Spaces:
Running
Running
File size: 25,409 Bytes
47b17cc | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 | """PrologEngine — Singleton wrapper around pyswip for X-bar syntactic analysis.
Thread safety: asyncio.Lock around all Prolog queries (pyswip is not thread-safe).
Loads all Prolog modules at init time. Uses serialize.pl for clean data marshalling
(pyswip returns compound terms as opaque strings; serialize.pl converts them to
nested lists of atoms that pyswip marshals cleanly to Python).
"""
from __future__ import annotations
import asyncio
import logging
from pathlib import Path
from typing import Any
from pyswip import Prolog
logger = logging.getLogger("grammar_engine")
_PROLOG_DIR = Path(__file__).resolve().parent.parent / "prolog"
_MODULES = [
"lexicon",
"agreement",
"subcategorization",
"binding",
"movement",
"isl_grammar",
"contrastive",
"xbar",
"tree_validation",
"compositional",
"chomsky_hierarchy",
"serialize",
]
def _flatten_cons_cells(node: Any) -> list | Any:
"""Recursively flatten pyswip cons cell representation.
pyswip marshals Prolog lists as nested ['[|]', head, tail] structures.
This converts them back to flat Python lists.
"""
if isinstance(node, list):
if len(node) == 3 and node[0] == '[|]':
head = _flatten_cons_cells(node[1])
tail = _flatten_cons_cells(node[2])
if isinstance(tail, list):
return [head] + tail
return [head]
return [_flatten_cons_cells(item) for item in node]
return node
def _extract_features(node: Any) -> dict[str, str] | None:
"""Extract feature dict from a Prolog feature list.
After flattening, feature lists look like:
[['=', 'person', 1], ['=', 'number', 'sg'], ...]
"""
if not isinstance(node, list):
return None
features: dict[str, str] = {}
for item in node:
if isinstance(item, list) and len(item) == 3 and item[0] == '=':
features[str(item[1])] = str(item[2])
if features:
return features
return None
def _nested_list_to_tree(node: Any) -> dict[str, Any] | None:
"""Convert serialize.pl nested list tree to dict.
Format from Prolog: [Label, Child1, Child2, ...]
Handles cons cell representation from pyswip.
"""
node = _flatten_cons_cells(node)
if node is None:
return None
if isinstance(node, str):
return {"label": node}
if isinstance(node, (int, float)):
return {"label": str(node)}
if isinstance(node, list) and len(node) >= 1:
label = str(node[0])
children: list[dict[str, Any]] = []
features: dict[str, str] | None = None
for arg in node[1:]:
if isinstance(arg, list):
# Check if it's a feature list [['=', k, v], ...]
feat = _extract_features(arg)
if feat:
features = feat
else:
child = _nested_list_to_tree(arg)
if child:
children.append(child)
elif isinstance(arg, str):
children.append({"label": arg})
elif isinstance(arg, (int, float)):
children.append({"label": str(arg)})
result: dict[str, Any] = {"label": label}
if children:
result["children"] = children
if features:
result["features"] = features
return result
return None
class PrologEngine:
"""Singleton wrapper around SWI-Prolog via pyswip."""
_instance: PrologEngine | None = None
_lock: asyncio.Lock
def __new__(cls) -> PrologEngine:
if cls._instance is None:
cls._instance = super().__new__(cls)
cls._instance._initialized = False
return cls._instance
def __init__(self) -> None:
if self._initialized:
return
self._prolog = Prolog()
self._lock = asyncio.Lock()
self._load_modules()
self._initialized = True
logger.info("PrologEngine initialized with %d modules", len(_MODULES))
def _load_modules(self) -> None:
for mod in _MODULES:
path = _PROLOG_DIR / f"{mod}.pl"
if not path.exists():
raise FileNotFoundError(f"Prolog module not found: {path}")
self._prolog.consult(str(path))
logger.debug("Loaded Prolog module: %s", mod)
# ------------------------------------------------------------------
# Public async methods
# ------------------------------------------------------------------
async def validate(self, gesture_ids: list[str]) -> dict[str, Any]:
async with self._lock:
return self._validate_sync(gesture_ids)
async def predict_next(self, gesture_ids: list[str]) -> list[dict[str, Any]]:
async with self._lock:
return self._predict_next_sync(gesture_ids)
async def detect_interference(self, gesture_ids: list[str]) -> list[dict[str, Any]]:
async with self._lock:
return self._detect_interference_sync(gesture_ids)
async def transform_isl_to_english(self, gesture_ids: list[str]) -> dict[str, Any]:
async with self._lock:
return self._transform_sync(gesture_ids)
async def get_parse_tree(self, gesture_ids: list[str]) -> dict[str, Any] | None:
async with self._lock:
return self._get_parse_tree_sync(gesture_ids)
async def compose_semantics(self, gesture_ids: list[str]) -> dict[str, Any]:
async with self._lock:
return self._compose_semantics_sync(gesture_ids)
async def get_grammar_capabilities(self) -> dict[str, Any]:
async with self._lock:
return self._get_grammar_capabilities_sync()
# ------------------------------------------------------------------
# Synchronous implementations (called under lock)
# ------------------------------------------------------------------
def _validate_sync(self, gesture_ids: list[str]) -> dict[str, Any]:
result: dict[str, Any] = {
"grammatical": False,
"parse_tree": None,
"agreement": None,
"theta": None,
"binding_violations": [],
"tense_resolution": "present",
"grammaticality_score": 0.0,
}
# 1. Parse tree
parse_tree = self._get_parse_tree_sync(gesture_ids)
if parse_tree:
result["parse_tree"] = parse_tree
result["grammatical"] = True
score = 1.0
else:
score = 0.3
# 2. Agreement
result["agreement"] = self._check_agreement_sync(gesture_ids)
if result["agreement"] and not result["agreement"].get("agrees", True):
score -= 0.2
# 3. Theta criterion
result["theta"] = self._check_theta_sync(gesture_ids)
if result["theta"] and not result["theta"].get("satisfied", True):
score -= 0.2
result["grammatical"] = False
# 4. Binding violations
result["binding_violations"] = self._check_binding_sync(gesture_ids)
if result["binding_violations"]:
score -= 0.1 * len(result["binding_violations"])
result["grammatical"] = False
# 5. Tense
result["tense_resolution"] = self._resolve_tense(gesture_ids)
result["grammaticality_score"] = max(0.0, min(1.0, score))
# 6. Compositional semantics (Frege's Principle)
result["semantics"] = self._compose_semantics_sync(gesture_ids)
return result
def _predict_next_sync(self, gesture_ids: list[str]) -> list[dict[str, Any]]:
valid_next: list[dict[str, Any]] = []
categories = self._categorize_ids(gesture_ids)
if not gesture_ids:
expected_cat = "d"
elif all(c == "subj" for c in categories):
expected_cat = "v"
elif "verb" in categories and "obj" not in categories:
expected_cat = "n"
else:
return []
if expected_cat == "d":
results = list(self._prolog.query(
"lexicon:lex(ID, Form, d, Feats), member(case=nom, Feats)"
))
# Deduplicate — pyswip may return multiple bindings
seen = set()
for r in results:
gid = str(r.get("ID", ""))
if gid not in seen:
seen.add(gid)
valid_next.append(self._build_lex_entry(gid, "d"))
elif expected_cat == "v":
results = list(self._prolog.query("lexicon:lex(ID, Form, v, _)"))
seen = set()
for r in results:
gid = str(r.get("ID", ""))
if gid not in seen:
seen.add(gid)
valid_next.append(self._build_lex_entry(gid, "v"))
elif expected_cat == "n":
results = list(self._prolog.query("lexicon:lex(ID, Form, n, _)"))
verb_id = self._find_verb(gesture_ids)
seen = set()
for r in results:
gid = str(r.get("ID", ""))
if gid not in seen:
seen.add(gid)
entry = self._build_lex_entry(gid, "n")
if verb_id:
role_results = list(self._prolog.query(
f"subcategorization:role_assignment({verb_id}, 2, Role)"
))
if role_results:
entry["theta_role"] = str(role_results[0].get("Role", ""))
valid_next.append(entry)
return valid_next
def _detect_interference_sync(self, gesture_ids: list[str]) -> list[dict[str, Any]]:
ids = self._to_prolog_list(gesture_ids)
query = f"serialize:serialize_interference({ids}, Patterns)"
results = list(self._prolog.query(query))
interferences: list[dict[str, Any]] = []
if results:
patterns = results[0].get("Patterns", [])
for p in patterns:
if isinstance(p, list) and len(p) >= 3:
interferences.append({
"type": str(p[0]),
"severity": str(p[1]),
"description": str(p[2]),
})
return interferences
def _transform_sync(self, gesture_ids: list[str]) -> dict[str, Any]:
ids = self._to_prolog_list(gesture_ids)
query = f"serialize:serialize_transform({ids}, Eng, TType, Ops)"
results = list(self._prolog.query(query))
if results:
r = results[0]
eng_order = r.get("Eng", gesture_ids)
if isinstance(eng_order, list):
eng_list = [str(x) for x in eng_order]
else:
eng_list = gesture_ids
transform_type = str(r.get("TType", "unknown"))
raw_ops = r.get("Ops", [])
operations: list[dict[str, str]] = []
if isinstance(raw_ops, list):
for op in raw_ops:
if isinstance(op, list) and len(op) >= 2:
operations.append({
"operation": str(op[0]),
"description": str(op[1]),
})
return {
"isl_order": gesture_ids,
"english_order": eng_list,
"transform": transform_type,
"movement_traces": operations,
}
return {
"isl_order": gesture_ids,
"english_order": gesture_ids,
"transform": "none",
"movement_traces": [],
}
def _get_parse_tree_sync(self, gesture_ids: list[str]) -> dict[str, Any] | None:
ids = self._to_prolog_list(gesture_ids)
query = f"serialize:serialize_tree({ids}, TreeList)"
results = list(self._prolog.query(query))
if results:
tree_list = results[0].get("TreeList")
if tree_list:
tree = _nested_list_to_tree(tree_list)
# Validate tree well-formedness (Partee et al., Ch 16)
tree_valid = self._validate_tree_sync(gesture_ids)
if tree and tree_valid is not None:
tree["_well_formed"] = tree_valid
return tree
return None
def _validate_tree_sync(self, gesture_ids: list[str]) -> dict[str, Any] | None:
"""Validate parse tree well-formedness using formal conditions.
Uses the raw Prolog parse tree (not serialized) for validation,
since well_formed_tree/1 operates on compound terms.
"""
ids = self._to_prolog_list(gesture_ids)
try:
# Query the raw parse tree and validate it directly
wf_query = (
f"xbar:parse_sentence({ids}, T, []), "
f"tree_validation:well_formed_tree(T)"
)
wf_results = list(self._prolog.query(wf_query))
is_well_formed = len(wf_results) > 0
# Get detailed report from raw tree
report_query = (
f"xbar:parse_sentence({ids}, T, []), "
f"tree_validation:validate_tree_structure(T, Report)"
)
report_results = list(self._prolog.query(report_query))
report: dict[str, Any] = {"well_formed": is_well_formed}
if report_results:
raw_report = report_results[0].get("Report", [])
raw_report = _flatten_cons_cells(raw_report)
if isinstance(raw_report, list):
for item in raw_report:
if isinstance(item, list) and len(item) == 2:
report[str(item[0])] = str(item[1]) if not isinstance(item[1], (int, float)) else item[1]
return report
except Exception as exc:
logger.warning("Tree validation failed: %s", exc)
return {"well_formed": False, "error": str(exc)}
# ------------------------------------------------------------------
# Agreement, Theta, Binding checks
# ------------------------------------------------------------------
def _check_agreement_sync(self, gesture_ids: list[str]) -> dict[str, Any] | None:
subj_id = None
verb_id = None
for gid in gesture_ids:
cat = self._get_category(gid)
if cat == "d" and subj_id is None:
subj_id = gid
elif cat == "v" and verb_id is None:
verb_id = gid
if not subj_id or not verb_id:
return None
query = (
f"serialize:serialize_agreement({subj_id}, {verb_id}, "
f"Agrees, InflForm, _)"
)
results = list(self._prolog.query(query))
if results:
r = results[0]
agrees = str(r.get("Agrees", "no")) == "yes"
inflected_form = str(r.get("InflForm", "unknown"))
return {
"agrees": agrees,
"inflected_form": inflected_form,
"subject_id": subj_id,
"verb_id": verb_id,
}
return None
def _check_theta_sync(self, gesture_ids: list[str]) -> dict[str, Any] | None:
verb_id = self._find_verb(gesture_ids)
if not verb_id:
return None
args = [gid for gid in gesture_ids if gid != verb_id]
args_list = self._to_prolog_list(args)
query = f"subcategorization:check_theta_criterion({verb_id}, {args_list}, Result)"
results = list(self._prolog.query(query))
if results:
result_str = str(results[0].get("Result", ""))
if result_str == "satisfied":
role_results = list(self._prolog.query(
f"lexicon:theta_grid({verb_id}, Roles)"
))
roles = []
if role_results:
roles_term = role_results[0].get("Roles", [])
if isinstance(roles_term, list):
roles = [str(r) for r in roles_term]
return {"satisfied": True, "roles": roles}
else:
# Parse violation(type, count) string
if "missing_args" in result_str:
return {"satisfied": False, "violation_type": "missing_args", "missing_count": 1}
elif "extra_args" in result_str:
return {"satisfied": False, "violation_type": "extra_args", "missing_count": 1}
return {"satisfied": False, "violation_type": "unknown", "missing_count": 0}
return None
def _check_binding_sync(self, gesture_ids: list[str]) -> list[dict[str, Any]]:
ids = self._to_prolog_list(gesture_ids)
query = f"binding:check_binding({ids}, local, Violations)"
results = list(self._prolog.query(query))
violations: list[dict[str, Any]] = []
if results:
raw = results[0].get("Violations", [])
if isinstance(raw, list):
for v in raw:
v_str = str(v)
if "principle_" in v_str:
# Parse violation(principle_x, id, message) string
violations.append({
"principle": "principle_b" if "principle_b" in v_str else
"principle_c" if "principle_c" in v_str else
"principle_a",
"gesture_id": "",
"message": v_str,
})
return violations
# ------------------------------------------------------------------
# Compositional Semantics (Partee et al., Ch 13)
# ------------------------------------------------------------------
def _compose_semantics_sync(self, gesture_ids: list[str]) -> dict[str, Any]:
"""Compute compositional semantic representation via lambda calculus."""
ids = self._to_prolog_list(gesture_ids)
try:
query = f"compositional:compose_sentence({ids}, Sem, Type)"
results = list(self._prolog.query(query))
if results:
r = results[0]
sem_raw = r.get("Sem", "unknown")
type_raw = r.get("Type", "unknown")
return {
"semantic_form": self._term_to_string(sem_raw),
"result_type": str(type_raw),
"complete": str(type_raw) == "t",
"gesture_ids": gesture_ids,
"gesture_types": self._get_semantic_types(gesture_ids),
}
return {
"semantic_form": "unknown",
"result_type": "unknown",
"complete": False,
"gesture_ids": gesture_ids,
"gesture_types": self._get_semantic_types(gesture_ids),
}
except Exception as exc:
logger.warning("Compositional semantics failed: %s", exc)
return {
"semantic_form": "error",
"result_type": "error",
"complete": False,
"error": str(exc),
}
def _get_semantic_types(self, gesture_ids: list[str]) -> list[dict[str, str]]:
"""Get semantic type for each gesture ID."""
types: list[dict[str, str]] = []
for gid in gesture_ids:
try:
results = list(self._prolog.query(
f"compositional:semantic_type({gid}, Type)"
))
if results:
types.append({
"gesture_id": gid,
"type": str(results[0].get("Type", "unknown")),
})
else:
types.append({"gesture_id": gid, "type": "unknown"})
except Exception:
types.append({"gesture_id": gid, "type": "error"})
return types
def _term_to_string(self, term: Any) -> str:
"""Convert a pyswip term to a readable string.
Cleans up pyswip's Functor representation and unwraps entity() wrappers.
"""
import re
if isinstance(term, (int, float)):
return str(term)
if isinstance(term, list):
flat = _flatten_cons_cells(term)
if isinstance(flat, list):
parts = [self._term_to_string(t) for t in flat]
return f"[{', '.join(parts)}]"
return self._term_to_string(flat)
# Handle pyswip Functor objects
if hasattr(term, 'name') and hasattr(term, 'args'):
name = str(term.name)
if name == 'entity' and term.args and len(list(term.args)) == 1:
return str(list(term.args)[0])
if term.args:
args_str = ", ".join(self._term_to_string(a) for a in term.args)
return f"{name}({args_str})"
return name
# String cleanup: pyswip sometimes returns stringified Functor refs
s = str(term)
# Clean up Functor(id,arity,name) → name
s = re.sub(r'Functor\(\d+,\d+,(\w+)\)', r'\1', s)
return s
# ------------------------------------------------------------------
# Chomsky Hierarchy (Partee et al., Ch 16)
# ------------------------------------------------------------------
def _get_grammar_capabilities_sync(self) -> dict[str, Any]:
"""Get MLAF's formal grammar classification report."""
try:
components: list[dict[str, Any]] = []
results = list(self._prolog.query(
"chomsky_hierarchy:grammar_class(Component, Type)"
))
for r in results:
comp = str(r.get("Component", ""))
typ = str(r.get("Type", ""))
level_results = list(self._prolog.query(
f"chomsky_hierarchy:chomsky_level({typ}, Level)"
))
level = int(level_results[0].get("Level", -1)) if level_results else -1
components.append({
"component": comp,
"chomsky_type": typ,
"level": level,
})
# Classify by level
by_level: dict[str, list[str]] = {
"type_3_regular": [],
"type_2_context_free": [],
"type_1_context_sensitive": [],
}
for c in components:
ct = c["chomsky_type"]
if ct in by_level:
by_level[ct].append(c["component"])
return {
"components": components,
"summary": {
"total": len(components),
"regular_count": len(by_level["type_3_regular"]),
"context_free_count": len(by_level["type_2_context_free"]),
"context_sensitive_count": len(by_level["type_1_context_sensitive"]),
},
"by_level": by_level,
"overall_power": "mildly_context_sensitive",
"note": "Natural languages are mildly context-sensitive (Joshi 1985). "
"MLAF uses CFG base + context-sensitive feature checking.",
}
except Exception as exc:
logger.warning("Grammar capabilities query failed: %s", exc)
return {"error": str(exc)}
# ------------------------------------------------------------------
# Helpers
# ------------------------------------------------------------------
def _to_prolog_list(self, ids: list[str]) -> str:
items = ", ".join(ids)
return f"[{items}]"
def _get_category(self, gid: str) -> str | None:
results = list(self._prolog.query(f"lexicon:lex({gid}, _, Cat, _)"))
if results:
return str(results[0].get("Cat", ""))
return None
def _find_verb(self, gesture_ids: list[str]) -> str | None:
for gid in gesture_ids:
if self._get_category(gid) == "v":
return gid
return None
def _build_lex_entry(self, gid: str, cat: str) -> dict[str, Any]:
results = list(self._prolog.query(f"lexicon:lex({gid}, Form, _, _)"))
form = str(results[0].get("Form", "")) if results else gid
return {
"grammar_id": gid,
"category": cat,
"phonological_form": form,
}
def _categorize_ids(self, gesture_ids: list[str]) -> list[str]:
categories = []
for gid in gesture_ids:
cat = self._get_category(gid)
if cat == "d":
categories.append("subj")
elif cat == "v":
categories.append("verb")
elif cat == "n":
categories.append("obj")
else:
categories.append("unknown")
return categories
def _resolve_tense(self, gesture_ids: list[str]) -> str:
for gid in gesture_ids:
results = list(self._prolog.query(
f"lexicon:lex({gid}, _, v, Feats), member(tense=T, Feats)"
))
if results:
return str(results[0].get("T", "present"))
return "present"
|