Skip to content
Switch branches/tags
Go to file



… filetags example demonstrating: controlled vocabulary file ~/.filetags, tagging multiple files at once, removing tags by prepending a minus character, tagging using the proposed number shortcuts, tab completion of tags via Tab, and mutually exclusive tags (switching from draft to final without removing draft).

This Python script adds or removes tags to file names in the following form:

  • “file without time stamp in name – tag2.txt
  • “file name with several tags – tag1 tag2.jpeg
  • “another example file name with multiple example tags – fun videos kids.mpeg
  • “2013-05-09 a file name with ISO date stamp in name – tag1.jpg
  • “2013-05-09T16.17 file name with time stamp – tag3.csv

The script accepts an arbitrary number of files (see your shell for possible length limitations).


Besides the fact that I am using ISO dates and times in file names (as shown in examples above), I am using tags with file names. To separate tags from the file name, I am using the separator “space dash dash space”.

For people familiar with Regular Expressions:

(<ISO date/time stamp>)?(?<descriptive file name>)?( -- <list of tags separated by spaces>)?.<file extension>

Tagging files this way requires a file renaming process. Adding (or removing) tag(s) to a set of file results in multiple renaming processes. Despite advanced renaming tools like vidir (from moreutils) it’s handy to have a tool that makes adding and removing tags as simple as possible.

You may like to add this tool to your image or file manager of choice. I added mine to geeqie which is my favorite image viewer on GNU/Linux.

Here is a 45 minute talk I gave at Linuxtage Graz 2018 presenting the idea of and workflows related to filetags and other handy tools for file management:

bin/2018-05-06 filetags demo slide for video preview with video button -- screenshots.png


This tool needs Python 3 to be installed.

You can install filetags either via pip which is the recommended way. Or you can install filetags using the source code, e.g., by cloning the GitHub repository of filetags.

Installation Via Pip

If you have installed Python 2 and Python 3 in parallel, make sure to use the correct pip version. You might need to use pip3 instead of pip. If you only have Python 3 installed, you don’t have to care ;-)

On Microsoft Windows (only), you are going to need pip install pypiwin32 as prerequisite. For easy Windows File Explorer integration, take a look at integratethis.

Now install filetags via pip: pip install filetags

You get updates by executing the very same pip command again.

Installation Via Source Code

If you use the GitHub sources (and not pip):

  • You need to install depending packages via: pip install -r requirements.txt
  • The executable is filetags/ You might want to create a symbolic link named “filetags” to that file.


usage: ./filetags/ [-h] [-t "STRING WITH TAGS"] [--remove] [-i]
                              [-R] [-s] [--hardlinks] [-f]
                              [--filebrowser PATH_TO_FILEBROWSER] [--tagtrees]
                              [--tagtrees-handle-no-tag "treeroot" | "ignore" | "FOLDERNAME"]
                              [--tagtrees-dir <existing_directory>]
                              [--tagtrees-depth TAGTREES_DEPTH] [--ln] [--la]
                              [--lu] [--tag-gardening] [-v] [-q] [--version]
                              [FILE [FILE ...]]

This tool adds or removes simple tags to/from file names.

Tags within file names are placed between the actual file name and
the file extension, separated with " -- ". Multiple tags are
separated with " ":
  Update for the Boss -- projectA presentation.pptx
  2013-05-16T15.31.42 Error message -- screenshot projectB.png

This easy to use tag system has a drawback: for tagging a larger
set of files with the same tag, you have to rename each file
separately. With this tool, this only requires one step.

Example usages:
  filetags --tags="presentation projectA" *.pptx
      … adds the tags "presentation" and "projectA" to all PPTX-files
  filetags --tags="presentation -projectA" *.pptx
      … adds the tag "presentation" to and removes tag "projectA" from all PPTX-files
  filetags -i *
      … ask for tag(s) and add them to all files in current folder
  filetags -r draft *report*
      … removes the tag "draft" from all files containing the word "report"

This tools is looking for the optional first text file named ".filetags" in
current and parent directories. Each of its lines is interpreted as a tag
for tag completion. Multiple tags per line are considered mutual exclusive.

