Indexer.java revision 240
0N/A/*
0N/A * CDDL HEADER START
0N/A *
0N/A * The contents of this file are subject to the terms of the
0N/A * Common Development and Distribution License (the "License").
0N/A * You may not use this file except in compliance with the License.
0N/A *
0N/A * See LICENSE.txt included in this distribution for the specific
0N/A * language governing permissions and limitations under the License.
0N/A *
0N/A * When distributing Covered Code, include this CDDL HEADER in each
0N/A * file and include the License file at LICENSE.txt.
0N/A * If applicable, add the following below this CDDL HEADER, with the
0N/A * fields enclosed by brackets "[]" replaced with your own identifying
0N/A * information: Portions Copyright [yyyy] [name of copyright owner]
0N/A *
0N/A * CDDL HEADER END
0N/A */
0N/A
0N/A/*
119N/A * Copyright 2007 Sun Microsystems, Inc. All rights reserved.
0N/A * Use is subject to license terms.
0N/A */
0N/Apackage org.opensolaris.opengrok.index;
65N/A
0N/Aimport java.awt.GraphicsEnvironment;
125N/Aimport java.io.BufferedReader;
125N/Aimport java.io.File;
125N/Aimport java.io.FileReader;
125N/Aimport java.io.IOException;
58N/Aimport java.net.InetAddress;
77N/Aimport java.text.ParseException;
125N/Aimport java.util.ArrayList;
125N/Aimport java.util.Collections;
125N/Aimport java.util.Comparator;
125N/Aimport java.util.List;
126N/Aimport org.opensolaris.opengrok.analysis.AnalyzerGuru;
204N/Aimport org.opensolaris.opengrok.analysis.FileAnalyzerFactory;
58N/Aimport org.opensolaris.opengrok.configuration.Project;
8N/Aimport org.opensolaris.opengrok.history.HistoryGuru;
30N/Aimport org.opensolaris.opengrok.configuration.RuntimeEnvironment;
240N/Aimport org.opensolaris.opengrok.history.ExternalRepository;
0N/Aimport org.opensolaris.opengrok.search.scope.MainFrame;
77N/Aimport org.opensolaris.opengrok.util.Getopt;
0N/A
0N/A/**
0N/A * Creates and updates an inverted source index
0N/A * as well as generates Xref, file stats etc., if specified
0N/A * in the options
0N/A */
0N/Apublic class Indexer {
0N/A private static String usage = "Usage: " +
205N/A "opengrok.jar [-qe] [-c ctagsToUse] [-H] [-R filename] [-W filename] [-U hostname:port] [-P] [-p project-path] [-w webapproot] [-i ignore_name [ -i ..]] [-n] [-s SRC_ROOT] [-d DATA_ROOT] [subtree .. ]\n" +
205N/A " opengrok.jar [-O | -l | -t] [-d DATA_ROOT]\n" +
0N/A "\t-q run quietly\n" +
119N/A "\t-v Print progress information\n" +
0N/A "\t-e economical - consumes less disk space\n" +
0N/A "\t-c path to ctags\n" +
58N/A "\t-R Read configuration from file\n" +
58N/A "\t-W Write the current running configuration\n" +
58N/A "\t-U Send configuration to hostname:port\n" +
58N/A "\t-P Generate a project for each toplevel directory\n" +
77N/A "\t-p Use the project specified by the project path as the default project\n" +
119N/A "\t-Q on/off Turn on / off quick context scan. By default only the first 32k\n" +
119N/A "\t of a file is scanned and a '[..all..]' link is inserted if the\n" +
119N/A "\t is bigger. Activating this option may slow down the server.\n" +
