TMSU is an application that allows you to organise your files by associating them with tags. It provides a tool for managing these tags and a virtual file- system to allow tag based access to your files.

TMSU's virtual file system does not store your files: it merely provides an alternative, tagged based view of your files stored elsewhere in the file- system. That way you have the freedom to choose the most suitable file-system for storage whilst still benefiting from tag based access.

A command overview and details on how to use each command are available via the integrated help:

$ tmsu help

Full documentation is maintained online on the wiki:

• http://bitbucket.org/oniony/tmsu/wiki

Binary builds for a limited number of architectures and operating system combinations are available:

• http://bitbucket.org/oniony/tmsu/downloads

You will need to ensure that both FUSE and Sqlite3 are installed for the program to function.

The following steps are for compiling from source.

1. Installing Go

   TMSU is written in the Go programming language. To compile from source you must first install Go and the packages that TMSU depends upon. You can get that from the Go website:

   • http://www.golang.org/

   Go can be installed per the instructions on the Go website or it may be available in the package management system that comes with your operating system.

   TMSU is currently built against the Go weekly builds. See VERSIONS for the latest version that is known to work with the dependent packages.

2. Install the dependent packages.

    $ go get github.com/mattn/go-sqlite3
    $ go get github.com/hanwen/go-fuse/fuse
   

3. Clone the TMSU respository:

    $ hg clone https://bitbucket.org/oniony/tmsu
   

4. Make the project

    $ cd tmsu
    $ make
   

   This will compile to 'bin/tmsu' within the working directory.

5. Install the project

    $ sudo make install
   

   This will install TMSU to '/usr/bin/tmsu'.

   It will also install the Zsh completion to '/usr/share/zsh/site-functions'.

   To change the paths used override the environment variables in the Makefile.

• Website: http://www.tmsu.org/
• Project: http://bitbucket.org/oniony/tmsu
• Wiki: http://bitbucket.org/oniony/tmsu/wiki
• Mailing list: http://groups.google.com/group/tmsu

TMSU is written in Go: http://www.golang.org/

TMSU itself is written and maintained by Paul Ruane paul@tmsu.org, however much of the functionality it provides is made possible by the FUSE and Sqlite3 libraries, their Go bindings and, of course, the Go language standard library.

• Feature: tag values.

• Added support for tag implications, e.g. tag 'a' implies 'b'. New 'imply' command for managing these.
• Added --force option to 'repair' command to remove missing files (and associated taggings) from the database.
• Added --from option to 'tag' command to allow tags to copied from one file to another. e.g. 'tmsu tag -f a b' will apply file b's tags to file a. ('tag -r -f a a' will recursively retag a directory's contents.)
• Added --directory option to 'status' command to stop it recursively processing directory contents.
• Added --print0 option to 'files' command to allow use with xargs.
• Added --count option to 'tags' and 'files' command to list tag/file count rather than names.
• Bug fixes and unit-test improvements.

• Removed use of Sqlite bulk insert functionality as the version of Sqlite in mainstream Linux distributions does not have this functionality.

• Fixed panic when using the 'rename' command with the incorrect number of arguments.
• 'status' command now handles case where file is replaced with directory or vice-versa.
• Fixed bug in 'status' command where mixing of relative and absolute files was missing some moved files.

This version aims to improve the performance of the program by removing auto- matic tag inheritence which has proven slow on larger databases and slow file- systems (e.g. network file-systems). Instead there is now a choice of how to handle directory contents:

• Add the nested files to the database using the --recursive option on the 'tag' and 'untag' commands.
• Dynamically discover directory contents using the --recursive option on 'files' command.

Adding directory contents using the --recursive option on the 'tag' command will be slower and will result in a larger database. However querying of the database will be faster as TMSU will not need to examine the file-system. This option also means that duplicate files will be identified.

Dynamically discovering directory conents using the --recursive option on the 'files' command will be slower as TMSU will need to scan the file-system and duplicate files will not be identified as these files are not added to the database. However tagging files will be faster and the resultant database will be smaller.

IMPORTANT: Please back up your database then upgrade it using the upgrade script. The 'repair' step may take a while to run as every file is reexamined to populate the new columns.