Verbose description:

positional arguments:
  FILE                  One or more files to tag

optional arguments:
  -h, --help            show this help message and exit
                        One or more tags (in quotes, separated by spaces) to
  --remove              Remove tags from (instead of adding to) file name(s)
  -i, --interactive     Interactive mode: ask for (a)dding or (r)emoving and
                        name of tag(s)
  -R, --recursive       Recursively go through the current directory and all
                        of its subdirectories. Implemented for --tag-gardening
                        and --tagtrees
  -s, --dryrun          Enable dryrun mode: just simulate what would happen,
                        do not modify files
  --hardlinks           Use hard links instead of symbolic links. This is
                        ignored on Windows systems. Note that renaming link
                        originals when tagging does not work with hardlinks.
  -f, --filter          Ask for list of tags and generate links in
                        "$HOME/.filetags_tagfilter" containing links to all
                        files with matching tags and start the filebrowser.
                        Target directory can be overridden by --tagtrees-dir.
  --filebrowser PATH_TO_FILEBROWSER
                        Use this option to override the tool to view/manage
                        files (for --filter; default: geeqie). Use "none" to
                        omit the default one.
  --tagtrees            This generates nested directories in
                        "$HOME/.filetags_tagfilter" for each combination of
                        tags up to a limit of 2. Target directory can be
                        overridden by --tagtrees-dir. Please note that this
                        may take long since it relates exponentially to the
                        number of tags involved. Can be combined with
                        --filter. See also and
  --tagtrees-handle-no-tag "treeroot" | "ignore" | "FOLDERNAME"
                        When tagtrees are created, this parameter defines how
                        to handle items that got no tag at all. The value
                        "treeroot" is the default behavior: items without a
                        tag are linked to the tagtrees root. The value
                        "ignore" will not link any non-tagged items at all.
                        Any other value is interpreted as a folder name within
                        the tagreees which is used to link all non-tagged
                        items to.
                        When the controlled vocabulary holds mutual exclusive
                        tags (multiple tags in one line) this option generates
                        directories in the tagtrees root that hold links to
                        items that have no single tag from those mutual
                        exclusive sets. For example, when "draft final" is
                        defined in the vocabulary, all items without "draft"
                        and "final" are linked to the "no-draft-final"
  --tagtrees-dir <existing_directory>
                        When tagtrees are created, this parameter overrides
                        the default target directory
                        "$HOME/.filetags_tagfilter" with a user-defined
                        one. It has to be an empty directory or a non-existing
                        directory which will be created. This also overrides
                        the default directory for --filter.
  --tagtrees-depth TAGTREES_DEPTH
                        When tagtrees are created, this parameter defines the
                        level of depth of the tagtree hierarchy. The default
                        value is 2. Please note that increasing the depth
                        increases the number of links exponentially.
                        Especially when running Windows (using lnk-files
                        instead of symbolic links) the performance is really
                        slow. Choose wisely.
  --ln, --list-tags-by-number
                        List all file-tags sorted by their number of use
  --la, --list-tags-by-alphabet
                        List all file-tags sorted by their name
  --lu, --list-tags-unknown-to-vocabulary
                        List all file-tags which are found in file names but
                        are not part of .filetags
  --tag-gardening       This is for getting an overview on tags that might
                        require to be renamed (typos, singular/plural, ...).
                        See also
  -v, --verbose         Enable verbose mode
  -q, --quiet           Enable quiet mode
  --version             Display version and exit

:copyright: (c) by Karl Voit <>
:license: GPL v3 or any later version
:bugreports: via github or <>
:version: 2018-08-02


filetags --tags foo a_file_name.txt

… adds tag “foo” such that it results in a_file_name -- foo.txt

filetags -i *.jpeg

… interactive mode: asking for list of tags (for the JPEG files) from the user

filetags --tags "foo bar" "file name 1.jpg" "file name 2 -- foo.txt" "file name 3 -- bar.csv"

… adds tag “foo” such that it results in …

"file name 1 -- foo bar.jpg"
"file name 2 -- foo bar.txt"
"file name 3 -- bar foo.csv"
filetags --remove --tags foo "foo a_file_name -- foo.txt"

… removes tag “foo” such that it results in foo a_file_name.txt

filetags --tag-gardening

… prints out a summary of tags in current and sub-folders used and tags that are most likely typos or abandoned

For --filter and --tagtrees examples see sections below.


  • 2013-05-16: first version on GitHub
  • 2014-12-21: --list-tags-by-number, --list-tags-by-alphabet, and --tag-gardening
  • 2015-01-02: tab completion for interactive tag input
    • Example: entering myt + pressing TAB completes the entered string to mytag if mytag is found in the vocabulary or existing file tags
  • 2015-12-11: shortcut numbers for removing tags
  • 2016-01-08: shortcut numbers for top nine tags for adding tags
    • Example: when filetags shows you Top nine previously used tags in this directory: with mytag(1) anothertag(2) oncemore(3), you don’t have to type in the tag names but use the numbers instead. Combinations of numbers are fine as well.
  • 2016-08-21: mutually exclusive tags: see chapter below
  • 2016-08-23: installable via pip install filetags
  • 2016-08-26: --filter option requires all tags to be matching
  • 2016-10-15: added tag gardening: vocabulary tags not used + tags not in vocabulary
  • 2016-10-16: interactively adding tags: omit already assigned tags in shortcuts and vocabulary
  • 2016-11-27: added existing shared tags to visual tags
  • 2017-02-06: better help text for --filter option
  • 2017-02-25: shortcut tags can be mixed with non-shortcut tags
    • Example: mytag 49 anothertag does add tags mytag and anothertag and the shortcut tags 4 and 9
  • 2017-04-09:
    • interactively removing tags via -tagname:
      • Example: the tag input tagname -removeme adds the tag tagname and removes the tag removeme from the filename(s)
    • try to find alternative filename if file not found
      • Example: if you try to tag file My file name.pdf which is not found, filetags tries to look for a different (unique and existing) filename that shares the same start of the file name such as My file name -- mytag.pdf. Very handy!
      • This happens a lof when you are interactively adding multiple tags one by one by simply re-executing the previous command line: the file name changes in between because of the previous tag(s) being added.
  • 2017-08-27: when tagging symbolic links whose source file has a matching file name, the source file gets the same tags as the symbolic link of it
    • This is especially useful when using the --filter option
  • 2017-08-28:
    • moved from optparse to argparse
    • removed option --tag (in favor to --tags)
    • added option shortcut for recursive: -R
    • renamed option --imageviewer to --filebrowser and enabled its functionality
    • added new feature --tagtrees
  • 2017-08-31:
    • improved screen output when renaming files
  • 2017-09-03:
    • --recursive option now works for linking files to tagtrees as well
    • corresponding .filetags files get linked to the output of tagtrees as well
  • 2017-11-11:
    • removed command line options -r, -d, and --delete
      • keeping --remove as the only option for removing tags
      • removing tags was overrepresented in the command line options, blocking them to be used for other useful commands
    • added =–tagtrees-handle-no-tag “treeroot” | “ignore” | “FOLDERNAME”=
    • added --tagtrees-link-missing-mutual-tagged-items
  • 2017-12-30:
    • added --tagtrees-dir <existing_directory>
      • overriding the default target directory for the tagtrees result
    • added --tagtrees-depth TAGTREES_DEPTH
      • allowing to override the default depth of tagtrees
      • use with care: especially on Windows a larger depth than 2 takes very long
    • tagtrees now work with Windows using lnk files
      • in contrast to symbolic links, that have rather poor performance though: generation of tagtrees take way longer than on Linux or macOS
  • 2018-01-30:
    • fixed the pip3 package
  • 2018-03-18:
    • added more detailed statistics on usage of tag groups when doing tag gardening
    • added internal data structure cache_of_files_with_metadata
  • 2018-04-05:
    • --tagtrees-dir can now be used for --filter
    • much deeper support for Windows .lnk files:
      • tagging lnk files within tagtrees also tag their original files
      • .filetags files can now be .lnk files as well
      • the unit tests now work on Windows and test some Windows specialities
  • 2018-04-18:
    • comments in .filetags files that contain the controlled vocabulary
  • 2018-04-25:
  • 2018-07-23: --tagtrees= can now be filtered with --filter
  • 2018-08-02: added option --hardlinks as an alternative for non-Windows systems
  • 2019-12-22: added manual file globbing for Windows because of #25

Get the most out of filetags: controlled vocabulary .filetags

This awesome tool is providing support for controlled vocabularies. When invoked for interactive tagging, it is looking for files named .filetags in the current working directory and its parent directories as well. The first file of this name found is read in. Each line represents one tag. Those tags are used for tag completion.

This is purely great: with tags within .filetags you don’t have to enter the tags entrirely: just type the first characters and press TAB (twice to show you all possibilities). You will be amazed how efficiently you are going to tag things! :-)

Of course, you can remove existing tags by prepending a - character to the tag: -tagname. This also works interactively using the tab completion feature.

You can use comments in .filetags files: everything after a # character is considered a comment. You can even add a comment after a tag like ”mytag # this is a test tag”.

Mutually exclusive tags

If you enter multiple tags in the same line in .filetags, they are interpreted as mutually exclusive tags. For example, if your .filetags contains the line winter spring summer autumn, filetags replaces any season-tag with the new one. So if you tag the file …

example file -- summer anothertag.txt

… with the tag winter, it gets renamed to …

example file -- winter anothertag.txt

… without having to manually remove the tag summer.

Common mutually exclusive tags are draft final or confidential internal public.


Consider you have a directory that contains hundreds of files.

If you want to retrieve a file whose tags you know, you can skim through all the files. However, filetags offers you a more elegant possibility: you can filter the files according to one or more tags.

For example, we take a look at following situation:

$HOME/my party/
|_ 2018-06-25 Party invitation -- scan correspondence.pdf
|_ 2018-07-31 Guest list -- correspondence.txt
|_ 2018-08-01T11.51.44 Uncle Bob arrives.jpg
|_ 2018-08-01T12.31.42 Sheila with her new boyfriend -- friends.jpg
|_ 2018-08-01T14.12.23 Start of BBQ with the big steak.jpg
|_ ...
|_ 2018-08-01T23.53.19 Even uncle Bob desides to go home -- fun.jpg
|_ 2018-08-05 Lessons learned for planning a party -- scan.pdf
|_ 2018-08-06 Thank-you letter Bob -- scan.pdf
|_ Bills/
  |_ 2018-07-30 Beverages by FreshYouUp -- scan taxes.pdf
  |_ 2018-08-03 Bill of the butcher -- scan taxes.pdf

Following command and interaction would generate following temporal link structure:

filetags --filter

User gets asked to enter one or more tags and she enters “scan”:

|_ 2018-06-25 Party invitation -- scan correspondence.pdf
|_ 2018-08-05 Lessons learned for planning a party -- scan.pdf
|_ 2018-08-06 Thank-you letter Bob -- scan.pdf

This way, our user is quickly able to skim through all scanned documents to locate the one desired to retrieve.

To locate all matching files in all sub-directories as well, the user is able to add the parameter --recursive

filetags --filter --recursive

… and chooses to enter the tag “scan” which would generate following temporal link structure:

|_ 2018-06-25 Party invitation -- scan correspondence.pdf
|_ 2018-08-05 Lessons learned for planning a party -- scan.pdf
|_ 2018-08-06 Thank-you letter Bob -- scan.pdf
|_ 2018-07-30 Beverages by FreshYouUp -- scan taxes.pdf
|_ 2018-08-03 Bill of the butcher -- scan taxes.pdf


This functions is somewhat sophisticated with regards to the background. If you’re really interested in the whole story behind the visualization/navigation of tags using TagTrees, feel free to read my PhD thesis about it on the tagstore webpage. It is surely a piece of work I am proud of and the general chapters of it are written so that the average person is perfectly well able to follow.

In short: this function takes the files of the current directory and generates hierarchies up to level of $maxdepth (by default 2, can be overridden via --tagtrees-depth) of all combinations of tags, linking all files according to their tags.

Consider having a file like:

My new car -- car hardware expensive.jpg

Now you generate the TagTrees, you’ll find links to this file within sub-directories of ~/.filetags, the default target directory: new/ and hardware/ and expensive/ and new/hardware/ and new/expensive/ and hardware/new/ and so on. You get the idea.

The default target directory can be overridden via --tagtrees-dir.

Therefore, within the folder new/expensive/ you will find all files that have at least the tags “new” and “expensive” in any order. This is really cool to have.

Files of the current directory that don’t have any tag at all, are linked directly to ~/.filetags so that you can find and tag them easily.

I personally, do use this feature within my image viewer of choice (geeqie). I mapped it to Shift-T because Shift-t is occupied by filetags for tagging of course. So when I am within my image viewer and I press Shift-T, TagTrees of the currently shown images are created. Then an additional image viewer window opens up for me, showing the resulting TagTrees. This way, I can quickly navigate through the tag combinations to easily interactively filter according to tags.

Please note: when you are tagging linked files within the TagTrees with filetags, only the current link gets updated with the new name. All other links to this modified filename within the other directories of the TagTrees gets broken. You have to re-create the TagTrees to update all the links after tagging files.

The option --tagtrees-handle-no-tag controls how files with no tags should be handled. When set to treeroot, untagged files are linked in the TagTrees target directory directly. The option ignore does not link them at all. The option FOLDERNAME links them to a directory named accordingly to the value which is a sub-directory of the TagTrees target directory.

With the option --tagtrees-link-missing-mutual-tagged-items you can control, whether or not there will be an additional TagTrees folder that contains all files which lack one of the mutually exclusive tags. Using the example winter spring summer autumn from above, all files that got none of those four tags get linked to a TagTrees directory named “no_winter_spring_summer_autumn”. This way, you can easily find and tag files that don’t participate in this set of mutually exclusive tags.

Using the example files from above:

$HOME/my party/
|_ 2018-06-25 Party invitation -- scan correspondence.pdf
|_ 2018-07-31 Guest list -- correspondence.txt
|_ 2018-08-01T11.51.44 Uncle Bob arrives.jpg
|_ 2018-08-01T12.31.42 Sheila with her new boyfriend -- friends.jpg
|_ 2018-08-01T14.12.23 Start of BBQ with the big steak.jpg
|_ ...
|_ 2018-08-01T23.53.19 Even uncle Bob desides to go home -- fun.jpg
|_ 2018-08-05 Lessons learned for planning a party -- scan.pdf
|_ 2018-08-06 Thank-you letter Bob -- scan.pdf
|_ Bills/
  |_ 2018-07-30 Beverages by FreshYouUp -- scan taxes.pdf
  |_ 2018-08-03 Bill of the butcher -- scan taxes.pdf

… and the command line …

filetags --tagtrees --tagtrees-handle-no-tag "has_no_tag" --tagtrees-depth 2 --recursive

… filetags generates the temporal link structure:

|_ scan/
  |_ 2018-06-25 Party invitation -- scan correspondence.pdf
  |_ 2018-08-05 Lessons learned for planning a party -- scan.pdf
  |_ 2018-08-06 Thank-you letter Bob -- scan.pdf
  |_ 2018-07-30 Beverages by FreshYouUp -- scan taxes.pdf
  |_ 2018-08-03 Bill of the butcher -- scan taxes.pdf
  |_ correspondence/
    |_ 2018-06-25 Party invitation -- scan correspondence.pdf
  |_ taxes/
    |_ 2018-07-30 Beverages by FreshYouUp -- scan taxes.pdf
    |_ 2018-08-03 Bill of the butcher -- scan taxes.pdf
|_ correspondence/
  |_ 2018-06-25 Party invitation -- scan correspondence.pdf
  |_ 2018-07-31 Guest list -- correspondence.txt
  |_ scan/
    |_ 2018-06-25 Party invitation -- scan correspondence.pdf
|_ friends/
  |_ 2018-08-01T12.31.42 Sheila with her new boyfriend -- friends.jpg
|_ fun/
  |_ 2018-08-01T23.53.19 Even uncle Bob desides to go home -- fun.jpg
|_ taxes/
  |_ 2018-07-30 Beverages by FreshYouUp -- scan taxes.pdf
  |_ 2018-08-03 Bill of the butcher -- scan taxes.pdf
  |_ scan/
    |_ 2018-07-30 Beverages by FreshYouUp -- scan taxes.pdf
    |_ 2018-08-03 Bill of the butcher -- scan taxes.pdf
|_ has_no_tag/
  |_ 2018-08-01T11.51.44 Uncle Bob arrives.jpg
  |_ 2018-08-01T14.12.23 Start of BBQ with the big steak.jpg
  |_ ...

This looks complicated because there are many links generated the user does not really need. The beauty of this solution is that the user is able to navigate to a file using a wide set of different paths (the TagTrees) and she is able to choose the one path that suits the current cognitive model.

For example, she might want to retrieve “the one document from the last party which she remembers of having scanned and which she used for the invitation correspondence”. With this mind-set, she most likely retrieves the document via $HOME/.filetags_tagfilter/scan/correspondence/ or $HOME/.filetags_tagfilter/correspondence/scan/ (does not matter which).

