Spaces:
Sleeping
Sleeping
File size: 10,648 Bytes
dbc6675 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 | """
Shortest-path tokenizer for Instance Learning Graphs.
Builds a histogram-style embedding from:
1. Predicate occurrence features
2. Goal/state predicate interaction features
3. Object-graph shortest path length histograms (up to max_path_length)
"""
import json
import logging
import os
import re
import sys
from collections import defaultdict, deque
import numpy as np
from tqdm import tqdm
from code.tokenization.base import TokenizationStrategy
logger = logging.getLogger(__name__)
_PREDICATE_REGEX = re.compile(r"\(([\w-]+(?: [\w-]+)*)\)")
def _progress_enabled() -> bool:
return bool(sys.stdout.isatty())
def _normalize_atoms(raw_atoms: list[str]) -> list[str]:
"""Normalize atom strings by removing parentheses and trimming spaces."""
normalized: list[str] = []
for atom in raw_atoms:
if not atom:
continue
matches = _PREDICATE_REGEX.findall(atom)
if matches:
normalized.extend(m.strip() for m in matches if m.strip())
continue
clean = atom.replace("(", "").replace(")", "").strip()
if clean:
normalized.append(clean)
return normalized
class ShortestPathTokenizer(TokenizationStrategy):
"""
Shortest-path kernel style tokenizer.
The embedding is a fixed-length nonnegative histogram over learned feature keys.
"""
def __init__(self, max_path_length: int = 5):
super().__init__(name="ShortestPath")
self.max_path_length = max_path_length
self._feature_keys: list[str] | None = None
self._feature_to_idx: dict[str, int] | None = None
def fit(
self,
domain_pddl_path: str,
train_states_dir: str,
train_pddl_dir: str,
) -> None:
import pddl
domain = pddl.parse_domain(domain_pddl_path)
all_feature_keys: set[str] = set()
train_files = sorted(
[f for f in os.listdir(train_states_dir) if f.endswith(".traj")]
)
for t_file in tqdm(
train_files,
desc=f" [{self.name}] Collecting features",
disable=(not _progress_enabled()),
):
prob_name = t_file.replace(".traj", "")
prob_pddl = os.path.join(train_pddl_dir, f"{prob_name}.pddl")
traj_path = os.path.join(train_states_dir, t_file)
if not os.path.exists(prob_pddl):
continue
try:
problem = pddl.parse_problem(prob_pddl)
objects = sorted(
{o.name for o in problem.objects}
| {o.name for o in domain.constants}
)
goal_atoms = self._extract_goal_atoms(problem)
with open(traj_path, "r") as f:
lines = f.readlines()
for line in lines:
state_atoms = _normalize_atoms([line.strip()])
features = self._extract_features(state_atoms, goal_atoms, objects)
all_feature_keys.update(features.keys())
except Exception:
continue
# Ensure dimensionality grows with max_path_length.
for d in range(1, self.max_path_length + 1):
all_feature_keys.add(f"sp_len:{d}")
all_feature_keys.add(f"goal_sp_len:{d}")
if not all_feature_keys:
raise RuntimeError("No features collected during ShortestPath fit.")
self._feature_keys = sorted(all_feature_keys)
self._feature_to_idx = {k: i for i, k in enumerate(self._feature_keys)}
self.embedding_dim = len(self._feature_keys)
self._is_fitted = True
logger.info(
f"[{self.name}] Fitted: {self.embedding_dim} features "
f"(max_path_length={self.max_path_length})"
)
def _extract_goal_atoms(self, problem) -> list[str]:
"""Extract goal atoms from a parsed pddl Problem object."""
import pddl.logic.predicates
goals = []
def visit(node):
if isinstance(node, pddl.logic.predicates.Predicate):
args = [
t.name if hasattr(t, "name") else str(t) for t in node.terms
]
goals.append(f"{node.name} {' '.join(args)}")
elif hasattr(node, "operands"):
for op in node.operands:
visit(op)
elif hasattr(node, "_operands"):
for op in node._operands:
visit(op)
visit(problem.goal)
return goals
def _build_object_graph(
self,
atoms: list[str],
objects: list[str],
) -> dict[str, set[str]]:
"""
Build an undirected object graph from binary predicates.
Unary predicates are handled separately via feature counts.
