Ref. #366 Also updates `CONTRIBUTING.md`. This was pretty straight-forward :)
8.1 KiB
Contributing to Catala
The project is open to external contributions, in the spirit of open source. If you want to open a pull request, please follow the instructions below.
To ask a question to the Catala team, please open an issue on this repository. You can also join the Zulip chat to ask any questions about the project.
If you want to contribute to the project on a longer-term basis, or if you have specific competences as a socio-fiscal lawyer or a programming language specialist, please contact the authors. The Catala team meets over visioconference once every week.
Please note that the copyright of this code is owned by Inria; by contributing, you disclaim all copyright interests in favor of Inria. Both the code for the compiler and the examples in this repository are distributed under the Apache2 license.
Writing Catala code
Before writing Catala code, please read the
tutorial. You can run the
programs of the tutorial yourself by following the instruction in the
README of the examples
directory. Then, it is suggested
that you create a new example directory again according to the instructions of
this README.
Let us now present the typical Catala workflow. First, you need to locate the legislative text that you want to use as a reference. Then, simply copy-paste the text into your source file.
First you will have to format the copy-pasted text using Catala headings and articles markers:
## Heading
### Sub-heading (the more '#', the less important)
#### Legislative atom
Please look at the code of other examples to see how to format things properly. While formatting the text, don't forget regularly to try and parse your example using for instance
make -C examples/foo foo.tex
make -C examples/foo foo.py
make -C examples/foo foo.ml
to see if you've made any syntax errors. Once the text formatting is done, you can start to annotate each legislative atom (article, provision, etc.) with some Catala code. To open up a code section in Catala, simply use
```catala
# In code sections, comments start with #
scope Foo:
<your code goes here>
```
While all the code sections are equivalent in terms of execution, you can mark some as "metadata" so that they are printed differently on lawyer-facing documents. Here's how it works:
```catala-metadata
declaration structure FooBar:
data foo content boolean
data bar content money
<your structure/enumeration/scope declarations goes here>
```
Again, make sure to regularly check that your example is parsing correctly. The error message from the compiler should help you debug the syntax if need be. You can also
live-test the programs you wrote by feeding them to the interpreter
(see the README of the examples
directory); this will
also type-check the programs, which is useful for debugging them.
Working on the compiler
The Catala compiler is a standard dune-managed OCaml project. You can look at the online OCaml documentation for the different modules' interfaces as well as high-level architecture documentation.
Installing and using nix
We provide an nix environement to develop the Catala compiler. It is available
after installing nix. You can then just
use nix develop
to enter the environment.
Example: adding a builtin function
The language provides a limited number of builtin functions, which are sometimes needed for things that can't easily be expressed in Catala itself; in case you need more, here is how one can be added:
- Choose a name wisely. Be ready to patch any code that already used the name for scope parameters, variables or structure fields, since it won't compile anymore.
- Add an element to the
builtin_expression
type insurface/ast.ml
- Add your builtin in the
builtins
list insurface/lexer.cppo.ml
, and with proper translations in all of the language-specific modulessurface/lexer_en.cppo.ml
,surface/lexer_fr.cppo.ml
, etc. Don't forget the macro at the beginning oflexer.cppo.ml
. - The rest can all be done by following the type errors downstream:
- Add a corresponding element to the lower-level AST in
shared_ast/definitions.ml
, typeOp.t
- Extend the generic operations on operators in
shared_ast/operators.ml
as well as the type information for the operator - Extend the translation accordingly in
desugared/from_surface.ml
- Extend the printer (
shared_ast/print.ml
) - Finally, provide the implementations:
- in
dcalc/interpreter.ml
, functionevaluate_operator
- in
../runtimes/ocaml/runtime.ml
- in
../runtimes/python/catala/src/catala/runtime.py
- in
- Add a corresponding element to the lower-level AST in
- Update the syntax guide in
doc/syntax/syntax.tex
with your new builtin
Internationalization of the Catala syntax
The Catala language should be adapted to any legislative text that follows a general-to-specifics statutes order. Therefore, there exists multiple versions of the Catala surface syntax, adapted to the language of the legislative text.
Currently, Catala supports English, French and Polish legislative text via the
--language=en
, --language=fr
or --language=pl
options.
To add support for a new language:
-
the basic syntax localisation is defined in
compiler/surface/lexer_xx.cppo.ml
wherexx
is the language code (en
,fr
...) -
copy the files from another language, e.g. english, then replace the strings with your translations. Be careful with the following:
- The file must be encoded in latin-1
- For a given token
FOO
, defineMS_FOO
to be the string version of the keyword. Due to the encoding, use\xNN
escape sequences for utf8 characters. - If the string contains spaces or non-latin1 characters, you need to define
MR_FOO
as well with a regular expression in sedlex format. Replace spaces with", space_plus, "
, and unicode characters with", 0xNNNN, "
whereNNNN
is the hexadecimal unicode codepoint.
Hint: You may get syntax errors with unhelpful locations because of
sedlex
. In that case the commandocamlc _build/default/compiler/surface/lexer_xx.ml
may point you to the source of the error. -
add your translation to the compilation rules:
- in
compiler/surface/dune
, copying anotherparser_xx.cppo.ml
rule - in the
extensions
list incompiler/driver.ml
- add a corresponding variant to
compiler/utils/cli.ml
backend_lang
, try to runmake build
and follow all type errors andmatch non exhaustive
warnings to be sure it is well handled everywhere.
- in
-
you may want to add syntax highlighting support, see
syntax_highlighting/
and the rules inMakefile
-
add examples and documentation!
Feel free to open a pull request for discussion even if you couldn't go through
all these steps, the lexer_xx.cppo.ml
file is the important part.
Example: writing custom backends as plugins
Catala has support for dynamically-loaded plugins to use as alternative
backends. See compiler/plugins
for examples, and the
documentation for more
detail.
Automatic formatting
Please ensure to submit commits formatted using the included ocamlformat
configuration. The make build
target should ensure that.
In case the formatting rules or ocamlformat version changed remotely, you can use this script to reformat your branch patch by patch before rebasing.
Hand-updating packages in the nix part
Requirements of catala that are not inside nixpkgs are available inside the .nix
directory of the repo. The main part is inside the .nix/packages.nix
, where all the packages are either added (because absent from nixpkgs) using ocamlPackage.callPackage
; or modified from nixpkgs, for instance cmdliner is currently pinned at version 1.1.0.