58N/A "\t-n Do not generate indexes\n" +
240N/A "\t-H Generate history cache for all external repositories\n" +
240N/A "\t-h /path/to/repos Generate history cache for the specified repos (absolute path from source root)\n" +
173N/A "\t-r on/off Turn on / off support for remote SCM systems\n" +
144N/A "\t-L laf Use \"laf\" as the look'n'feel for the webapp\n" +
0N/A "\t-w root URL of the webapp, default is /source\n" +
0N/A "\t-i ignore named files or directories\n" +
126N/A "\t-A ext:analyzer Files with extension ext should be analyzed with the named class\n" +
99N/A "\t-m Maximum words in a file to index\n" +
125N/A "\t-a on/off Allow or disallow leading wildcards in a search\n" +
77N/A "\t-S Search and add \"External\" repositories (Mercurial etc)\n" +
0N/A "\t-s SRC_ROOT is root directory of source tree\n" +
0N/A "\t default: last used SRC_ROOT\n" +
205N/A "\t-d DATA_ROOT - is where output of indexer is stored\n" +
0N/A "\tsubtree - only specified files or directories under SRC_ROOT are processed\n" +
0N/A "\t if not specified all files under SRC_ROOT are processed\n" +
207N/A "\n" +
119N/A "\t-l list all files in the index\n" +
0N/A "\t-t lists tokens occuring more than 5 times. Useful for building a unix dictionary\n" +
0N/A "\n Eg. java -jar opengrok.jar -s /usr/include /var/tmp/opengrok_data rpc";
77N/A
240N/A private static String options = "d:r:a:qec:Q:R:W:U:Pp:nHh:w:i:Ss:ltvm:A:L:";
77N/A
65N/A /**
65N/A * Program entry point
65N/A * @param argv argument vector
65N/A */
0N/A public static void main(String argv[]) {
30N/A RuntimeEnvironment env = RuntimeEnvironment.getInstance();
58N/A boolean runIndex = true;
112N/A
0N/A if(argv.length == 0) {
0N/A if (GraphicsEnvironment.isHeadless()) {
0N/A System.err.println("No display available for the Graphical User Interface");
0N/A System.err.println(usage);
0N/A System.exit(1);
0N/A } else {
0N/A MainFrame.main(argv);
0N/A }
0N/A //Run Scope GUI here I am running Indexing GUI for testing
0N/A //new IndexerWizard(null).setVisible(true);
0N/A } else {
11N/A boolean searchRepositories = false;
0N/A ArrayList<String> subFiles = new ArrayList<String>();
240N/A ArrayList<String> repositories = new ArrayList<String>();
58N/A String configFilename = null;
58N/A String configHost = null;
58N/A boolean addProjects = false;
58N/A boolean refreshHistory = false;
77N/A String defaultProject = null;
207N/A boolean listFiles = false;
207N/A boolean createDict = false;
77N/A
77N/A // Parse command line options:
77N/A Getopt getopt = new Getopt(argv, options);
112N/A
77N/A try {
77N/A getopt.parse();
77N/A } catch (ParseException ex) {
77N/A System.err.println("OpenGrok: " + ex.getMessage());
77N/A System.err.println(usage);
77N/A System.exit(1);
77N/A }
77N/A
0N/A try{
77N/A int cmd;
111N/A
111N/A // We need to read the configuration file first, since we
111N/A // will try to overwrite options..
111N/A while ((cmd = getopt.getOpt()) != -1) {
111N/A if (cmd == 'R') {
111N/A env.readConfiguration(new File(getopt.getOptarg()));
111N/A break;
111N/A }
111N/A }
111N/A
111N/A // Now we can handle all the other options..
