Table of Contents 1 Introduction 2 Use cases 3 Features 4 Dependencies 4.1 Recommended dependencies 4.2 For building Tracker's Deskbar-applet backend 5 Compilation 5.1 Notes on Solaris 5.2 Compile options 6 Running Tracker 6.1 Usage 6.2 Setting inotify watch limit 6.3 Tracker files 7 Integration with other tools 7.1 Nautilus search 7.2 Deskbar applet 8 Tracker tools 1 Introduction Tracker is a powerful desktop-neutral first class object database, tag/metadata database, search tool and indexer. Tracker is also extremely fast and very memory-efficient when compared with some other competing frameworks and has by far the fastest and most memory-efficient Nautilus search and Deskbar backends currently available. It consists of a common object database that allows entities to have an almost infinte number of properties, metadata (both embedded/harvested as well as user definable), a comprehensive database of keywords/tags and links to other entities. It provides additional features for file-based objects including context linking and audit trails for a file object. It has the ability to index, store, harvest metadata, retrieve and search all types of files and other first class objects. Supported first class objects include: * Files, Documents, Music, Images, Videos, Applications, Emails, Conversations, Playlists Planned support: * Appointments, Contacts, Projects, Tasks, Bookmarks, Notes, Firefox Web History All discussion related to tracker happens on the Tracker mailing list http://mail.gnome.org/mailman/listinfo/tracker-list IRC channel #tracker on: irc.gimp.net Bugs should be filed at: http://bugzilla.gnome.org. More infomation on Tracker can be found at http://tracker-project.org. Repository (first is readonly, second is write access) git://git.gnome.org/tracker ssh://@git.gnome.org/git/tracker 2 Use Cases Tracker is the most powerful open source metadata database and indexer framework currently available and because it is built around a combination indexer and sql database and not a dedicated indexer, it has much more powerful use cases: * Provide search and indexing facilities similiar to those on other systems (Windows Vista and Mac OS X). * Common database storage for all first class objects (e.g. a common music/photo/contacts/email/bookmarks/history database) complete with additional metadata and tags/keywords. * Comprehensive one stop solution for all applications needing an object database, powerful search (via RDF Query), first class methods, related metadata and user-definable metadata/tags. * Can provide a full semantic desktop with metadata everywhere. * Can provide powerful criteria-based searching suitable for creating smart file dialogs and vfolder systems. * Can provide a more intelligent desktop using statistical metadata. 3 Features * Desktop-neutral design (it's a freedesktop product built around other freedesktop technologies like D-Bus and XDGMime but contains no GNOME-specific dependencies besides GLib). * Very memory efficient and non-leaking (typical RAM usage 4 - 30 MB). Unlike some other indexers, Tracker is designed and built to run well on systems with lower memory (256MB or less). It should even be efficient enough to use on some mobile devices. * Non-bloated and written in C for maximum efficiency. * Small size and minimal dependencies makes it easy to bundle into various distros, including live cds. * Fast indexing and unobtrusive - no need to index stuff overnight. Tracker runs at nice+10 so it should have a minimal impact on your system. With the addition of detection of mouse and keyboard events via tracker-applet (described below), there is an option to auto-pause indexing in order to improve responsiveness. This is in addition to Tracker's built-in check if there's heavy disk I/O in order to auto-pause, so not to slow other processes. * Provides option to disable indexing when running on battery. * Provides option to index removable devices. * Implements the freedesktop specification for metadata (http://freedesktop.org/wiki/Standards/shared-filemetadata-spec). * Extracts embedded File, Image, Document and Audio type metadata from files. * Supports the WC3's RDF Query syntax for querying metadata * Provides support for both free text search (like Beagle/Google) as well as structured searches using RDF Query. * Responds in real time to file system changes to keep its metadata database up to date and in sync. * Fully extensible with custom metadata - you can store, retrieve, register and search via RDF Query all your own custom metadata. * Can extract a file's contents as plain text and index them. * Provides text filters for PDF, MS Office, OpenOffice (all versions), HTML and PS files. * Can provide thumbnailing on the fly. * It auto-pauses indexing when running low on diskspace. 4 Dependencies 4.1 Recommended dependencies * GStreamer 0.10 + plugins for audio/video file indexing * xsltproc * w3m * wv 1.0.2 * poppler (pdftotext) * odt2txt 0.4 (indexing OpenOffice/ODF documents) * libvorbis * libpng * libexif * libgsf * libglade 2.5 * libxml2 * libxml2 (for extracting html/xml content) * unac (accent stripper) * exempi * hal 0.5 (for detection of removable devices, mounted directories, as well as whether the computer is running on battery) * GTK and GNOME stack (for GUI tools) * totem-plparser (for playlist extraction) 4.2 For building Tracker's Deskbar-applet backend * python-dev 2.3 * python-gtk2-dev 2.3 * deskbar-applet 2.16 5 Compilation To compile and install Tracker, use the following commands : ./configure --prefix=/usr --sysconfdir=/etc --localstatedir=/var make sudo make install If you install using any other prefix, you might have problems with files not being installed correctly. (You may need to copy and amend the dbus service file to the correct directory and/or might need to update ld_conf if you install into non-standard directories.) 