"""
graph: dict[str, set[str]] = defaultdict(set)
for o in objects:
graph[o] # ensure key exists
for atom in atoms:
parts = atom.split()
if not parts:
continue
args = parts[1:]
if len(args) == 2:
a, b = args
graph[a].add(b)
graph[b].add(a)
elif len(args) == 1:
arg = args[0]
graph[arg] # ensure singleton nodes are present
return graph
def _shortest_path_lengths(
self,
graph: dict[str, set[str]],
start: str,
) -> dict[str, int]:
"""BFS shortest path lengths from one start node."""
dist = {start: 0}
queue = deque([start])
while queue:
cur = queue.popleft()
cur_d = dist[cur]
if cur_d >= self.max_path_length:
continue
for nxt in graph.get(cur, ()):
if nxt in dist:
continue
dist[nxt] = cur_d + 1
queue.append(nxt)
return dist
def _path_histogram(
self,
graph: dict[str, set[str]],
objects: list[str],
prefix: str,
) -> dict[str, float]:
"""Histogram of pair shortest-path lengths up to max_path_length."""
hist: dict[str, float] = {}
objs = list(dict.fromkeys(objects)) # stable unique
for i, src in enumerate(objs):
dmap = self._shortest_path_lengths(graph, src)
for dst in objs[i + 1 :]:
d = dmap.get(dst)
if d is None:
continue
if 1 <= d <= self.max_path_length:
key = f"{prefix}_sp_len:{d}"
hist[key] = hist.get(key, 0.0) + 1.0
return hist
def _extract_features(
self,
state_atoms: list[str],
goal_atoms: list[str],
objects: list[str],
) -> dict[str, float]:
"""Extract nonnegative histogram features."""
features: dict[str, float] = {}
# Predicate count features.
state_preds = []
for atom in state_atoms:
parts = atom.split()
if not parts:
continue
pred = parts[0].lower()
state_preds.append(pred)
features[f"state_pred:{pred}"] = features.get(f"state_pred:{pred}", 0.0) + 1.0
features[f"state_arity:{len(parts) - 1}"] = (
features.get(f"state_arity:{len(parts) - 1}", 0.0) + 1.0
)
goal_preds = []
for atom in goal_atoms:
parts = atom.split()
if not parts:
continue
pred = parts[0].lower()
goal_preds.append(pred)
features[f"goal_pred:{pred}"] = features.get(f"goal_pred:{pred}", 0.0) + 1.0
features[f"goal_arity:{len(parts) - 1}"] = (
features.get(f"goal_arity:{len(parts) - 1}", 0.0) + 1.0
)
state_set = set(state_preds)
goal_set = set(goal_preds)
for pred in sorted(state_set & goal_set):
features[f"state_goal_overlap:{pred}"] = 1.0
for pred in sorted(goal_set - state_set):
features[f"goal_missing:{pred}"] = 1.0
# Graph shortest-path histograms.
state_graph = self._build_object_graph(state_atoms, objects)
goal_graph = self._build_object_graph(goal_atoms, objects)
features.update(self._path_histogram(state_graph, objects, prefix=""))
features.update(self._path_histogram(goal_graph, objects, prefix="goal"))
# Structural scalars.
features["n_objects"] = float(len(objects))
features["n_state_atoms"] = float(len(state_atoms))
features["n_goal_atoms"] = float(len(goal_atoms))
return features
def _features_to_vector(self, features: dict[str, float]) -> np.ndarray:
vec = np.zeros(len(self._feature_keys), dtype=np.float32)
for key, value in features.items():
idx = self._feature_to_idx.get(key)
if idx is not None:
vec[idx] = float(value)
return vec
def transform_state(
self,
state_atoms: list[str],
goal_atoms: list[str],
objects: list[str],
) -> np.ndarray:
self._check_fitted()
parsed_state = _normalize_atoms(state_atoms)
parsed_goal = _normalize_atoms(goal_atoms)
features = self._extract_features(parsed_state, parsed_goal, objects)
return self._features_to_vector(features)
def transform_goal(
self,
goal_atoms: list[str],
objects: list[str],
) -> np.ndarray:
return self.transform_state([], goal_atoms, objects)
def get_embedding_dim(self) -> int:
self._check_fitted()
return int(self.embedding_dim)
def save_vocabulary(self, filepath: str) -> None:
self._check_fitted()
data = {
"feature_keys": self._feature_keys,
"max_path_length": self.max_path_length,
}
with open(filepath, "w") as f:
json.dump(data, f, indent=2)
logger.info(f"[{self.name}] Saved vocabulary to {filepath}")
def load_vocabulary(self, filepath: str) -> None:
with open(filepath, "r") as f:
data = json.load(f)
self._feature_keys = list(data["feature_keys"])
self._feature_to_idx = {k: i for i, k in enumerate(self._feature_keys)}
self.max_path_length = int(data.get("max_path_length", self.max_path_length))
self.embedding_dim = len(self._feature_keys)
self._is_fitted = True
logger.info(f"[{self.name}] Loaded vocabulary from {filepath}")
|