The large number of other TagTrees can be ignored for this retrieval task.

Another retrieval task example would be “all photos that do have no tag in order to continue tagging the photos”. In this example, the user visits $HOME/.filetags_tagfilter/has_no_tag/, fires her image viewer (which has filetags integrated already - see below) and continues with the tagging activity. Since filetags synchronizes the tags within TagTrees linked files and the original files, the original files get renamed accordingly.

Bonus: Using tags to specify a sub-set of photographs

You know the problem: got back from Paris and you can not show 937 image files to your friends. It’s just too much.

My solution: I tag to define selections. For example, I am using sel (“selection”) for the ultimate cool photographs using filetags, of course.

Within geeqie, which is my preferred image viewer, I redefined F to call filetags with its --filter parameter. Now I get asked to enter one or more tags to filter the current folder. For presenting only the files that were tagged with sel, I enter sel and confirm with Enter.

This creates a temporary folder with symbolic links to all photographs of the current folder that contain the tag sel and it starts a new (additional) instance of geeqie.

In short: after returning from a trip, I mark all “cool” photographs within geeqie, choose t and tag them with sel (described in previous section). For showing only sel images, I just press F, enter sel and instead of 937 photographs, my friends just have to watch the best 50 or so. :-)

Watch this 45 minute talk on how I am using this (and other) features.

Integration Into Common Tools

If your system has Python 3 installed, you can start using filetags right away in any command line environment.

However, users do want to integrate tools like filetags also in various GUI tools.

The file explains integration in some tools that allow external commands being added:

  • geeqie, a GNU/Linux image viewer I am using
  • Thunar is a popular GNU/Linux file browser for the xfce environment
  • Windows Explorer
  • FreeCommander, my recommendated alternative to Windows explorer
  • Dired, the GNU/Emacs file manager

If you have integrated filetags in additional commonly used tools, please send me a short how-to so that others are able to get the most out of filetags as well.

Related Tools and Workflows

This tool is part of a tool-set which I use to manage my digital files such as photographs. My work-flows are described in this blog posting you might like to read and in the video which is linked above.

In short:

For tagging, please refer to filetags and its documentation.

See date2name for easily adding ISO time-stamps or date-stamps to files.

For easily naming and tagging files within file browsers that allow integration of external tools, see appendfilename (once more) and filetags.

Moving to the archive folders is done using move2archive.

Having tagged photographs gives you many advantages. For example, I automatically choose my desktop background image according to the current season.

Files containing an ISO time/date-stamp gets indexed by the filename-module of Memacs.

How to Thank Me

I’m glad you like my tools. If you want to support me:

  • Send old-fashioned postcard per snailmail - I love personal feedback!
  • Send feature wishes or improvements as an issue on GitHub
  • Create issues on GitHub for bugs
  • Contribute merge requests for bug fixes
  • Check out my other cool projects on GitHub

Exhaustive List of All Features

This section is an exhaustive list of features of filetags. You might skip this when you’re a first-time user in order not to get irritated for simple use-cases only.

This section is particularily helpful for re-implementing filetags functionality and for power-users which are interested in the advanced functions provided by this tool.


Some file name.jpegtagging with fooSome file name -- foo.jpegTag separator is added automatically
Some file nametagging with fooSome file name -- fooThere is no need for a file extension
Some file name -- foo.jpegtagging with barSome file name -- foo bar.jpegbar becomes last tag
Some file name.jpeg.lnktagging with barSome file name -- bar.jpeg.lnkThe .lnk extension is taken into account
Some file name -- bar.jpeguntagging barSome file name.jpegTag separator is removed
Some file name -- foo bar.jpeguntagging fooSome file name -- bar.jpegTag order stays same when removing
  • filetags may be used
    1. interactively (via --interactive or missing “action” command line parameters) from command line or
    2. in a script using command line parameters.
  • filetags offers a --dryrun option which does not modify any file or directory.
  • Added tag(s) get appended as last tag(s).
  • When removing tags, their relative order is preserved.
  • When modifying any file that is a symbolic link or a Windows .LNK file to a file that has the same basename (file name without path), the linked/original file gets modified as well.
    • This comes very handy when working within TagTrees (see below).
    • However, when modifying links which do not share the same base-name with its link source, the link might become a broken one (depending on the link technology used).
  • When un-tagging tags from files that do not have those tags, it is silently ignored.
  • [ ] FIXXME: describe find_unique_alternative_to_file(filename) and implications
  • FUTURE: support for tagging folders/directories · Issue #13 · novoid/filetags · GitHub
  • FUTURE: Files within tagged directories do inherit the tags for all relevant features · Issue #14 · novoid/f…
    • Inheritance applies to many features such as “don’t tag a file with a tag from any parent directory” and so forth.
    • Not that simple to decide each use-case. This is a hard nut to crack with many complex things to take care of.
  • FUTURE: CV: add CLI option that prevents users from using tags that are not part of the used .filetags file …
    • Enforcing CVs is a good practice IMHO.

