Indexer.java revision 119
494N/A/*
494N/A * CDDL HEADER START
494N/A *
494N/A * The contents of this file are subject to the terms of the
494N/A * Common Development and Distribution License (the "License").
494N/A * You may not use this file except in compliance with the License.
494N/A *
494N/A * See LICENSE.txt included in this distribution for the specific
494N/A * language governing permissions and limitations under the License.
494N/A *
494N/A * When distributing Covered Code, include this CDDL HEADER in each
494N/A * file and include the License file at LICENSE.txt.
494N/A * If applicable, add the following below this CDDL HEADER, with the
494N/A * fields enclosed by brackets "[]" replaced with your own identifying
494N/A * information: Portions Copyright [yyyy] [name of copyright owner]
494N/A *
494N/A * CDDL HEADER END
494N/A */
494N/A
494N/A/*
494N/A * Copyright 2007 Sun Microsystems, Inc. All rights reserved.
494N/A * Use is subject to license terms.
494N/A */
494N/A
494N/Apackage org.opensolaris.opengrok.index;
697N/A
697N/Aimport java.awt.GraphicsEnvironment;
697N/Aimport java.io.*;
697N/Aimport java.net.InetAddress;
697N/Aimport java.text.ParseException;
697N/Aimport java.util.*;
697N/Aimport org.apache.lucene.document.*;
494N/Aimport org.apache.lucene.index.*;
697N/Aimport org.opensolaris.opengrok.analysis.*;
700N/Aimport org.opensolaris.opengrok.configuration.Project;
697N/Aimport org.opensolaris.opengrok.history.ExternalRepository;
697N/Aimport org.opensolaris.opengrok.history.HistoryGuru;
697N/Aimport org.opensolaris.opengrok.configuration.RuntimeEnvironment;
494N/Aimport org.opensolaris.opengrok.search.scope.MainFrame;
494N/Aimport org.opensolaris.opengrok.util.Getopt;
494N/A
494N/A/**
494N/A * Creates and updates an inverted source index
494N/A * as well as generates Xref, file stats etc., if specified
494N/A * in the options
494N/A */
494N/Apublic class Indexer {
697N/A private static String usage = "Usage: " +
697N/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] DATA_ROOT [subtree .. ]\n" +
697N/A " opengrok.jar [-O | -l | -t] DATA_ROOT\n" +
697N/A "\t-q run quietly\n" +
697N/A "\t-v Print progress information\n" +
697N/A "\t-e economical - consumes less disk space\n" +
697N/A "\t-c path to ctags\n" +
700N/A "\t-R Read configuration from file\n" +
697N/A "\t-W Write the current running configuration\n" +
697N/A "\t-U Send configuration to hostname:port\n" +
697N/A "\t-P Generate a project for each toplevel directory\n" +
697N/A "\t-p Use the project specified by the project path as the default project\n" +
697N/A "\t-Q on/off Turn on / off quick context scan. By default only the first 32k\n" +
697N/A "\t of a file is scanned and a '[..all..]' link is inserted if the\n" +
697N/A "\t is bigger. Activating this option may slow down the server.\n" +
697N/A "\t-n Do not generate indexes\n" +
697N/A "\t-H Generate history cache for external repositories\n" +
697N/A "\t-w root URL of the webapp, default is /source\n" +
697N/A "\t-i ignore named files or directories\n" +
697N/A "\t-m Maximum words in a file to index\n" +
697N/A "\t-S Search and add \"External\" repositories (Mercurial etc)\n" +
697N/A "\t-s SRC_ROOT is root directory of source tree\n" +
697N/A "\t default: last used SRC_ROOT\n" +
697N/A "\tDATA_ROOT - is where output of indexer is stored\n" +
697N/A "\tsubtree - only specified files or directories under SRC_ROOT are processed\n" +
697N/A "\t if not specified all files under SRC_ROOT are processed\n" +
697N/A "\n\t-O optimize the index \n" +
697N/A "\t-l list all files in the index\n" +
697N/A "\t-D list the index uid's\n" +
697N/A "\t-t lists tokens occuring more than 5 times. Useful for building a unix dictionary\n" +
697N/A "\n Eg. java -jar opengrok.jar -s /usr/include /var/tmp/opengrok_data rpc";
697N/A
697N/A private static String options = "qec:Q:R:W:U:Pp:nHw:i:Ss:O:l:t:vD:m:";
697N/A
697N/A /**
697N/A * Program entry point
697N/A * @param argv argument vector
697N/A */
697N/A public static void main(String argv[]) {
697N/A RuntimeEnvironment env = RuntimeEnvironment.getInstance();
697N/A boolean runIndex = true;
697N/A
697N/A if(argv.length == 0) {
697N/A if (GraphicsEnvironment.isHeadless()) {
494N/A System.err.println("No display available for the Graphical User Interface");
494N/A System.err.println(usage);
494N/A System.exit(1);
494N/A } else {
494N/A MainFrame.main(argv);
494N/A }
494N/A //Run Scope GUI here I am running Indexing GUI for testing
494N/A //new IndexerWizard(null).setVisible(true);
494N/A } else {
494N/A boolean searchRepositories = false;
494N/A ArrayList<String> subFiles = new ArrayList<String>();
494N/A String configFilename = null;
494N/A String configHost = null;
494N/A boolean addProjects = false;
494N/A boolean refreshHistory = false;
494N/A String defaultProject = null;
494N/A
494N/A // Parse command line options:
494N/A Getopt getopt = new Getopt(argv, options);
494N/A
494N/A try {
494N/A getopt.parse();
494N/A } catch (ParseException ex) {
494N/A System.err.println("OpenGrok: " + ex.getMessage());
494N/A System.err.println(usage);
494N/A System.exit(1);
494N/A }
494N/A
494N/A try{
494N/A int cmd;
494N/A
494N/A // We need to read the configuration file first, since we
494N/A // will try to overwrite options..