IMPORTANT: If you have been following 'tip' there is a separate upgrade script called tip_to_0.1.0.sql. Due to a bug in the previous version please re-run this even if you have previously upgraded otherwise you may end up with duplicate entries in the file_tag table. Please read through the script before running it.

$ cp ~/.tmsu/default.db ~/.tmsu/default.db.bak  # back up
$ sqlite3 -init misc/db-upgrades/0.0.9_to_0.1.0.sql ~/.tmsu/default.db
$ tmsu repair

• 'files' command no longer finds files that inherit a tag from the file- system on-the-fly unless run with the new --recursive (-r) option. New --directory (-d) and --file (-f) options to limit output to just files or directories. In addition the new --top (-t) and --leaf (-l) options show the top-most matching entries (excluding items from matching directories) or bottom-most matching entries (excluding parent directories) respectevly.
• 'tag' and 'untag' now sport a --recursive (-r) option for tagging or untagging directory contents.
• Improved command-line parsing: now supports global options, short options and mixed option ordering. Options with arguments can now be specified any- where on the command line.
• 'repair' command rewritten to fix bugs.
• Fingerprints for directories no longer calculated. (Recursively tag files instead to detect file duplicates.)
• Removed the 'export' command. (Sqlite tooling has better facilities.)
• Tags containing '/' are no longer legal.
• Fixed bug that prevented file lists from being shown sorted alphanumerically.
• The 'tag' command no longer identifies modified files. (Use 'repair' instead.)
• The 'mount' command now has a '--allow-other' option which allows other users to access the mounted file-system.
• Updated Zsh completion.
• Improved error messages.
• Improved unit-test coverage.
• Minor bug fixes.

• Fixed bug which caused process hosting the virtual file-system to crash if a non-existant tag directory is 'stat'ed.
• Untagged files now inherit parent directory tags.

Files can now be tagged within tagged directories. Files within tagged directories will inherit the directory's tags.

• Fixed bug with 'untag' command when non-existant tag is specified.
• Updated with respect to go-fuse API change.
• 'mount' command now lists mount points if invoked without arguments.
• Improved 'mount' command help.
• 'rename' command now validates destination tag name.
• Fixed bug with 'unmount --all' returning an error if there are no mounts.
• Removed dependency upon 'pgrep'; now accesses proc-fs directotly for mount information.
• 'files' command will now show files that inherit the specified tags ('--explicit' option turns this off).
• 'tags' command will now shows inherited tags ('--explicit' turns this off).
• 'status' command now reports inherited tags.
• 'stats' command formatting updated.
• Other minor bug fixes.

Files larger than 5MB now use a different fingerprinting algorithm where the fingerprint is produced by taking a 500KB of the start, middle and end of the file. This should dramatically improve performance, especially on slow file- systems.

NOTE: it is advisable to run 'repair' after upgrading to this version to update large files with fingerprints produced using the new algorithm.

IMPORTANT: This version adds a column to one of the database tables to record the file's modification timestamep. The new code will not work with an existing TMSU database until it has been upgraded.

To upgrade the database, run the upgrade script using the Sqlite3 tooling:

$ sqlite3 -init misc/db-upgrade/0.0.5_to_0.0.6.sql

It is also advisable to run 'repair' after upgrading which will populate the new column and also fix the directory fingerprints which have a new algorithm.

• Upgraded to Go 1.
• Added 'repair' command to fix up database when files are moved or modified.
• Added 'version' command.
• Added 'copy' command (duplicates a tag).
• Added modification timestamp which is used in preference to fingerprint.
• Command output now includes 'tmsu' to make it clear where output is coming from when piping.
• Relative paths now calculated more accurately.
• Zsh completion now supports tags containing colons.
• 'status' command performance and functionality improvements.
• Added directory fingerprinting.
• Added 'repair' command to Zsh completion.
• The 'files' command now allows tags to be excluded by prefixing them with a minus, e.g. -jazz.

Copyright 2011 Paul Ruane

Copying and distribution of this file, with or without modification, are permitted in any medium without royalty provided the copyright notice and this notice are preserved. This file is offered as-is, without any warranty.