111N/A getopt.reset();
77N/A while ((cmd = getopt.getOpt()) != -1) {
77N/A switch (cmd) {
77N/A case 'l':
207N/A listFiles = true;
207N/A runIndex = false;
77N/A break;
77N/A case 't':
207N/A createDict = true;
207N/A runIndex = false;
77N/A break;
77N/A
99N/A case 'q': env.setVerbose(false); break;
77N/A case 'e': env.setGenerateHtml(false); break;
77N/A case 'P': addProjects = true; break;
77N/A case 'p': defaultProject = getopt.getOptarg(); break;
77N/A case 'c': env.setCtags(getopt.getOptarg()); break;
77N/A case 'w': {
77N/A String webapp = getopt.getOptarg();
77N/A if (webapp.startsWith("/") || webapp.startsWith("http")) {
77N/A ;
0N/A } else {
77N/A webapp = "/" + webapp;
77N/A }
77N/A if (webapp.endsWith("/")) {
77N/A env.setUrlPrefix(webapp + "s?");
77N/A } else {
77N/A env.setUrlPrefix(webapp + "/s?");
77N/A }
77N/A }
77N/A break;
77N/A case 'W': configFilename = getopt.getOptarg(); break;
77N/A case 'U': configHost = getopt.getOptarg(); break;
111N/A case 'R':
111N/A // already handled
77N/A break;
77N/A case 'n': runIndex = false; break;
77N/A case 'H': refreshHistory = true; break;
240N/A case 'h' : repositories.add(getopt.getOptarg()); break;
173N/A case 'r': {
173N/A if (getopt.getOptarg().equalsIgnoreCase("on")) {
173N/A env.setRemoteScmSupported(true);
173N/A } else if (getopt.getOptarg().equalsIgnoreCase("off")) {
173N/A env.setRemoteScmSupported(true);
173N/A } else {
173N/A System.err.println("ERROR: You should pass either \"on\" or \"off\" as argument to -r");
173N/A System.err.println(" Ex: \"-r on\" will allow retrival for remote SCM systems");
173N/A System.err.println(" \"-Q off\" will ignore SCM for remote systems");
173N/A }
173N/A }
173N/A break;
99N/A case 'v': env.setVerbose(true); break;
77N/A
77N/A case 's': {
77N/A File file = new File(getopt.getOptarg());
77N/A if (!file.isDirectory()) {
77N/A System.err.println("ERROR: No such directory: " + file.toString());
0N/A System.exit(1);
0N/A }
77N/A
125N/A env.setSourceRootFile(file);
77N/A break;
77N/A }
205N/A case 'd':
205N/A env.setDataRoot(getopt.getOptarg());
205N/A break;
112N/A case 'i':
112N/A env.getIgnoredNames().add(getopt.getOptarg());
112N/A break;
77N/A case 'S' : searchRepositories = true; break;
106N/A case 'Q' :
119N/A if (getopt.getOptarg().equalsIgnoreCase("on")) {
106N/A env.setQuickContextScan(true);
119N/A } else if (getopt.getOptarg().equalsIgnoreCase("off")) {
106N/A env.setQuickContextScan(false);
119N/A } else {
119N/A System.err.println("ERROR: You should pass either \"on\" or \"off\" as argument to -Q");
119N/A System.err.println(" Ex: \"-Q on\" will just scan a \"chunk\" of the file and insert \"[..all..]\"");
119N/A System.err.println(" \"-Q off\" will try to build a more accurate list by reading the complete file.");
106N/A }
106N/A
106N/A break;
99N/A case 'm' : {
99N/A try {
99N/A env.setIndexWordLimit(Integer.parseInt(getopt.getOptarg()));
99N/A } catch (NumberFormatException exp) {
99N/A System.err.println("ERROR: Failed to parse argument to \"-m\": " + exp.getMessage());
99N/A System.exit(1);
99N/A }
99N/A break;
99N/A }
125N/A case 'a' :
125N/A if (getopt.getOptarg().equalsIgnoreCase("on")) {
125N/A env.setAllowLeadingWildcard(true);
125N/A } else if (getopt.getOptarg().equalsIgnoreCase("off")) {
125N/A env.setAllowLeadingWildcard(false);
125N/A } else {
125N/A System.err.println("ERROR: You should pass either \"on\" or \"off\" as argument to -a");
