Read this article about Puzzle Driven Development. Check also patent application US 12/840,306
Also, check 0pdd.com: a hosted service, where this command line tool works for you.
Read PDD in Action and watch this webinar.
First, make sure Ruby 2.6+ and libmagic
are installed. Then, install our gem:
$ gem install pdd
Run it locally and read its output:
$ pdd --help
You can exclude & include certain number of files from the search via these options:
$ pdd --exclude glob
You can skip any file(s) with a name suffix that matches the pattern glob, using wildcard matching; a name suffix is either the whole path and name, or reg expr, for example:
$ pdd --exclude src/**/*.java --exclude target/**/*
$ pdd --exclude src/**/*.java # exclude .java files in src/
$ pdd --exclude src/**/* # exclude all files in src/
You can include too:
$ pdd --include glob
Search only files whose name matches glob, using wildcard matching as described under --exclude
.
If contradictory --include
and --exclude
options are given, the last matching one wins.
If no --include
or --exclude
options are given, all files from working directory are included, example:
$ pdd --include src/**/*.py # include only .py files in src/
$ pdd --include src/**/* # include all files in src/
Full command format is (all parameters are optional):
$ pdd [--verbose] [--quiet] [--remove] [--skip-gitignore] [--skip-errors] \
[--source <project_dir_path>] [--file puzzles_file.xml] [--include src/**/*.py] \
[--format xml|html] [--rule min-words:5] [--exclude src/**/*.java]
Parameter | Description |
---|---|
--verbose | Enable verbose (debug) mode. --file must be used in case of using this option |
--quiet | Disable logs |
--remove | Remove all found puzzles from the source code |
--skip-gitignore | Don't look into .gitignore for excludes |
--skip-errors | Suppress error as warning and skip badly formatted puzzles (do not skip broken rules) |
--source | Source directory to parse ("." by default) |
--file puzzles.xml | File to save report into (xml of html) (displayed in console by default) |
--include *.py | Glob pattern to include (can be used several times) |
--exclude *.java | Glob pattern to exclude (can be used several times) |
--format xml | Format of the report xml or html (xml is default) |
--rule min-words:5 | Rule to apply (can be used several times), described later |
💡 There is an option to create a .pdd file in your project and save all required parameters in it. File example you can see in this project.
Every puzzle has to be formatted like this (pay attention to the leading space in every consecutive line):
/**
* @todo #[issue#]<:[time]></[role]> <[description]>
*/
[related code]
[]
- Replace with appropriate data (see text enclosed in brackets)
<>
- Optional (enclosed data can be left out)
Example:
/**
* @todo #234:15m/DEV This is something to do later
* in one of the next releases. I can't figure out
* how to implement it now, that's why the puzzle.
* The text can be so long, as needed, just use
* the same anount of spaces, as the second line.
* This text will be not a part of the puzzle, as
* it has less spaces.
*/
void sendEmail() {
throw new UnsupportedOperationException();
}
If you use it in combination with 0pdd, after processing this text, the issue titled "File.java:10-13: This is something to do later in one of ..." will be created. The specified markers will be included in the issues body along with some predefined text. If your comment is longer than 40 characters, it will be truncated in the title.
Note: if you create several puzzle duplicates (same text after puzzle keyword), pdd will fail to parse puzzles and produce an error with duplicates list.
There are 3 supported keywords, one of which must precede the mandatory
puzzle marker. They are @todo
, TODO
and TODO:
.
As an example, it starts with @todo
, followed by a space and a mandatory
puzzle marker. Possible formats of puzzle markers (it doesn't matter what the
line starts with and where it is located,
as long as you have one of the 3 supported keywords right in front
of the mandatory marker):
// @todo #224 Puzzle description
# @todo #55:45min Puzzle description
@todo #67/DES Puzzle description
;; @todo #678:40m/DEV Puzzle description
// TODO: #TEST-21:30min Puzzle description
Here DES
and DEV
are the roles of people who must fix that puzzles;
45min
and 40m
is the amount of time the puzzle should take;
224
, 55
, 67
, 678
and TEST-21
are the IDs of the tickets
these puzzles are coming from.
Markers are absolutely necessary for all puzzles, because they allow
us to build a hierarchical dependency tree of all puzzles, like
this one,
for example. Technically, of course, you can abuse the system
and put a dummy #1
marker everywhere.
For multiline puzzles there are two important things:
- prefix - any optional text followed by space before puzzle keyword (todo). It should be the same for all lines of puzzle description.
- \ symbol can be used to logically divide puzzle description. prefix should be presented with it.
Examples:
<!--
~ if comment should be started and closed by special symbols, then place them in
~ a separate lines
~ Any symbol can be used as a prefix, it will be excluded from the text.
~ But do not forget about the space before puzzle keyword.
~
~ @todo #34 Description can be as long as needed.
~ Just use at least the same amount of the spaces, as on the first line.
~ It will be added to description.
-->
/**
* @todo #36 Multiline text can use the same prefix in all lines or the same
* amount of spaces.
* So this will be added to the puzzle description. If you want to divide the
* puzzle logically by empty line, just add a backspace to that line
* \
* and continue the text after.
*
* This line is not part of the puzzle, because the line before does not contain
* prefix.
*/
You can specify post-parsing rules for your puzzles, in command line, for example:
$ pdd --rule min-estimate:60 --rule max-estimate:120
These two parameters will add two post-parsing rules min-estimate
and max-estimate
with parameters. Each rule may have an optional
parameter specified after a colon.
Here is a list of rules available now:
-
min-estimate:15
blocks all puzzles that don't have an estimate or their estimates are less than 15 minutes. -
max-estimate:120
blocks all puzzles with estimates over 120 minutes. -
available-roles:DEV,IMP,DES
specifies a list of roles that are allowed in puzzles. Puzzles without explicitly specified roles will be rejected. -
min-words:5
blocks puzzles with descriptions shorter than five words. -
max-duplicates:1
blocks more than one duplicate of any puzzle. This rule is used by default and you can't configure it at the moment, it must always be set to1
.
💡 You can put all command line options into .pdd
file. The options from the
file will be used first. Command line options may be added on top of them.
See, how it is done in yegor256/0pdd.
The XML produced will look approximately like this (here is a real example):
<puzzles>
<puzzle>
<ticket>516</ticket>
<estimate>15</estimate>
<role>DEV</role>
<id>516-ffc97ad1</id>
<lines>61-63</lines>
<body>This has to be fixed later...</body>
<file>src/test/java/org/takes/SomeTest.java</file>
<author>Yegor Bugayenko</author>
<email>yegor256@gmail.com</email>
<time>2018-01-01T21:09:03Z</time>
</puzzle>
</puzzles>
NOTE: puzzles are saved with utf-8 encoding
XSD Schema is here. The most interesting parts of each puzzle are:
-
ticket
is a ticket name puzzle marker starts from, in most cases it will be the number of GitHub issue. -
estimate
is the amount of minutes the puzzle is supposed to take. -
id
is a unique ID of the puzzle. It is calculated by the internal algorithm that takes into account only the text of the puzzle. Thus, if you move the puzzle from one file to another, the ID won't change. Also, changing the location of a puzzle inside a file won't change its ID. -
lines
is where the puzzle is found, inside the file.
For Debian/Ubuntu:
$ apt install libmagic-dev
For Mac:
$ brew install libmagic
Unfortunately, there is no easy way to install on Windows, try to use WSL or Docker.
Read these guidelines. Make sure your build is green before you contribute your pull request. You will need to have Ruby 2.7+ and Bundler installed. Then:
$ bundle install
$ bundle exec rake
Next, install and run overcommit to install hooks (required once)
$ gem install overcommit -v '=0.58.0'
$ overcommit --install
If it's clean and you don't see any error messages, submit your pull request.
This is how you run the tool locally to test how it works:
$ ./bin/pdd --help
To run a single unit test:
$ bundle exec ruby test/test_many.rb