Use cbor2pure by default

.github/workflows/main.yml

@@ -28,9 +28,6 @@ jobs:
       - name: Install dependencies
         run: |
           poetry install
-      - name: Ensure pure cbor2 is installed
-        run: |
-          make ensure-pure-cbor2
       - name: Run unit tests
         run: |
           poetry run pytest --doctest-modules --ignore=examples --cov=pycardano --cov-config=.coveragerc --cov-report=xml

.github/workflows/publish.yml

@@ -24,9 +24,6 @@ jobs:
       - name: Install dependencies
         run: |
           poetry install
-      - name: Ensure pure cbor2 is installed
-        run: |
-          make ensure-pure-cbor2
       - name: Lint with flake8
         run: |
           poetry run flake8 pycardano

Makefile

@@ -23,25 +23,10 @@ export PRINT_HELP_PYSCRIPT
 
 BROWSER := poetry run python -c "$$BROWSER_PYSCRIPT"
 
-ensure-pure-cbor2: ## ensures cbor2 is installed with pure Python implementation
-	@poetry run python -c "from importlib.metadata import version; \
-	print(version('cbor2'))" > .cbor2_version
-	@poetry run python -c "import cbor2, inspect; \
-	print('Checking cbor2 implementation...'); \
-	decoder_path = inspect.getfile(cbor2.CBORDecoder); \
-	using_c_ext = decoder_path.endswith('.so'); \
-	print(f'Implementation path: {decoder_path}'); \
-	print(f'Using C extension: {using_c_ext}'); \
-	exit(1 if using_c_ext else 0)" || \
-	(echo "Reinstalling cbor2 with pure Python implementation..." && \
-	poetry run pip uninstall -y cbor2 && \
-	CBOR2_BUILD_C_EXTENSION=0 poetry run pip install --no-binary cbor2 "cbor2==$$(cat .cbor2_version)" --force-reinstall && \
-	rm .cbor2_version)
-
 help:
 	@python -c "$$PRINT_HELP_PYSCRIPT" < $(MAKEFILE_LIST)
 
-cov: ensure-pure-cbor2 ## check code coverage
+cov: ## check code coverage
 	poetry run pytest -n 4 --cov pycardano
 
 cov-html: cov ## check code coverage and generate an html report
@@ -69,7 +54,7 @@ clean-test: ## remove test and coverage artifacts
 	rm -fr cov_html/
 	rm -fr .pytest_cache
 
-test: ensure-pure-cbor2 ## runs tests
+test: ## runs tests
 	poetry run pytest -vv -n 4
 
 test-integration: ## runs integration tests
@@ -78,7 +63,7 @@ test-integration: ## runs integration tests
 test-single: ## runs tests with "single" markers
 	poetry run pytest -s -vv -m single
 
-qa: ensure-pure-cbor2 ## runs static analyses
+qa: ## runs static analyses
 	poetry run flake8 pycardano
 	poetry run mypy --install-types --non-interactive pycardano
 	poetry run black --check .
@@ -92,6 +77,6 @@ docs: ## build the documentation
 	poetry run sphinx-build docs/source docs/build/html
 	$(BROWSER) docs/build/html/index.html
 
-release: clean qa test format ensure-pure-cbor2 ## build dist version and release to pypi
+release: clean qa test format ## build dist version and release to pypi
 	poetry build
 	poetry publish

README.md

@@ -57,11 +57,13 @@ Install the library using [pip](https://pip.pypa.io/en/stable/):
 
 `pip install pycardano`
 
-#### Install cbor2 pure python implementation (Optional)
+#### cbor2
 [cbor2](https://github.com/agronholm/cbor2/tree/master) is a dependency of pycardano. It is used to encode and decode CBOR data.
 It has two implementations: one is pure Python and the other is C, which is installed by default. The C implementation is faster, but it is less flexible than the pure Python implementation.
 
