Prefer conflicting causes

Author: notatallshaw

src/pip/_internal/resolution/resolvelib/provider.py

@@ -75,6 +75,100 @@ def _get_with_identifier(
     return default
 
 
+def conflicting_causes(
+    causes: Sequence["PreferenceInformation"],
+) -> Sequence["PreferenceInformation"]:
+    """Given causes return which causes conflict with each other
+    For each cause check one of two things:
+        1. If it's specifier conflicts with another causes parent version
+        2. If it's specifier conflicts with another causes specifier
+
+    Any causes which match this criteria are returned as conflicting causes
+    """
+    conflicting_ids: set[int] = set()
+
+    # Build a relationship between causes, cause ids, and cause parent names
+    causes_id_and_parents_by_name: dict[
+        str, list[tuple[int, Candidate]]
+    ] = collections.defaultdict(list)
+    causes_by_id = {id(c): c for c in causes}
+    for cause_id, cause in causes_by_id.items():
+        if cause.parent:
+            causes_id_and_parents_by_name[cause.parent.name].append(
+                (cause_id, cause.parent)
+            )
+
+    # From 1, check if each cause's specifier conflicts
+    # with another causes parent's version
+    for cause_id, cause in causes_by_id.items():
+        if cause_id in conflicting_ids:
+            continue
+
+        cause_id_and_parents = causes_id_and_parents_by_name.get(cause.requirement.name)
+        if not cause_id_and_parents:
+            continue
+
+        conflicting_alternative_cause_ids: set[int] = set()
+        for alternative_cause_id, parent in cause_id_and_parents:
+            if not cause.requirement.is_satisfied_by(parent):
+                conflicting_alternative_cause_ids.add(alternative_cause_id)
+
+        if conflicting_alternative_cause_ids:
+            conflicting_ids.add(cause_id)
+            conflicting_ids.update(conflicting_alternative_cause_ids)
+
+    # For comparing if two specifiers conflict first group causes
+    # by name, as comparing specifiers is O(n^2) so comparing the
+    # smaller groups is more efficent
+    causes_by_name: dict[str, list["PreferenceInformation"]] = collections.defaultdict(
+        list
+    )
+    for cause in causes:
+        causes_by_name[cause.requirement.name].append(cause)
+
+    # From 2, check if each cause's specifier conflicts
+    # with another cause specifier
+    for causes_list in causes_by_name.values():
+        if len(causes_list) < 2:
+            continue
+
+        while causes_list:
+            cause = causes_list.pop()
+            for i, alternative_cause in enumerate(causes_list):
+                candidate = cause.requirement.get_candidate_lookup()[1]
+                if candidate is None:
+                    continue
+                specifier = candidate.specifier
+
+                # Specifiers which provide no restrictions can be skipped
+                if len(specifier) == 0:
+                    continue
+
+                alternative_candidate = (
+                    alternative_cause.requirement.get_candidate_lookup()[1]
+                )
+                if alternative_candidate is None:
+                    continue
+
+                alternative_specifier = alternative_candidate.specifier
+
+                # Alternative specifiers which provide no
+                # restrictions can be skipped
+                if len(alternative_specifier) == 0:
+                    continue
+
+                # If intersection of specifiers are empty they are
+                # impossibe to fill and therefore conflicting
+                specifier_intersection = specifier and alternative_specifier
+                if len(specifier_intersection) == 0:
+                    conflicting_ids.add(id(cause))
+                    conflicting_ids.add(id(causes_list.pop(i)))
+
+    return [
+        cause for cause_id, cause in causes_by_id.items() if cause_id in conflicting_ids
+    ]
+
+
 class PipProvider(_ProviderBase):
     """Pip's provider implementation for resolvelib.
 
@@ -243,11 +337,18 @@ def filter_unsatisfied_names(
         causes: Sequence["PreferenceInformation"],
     ) -> Iterable[str]:
         """
-        Prefer backtracking on unsatisfied names that are causes
+        Prefer backtracking on unsatisfied names that are conficting
+        causes, or secondly are causes
         """
         if not causes:
             return unsatisfied_names
 
+        # Check if backtrack causes are conflicting and prefer them
+        if len(causes) > 2:
+            _conflicting_causes = conflicting_causes(causes)
+            if len(_conflicting_causes) > 1:
+                causes = _conflicting_causes
+
         # Extract the causes and parents names
         causes_names = set()
         for cause in causes: