Dreamdir – the Unix dream journal format
Dreamdir is a format for storing a dream journal inspired by the popular Maildir mailbox format and the RFC 2822 email format itself.
Dreamdir follows the Unix philosophy, as follows:
- All dreams are stored in plain text formatted in a consistent way. The format is human-readable and is normally read and written directly.
- Dreamdir is first and foremost a format, not an application. Simple code designed to do one thing well is provided to make working with the format easier.
- In addition to the provided code, standard Unix text-processing tools and simple scripts can easily manipulate dreams stored in a Dreamdir.
Data format:
- Each dream is stored in its own plain text file.
- Each dream file begins with a list of data fields (called headers), which can store things like the date, the names of people who appeared in the dream, and tags.
- The remainder of the dream file is the text of your entry.
Application:
- The
dr
script offers powerful and intuitive search functionality and several other useful tools to create new dreams, combine tags, and so on. It easily integrates with other Unix tools. - Example user scripts are provided for tasks like graphing day-to-day recall and generating indexes. A small Python library is also available.
- Free and open source: the Dreamdir format itself and this documentation are public domain, and all code is provided under the MIT license (see the LICENSE file for details).
Dreamdir scripts are permitted to assume that the rules specified in this
section are followed, so you should definitely follow them. Conformance to the
most basic rules can be checked with the dr check-validity
command.
The easiest way to learn the format is to look at an example dream file (people’s names are censored here but are written out in full in the actual file format):
Id: 00952
Date: 2015-11-19
Time: 6:02
People: M., Md., T.
Places: Wal-Mart
Tags: store, radio, cottage, enchantment, police
I'm in some kind of large store which sells food, perhaps a supermarket or
Wal-Mart or something of that nature. I have found two small individual
packages of applesauce and eaten them, and I now am wondering if that's
okay and how I will pay for them. I'm with M here. Near the back of the
store, there's a small table with a store walkie-talkie on it. I pick this
up, and though I'm not sure exactly how this happened, I soon get into a
heated argument with a store manager over the radio. It starts off being
about the applesauce in some fashion, but it soon turns into a tirade about
consumerism and the advertising in their store. Eventually the manager
gets so pissed at what I'm saying (though I wasn't *trying* to upset him, I
also wasn't trying to be careful) that he tells me he's going to have the
police come and arrest me. I don't question that he has the authority to
do this, and so M and I decide that we should leave quickly before they
arrive. We head out of the store; as a final "screw you" I grab a Snickers
bar that's sitting on a table as I leave.
[remainder of dream clipped]
At the top you see the headers.
Each header is separated from its values by a colon followed by a hard tab.
(The requirement of a tab makes commands like grep
easier to use,
since a colon will not be followed by a tab in running text.)
For headers that can take multiple values,
such as People, Places, and Tags,
individual values are separated by a comma and a space.
Two headers are required to form a valid dream file: the Id number and the Date. All other headers are optional. There are no rules about what constitutes a valid header name (except that of course it cannot contain a colon followed by a hard tab), and it is fine for some dreams to not have a particular optional header at all. The headers can come in any order. A blank line follows the headers.
The two required headers are fussier, as follows:
- Id: Dreamdir uses fixed-width five-digit ID numbers,
beginning at
00001
and increasing for each dream up to99999
. (If you manage to record 100,000 dreams, updates to the program and a beer are on me!) - Date: Dreamdir scripts expect ISO 8601-formatted dates (YYYY-MM-DD).
Beneath the headers, following a blank line, comes the text of the dream. As long as you don’t begin any later line with a header (i.e., a line containing a colon immediately followed by a hard tab), you can do anything you like here, though you may wish to look at the “Formatting guidelines” section, below.
- Emphasis: Use
*single stars*
or_underlines_
around the area to be emphasized. - Commentary: Place notes that are not actually part of the dream in
[square brackets]
. The commentary may continue over multiple lines. - Verbatim quoting: Place text that is directly quoted from some earlier form of notes (such as a notebook you scribbled in in bed) in
`backticks`
. - Lucid sections: If you lucid dream, you can place sections where you knew you were dreaming in
{curly braces}
. I also use the headerLucid: 1
to make it easier to find these dreams.
Emphasis and verbatim quoting are recognized by the syntax highlighting functionality. Lucid sections and commentary are recognized by the word count scripts (word counts can be split into “normal,” “lucid,” and “notes”).
I use one physical line per paragraph and double-space between sentences, but I don’t see those conventions ever being expected by any code.
As mentioned earlier, you can use any headers you like as long as you include the ID number and the date. As a starting point, here are the ones I currently use:
- People: Comma-separated list of waking-life people who appeared in the dream.
- Places: Ditto for places with proper names and general geographic regions.
- Tags: List of motifs, categories, and other elements that are useful to track across multiple dreams but don’t fit into other headers.
- Title: A title...
- Time: If I had a clock handy and remembered to write it down, the time at which I woke up from the dream.
- Lucid: Included and with a value of 1 if the dream was lucid at any point.
All dreams are kept in the root of the dreamdir.
A dream’s filename is its ID number with a .dre
extension, e.g., 00592.dre
.
-rw------- 1 soren soren 623 Dec 14 2014 00001.dre
-rw-rw-r-- 1 soren soren 210 Feb 14 19:00 00002.dre
[...]
-rw------- 1 soren soren 3075 Apr 28 21:25 01227.dre
-rw------- 1 soren soren 1600 Apr 27 13:21 01228.dre
-rw-rw-r-- 1 soren soren 0 Apr 29 15:56 .dreamdir
Note the .dreamdir
file, which marks this directory as a Dreamdir.
