Skip to main content

Very basic btor2 parser, circuit miter, and code optimizer.

Project description

btor2-opt

Very basic btor2 parser, circuit miter, and code optimizer.

Install

pip install btor2-opt

For nix users

nix-shell -p 'python3.withPackages(ps: with ps; [ black build bumpver click colorama isort lexid looseversion mypy-extensions packaging pathspec pip-tools platformdirs pyproject-hooks toml tomli typing-extensions wheel])'

Test

Unit tests can be run using:

python -m unittest tests/test.py

Overview

This repo contains two main scripts:

  • btoropt: Takes a .btor2 file and a list of pass names as argument and prints out the transformed result.
  • btormiter: Takes a .fir file as input and runs it through firrtl and firtool to obtain two .btor2 designs which are then merged into a single miter circuit before being returned to the user. Note that this requires having firrtl and firtool in your path, specifically a version of firtool that has the --btor2 flag.

The rest of the code can be found in the src folder, which contains a basic parser for btor2 (not entirely complete, but supports anything firtool --btor2 can produce), an internal representation of the language and a simple pass infrastructure, where you can add any of you custom passes.

Supported Instructions

This compiler currently supports the following btor2 instructions:

Instruction Description
<lid> sort <type> <width> Declares a type
<lid> input <sid> <name> Declares an input
<lid> output <out> Declares an output
<lid> bad <cond> Checks the inversion of a condition
<lid> constraint <cond> Assumes a condition
<lid> zero <sid> Declares a 0 constant
<lid> one <sid> Declares a 1 constant
<lid> ones <sid> Declares a bit-string of 1s
<lid> not <sid> <cond> Negates a condition
<lid> constd <sid> <val> Declares a decimal constant
<lid> consth <sid> <val> Declares a hexadecimal constant
<lid> const <sid> <val> Declares a binary constant
<lid> state <sid> <name> Declares a stateful element
<lid> init <sid> <state> <val> Initializes a state
<lid> next <sid> <state> <next> Sets the transition logic of a state
<lid> slice <sid> <op> <hb> <lb> Extracts bits [hb:lb] from a result
<lid> ite <sid> <cond> <t> <f> If-then-else expression
<lid> implies <sid> <lhs> <rhs> Logical implication
<lid> iff <sid> <lhs> <rhs> If and only if expression
<lid> add/sub/mul <sid> <l> <r> Binary operation
<lid> {s,u}div <sid> <l> <r> Signed or unsigned division
<lid> smod <sid> <l> <r> Signed modulo
<lid> s{l,r}l <sid> <l> <r> Logical shift left/right
<lid> sra <sid> <l> <r> Arithmetic shift right
<lid> and/or/xor <sid> <l> <r> Binary logical operators
<lid> concat <sid> <l> <r> Concatenate two results
<lid> eq/neq <sid> <l> <r> Equality comparators
<lid> {s,u}gt <sid> <l> <r> Signed/Unsigned l > r
<lid> {s,u}gte <sid> <l> <r> Signed/Unsigned l ≥ r
<lid> {s,u}lt <sid> <l> <r> Signed/Unsigned l < r
<lid> {s,u}lte <sid> <l> <r> Signed/Unsigned l ≤ r
<lid> uext <sid> <opid> <w> <name> Unsigned width extension / aliasing

Adding a Pass

Simply create a new class (as its own file) in src/passes that inherits from Pass. Then in the constructor, make sure you give it a name. The pass's logic itself is written by overiding the run(p: list[Instruction]) -> list[Instruction] method. The pass must then be imported in src/passes/passes.py and instantiated in the all_passes list. Passes are grouped either in transforms, which contain all of the passes that transform the AST, and validation, which contains all of the passes used to gurantee the syntactic correctness of the output program.

Here is a simple example pass that renames all inputs to "inp_n".

# Example pass: Simply renames all inputs to inp_<pos>
class RenameInputs(Pass):
    def __init__(self):
        super().__init__("rename-inputs")

    # I chose to have this pass not modify p in place
    # you can also simply modify p and return it
    def run(self, p: list[Instruction]) -> list[Instruction]:
        i = 0
        res = []
        for inst in p:
            if isinstance(inst, Input):
                res.append(Input(inst.lid, inst.sort, f"inp_{i}"))
                i += 1
            else:
                res.append(inst)
        return res

# Make sure to add an instance of the pass to the all_passes array
all_passes = [RenameInputs()]

This pass can then be called by running:

btoropt ex.btor2 rename-inputs

Custom BTOR2 Extensions

btoropt currently supports custom extensions to the standard btor2 format, enabling the expression of modularity. In order to maintin inter-operability with standard btor2 files, this extension must be explicitly enabled using the --modular flag. These extensions are simply syntactic sugar to enable parallelism and can be lowered to standard btor2. This makes it possible to express the following custom structures:

Structure Description
module <name> {...} Declares a named region of instructions
contract <module_name> {...} Declares a contract for a module

Module bodies support all standard btor2 along with the following custom instructions:

Module Instructions Description
<lid> ref <module_name> <lid_in_module> References an instruction from another module
<lid> inst <module_name> Creates an instance of a module
<lid> set <instance_lid> <ref_lid> <local_lid> Sets an instance reference to a local instruction

Contract bodies support all standard btor2 along with the following custom instructions:

Contract Instructions Description
<lid> ref <module_name> <lid_in_module> References an instruction from the contract's module
<lid> prec <cond_lid> Declares a precondition
<lid> post <cond_lid> Declares a postcondition

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

btor2_opt-0.3.0.tar.gz (55.7 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

btor2_opt-0.3.0-py3-none-any.whl (47.3 kB view details)

Uploaded Python 3

File details

Details for the file btor2_opt-0.3.0.tar.gz.

File metadata

  • Download URL: btor2_opt-0.3.0.tar.gz
  • Upload date:
  • Size: 55.7 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.11.9

File hashes

Hashes for btor2_opt-0.3.0.tar.gz
Algorithm Hash digest
SHA256 c2d1c0c947b8bdf5c70d9726e723452984952fa5e490a662167983b4fe7b2a96
MD5 24a57911931d245c010226eeb3566aba
BLAKE2b-256 55a7906a056e8d449b4bad49469bb146bb3274a07ae4de86348fbaaf5c42d51a

See more details on using hashes here.

File details

Details for the file btor2_opt-0.3.0-py3-none-any.whl.

File metadata

  • Download URL: btor2_opt-0.3.0-py3-none-any.whl
  • Upload date:
  • Size: 47.3 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.11.9

File hashes

Hashes for btor2_opt-0.3.0-py3-none-any.whl
Algorithm Hash digest
SHA256 e77bc4f5fef4ebb7b2e27c44ea52d9364d3f88920c73e33b5ab1b6640b0edf8c
MD5 c442cf355d9784bab127f3c0ee8f824c
BLAKE2b-256 6bd496e3abbb8654cd363019b4ae7b1c2e054cda072325691b493cda2d5bb1ca

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page