aniva/Pantograph
aniva
/
Pantograph
1
1
Fork 0
Code Issues 13 Pull Requests 4 Packages Projects Releases Wiki Activity

feat: add state specific tactic feedback

This commit is contained in:
Simon 2024-10-29 14:16:38 +00:00 committed by Leni Aniva
parent 93ecd0d5ad
commit b89f3a19bf
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
pantograph/search.py
View File

@ -15,6 +15,7 @@ class SearchState:
parent: Optional[int]
parent_goal_id: Optional[int]
priorities: list[float]
tactic_feedback: Optional[str] = None
def __post_init__(self):
assert len(self.priorities) == len(self.state.goals)
@ -48,7 +49,6 @@ class Agent:
"""
An agent interface for proof search
"""
tactic_feedback: Optional[str] = None
@abstractmethod
def next_tactic(
@ -126,7 +126,7 @@ class Agent:
print(f"Next tactic: {tactic}")
if not tactic:
# resets the feedback
self.tactic_feedback = None
search_state.tactic_feedback = None
# pop the current state and continue to the next
search_stack.pop(-1)
if not search_stack:
@ -163,7 +163,7 @@ class Agent:
except TacticFailure as t:
if verbose:
print(f"Tactic failed: {t}")
self.tactic_feedback = str(t)
search_state.tactic_feedback = str(t)
# try the next tactic. this one failed
except ServerError as e:
raise RuntimeError(f"While executing tactic: {tactic}") from e