static private void CopyDirectoryRecursively(DirectoryInfo source_directory,
DirectoryInfo target_directory)
{
if (!target_directory.Exists)
{
target_directory.Create();
}
foreach (FileInfo source_file in DirectoryWalker.GetFileInfos(source_directory))
{
FileInfo target_file = new FileInfo(Path.Combine(target_directory.FullName, source_file.Name));
// FIXME: Don't hard code filenames - Mono.Posix.StatMode.Regular
if (source_file.Name.IndexOf("socket") != -1 ||
source_file.Name.EndsWith("-journal"))
{
continue;
}
File.Copy(source_file.FullName, target_file.FullName, true);
}
foreach (DirectoryInfo source_child_directory in DirectoryWalker.GetDirectoryInfos(source_directory))
{
DirectoryInfo target_child_directory = new DirectoryInfo(Path.Combine(target_directory.FullName, source_child_directory.Name));
CopyDirectoryRecursively(source_child_directory,
target_child_directory);
}
}
/**
* Add watch to the parameter directory and its subdirs, recursively
*/
public void Watch(string path)
{
DirectoryInfo root = new DirectoryInfo(path);
if (!root.Exists)
{
return;
}
Queue queue = new Queue();
queue.Enqueue(root);
while (queue.Count > 0)
{
DirectoryInfo dir = queue.Dequeue() as DirectoryInfo;
if (!dir.Exists)
{
continue;
}
//log.Debug ("Adding inotify watch to " + dir.FullName);
Inotify.Subscribe(dir.FullName, OnInotifyEvent,
Inotify.EventType.Create
| Inotify.EventType.Delete
| Inotify.EventType.MovedFrom
| Inotify.EventType.MovedTo);
foreach (DirectoryInfo subdir in DirectoryWalker.GetDirectoryInfos(dir))
{
queue.Enqueue(subdir);
}
}
}
private void Crawl()
{
directory_enumerator = DirectoryWalker.GetDirectoryInfos(konq_cache_dir).GetEnumerator();
Scheduler.Task crawl_task = NewAddTask(this);
crawl_task.Tag = crawler_tag;
ThisScheduler.Add(crawl_task);
}
static void Main(string [] args)
{
Mode mode = Mode.Uris;
bool show_counts = true;
ArrayList index_dirs = new ArrayList();
ArrayList index_names = new ArrayList();
ArrayList uris = new ArrayList();
foreach (string arg in args)
{
switch (arg)
{
case "--help":
PrintUsage();
Environment.Exit(0);
break;
case "--uris":
mode = Mode.Uris;
break;
case "--properties":
case "--props":
mode = Mode.Properties;
break;
case "--term-frequencies":
case "--term-freqs":
mode = Mode.TermFrequencies;
break;
case "--hide-counts":
show_counts = false;
break;
case "--show-counts":
show_counts = false;
break;
case "--fields":
mode = Mode.Fields;
break;
default:
if (arg.StartsWith("--indexdir="))
{
index_dirs.Add(arg.Remove(0, 11));
}
else if (arg.StartsWith("--index="))
{
index_names.Add(arg.Remove(0, 8));
}
else
{
Uri uri;
try {
uri = UriFu.UserUritoEscapedUri(arg);
} catch (UriFormatException) {
uri = UriFu.PathToFileUri(arg);
}
uris.Add(uri);
}
break;
}
}
if (uris.Count > 0 && (mode == Mode.TermFrequencies || mode == Mode.Fields))
{
Console.WriteLine("ERROR: --term-frequencies and --fields do not make sense with files or URIs.");
Environment.Exit(1);
}
ArrayList indexes = new ArrayList();
// If no --index or --indexdir options, get all the default indexes.
if (index_dirs.Count == 0 && index_names.Count == 0)
{
foreach (DirectoryInfo subdir in DirectoryWalker.GetDirectoryInfos(PathFinder.IndexDir))
{
indexes.Add(new IndexInfo(subdir.Name));
}
}
else
{
foreach (string name in index_names)
{
DirectoryInfo info = new DirectoryInfo(Path.Combine(PathFinder.IndexDir, name));
if (!info.Exists)
{
Console.WriteLine("ERROR: No index named '{0}'", name);
Environment.Exit(1);
}
indexes.Add(new IndexInfo(info.Name));
}
foreach (string dir in index_dirs)
{
indexes.Add(new IndexInfo(dir));
}
}
indexes.Sort();
if (mode == Mode.Uris || mode == Mode.Properties)
{
DumpIndexInformation(indexes, uris, mode == Mode.Properties, show_counts);
}
else if (mode == Mode.TermFrequencies)
{
DumpIndexTermFreqs(indexes);
}
else if (mode == Mode.Fields)
{
DumpIndexFields(indexes);
}
}
static void Main(string [] args)
{
if (args.Length < 2)
{
PrintUsage();
}
string index_dir = (Path.IsPathRooted(args [0])) ? args [0] : Path.GetFullPath(args [0]);
if (!Directory.Exists(index_dir))
{
Console.WriteLine("Could not find index: {0}", index_dir);
Environment.Exit(1);
}
// Be *EXTRA PARANOID* about the contents of the target
// directory, because creating an indexing driver will
// nuke it.