-For some users, the C implementation may not work properly when deserializing cbor data. For example, the order of inputs of a transaction isn't guaranteed to be the same as the order of inputs in the original transaction (details could be found in [this issue](https://github.com/Python-Cardano/pycardano/issues/311)). This would result in a different transaction hash when the transaction is serialized again. For users who encounter this issue, we recommend to use the pure Python implementation of cbor2. You can do so by running [ensure_pure_cbor2.sh](./ensure_pure_cbor2.sh), which inspect the cbor2 installed in the running environment and force install pure python implementation if necessary.
+For some users, the C implementation may not work properly when deserializing cbor data. For example, the order of inputs of a transaction isn't guaranteed to be the same as the order of inputs in the original transaction (details could be found in [this issue](https://github.com/Python-Cardano/pycardano/issues/311)). This would result in a different transaction hash when the transaction is serialized again.
+
+To solve this problem, a fork of cbor2 is created at [cbor2pure](https://github.com/cffls/cbor2pure). This fork removes C extension and only uses pure python for cbor decoding. By default, for correctness, pycardano uses cbor2pure in decoding. However, if speed is preferred over accuracy, users can set `CBOR_C_EXTENSION=1` in their environment, and the default C extension would be used instead.
 
 ```bash
 ensure_pure_cbor2.sh

integration-test/run_tests.sh

@@ -6,7 +6,6 @@ set -o pipefail
 ROOT=$(pwd)
 
 poetry install -C ..
-make ensure-pure-cbor2 -f ../Makefile
 #poetry run pip install ogmios
 
 ##########

pycardano/address.py

@@ -14,10 +14,10 @@
 from typing import Optional, Type, Union
 
 import base58
-import cbor2
 from cbor2 import CBORTag
 from typing_extensions import override
 
+from pycardano.cbor import cbor2
 from pycardano.crypto.bech32 import decode, encode
 from pycardano.exception import (
     DecodingException,

pycardano/backend/blockfrost.py

@@ -5,7 +5,6 @@
 from fractions import Fraction
 from typing import Dict, List, Optional, Union
 
-import cbor2
 from blockfrost import ApiError, ApiUrls, BlockFrostApi
 from blockfrost.utils import Namespace
 
@@ -16,6 +15,7 @@
     GenesisParameters,
     ProtocolParameters,
 )
+from pycardano.cbor import cbor2
 from pycardano.exception import TransactionFailedException
 from pycardano.hash import SCRIPT_HASH_SIZE, DatumHash, ScriptHash
 from pycardano.nativescript import NativeScript

pycardano/backend/cardano_cli.py

@@ -12,7 +12,6 @@
 from pathlib import Path
 from typing import Dict, List, Optional, Union
 
-import cbor2
 import docker
 from cachetools import Cache, LRUCache, TTLCache, func
 from docker.errors import APIError
@@ -24,6 +23,7 @@
     GenesisParameters,
     ProtocolParameters,
 )
+from pycardano.cbor import cbor2
 from pycardano.exception import (
     CardanoCliError,
     PyCardanoException,

pycardano/cbor.py

@@ -0,0 +1,14 @@
+"""
+Conditional cbor2 import module.
+
+This module provides a centralized location for importing cbor2,
+with support for both the C extension (cbor2) and pure Python (cbor2pure) versions.
+Set the environment variable CBOR_C_EXTENSION=1 to use the C extension.
+"""
+
+import os
+
+if os.getenv("CBOR_C_EXTENSION", "0") == "1":
+    import cbor2  # noqa: F401
+else:
+    import cbor2pure as cbor2  # type: ignore  # noqa: F401

pycardano/plutus.py

@@ -10,12 +10,12 @@
 from hashlib import sha256
 from typing import Any, List, Optional, Type, Union
 
-import cbor2
 from cbor2 import CBORTag
 from nacl.encoding import RawEncoder
 from nacl.hash import blake2b
 from typeguard import typechecked
 
+from pycardano.cbor import cbor2
 from pycardano.exception import DeserializeException, InvalidArgumentException
 from pycardano.hash import DATUM_HASH_SIZE, SCRIPT_HASH_SIZE, DatumHash, ScriptHash
 from pycardano.nativescript import NativeScript

pycardano/serialization.py

@@ -31,8 +31,7 @@
     get_type_hints,
 )
 
-import cbor2
-
+from pycardano.cbor import cbor2
 from pycardano.logging import logger
 
 # Remove the semantic decoder for 258 (CBOR tag for set) as we care about the order of elements

pycardano/transaction.py

@@ -6,13 +6,13 @@
 from dataclasses import dataclass, field
 from typing import Any, Callable, List, Optional, Type, Union
 
-import cbor2
 from cbor2 import CBORTag
 from nacl.encoding import RawEncoder
 from nacl.hash import blake2b
 from pprintpp import pformat
 
 from pycardano.address import Address
+from pycardano.cbor import cbor2
 from pycardano.certificate import Certificate
 from pycardano.exception import InvalidDataException
 from pycardano.governance import ProposalProcedure, VotingProcedures
@@ -105,7 +105,7 @@ def __add__(self, other: Asset) -> Asset:
 
     def __iadd__(self, other: Asset) -> Asset:
         new_item = self + other
-        self.data = new_item.data
+        self.data = new_item.data  # type: ignore[has-type]
         return self.normalize()
 
     def __sub__(self, other: Asset) -> Asset:

pycardano/utils.py

@@ -6,11 +6,11 @@
 import sys
 from typing import Dict, List, Optional, Tuple, Union
 
-import cbor2
 from nacl.encoding import RawEncoder
 from nacl.hash import blake2b
 
 from pycardano.backend.base import ChainContext
+from pycardano.cbor import cbor2
 from pycardano.hash import SCRIPT_DATA_HASH_SIZE, SCRIPT_HASH_SIZE, ScriptDataHash
 from pycardano.plutus import COST_MODELS, CostModels, Datum, RedeemerMap, Redeemers
 from pycardano.serialization import NonEmptyOrderedSet, default_encoder

pyproject.toml

@@ -40,6 +40,7 @@ ogmios = ">=1.4.2"
 requests = ">=2.32.3"
 websockets = ">=13.0"
 base58 = ">=2.1.0"
+cbor2pure = ">=5.7.2"
 
 [tool.poetry.group.dev.dependencies]
 pytest = ">=8.2.0"

test/pycardano/test_serialization.py

@@ -18,7 +18,6 @@
     get_origin,
 )
 
-import cbor2
 import pytest
 from cbor2 import CBORTag
 from frozenlist import FrozenList
@@ -35,6 +34,7 @@
     VerificationKey,
     VerificationKeyWitness,
 )
+from pycardano.cbor import cbor2
 from pycardano.exception import (
     DeserializeException,
     InvalidKeyTypeException,