5.1 Notes on Solaris To compile Tracker with GCC on Solaris uses the following commands : CFLAGS="-D_POSIX_PTHREAD_SEMANTICS" ./configure \ --prefix=/usr \ --sysconfdir=/etc \ --localstatedir=/var \ --with-pic make sudo make install To compile Tracker with SUN Studio on Solaris uses the following commands, because there are some problems to compile exiv2 using SUN C++ compiler : CFLAGS="-D_POSIX_PTHREAD_SEMANTICS" ./configure \ --prefix=/usr \ --sysconfdir=/etc \ --localstatedir=/var \ --with-pic \ --disable-exiv2 make sudo make install 5.2 Compile Options Tracker has several compiler options to enable/disable certain features. You can get a full listing by running ./configure --help 6 Running Tracker 6.1 Usage To run Tracker, you can manually start the Tracker daemon, trackerd. This is run for you if you issue a DBus call to the daemon API. You can also pass a directory root to be indexed as a command line parameter if you dont want your entire home directory indexed, e.g. "trackerd -i /home/jamie/Documents" (if you want your home directory indexed when explicily specifying index directory roots then you must add your home directory to the arguments: e.g. trackerd -i /home/jamie -i /mnt/share) You can disable indexing by passing "--no-indexing" You can enable a low memory usage mode (recommended for machines with less than 256MB of RAM) by passing "--enable-low-memory" You can artificially throttle indexing by passing "--throttle=VALUE" where VALUE is in the range 0-20 (with 0, the default, being fastest and 20 being slowest). Default is 0. You should only change this value if you want to prevent noisy fans or hot laptops arising from cpu intensive indexing. Tracker should have a negligible impact on the system (as it is scheduled) so you can safely work with it on full throttle without experiencing slow downs. You can specify directory roots to be excluded from being watched or indexed by passing "--exclude=DIRECTORY" for each directory root. You can specify logging verbosity by passing "--verbosity". Valid values are from 0 to 3, ranging from least to most verbose respectively. Yet another option is "--language" which allows for specifying the language to use for stemmer and stop-words list. All the above options (and more) can be set by editing Tracker config file "~/.config/tracker/tracker.cfg" which is created with specific defaults when non-existent (e.g. when trackerd is ran for the first time). Ensure that you restart trackerd for the changes to take effect. "tracker.cfg" also provides options that allows Tracker to only index a subset of your home directory as well as other folders not in your home directory by setting WatchDirectoryRoots to a semicolon-delimited list of directories (full path required!) e.g.: "WatchDirectoryRoots=directory1;directory2;directory3"" An additional option is "--reindex" which indexes user data from scratch, removing the need to delete Tracker's database manually. Keywords and metadata definitions are preserved however. On the first run, Tracker will automatically create a new database and start populating it with metadata by browsing through the user's home directory and/or the root folder(s) specified. On subsequent runs, Tracker will start up much much faster and will only ever incrementally index files (IE files that have changed since last index). If installed correctly, the Tracker daemon (trackerd) can also be started automatically via Dbus activation (e.g. by running tracker-search SEARCHTERM) 6.2 Setting Inotify Watch Limit When watching large numbers of folders, its ppossible to exceed the default number of inotify watches. In order to get real time updates when this value is exceeded it is necessary to increase the number of allowed watches. This can be done as follows: 1. Add this line to /etc/sysctl.conf: "fs.inotify.max_user_watches = (number of folders to be watched; default used to be 8192 and now is 524288)" 2. Reboot the system OR (on a Debian-like system) run "sudo /etc/init.d/procps restart" 6.3 Tracker files Here are some of the files that Tracker creates and uses for its operation, apart from "~/.config/tracker/tracker.cfg" which is stated in Sec 6.1 above: * "~/.local/share/tracker" is used for non-expendable content, like keywords and metadata definitions. * "~/.cache/tracker" is used for the expendable indexes and expendable metadata that can be rebuilt if deleted (this is the purpose of the ".cache" - its more a permanent tmp directory than sys tmp but can be deleted if more disk space is needed). * The system tmp ("/var/tmp" and "/tmp") is used for short-lived session data. 7 Integration with other tools 7.1 Nautilus Search Once you have installed Tracker and have some indexed contents, you should now compile Nautilus (ver 2.13.4 or higher) which should auto-detect that Tracker is installed and automatically compile in Tracker support. You are now ready to appreciate a powerful and super efficient C-based indexer in all its glory... happy hunting! To make sure trackerd always starts when you login to GNOME, you will need to add it to gnome-session (select sessions from preferences menu, select startup program tab and then add /usr/bin/trackerd). For non-GNOME installations, see the desktop docs for how to achieve similar. 7.2 Deskbar applet Tracker is also integrated into GNOME's deskbar applet. Here are the compile options on how to get it built: --enable-deskbar-applet=(auto/handler/module) Enables Deskbar-applet support; 'auto' should do since it automatically chooses whether to install the 'handler' (for Deskbar-applet >=2.16) or the 'module' (for Deskbar-applet >=2.19) --with-deskbar-applet-dir=(directory) This sets where Deskbar-applet should find the tracker-handler; this should be automatically detected, perhaps in /usr/lib/deskbar-applet/{handlers,modules-2.20-compatible} 8 Tracker Tools The Tracker distribution comes with a number of useful utilities which are listed at the bottom of trackerd man page, "man trackerd". It's recommended that you use the "--help" command line switch (e.g., "tracker-services --help") for more up-to-date usage information.