125N/A System.err.println(" Ex: \"-a on\" will allow a search to start with a wildcard");
125N/A System.err.println(" \"-a off\" will disallow a search to start with a wildcard");
126N/A System.exit(1);
125N/A }
125N/A
125N/A break;
126N/A
126N/A case 'A': {
126N/A String[] arg = getopt.getOptarg().split(":");
126N/A if (arg.length != 2) {
126N/A System.err.println("ERROR: You must specify: -A extension:class");
126N/A System.err.println(" Ex: -A foo:org.opensolaris.opengrok.analysis.c.CAnalyzer");
126N/A System.err.println(" will use the C analyzer for all files ending with .foo");
126N/A System.err.println(" Ex: -A c:-");
126N/A System.err.println(" will disable the c-analyzer for for all files ending with .c");
126N/A System.exit(1);
126N/A }
126N/A
126N/A arg[0] = arg[0].substring(arg[0].lastIndexOf('.') + 1).toUpperCase();
126N/A if (arg[1].equals("-")) {
126N/A AnalyzerGuru.addExtension(arg[0], null);
126N/A break;
126N/A }
126N/A
126N/A try {
210N/A AnalyzerGuru.addExtension(
210N/A arg[0],
210N/A AnalyzerGuru.findFactory(arg[1]));
210N/A } catch (Exception e) {
210N/A System.err.println("Unable to use " + arg[1] +
210N/A " as a FileAnalyzerFactory");
210N/A e.printStackTrace();
126N/A System.exit(1);
126N/A }
126N/A }
126N/A break;
144N/A case 'L' :
144N/A env.setWebappLAF(getopt.getOptarg());
144N/A break;
77N/A default:
99N/A System.err.println("Unknown option: " + (char)cmd);
0N/A System.exit(1);
0N/A }
0N/A }
77N/A
77N/A int optind = getopt.getOptind();
205N/A if (optind != -1) {
77N/A while (optind < argv.length) {
77N/A subFiles.add(argv[optind]);
77N/A ++optind;
77N/A }
77N/A }
58N/A
58N/A if (env.getDataRootPath() == null) {
77N/A System.err.println("ERROR: Please specify a DATA ROOT path");
77N/A System.err.println(usage);
0N/A System.exit(1);
0N/A }
77N/A
58N/A if (env.getSourceRootFile() == null) {
58N/A File srcConfig = new File(env.getDataRootPath(), "SRC_ROOT");
58N/A String line = null;
0N/A if(srcConfig.exists()) {
0N/A try {
0N/A BufferedReader sr = new BufferedReader(new FileReader(srcConfig));
58N/A line = sr.readLine();
0N/A sr.close();
0N/A } catch (IOException e) {
0N/A }
0N/A }
58N/A if(line == null) {
0N/A System.err.println("ERROR: please specify a SRC_ROOT with option -s !");
0N/A System.err.println(usage);
0N/A System.exit(1);
0N/A }
58N/A env.setSourceRoot(line);
77N/A
58N/A if (!env.getSourceRootFile().isDirectory()) {
58N/A System.err.println("ERROR: No such directory:" + line);
0N/A System.err.println(usage);
0N/A System.exit(1);
0N/A }
0N/A }
77N/A
99N/A if (!env.validateExuberantCtags()) {
0N/A System.exit(1);
0N/A }
77N/A
11N/A if (searchRepositories) {
99N/A if (env.isVerbose()) {
99N/A System.out.println("Scanning for repositories...");
99N/A }
99N/A env.getRepositories().clear();
11N/A long start = System.currentTimeMillis();
58N/A HistoryGuru.getInstance().addExternalRepositories(env.getSourceRootPath());
11N/A long time = (System.currentTimeMillis() - start) / 1000;
99N/A if (env.isVerbose()) {
99N/A System.out.println("Done searching for repositories (" + time + "s)");
99N/A }
11N/A }
77N/A
58N/A if (addProjects) {
77N/A File files[] = env.getSourceRootFile().listFiles();
70N/A List<Project> projects = env.getProjects();
70N/A projects.clear();
58N/A for (File file : files) {
77N/A if (!file.getName().startsWith(".") && file.isDirectory()) {
70N/A projects.add(new Project(file.getName(), "/" + file.getName()));
58N/A }
58N/A }
77N/A
77N/A // The projects should be sorted...
