add support for string comparisons with in/not in (#722)

This allows markers like `"tegra" in platform_release`.

Co-authored-by: Randy Döring <30527984+radoering@users.noreply.github.com>

Author: npapapietro

src/poetry/core/constraints/generic/parser.py

@@ -16,6 +16,15 @@
 
 
 BASIC_CONSTRAINT = re.compile(r"^(!?==?)?\s*([^\s]+?)\s*$")
+STR_CMP_CONSTRAINT = re.compile(
+    r"""(?ix)^ # case insensitive and verbose mode
+    (?P<quote>['"]) # Single or double quotes
+    (?P<value>.+?) # The value itself inside quotes
+    \1 # Closing single of double quote
+    \s* # Space
+    (?P<op>(not\sin|in)) # Literal match of 'in' or 'not in'
+    $"""
+)
 
 
 @functools.lru_cache(maxsize=None)
@@ -26,9 +35,7 @@ def parse_constraint(constraints: str) -> BaseConstraint:
     or_constraints = re.split(r"\s*\|\|?\s*", constraints.strip())
     or_groups = []
     for constraints in or_constraints:
-        and_constraints = re.split(
-            r"(?<!^)(?<![=>< ,]) *(?<!-)[, ](?!-) *(?!,|$)", constraints
-        )
+        and_constraints = re.split(r"\s*,\s*", constraints)
         constraint_objects = []
 
         if len(and_constraints) > 1:
@@ -53,9 +60,15 @@ def parse_constraint(constraints: str) -> BaseConstraint:
 
 
 def parse_single_constraint(constraint: str) -> Constraint:
+    # string comparator
+    if m := STR_CMP_CONSTRAINT.match(constraint):
+        op = m.group("op")
+        value = m.group("value").strip()
+        return Constraint(value, op)
+
     # Basic comparator
-    m = BASIC_CONSTRAINT.match(constraint)
-    if m:
+
+    if m := BASIC_CONSTRAINT.match(constraint):
         op = m.group(1)
         if op is None:
             op = "=="

src/poetry/core/version/markers.py

@@ -19,6 +19,7 @@
 from poetry.core.constraints.generic import Constraint
 from poetry.core.constraints.generic import MultiConstraint
 from poetry.core.constraints.generic import UnionConstraint
+from poetry.core.constraints.generic.parser import STR_CMP_CONSTRAINT
 from poetry.core.constraints.version import VersionConstraint
 from poetry.core.constraints.version import VersionUnion
 from poetry.core.constraints.version.exceptions import ParseConstraintError
@@ -336,7 +337,11 @@ def __hash__(self) -> int:
 
 
 class SingleMarker(SingleMarkerLike[Union[BaseConstraint, VersionConstraint]]):
-    _CONSTRAINT_RE = re.compile(r"(?i)^(~=|!=|>=?|<=?|==?=?|in|not in)?\s*(.+)$")
+    _CONSTRAINT_RE_PATTERN_1 = re.compile(
+        r"(?i)^(?P<op>~=|!=|>=?|<=?|==?=?|not in|in)?\s*(?P<value>.+)$"
+    )
+    _CONSTRAINT_RE_PATTERN_2 = STR_CMP_CONSTRAINT
+
     VALUE_SEPARATOR_RE = re.compile("[ ,|]+")
     _VERSION_LIKE_MARKER_NAME: ClassVar[set[str]] = {
         "python_version",
@@ -345,7 +350,10 @@ class SingleMarker(SingleMarkerLike[Union[BaseConstraint, VersionConstraint]]):
     }
 
     def __init__(
-        self, name: str, constraint: str | BaseConstraint | VersionConstraint
+        self,
+        name: str,
+        constraint: str | BaseConstraint | VersionConstraint,
+        swapped_name_value: bool = False,
     ) -> None:
         from poetry.core.constraints.generic import (
             parse_constraint as parse_generic_constraint,
@@ -355,20 +363,29 @@ def __init__(
         parsed_constraint: BaseConstraint | VersionConstraint
         parser: Callable[[str], BaseConstraint | VersionConstraint]
         original_constraint_string = constraint_string = str(constraint)
+        self._swapped_name_value: bool = swapped_name_value
+
+        if swapped_name_value:
+            pattern = self._CONSTRAINT_RE_PATTERN_2
+        else:
+            pattern = self._CONSTRAINT_RE_PATTERN_1
 
-        # Extract operator and value
-        m = self._CONSTRAINT_RE.match(constraint_string)
+        m = pattern.match(constraint_string)
         if m is None:
             raise InvalidMarker(f"Invalid marker for '{name}': {constraint_string}")
 
-        self._operator = m.group(1)
+        self._operator = m.group("op")
         if self._operator is None:
             self._operator = "=="
 
-        self._value = m.group(2)
+        self._value = m.group("value")
         parser = parse_generic_constraint
 
-        if name in self._VERSION_LIKE_MARKER_NAME:
+        if swapped_name_value and name not in PYTHON_VERSION_MARKERS:
+            # Something like `"tegra" in platform_release`
+            # or `"arm" not in platform_version`.
+            pass
+        elif name in self._VERSION_LIKE_MARKER_NAME:
             parser = parse_marker_version_constraint
 
             if self._operator in {"in", "not in"}:
@@ -472,7 +489,11 @@ def invert(self) -> BaseMarker:
             # We should never go there
             raise RuntimeError(f"Invalid marker operator '{self._operator}'")
 
