Math.max(1, StringUtils.countLines(content.substring(0, pos)));
+
+ // create fragment tag with line number and language
+ String lang = "";
+ String ext = StringUtils.getFileExtension(result.path).toLowerCase();
+ if (!StringUtils.isEmpty(ext)) {
+ // maintain leading space!
+ lang = " lang-" + ext;
+ }
+ tag = MessageFormat.format("<pre class=\"prettyprint linenums:{0,number,0}{1}\">", line, lang);
+
+ }
+ sb.append(tag);
+
// replace the artificial delimiter with html tags
- String html = fragment.replace(termTag, "<span class=\"highlight\">").replace(termTagEnd, "</span>");
+ String html = StringUtils.escapeForHtml(fragment, false);
+ html = html.replace(termTag, "<span class=\"highlight\">").replace(termTagEnd, "</span>");
sb.append(html);
sb.append("</pre>");
if (i < len - 1) {
@@ -1148,31 +1297,7 @@
}
}
return sb.toString();
- }
-
- /**
- * Returns the appropriate tag for a fragment. Commit messages are visually
- * differentiated from blob fragments.
- *
- * @param result
- * @param fragment
- * @param content
- * @return an html tag appropriate for the fragment
- */
- private String getPreTag(SearchResult result, String fragment, String content) {
- String pre = "<pre class=\"text\">";
- if (SearchObjectType.blob == result.type) {
- int line = StringUtils.countLines(content.substring(0, content.indexOf(fragment)));
- int lastDot = result.path.lastIndexOf('.');
- if (lastDot > -1) {
- String ext = result.path.substring(lastDot + 1).toLowerCase();
- pre = MessageFormat.format("<pre class=\"prettyprint linenums:{0,number,0} lang-{1}\">", line, ext);
- } else {
- pre = MessageFormat.format("<pre class=\"prettyprint linenums:{0,number,0}\">", line);
- }
- }
- return pre;
- }
+ }
/**
* Simple class to track the results of an index update.
--
Gitblit v1.9.1