494N/A while ((cmd = getopt.getOpt()) != -1) {
494N/A if (cmd == 'R') {
494N/A env.readConfiguration(new File(getopt.getOptarg()));
494N/A break;
494N/A }
494N/A }
494N/A
494N/A // Now we can handle all the other options..
494N/A getopt.reset();
494N/A while ((cmd = getopt.getOpt()) != -1) {
494N/A switch (cmd) {
494N/A case 'D' :
494N/A Index.dumpU(new File(getopt.getOptarg()));
494N/A System.exit(0);
494N/A break;
494N/A case 'O':
494N/A Index.doOptimize(new File(getopt.getOptarg()));
494N/A System.exit(0);
494N/A break;
494N/A case 'l':
494N/A Index.doList(new File(getopt.getOptarg()));
494N/A System.exit(0);
494N/A break;
494N/A case 't':
494N/A Index.doDict(new File(getopt.getOptarg()));
494N/A System.exit(0);
494N/A break;
494N/A
494N/A case 'q': env.setVerbose(false); break;
494N/A case 'e': env.setGenerateHtml(false); break;
494N/A case 'P': addProjects = true; break;
494N/A case 'p': defaultProject = getopt.getOptarg(); break;
697N/A case 'c': env.setCtags(getopt.getOptarg()); break;
697N/A case 'w': {
697N/A String webapp = getopt.getOptarg();
697N/A if (webapp.startsWith("/") || webapp.startsWith("http")) {
697N/A ;
697N/A } else {
697N/A webapp = "/" + webapp;
700N/A }
697N/A if (webapp.endsWith("/")) {
700N/A env.setUrlPrefix(webapp + "s?");
700N/A } else {
700N/A env.setUrlPrefix(webapp + "/s?");
700N/A }
700N/A }
700N/A break;
697N/A case 'W': configFilename = getopt.getOptarg(); break;
700N/A case 'U': configHost = getopt.getOptarg(); break;
700N/A case 'R':
700N/A // already handled
700N/A break;
700N/A case 'n': runIndex = false; break;
697N/A case 'H': refreshHistory = true; break;
700N/A case 'v': env.setVerbose(true); break;
700N/A
700N/A case 's': {
700N/A File file = new File(getopt.getOptarg());
700N/A if (!file.isDirectory()) {
697N/A System.err.println("ERROR: No such directory: " + file.toString());
700N/A System.exit(1);
700N/A }
700N/A
697N/A env.setSourceRoot(file);
700N/A break;
700N/A }
700N/A
700N/A case 'i':
700N/A env.getIgnoredNames().add(getopt.getOptarg());
700N/A break;
700N/A case 'S' : searchRepositories = true; break;
700N/A case 'Q' :
700N/A if (getopt.getOptarg().equalsIgnoreCase("on")) {
700N/A env.setQuickContextScan(true);
700N/A } else if (getopt.getOptarg().equalsIgnoreCase("off")) {
700N/A env.setQuickContextScan(false);
700N/A } else {
697N/A System.err.println("ERROR: You should pass either \"on\" or \"off\" as argument to -Q");
700N/A System.err.println(" Ex: \"-Q on\" will just scan a \"chunk\" of the file and insert \"[..all..]\"");
700N/A System.err.println(" \"-Q off\" will try to build a more accurate list by reading the complete file.");
700N/A }
700N/A
700N/A break;
700N/A case 'm' : {
700N/A try {
700N/A env.setIndexWordLimit(Integer.parseInt(getopt.getOptarg()));
700N/A } catch (NumberFormatException exp) {
700N/A System.err.println("ERROR: Failed to parse argument to \"-m\": " + exp.getMessage());
700N/A System.exit(1);
700N/A }
700N/A break;
697N/A }
697N/A default:
System.err.println("Unknown option: " + (char)cmd);
System.exit(1);
}
}
int optind = getopt.getOptind();
if (optind != -1) {
if (optind < argv.length) {
env.setDataRoot(argv[optind]);
++optind;
}
while (optind < argv.length) {
subFiles.add(argv[optind]);
++optind;
}
}
if (env.getDataRootPath() == null) {
System.err.println("ERROR: Please specify a DATA ROOT path");
System.err.println(usage);
System.exit(1);