99N/A Collections.sort(projects, new Comparator<Project>() {
99N/A public int compare(Project p1, Project p2){
99N/A String s1 = p1.getDescription();
99N/A String s2 = p2.getDescription();
77N/A
77N/A int ret;
77N/A if (s1 == null) {
77N/A ret = (s2 == null) ? 0 : 1;
77N/A } else {
77N/A ret = s1.compareTo(s2);
77N/A }
77N/A return ret;
77N/A }
77N/A });
58N/A }
77N/A
77N/A if (defaultProject != null) {
77N/A for (Project p : env.getProjects()) {
77N/A if (p.getPath().equals(defaultProject)) {
77N/A env.setDefaultProject(p);
77N/A break;
77N/A }
77N/A }
77N/A }
99N/A
58N/A if (configFilename != null) {
99N/A if (env.isVerbose()) {
99N/A System.out.println("Writing configuration to " + configFilename);
99N/A System.out.flush();
99N/A }
58N/A env.writeConfiguration(new File(configFilename));
99N/A if (env.isVerbose()) {
99N/A System.out.println("Done...");
99N/A System.out.flush();
99N/A }
58N/A }
112N/A
58N/A if (refreshHistory) {
203N/A HistoryGuru.getInstance().createCache();
240N/A } else if (repositories.size() > 0) {
240N/A HistoryGuru.getInstance().createCache(repositories);
58N/A }
58N/A
207N/A if (listFiles) {
207N/A IndexDatabase.listAllFiles(subFiles);
207N/A }
207N/A
207N/A if (createDict) {
207N/A IndexDatabase.listFrequentTokens(subFiles);
207N/A }
207N/A
58N/A if (runIndex) {
207N/A IndexChangedListener progress = new DefaultIndexChangedListener();
207N/A if (subFiles.isEmpty() || !env.hasProjects()) {
207N/A IndexDatabase.updateAll(progress);
207N/A } else {
207N/A for (String path : subFiles) {
207N/A Project project = Project.getProject(path);
207N/A if (project == null) {
207N/A System.err.println("Warning: Could not find a project for \"" + path + "\"");
207N/A } else {
207N/A IndexDatabase db = new IndexDatabase(project);
207N/A db.addIndexChangedListener(progress);
207N/A db.update();
207N/A }
207N/A }
207N/A }
58N/A }
58N/A
58N/A if (configHost != null) {
58N/A String[] cfg = configHost.split(":");
99N/A if (env.isVerbose()) {
99N/A System.out.println("Send configuration to: " + configHost);
99N/A }
58N/A
58N/A if (cfg.length == 2) {
58N/A try {
58N/A InetAddress host = InetAddress.getByName(cfg[0]);
58N/A RuntimeEnvironment.getInstance().writeConfiguration(host, Integer.parseInt(cfg[1]));
58N/A } catch (Exception ex) {
58N/A System.err.println("Failed to send configuration to " + configHost);
58N/A ex.printStackTrace();
58N/A }
58N/A } else {
58N/A System.err.println("Syntax error: ");
58N/A for (String s : cfg) {
58N/A System.err.print("[" + s + "]");
58N/A }
58N/A System.err.println();
58N/A }
99N/A if (env.isVerbose()) {
99N/A System.out.println("Configuration successfully updated");
99N/A }
58N/A }
58N/A } catch (Exception e) {
0N/A System.err.println("Error: [ main ] " + e);
99N/A if (env.isVerbose()) e.printStackTrace();
0N/A System.exit(1);
0N/A }
0N/A }
0N/A }
0N/A}