Interactive Mode

  • Print used tags of selected file(s).
    • For multiple files, show only the tags that are used within all selected files.
  • filetags dialog shows up to nine topmost used tags (sorted by number of usage) used for files within the current directory.
    • E.g., draft(1) projectX(2) customer(3) bill(4)
    • You can use 0-9 as shortcuts to select those tags.
      • You can concatenate shortcut numbers without spaces in-between: 143 foo tags with the shortcuts number 1, 3 and 4 and adds new tag foo.
        • With the example above, it is equivalent to tagging with: draft bill customer foo or draft 4 3 foo.
  • You can un-tag tags that appear in file name using the minus prefix.
    • E.g., -foo un-tags the tag foo.
    • Auto-completion is provided to un-tag existing tags.
  • Tags from the CV (within .filetags files) and from tags used in the current directory can be auto-completed via TAB.
    • Already used tags are not available for completion.
  • Multiple Files
    • You can tag/un-tag multiple selected files at once.
      • Selected files containing the tag(s) to tag are not modified and no tags get duplicated.
      • Selected files not containing the tag(s) to un-tag are not modified.
    • Tag suggestions for un-tagging contain the common tags of selected files.
  • Tagging dialog can be aborted any time via Ctrl-c.

Controlled Vocabulary (CV)

Please read this first in order to understand CVs.

  • CV is read from .filetags files.
    • One tag per line: simple tag
    • Multiple tags per line, separated via spaces: a group of mutually exclusive tags
      • E.g., draft final approved
        • When tagging My report -- draft.txt with final, draft gets replaced by final without the user un-tagging it before.
        • filetags does not prevent user from manually tagging files with two or more mutually exclusive tags.
    • The order of priority to locate “matching” .filetags files is:
      1. Current directory of the first file to tag/un-tag.
      2. Any higher-level directory from the current directory of the first file to tag/un-tag.
      3. .filetags file from the HOME directory.
    • .filetags files may be links (hardlinks, symbolic links or even Windows .LNK files)
  • Comments within .filetags files begin with one or more # characters that may be prepended by one or more spaces.
  • PLANNED: .filetags files may include other .filetags files via #include <relative or absolute path to another file>
  • FUTURE: CV: .filetags may contain mandatory options · Issue #17 · novoid/filetags · GitHub
    • Probably a nice to have for different default-behavior in different sub-hierarchies of the file system.


This function is very handy for filtering groups of photographs within a large set of photographs as described here.

  • The user defines one or more tags whose files are linked to a target directory.
  • Any “matching” .filetags file is linked to the target directory.
  • A populated target directory is never overwritten.
  • The default target directory is .filetags_tagfilter and might be changed by --tagtrees-dir.
  • When started interactively, a file browser is opened showing the target directory.
    • The file browser tool might be overwritten with --filebrowser.
  • The --recursive option is taken into account accordingly.

Features Related to TagTrees

The TagTrees concept was developed by me during my PhD thesis (PDF) when developing with the tagstore research platform.

Please note that in future, all functions related to TagTrees will be moved into a separate tool named tagtrees.

Tag Gardening

Just invoke filetags --tag-gardening or filetags --recursive --tag-gardening and read its output to learn about helpful analysis results to curate your tags. My personal favorites are:

  • I am able to find typos in tags (tag count is low and similar tags are found).
  • I can determine tags I seldom use and therefore might be removed from CVs.
  • Statistics on tag usage like, e.g.:
    • Distribution of mutually exclusive tag options.
    • Fraction of files that are not tagged.
  • Tags I have used which are not in my CVs.
  • Unused tags.

Local Variables