in-house bytes pkgbuild parser

src/ahriman/core/exceptions.py

@@ -95,19 +95,6 @@ class DuplicateRunError(RuntimeError):
             self, "Another application instance is run. This error can be suppressed by using --force flag.")
 
 
-class EncodeError(ValueError):
-    """
-    exception used for bytes encoding errors
-    """
-
-    def __init__(self, encodings: list[str]) -> None:
-        """
-        Args:
-            encodings(list[str]): list of encodings tried
-        """
-        ValueError.__init__(self, f"Could not encode bytes by using {encodings}")
-
-
 class ExitCode(RuntimeError):
     """
     special exception which has to be thrown to return non-zero status without error message

src/ahriman/models/pkgbuild.py

@@ -24,7 +24,6 @@ from pathlib import Path
 from typing import Any, ClassVar, IO, Self
 
 from ahriman.core.alpm.pkgbuild_parser import PkgbuildParser, PkgbuildToken
-from ahriman.core.exceptions import EncodeError
 from ahriman.models.pkgbuild_patch import PkgbuildPatch
 
 
@@ -34,13 +33,13 @@ class Pkgbuild(Mapping[str, Any]):
     model and proxy for PKGBUILD properties
 
     Attributes:
-        DEFAULT_ENCODINGS(list[str]): (class attribute) list of encoding to be applied on the file content
+        DEFAULT_ENCODINGS(str): (class attribute) default encoding to be applied on the file content
         fields(dict[str, PkgbuildPatch]): PKGBUILD fields
     """
 
     fields: dict[str, PkgbuildPatch]
 
-    DEFAULT_ENCODINGS: ClassVar[list[str]] = ["utf8", "latin-1"]
+    DEFAULT_ENCODINGS: ClassVar[str] = "utf8"
 
     @property
     def variables(self) -> dict[str, str]:
@@ -58,13 +57,13 @@ class Pkgbuild(Mapping[str, Any]):
         }
 
     @classmethod
-    def from_file(cls, path: Path, encodings: list[str] | None = None) -> Self:
+    def from_file(cls, path: Path, encoding: str | None = None) -> Self:
         """
         parse PKGBUILD from the file
 
         Args:
             path(Path): path to the PKGBUILD file
-            encodings(list[str] | None, optional): the encoding of the file (Default value = None)
+            encoding(str | None, optional): the encoding of the file (Default value = None)
 
         Returns:
             Self: constructed instance of self
@@ -77,15 +76,10 @@ class Pkgbuild(Mapping[str, Any]):
             content = input_file.read()
 
         # decode bytes content based on either
-        encodings = encodings or cls.DEFAULT_ENCODINGS
-        for encoding in encodings:
-            try:
-                io = StringIO(content.decode(encoding))
-                return cls.from_io(io)
-            except ValueError:
-                pass
+        encoding = encoding or cls.DEFAULT_ENCODINGS
+        io = StringIO(content.decode(encoding, errors="backslashreplace"))
 
-        raise EncodeError(encodings)
+        return cls.from_io(io)
 
     @classmethod
     def from_io(cls, stream: IO[str]) -> Self:

tests/ahriman/models/test_pkgbuild.py

@@ -3,9 +3,7 @@ import pytest
 from io import BytesIO, StringIO
 from pathlib import Path
 from pytest_mock import MockerFixture
-from unittest.mock import MagicMock
 
-from ahriman.core.exceptions import EncodeError
 from ahriman.models.pkgbuild import Pkgbuild
 from ahriman.models.pkgbuild_patch import PkgbuildPatch
 
@@ -46,18 +44,6 @@ def test_from_file_latin(pkgbuild_ahriman: Pkgbuild, mocker: MockerFixture) -> N
     load_mock.assert_called_once_with(pytest.helpers.anyvar(int))
 
 
-def test_from_file_unknown_encoding(pkgbuild_ahriman: Pkgbuild, mocker: MockerFixture) -> None:
-    """
-    must raise exception when encoding is unknown
-    """
-    open_mock = mocker.patch("pathlib.Path.open")
-    io_mock = open_mock.return_value.__enter__.return_value = MagicMock()
-    io_mock.read.return_value.decode.side_effect = EncodeError(pkgbuild_ahriman.DEFAULT_ENCODINGS)
-
-    with pytest.raises(EncodeError):
-        assert Pkgbuild.from_file(Path("local"))
-
-
 def test_from_io(pkgbuild_ahriman: Pkgbuild, mocker: MockerFixture) -> None:
     """
     must correctly load from io

tox.ini

@@ -1,6 +1,6 @@
 [tox]
 envlist = check, tests
-isolated_build = True
+isolated_build = true
 labels =
 	release = version, docs, publish
 dependencies = -e .[journald,pacman,s3,shell,stats,validator,web]
@@ -27,7 +27,9 @@ description = Run common checks like linter, mypy, etc
 deps =
 	{[tox]dependencies}
 	-e .[check]
+pip_pre = true
 setenv =
+	CFLAGS="-Wno-unterminated-string-initialization"
 	MYPYPATH=src
 commands =
 	autopep8 --exit-code --max-line-length 120 -aa -i -j 0 -r "src/{[tox]project_name}" "tests/{[tox]project_name}"
@@ -65,7 +67,7 @@ description = Generate html documentation
 deps =
 	{[tox]dependencies}
 	-e .[docs]
-recreate = True
+recreate = true
 commands =
 	sphinx-build -b html -a -j auto -W docs {envtmpdir}{/}html
 
@@ -89,6 +91,7 @@ description = Run tests
 deps =
 	{[tox]dependencies}
 	-e .[tests]
+pip_pre = true
 commands =
 	pytest {posargs}