}
if (env.getSourceRootFile() == null) {
File srcConfig = new File(env.getDataRootPath(), "SRC_ROOT");
String line = null;
if(srcConfig.exists()) {
try {
BufferedReader sr = new BufferedReader(new FileReader(srcConfig));
line = sr.readLine();
sr.close();
} catch (IOException e) {
}
}
if(line == null) {
System.err.println("ERROR: please specify a SRC_ROOT with option -s !");
System.err.println(usage);
System.exit(1);
}
env.setSourceRoot(line);
if (!env.getSourceRootFile().isDirectory()) {
System.err.println("ERROR: No such directory:" + line);
System.err.println(usage);
System.exit(1);
}
}
if (!env.validateExuberantCtags()) {
System.exit(1);
}
if (searchRepositories) {
if (env.isVerbose()) {
System.out.println("Scanning for repositories...");
}
env.getRepositories().clear();
long start = System.currentTimeMillis();
HistoryGuru.getInstance().addExternalRepositories(env.getSourceRootPath());
long time = (System.currentTimeMillis() - start) / 1000;
if (env.isVerbose()) {
System.out.println("Done searching for repositories (" + time + "s)");
}
}
if (addProjects) {
File files[] = env.getSourceRootFile().listFiles();
List<Project> projects = env.getProjects();
projects.clear();
for (File file : files) {
if (!file.getName().startsWith(".") && file.isDirectory()) {
projects.add(new Project(file.getName(), "/" + file.getName()));
}
}
// The projects should be sorted...
Collections.sort(projects, new Comparator<Project>() {
public int compare(Project p1, Project p2){
String s1 = p1.getDescription();
String s2 = p2.getDescription();
int ret;
if (s1 == null) {
ret = (s2 == null) ? 0 : 1;
} else {
ret = s1.compareTo(s2);
}
return ret;
}
});
}
if (defaultProject != null) {
for (Project p : env.getProjects()) {
if (p.getPath().equals(defaultProject)) {
env.setDefaultProject(p);
break;
}
}
}
if (configFilename != null) {
if (env.isVerbose()) {
System.out.println("Writing configuration to " + configFilename);
System.out.flush();
}
env.writeConfiguration(new File(configFilename));
if (env.isVerbose()) {
System.out.println("Done...");
System.out.flush();
}
}
if (refreshHistory) {
for (Map.Entry<String, ExternalRepository> entry : RuntimeEnvironment.getInstance().getRepositories().entrySet()) {
try {
entry.getValue().createCache();
} catch (Exception e) {
System.err.println("Failed to generate history cache.");
e.printStackTrace();
}
}
}
if (runIndex) {
Index idx = new Index(env.isVerbose() ? new StandardPrinter(System.out) : new NullPrinter(), new StandardPrinter(System.err));
idx.runIndexer(env.getDataRootFile(), env.getSourceRootFile(), subFiles, !env.isGenerateHtml());
}
if (configHost != null) {
String[] cfg = configHost.split(":");
if (env.isVerbose()) {
System.out.println("Send configuration to: " + configHost);
}
if (cfg.length == 2) {
try {
InetAddress host = InetAddress.getByName(cfg[0]);
RuntimeEnvironment.getInstance().writeConfiguration(host, Integer.parseInt(cfg[1]));
} catch (Exception ex) {
System.err.println("Failed to send configuration to " + configHost);
ex.printStackTrace();
}
} else {
System.err.println("Syntax error: ");
for (String s : cfg) {
System.err.print("[" + s + "]");
}
System.err.println();
}
if (env.isVerbose()) {
System.out.println("Configuration successfully updated");
}
}
} catch (Exception e) {
System.err.println("Error: [ main ] " + e);
if (env.isVerbose()) e.printStackTrace();
System.exit(1);
}
}
}
}