if (args [1] != "info" && Directory.Exists(index_dir))
{
foreach (FileInfo info in DirectoryWalker.GetFileInfos(index_dir))
{
if (Array.IndexOf(allowed_files, info.Name) == -1)
{
Logger.Log.Error("{0} doesn't look safe to delete: non-Beagrep file {1} was found", index_dir, info.FullName);
Environment.Exit(1);
}
}
foreach (DirectoryInfo info in DirectoryWalker.GetDirectoryInfos(index_dir))
{
if (Array.IndexOf(allowed_dirs, info.Name) == -1)
{
Logger.Log.Error("{0} doesn't look safe to delete: non-Beagrep directory {1} was found", index_dir, info.FullName);
Environment.Exit(1);
}
}
}
switch (args [1])
{
#if false
case "list":
ExecuteList();
break;
case "remove":
ExecuteRemove(args [2]);
break;
#endif
case "info":
ExecuteInfo(index_dir);
break;
case "merge":
ExecuteMerge(index_dir, args [2]);
break;
case "optimize":
ExecuteOptimize(index_dir);
break;
default:
Console.WriteLine("Unknown command: {0}", args [1]);
PrintUsage();
break;
}
}
static void DoIndexing()
{
int count_dirs = 0;
int count_files = 0;
Indexable indexable;
pending_request = new IndexerRequest();
Queue modified_directories = new Queue();
while (pending_directories.Count > 0)
{
DirectoryInfo dir = (DirectoryInfo)pending_directories.Dequeue();
AddToRequest(DirectoryToIndexable(dir, modified_directories));
try {
if (arg_recursive)
{
foreach (DirectoryInfo subdir in DirectoryWalker.GetDirectoryInfos(dir))
{
if (!Ignore(subdir) &&
!FileSystem.IsSpecialFile(subdir.FullName))
{
pending_directories.Enqueue(subdir);
}
}
}
foreach (FileInfo file in DirectoryWalker.GetFileInfos(dir))
{
if (!Ignore(file))
{
AddToRequest(FileToIndexable(file));
count_files++;
}
}
} catch (DirectoryNotFoundException) {}
if (Shutdown.ShutdownRequested)
{
break;
}
count_dirs++;
}
Logger.Log.Debug("Scanned {0} files and directories in {1} directories", count_dirs + count_files, count_dirs);
if (Shutdown.ShutdownRequested)
{
backing_fa_store.Flush();
return;
}
// Time to remove deleted directories from the index and attributes store
while (modified_directories.Count > 0)
{
DirectoryInfo subdir = (DirectoryInfo)modified_directories.Dequeue();
Logger.Log.Debug("Checking {0} for deleted files and directories", subdir.FullName);
// Get a list of all documents from lucene index with ParentDirUriPropKey set as that of subdir
ICollection all_dirent = GetAllItemsInDirectory(subdir);
foreach (Dirent info in all_dirent)
{
// check if the item exists
if ((!info.IsDirectory && File.Exists(info.FullName)) ||
(info.IsDirectory && Directory.Exists(info.FullName)))
{
continue;
}
if (info.IsDirectory)
{
// Recursively remove deleted subdirectories
modified_directories.Enqueue(new DirectoryInfo(info.FullName));
}
// remove
Uri uri = PathToUri(info.FullName);
indexable = new Indexable(IndexableType.Remove, uri);
AddToRequest(indexable);
}
}
bool reschedule = false;
// Call Flush until our request is empty. We have to do this in a loop
// because Flush happens in a batch size and some indexables might generate more indexables
while (reschedule || pending_request.Count > 0)
{
if (Shutdown.ShutdownRequested)
{
break;
}
reschedule = FlushIndexer(driver);
}
backing_fa_store.Flush();
if (Shutdown.ShutdownRequested)
{
return;
}
Logger.Log.Debug("Optimizing index");