The content is currently unimportant,
but scripts may check for this file to ensure they’re working in a dreamdir.
dr
uses the current directory as the dreamdir
if it contains that .dreamdir
file.
Otherwise, it checks the environment variable DREAMDIR
for a path and uses that location if it’s set to a valid dreamdir.
If neither of these locations are valid dreamdirs,
dr
will exit without doing anything.
Begin by copying the dr
script to somewhere on your system path.
If you want to use the word-count
functionality,
which differs from the standard wc
in that it ignores headers
and can count marked lucid and notes sections separately,
change into the drwc
directory and run make
(this requires gcc
),
then install the compiled drwc
binary to your system path as well.
If you want to use the ctags generation functionality,
make sure you have python3
available on your PATH.
Finally, create a directory for your dreamdir and use touch .dreamdir
to mark the directory as a dreamdir.
You should now be able to use dr new
to create your first dream file.
You may also wish to set the environment variable DREAMDIR
to the path to your dreamdir;
this way you can run dr
from anywhere in your filesystem.
I keep my dreamdir under git
control
to keep track of any revisions I make to headers and dreams
and as an extra backup against scripting and PEBKAC errors.
You may wish to do likewise.
dr
works great on MacOS,
but you will have to install newer versions of a couple of utilities.
Installation through Homebrew is recommended.
After following the steps above:
brew install bash python bats
Installation of bats
is not required,
but it allows you to run tests/test_dr
to confirm you have all the right utilities
and your installation works correctly,
which may otherwise be difficult to do:
cd dreamdir
tests/test_dr
Utilities needed by dr that are not part of POSIX:
bash
4.0+python3
(needed forregenerate-tags
)
dr tries to stick to POSIX standards,
but we have limited ability to test that we have done so
across multiple systems.
If something in dr
doesn’t work right in your version of a POSIX utility,
or if a system in common use doesn’t comply with the relevant aspects
of a POSIX standard dr
relies upon,
please submit a bug report.
(dr also uses seq
, tac
, and shuf
when available,
but includes slower, more memory-intensive fallback methods
for when GNU coreutils aren’t present,
so they are not necessary.)
A number of example scripts, largely written in Python, are provided in the
scripts/
directory of this repository; you may wish to use some of these as
models for building your own scripts. Of particular note is ddirparse.py
,
which is a general library for use in developing Dreamdir scripts.
I don’t make any guarantees about the general applicability or correctness of these scripts. Some have not been tested recently. You must read through the code of any script that looks interesting before using it; you may find there are still file paths, system-specific constants, or other surprises lurking somewhere.
For those who use vim, syntax highlighting and ftplugin files are located in
the vim/
directory; you can install these to your ~/.vim
directory directly
or use your favorite plugin manager.
You may want to remove the setlocal cpoptions+=J
(:help cpo-J
) line from
vim/ftplugin/dream.vim
if you don’t want to double-space between sentences
(see the “Formatting guidelines” section above).
Ctags functionality is included in dreamdir. The tags file contains tags for all dream ID numbers and all header values. You can do a number of handy things with this in vim. Among others:
- With a dream open, jump to a different dream N with
:ta N
, e.g.,:ta 900
. - With the cursor on a dream number (see #900), press
Control-]
to open that dream. - With a header value highlighted in visual mode, or with the cursor on top of
a single-word header value, press
g]
to show all matching tags. This will show all the dreams that use that header value, along with the whole header line and the Title header of the matching dream (if available). - Search for header values containing foo with
:tjump /foo
. Along with being a handy way to do a quick search without having to leave your editor and rundr
again, this is an easy way to answer questions about what tags you’ve previously used when you’re thinking of tags for the dream you’re currently writing up.
Tags are stored in the file .dreams.ctags
in your dreamdir folder. They need
to be initially created as well as updated periodically by running dr regenerate-tags
; any changes to headers or new dreams will not be known to Vim
until the tags file is updated.
The dr
script provides convenient tools to manage your dreamdir. The script
assumes the current directory is your dreamdir, so it may be run like this:
$ cd ~/dreams
$ dr new
Alternatively, dr
recognizes the environment variable DREAMDIR
as the path
to your dreamdir. If this variable is defined and you are not currently in a
dreamdir when you run dr
, that directory will be used instead.
Detailed help on all the functions and options you have can be obtained at the
command line. Type dr help
for the basics, and it will refer you to other help pages
(e.g., dr help search
) where appropriate.
Note that all commands can be abbreviated by their initials; find
is f
, for
example, and dump-headers
is dh
. Keywords in search expressions can be
abbreviated similarly, so to find dreams tagged with Maud
in the People
field, we can use dr f t People Maud
as well as dr find tagged People Maud
.
The long names are normally used throughout the documentation for clarity, and
they are easier to remember when you’re getting started, but once you know what
you’re doing you can save quite a few keystrokes this way.
Please post bugs on the Github issue tracker; if you prefer you can email me at
[email protected]
. If you have a problem with dr
, please mention
your operating system and version of bash.
Improvements and pull requests are welcome as long as you release your code
under the MIT license and they are consistent with the project’s philosophy.
Please make sure that tests/pre-commit-hook.sh
exits successfully before
submitting any modifications to dr
; this requires
shellcheck,
BATS, and GCC. If you have modified the
behavior of dr
, you may need to change the tests in tests/test_dr
; they are
pretty easy to figure out. If you don’t want to install shellcheck
(it
requires Haskell and takes some time to install), you can paste the code on the
shellcheck website and then run tests/test_dr
and cd scripts && make clean && make
manually to finish the pre-commit tests.