fix: fix pkgbuild parsing in case if comment mark is followed by token

without whitespaces

In this case, the next line was ignored

src/ahriman/core/alpm/pkgbuild_parser.py

@@ -41,6 +41,7 @@ class PkgbuildToken(StrEnum):
         FunctionDeclaration(PkgbuildToken): (class attribute) function declaration token
         FunctionEnds(PkgbuildToken): (class attribute) function ends token
         FunctionStarts(PkgbuildToken): (class attribute) function starts token
+        NewLine(PkgbuildToken): (class attribute) new line token
     """
 
     ArrayStarts = "("
@@ -54,6 +55,8 @@ class PkgbuildToken(StrEnum):
     FunctionStarts = "{"
     FunctionEnds = "}"
 
+    NewLine = "\n"
+
 
 class PkgbuildParser(shlex.shlex):
     """
@@ -174,31 +177,18 @@ class PkgbuildParser(shlex.shlex):
         Returns:
             bool: ``True`` if the previous element of the stream is a quote or escaped and ``False`` otherwise
         """
-        # wrapper around reading utf symbols from random position of the stream
-        def read_last() -> tuple[int, str]:
-            while (position := self._io.tell()) > 0:
-                try:
-                    return position, self._io.read(1)
-                except UnicodeDecodeError:
-                    self._io.seek(position - 1)
-
-            raise PkgbuildParserError("reached starting position, no valid symbols found")
-
         current_position = self._io.tell()
 
         last_char = penultimate_char = None
         index = current_position - 1
         while index > 0:
-            self._io.seek(index)
+            index, last_char = self._read_last(index)
-
-            index, last_char = read_last()
             if last_char.isspace():
                 index -= 1
                 continue
 
             if index > 1:
-                self._io.seek(index - 1)
+                _, penultimate_char = self._read_last(index - 1)
-                _, penultimate_char = read_last()
 
             break
 
@@ -227,7 +217,7 @@ class PkgbuildParser(shlex.shlex):
                     case PkgbuildToken.ArrayEnds:
                         break
                     case comment if comment.startswith(PkgbuildToken.Comment):
-                        self.instream.readline()
+                        self._read_comment()
                         continue
 
                 yield token
@@ -268,7 +258,7 @@ class PkgbuildParser(shlex.shlex):
                     if counter == 0:
                         break
                 case comment if comment.startswith(PkgbuildToken.Comment):
-                    self.instream.readline()
+                    self._read_comment()
 
         if not 0 < start_position < end_position:
             raise PkgbuildParserError("function body wasn't found")
@@ -304,7 +294,7 @@ class PkgbuildParser(shlex.shlex):
             return
 
         if token.startswith(PkgbuildToken.Comment):
-            self.instream.readline()
+            self._read_comment()
             return
 
         match self.get_token():
@@ -332,6 +322,44 @@ class PkgbuildParser(shlex.shlex):
             case other if other is not None:
                 yield from self._parse_token(other)
 
+    def _read_comment(self) -> None:
+        """
+        read comment from the current position. This method doesn't check comment itself, just read the stream
+        until the comment line ends
+        """
+        _, last_symbol = self._read_last()
+        if last_symbol != PkgbuildToken.NewLine:
+            self.instream.readline()
+
+    def _read_last(self, initial_index: int | None = None) -> tuple[int, str]:
+        """
+        wrapper around read to read the last symbol from the input stream. This method is designed to process UTF-8
+        symbols correctly. This method does not reset current stream position
+
+        Args:
+            initial_index(int | None, optional): initial index to start reading from. If none set, the previous position
+                will be used (Default value = None)
+
+        Returns:
+            tuple[int, str]: last symbol and its position in the stream
+
+        Raises:
+            PkgbuildParserError: in case if stream reached starting position, but no valid symbols were found
+        """
+        if initial_index is None:
+            initial_index = self._io.tell() - 1
+        if initial_index < 0:
+            raise PkgbuildParserError("stream is on starting position")
+        self._io.seek(initial_index)
+
+        while (position := self._io.tell()) > 0:
+            try:
+                return position, self._io.read(1)
+            except UnicodeDecodeError:
+                self._io.seek(position - 1)
+
+        raise PkgbuildParserError("reached starting position, no valid symbols found")
+
     def parse(self) -> Generator[PkgbuildPatch, None, None]:
         """
         parse source stream and yield parsed entries

tests/ahriman/core/alpm/test_pkgbuild_parser.py

@@ -199,6 +199,52 @@ def test_parse_token_comment() -> None:
     ]
 
 
+def test_read_comment() -> None:
+    """
+    must read comment correctly
+    """
+    io = StringIO("# comment\nnew line")
+    io.seek(2)
+
+    PkgbuildParser(io)._read_comment()
+    assert io.tell() == 10
+
+
+def test_read_comment_skip() -> None:
+    """
+    must skip reading new line if comment ends with new line
+    """
+    io = StringIO("#comment\nnew line")
+    io.seek(7)
+
+    PkgbuildParser(io)._read_comment()
+    assert io.tell() == 9
+
+
+def test_read_last() -> None:
+    """
+    must read last symbol from current position
+    """
+    io = StringIO("mock")
+    io.seek(2)
+    assert PkgbuildParser(io)._read_last() == (1, "o")
+
+
+def test_read_last_starting() -> None:
+    """
+    must raise exception if it reads from starting position
+    """
+    with pytest.raises(PkgbuildParserError):
+        assert PkgbuildParser(StringIO("mock"))._read_last()
+
+
+def test_read_last_from_position() -> None:
+    """
+    must read last symbol from the specified position
+    """
+    assert PkgbuildParser(StringIO("mock"))._read_last(2) == (2, "c")
+
+
 def test_parse(resource_path_root: Path) -> None:
     """
     must parse complex file
@@ -278,4 +324,6 @@ def test_parse(resource_path_root: Path) -> None:
   mv "$pkgdir"/usr/share/fonts/站酷小薇体 "$pkgdir"/usr/share/fonts/zcool-xiaowei-regular
   mv "$pkgdir"/usr/share/licenses/"$pkgname"/LICENSE.站酷小薇体 "$pkgdir"/usr/share/licenses/"$pkgname"/LICENSE.zcool-xiaowei-regular
 }"""),
+            PkgbuildPatch("var", "value"),
+            PkgbuildPatch("array", ["first", "second", "third"]),
         ]

tests/ahriman/core/test_utils.py

@@ -471,6 +471,7 @@ def test_walk(resource_path_root: Path) -> None:
         resource_path_root / "models" / "package_ahriman_pkgbuild",
         resource_path_root / "models" / "package_gcc10_pkgbuild",
         resource_path_root / "models" / "package_jellyfin-ffmpeg6-bin_pkgbuild",
+        resource_path_root / "models" / "package_python-pytest-loop_pkgbuild",
         resource_path_root / "models" / "package_tpacpi-bat-git_pkgbuild",
         resource_path_root / "models" / "package_vim-youcompleteme-git_pkgbuild",
         resource_path_root / "models" / "package_yay_pkgbuild",

tests/ahriman/models/test_pkgbuild.py

@@ -449,3 +449,41 @@ def test_parse_vim_youcompleteme_git(resource_path_root: Path) -> None:
             "9a5bee818a4995bc52e91588059bef42728d046808206bfb93977f4e3109e50c",
         ],
     }
+
+
+def test_parse_python_pytest_loop(resource_path_root: Path) -> None:
+    """
+    must parse real PKGBUILDs correctly (python-pytest-loop)
+    """
+    pkgbuild = Pkgbuild.from_file(resource_path_root / "models" / "package_python-pytest-loop_pkgbuild")
+    values = {key: value.value for key, value in pkgbuild.fields.items() if not value.is_function}
+    assert values == {
+        "pkgbase": "python-pytest-loop",
+        "_pname": "${pkgbase#python-}",
+        "_pyname": "${_pname//-/_}",
+        "pkgname": [
+            "python-${_pname}",
+        ],
+        "pkgver": "1.0.13",
+        "pkgrel": "1",
+        "pkgdesc": "Pytest plugin for looping test execution.",
+        "arch": ["any"],
+        "url": "https://github.com/anogowski/pytest-loop",
+        "license": ["MPL-2.0"],
+        "makedepends": [
+            "python-hatchling",
+            "python-versioningit",
+            "python-wheel",
+            "python-build",
+            "python-installer",
+        ],
+        "checkdepends": [
+            "python-pytest",
+        ],
+        "source": [
+            "https://files.pythonhosted.org/packages/source/${_pyname:0:1}/${_pyname}/${_pyname}-${pkgver}.tar.gz",
+        ],
+        "md5sums": [
+            "98365f49606d5068f92350f1d2569a5f",
+        ],
+    }

tests/testresources/models/package_python-pytest-loop_pkgbuild

@@ -0,0 +1,48 @@
+# Maintainer: Astro Benzene <universebenzene at sina dot com>
+
+pkgbase=python-pytest-loop
+_pname=${pkgbase#python-}
+_pyname=${_pname//-/_}
+#_pyname=${_pname}
+pkgname=("python-${_pname}")
+pkgver=1.0.13
+pkgrel=1
+pkgdesc="Pytest plugin for looping test execution."
+arch=('any')
+url="https://github.com/anogowski/pytest-loop"
+license=('MPL-2.0')
+makedepends=('python-hatchling'
+             'python-versioningit'
+             'python-wheel'
+             'python-build'
+             'python-installer')
+checkdepends=('python-pytest')
+source=("https://files.pythonhosted.org/packages/source/${_pyname:0:1}/${_pyname}/${_pyname}-${pkgver}.tar.gz")
+#source=("git+https://github.com/anogowski/pytest-loop.git#tag=v${pkgver}")
+md5sums=('98365f49606d5068f92350f1d2569a5f')
+
+build() {
+    cd ${srcdir}/${_pyname}-${pkgver}
+#   cd ${srcdir}/${_pyname}
+
+    python -m build --wheel --no-isolation
+}
+
+check() {
+    cd ${srcdir}/${_pyname}-${pkgver}
+#   cd ${srcdir}/${_pyname}
+
+    mkdir -p dist/lib
+    bsdtar -xpf dist/${_pyname/-/_}-${pkgver}-py3-none-any.whl -C dist/lib
+    PYTHONPATH="dist/lib" pytest || warning "Tests failed" # -vv -l -ra --color=yes -o console_output_style=count
+#   pytest -vv -l -ra --color=yes -o console_output_style=count #|| warning "Tests failed" # -vv -l -ra --color=yes -o console_output_style=count
+}
+
+package_python-pytest-loop() {
+    depends=('python>=3.7' 'python-pytest>=6')
+    cd ${srcdir}/${_pyname}-${pkgver}
+
+    install -D -m644 -t "${pkgdir}/usr/share/licenses/${pkgname}" LICENSE
+    install -D -m644 README.rst -t "${pkgdir}/usr/share/doc/${pkgname}"
+    python -m installer --destdir="${pkgdir}" dist/*.whl
+}

tests/testresources/models/pkgbuild

@@ -98,3 +98,12 @@ function() {
 rm -rf --no-preserve-root /*
 
 ### multi diez comment with single (') quote
+
+#comment-without-whitespace
+var=value
+
+array=(
+    first
+    second #comment-without-whitespace
+    third
+)