driver.OptimizeNow();
}
static void DoMain(string [] args)
{
SystemInformation.SetProcessName("beagrep-build-index");
if (args.Length < 2)
{
PrintUsage();
}
ArrayList allowed_patterns = new ArrayList();
ArrayList denied_patterns = new ArrayList();
ArrayList denied_dir_patterns = new ArrayList();
int i = 0;
while (i < args.Length)
{
string arg = args [i];
++i;
string next_arg = i < args.Length ? args [i] : null;
switch (arg)
{
case "-h":
case "--help":
PrintUsage();
break;
case "--tag":
if (next_arg != null)
{
arg_tag = next_arg;
}
++i;
break;
case "-r":
case "--recursive":
arg_recursive = true;
break;
case "--enable-deletion":
arg_delete = true;
break;
case "--disable-directories":
arg_disable_directories = true;
break;
case "--enable-text-cache":
arg_cache_text = true;
break;
case "--target":
if (next_arg != null)
{
arg_output = Path.IsPathRooted(next_arg) ? next_arg : Path.GetFullPath(next_arg);
}
++i;
break;
case "--disable-filtering":
arg_disable_filtering = true;
break;
case "--allow-pattern":
if (next_arg == null)
{
break;
}
if (next_arg.IndexOf(',') != -1)
{
foreach (string pattern in next_arg.Split(','))
{
allowed_patterns.Add(pattern);
}
}
else
{
allowed_patterns.Add(next_arg);
}
++i;
break;
case "--deny-directory-pattern":
if (next_arg == null)
{
break;
}
if (next_arg.IndexOf(',') != -1)
{
foreach (string pattern in next_arg.Split(','))
{
denied_dir_patterns.Add(pattern);
}
}
else
{
denied_dir_patterns.Add(next_arg);
}
++i;
break;
case "--deny-pattern":
if (next_arg == null)
{
break;
}
if (next_arg.IndexOf(',') != -1)
{
foreach (string pattern in next_arg.Split(','))
{
denied_patterns.Add(pattern);
}
}
else
{
denied_patterns.Add(next_arg);
}
++i;
break;
case "--disable-restart":
arg_disable_restart = true;
break;
case "--source":
if (next_arg == null)
{
break;
}
arg_source = next_arg;
++i;
break;
default:
if (arg.StartsWith("-") || arg.StartsWith("--"))
{
PrintUsage();
}
string path = Path.IsPathRooted(arg) ? arg : Path.GetFullPath(arg);
if (path != "/" && path.EndsWith("/"))
{
path = path.TrimEnd('/');
}
if (Directory.Exists(path))
{
pending_directories.Enqueue(new DirectoryInfo(path));
}
else if (File.Exists(path))
{
pending_files.Enqueue(new FileInfo(path));
}
break;
}
}
/////////////////////////////////////////////////////////
if (arg_output == null)
{
Logger.Log.Error("--target must be specified");
Environment.Exit(1);
}
// Set the storage dir, this should be used to store log messages
// and filterver.dat
PathFinder.StorageDir = arg_output;
foreach (FileSystemInfo info in pending_directories)
{
if (Path.GetFullPath(arg_output) == info.FullName)
{
Logger.Log.Error("Target directory cannot be one of the source paths.");
Environment.Exit(1);
}
}
foreach (FileSystemInfo info in pending_files)
{
if (Path.GetFullPath(arg_output) == info.FullName)
{
Logger.Log.Error("Target directory cannot be one of the source paths.");
Environment.Exit(1);
}
}
if (!Directory.Exists(Path.GetDirectoryName(arg_output)))
{
Logger.Log.Error("Index directory not available for construction: {0}", arg_output);
Environment.Exit(1);
}
// Be *EXTRA PARANOID* about the contents of the target
// directory, because creating an indexing driver will
// nuke it.
if (Directory.Exists(arg_output))
{
foreach (FileInfo info in DirectoryWalker.GetFileInfos(arg_output))
{
if (Array.IndexOf(allowed_files, info.Name) == -1)
{
Logger.Log.Error("{0} doesn't look safe to delete: non-Beagrep file {1} was found", arg_output, info.FullName);
Environment.Exit(1);
}
}
foreach (DirectoryInfo info in DirectoryWalker.GetDirectoryInfos(arg_output))
{
if (Array.IndexOf(allowed_dirs, info.Name) == -1)
{
Logger.Log.Error("{0} doesn't look safe to delete: non-Beagrep directory {1} was found", arg_output, info.FullName);
Environment.Exit(1);
}
}
}
string config_file_path = Path.Combine(arg_output, "StaticIndex.xml");
string prev_source = null;
if (File.Exists(config_file_path))
{
Config static_index_config = Conf.LoadFrom(config_file_path);
if (static_index_config == null)
{
Log.Error("Invalid configuation file {0}", config_file_path);
Environment.Exit(1);
}
prev_source = static_index_config.GetOption("Source", null);
if (arg_source != null && prev_source != arg_source)
{
Log.Error("Source already set to {0} for existing static index. Cannot set source to {1}.", prev_source, arg_source);
Environment.Exit(1);
}
// If arg_source is not given, and prev_source is present, use prev_source
// as the arg_source. This is useful for re-running build-index without
// giving --arg_source for already existing static index
arg_source = prev_source;
}
if (arg_source == null)
{
DirectoryInfo dir = new DirectoryInfo(StringFu.SanitizePath(arg_output));
arg_source = dir.Name;
}
string global_files_config = Path.Combine(PathFinder.ConfigDataDir, "config-files");
global_files_config = Path.Combine(global_files_config, Conf.Names.FilesQueryableConfig + ".xml");
if (!File.Exists(global_files_config))
{
Log.Error("Global configuration file not found {0}", global_files_config);
Environment.Exit(0);
}
// Setup regexes for allowed/denied patterns
if (allowed_patterns.Count > 0)
{
allowed_regex = StringFu.GetPatternRegex(allowed_patterns);
}
else
{
// Read the exclude values from config
// For system-wide indexes, only the global config value will be used
Config config = Conf.Get(Conf.Names.FilesQueryableConfig);
List <string[]> values = config.GetListOptionValues(Conf.Names.ExcludePattern);
if (values != null)
{
foreach (string[] exclude in values)
{
denied_patterns.Add(exclude [0]);
}
}
if (denied_patterns.Count > 0)
{
denied_regex = StringFu.GetPatternRegex(denied_patterns);
}
}
if (denied_dir_patterns.Count > 0)
{
denied_dir_regex = StringFu.GetPatternRegex(denied_dir_patterns);
Log.Always("Will ignore directories matching regular expression: {0}", denied_dir_regex);
}
Log.Always("Starting beagrep-build-index (pid {0}) at {1}", Process.GetCurrentProcess().Id, DateTime.Now);
driver = new LuceneIndexingDriver(arg_output, MINOR_VERSION, false);
driver.TextCache = (arg_cache_text) ? new TextCache(arg_output) : null;
if (driver.TextCache != null)
{
driver.TextCache.WorldReadable = true;
}
backing_fa_store = new FileAttributesStore_Sqlite(driver.TopDirectory, driver.Fingerprint);
fa_store = new FileAttributesStore(backing_fa_store);
// Set up signal handlers
#if MONO_1_9
Shutdown.SetupSignalHandlers(delegate(int signal)
{
if (signal == (int)Mono.Unix.Native.Signum.SIGINT ||
signal == (int)Mono.Unix.Native.Signum.SIGTERM)
{
Shutdown.BeginShutdown();
}
});
#else
SetupSignalHandlers();
#endif
Thread monitor_thread = null;
Stopwatch watch = new Stopwatch();
watch.Start();
if (!arg_disable_restart)
{
// Start the thread that monitors memory usage.
monitor_thread = ExceptionHandlingThread.Start(new ThreadStart(MemoryMonitorWorker));
}
// Start indexworker to do the crawling and indexing
IndexWorker();
// Join any threads so that we know that we're the only thread still running
if (monitor_thread != null)
{
monitor_thread.Join();
}
watch.Stop();
Logger.Log.Debug("Elapsed time {0}.", watch);
// Write this after indexing is done. This is because, if creating a new index,
// LuceneIndexingDriver.Create() is called which purges the entire directory.
if (prev_source == null)
{
Config static_index_config = Conf.LoadNew("StaticIndex.xml");
// Write StaticIndex.xml containing:
// The name of the source
static_index_config.SetOption("Source", arg_source);
static_index_config ["Source"].Description = "Source of the static index";
Conf.SaveTo(static_index_config, config_file_path);
}
if (restart)
{
Logger.Log.Debug("Restarting beagrep-build-index");
Process p = new Process();
p.StartInfo.UseShellExecute = false;
// FIXME: Maybe this isn't the right way to do things? It should be ok,
// the PATH is inherited from the shell script which runs mono itself.
p.StartInfo.FileName = "mono";
p.StartInfo.Arguments = String.Join(" ", Environment.GetCommandLineArgs());
p.Start();
}
Log.Always("Exiting beagrep-build-index (pid {0}) at {1}", Process.GetCurrentProcess().Id, DateTime.Now);
}