-        return parse_marker(f"{self._name} {operator} '{self._value}'")
+        if self._swapped_name_value:
+            constraint = f'"{self._value}" {operator} {self._name}'
+        else:
+            constraint = f'{self._name} {operator} "{self._value}"'
+        return parse_marker(constraint)
 
     def __eq__(self, other: object) -> bool:
         if not isinstance(other, SingleMarker):
@@ -484,6 +505,8 @@ def __hash__(self) -> int:
         return hash(self._key)
 
     def __str__(self) -> str:
+        if self._swapped_name_value:
+            return f'"{self._value}" {self._operator} {self._name}'
         return f'{self._name} {self._operator} "{self._value}"'
 
 
@@ -961,11 +984,21 @@ def _compact_markers(
 
         elif token.data == f"{tree_prefix}item":
             name, op, value = token.children
-            if value.type == f"{tree_prefix}MARKER_NAME":
+            swapped_name_value = value.type == f"{tree_prefix}MARKER_NAME"
+            stringed_value = name.type in {
+                f"{tree_prefix}ESCAPED_STRING",
+                f"{tree_prefix}SINGLE_QUOTED_STRING",
+            }
+            if swapped_name_value:
                 name, value = value, name
 
             value = value[1:-1]
-            sub_marker = SingleMarker(str(name), f"{op}{value}")
+
+            sub_marker = SingleMarker(
+                str(name),
+                f'"{value}" {op}' if stringed_value else f"{op}{value}",
+                swapped_name_value=swapped_name_value,
+            )
             groups[-1].append(sub_marker)
 
         elif token.data == f"{tree_prefix}BOOL_OP" and token.children[0] == "or":

tests/constraints/generic/test_main.py

@@ -18,6 +18,8 @@
         ("==win32", Constraint("win32", "=")),
         ("!=win32", Constraint("win32", "!=")),
         ("!= win32", Constraint("win32", "!=")),
+        ("'tegra' not in", Constraint("tegra", "not in")),
+        ("'tegra' in", Constraint("tegra", "in")),
     ],
 )
 def test_parse_constraint(input: str, constraint: AnyConstraint | Constraint) -> None:
@@ -39,6 +41,13 @@ def test_parse_constraint(input: str, constraint: AnyConstraint | Constraint) ->
                 Constraint("linux2", "!="),
             ),
         ),
+        (
+            "'tegra' not in,'rpi-v8' not in",
+            MultiConstraint(
+                Constraint("tegra", "not in"),
+                Constraint("rpi-v8", "not in"),
+            ),
+        ),
     ],
 )
 def test_parse_constraint_multi(input: str, constraint: MultiConstraint) -> None:
@@ -53,6 +62,10 @@ def test_parse_constraint_multi(input: str, constraint: MultiConstraint) -> None
             "win32 || !=linux2",
             UnionConstraint(Constraint("win32"), Constraint("linux2", "!=")),
         ),
+        (
+            "'tegra' in || 'rpi-v8' in",
+            UnionConstraint(Constraint("tegra", "in"), Constraint("rpi-v8", "in")),
+        ),
     ],
 )
 def test_parse_constraint_union(input: str, constraint: UnionConstraint) -> None:

tests/version/test_markers.py

@@ -62,6 +62,11 @@
         'extra == "a" or extra != "b"',
         'extra != "a" or extra == "b"',
         'extra != "a" or extra != "b"',
+        # String comparison markers
+        '"tegra" in platform_release',
+        '"tegra" not in platform_release',
+        '"tegra" in platform_release or "rpi-v8" in platform_release',
+        '"tegra" not in platform_release and "rpi-v8" not in platform_release',
     ],
 )
 def test_parse_marker(marker: str) -> None:
@@ -110,6 +115,10 @@ def test_parse_marker(marker: str) -> None:
             "platform_machine",
             "!=aarch64, !=loongarch64",
         ),
+        ('"tegra" not in platform_release', "platform_release", "'tegra' not in"),
+        ('"rpi-v8" in platform_release', "platform_release", "'rpi-v8' in"),
+        ('"arm" not in platform_version', "platform_version", "'arm' not in"),
+        ('"arm" in platform_version', "platform_version", "'arm' in"),
     ],
 )
 def test_parse_single_marker(
@@ -1300,6 +1309,8 @@ def test_union_of_multi_with_a_containing_single() -> None:
             'python_full_version ~= "3.6.3"',
             'python_full_version < "3.6.3" or python_full_version >= "3.7.0"',
         ),
+        ('"tegra" in platform_release', '"tegra" not in platform_release'),
+        ('"tegra" not in platform_release', '"tegra" in platform_release'),
     ],
 )
 def test_invert(marker: str, inverse: str) -> None:

tests/version/test_requirements.py

@@ -109,6 +109,18 @@ def assert_requirement(
                 ),
             },
         ),
+        (
+            (
+                'foo (>=1.2.3) ; "tegra" not in platform_release and python_version >= "3.10"'
+            ),
+            {
+                "name": "foo",
+                "constraint": ">=1.2.3",
+                "marker": (
+                    '"tegra" not in platform_release and python_version >= "3.10"'
+                ),
+            },
+        ),
     ],
 )
 def test_requirement(string: str, expected: dict[str, Any]) -> None: