# v0.2.17 # { "Depends": "py-genlayer:1jb45aa8ynh2a9c9xn3b7qqh8sm5q93hwfp7jqmwsfhh8jpz09h6" } import datetime import json import typing from dataclasses import dataclass from genlayer import * # ============================================================================ # Constants — statuses, sides, limits # ============================================================================ # Market lifecycle statuses. Stored as small ints for cheap storage; # translated to strings at the view boundary. STATUS_OPEN: int = 0 # accepting stakes STATUS_RESOLVING: int = 1 # resolution requested, adjudication ran at least once STATUS_RESOLVED_YES: int = 2 # full chain verified — YES stakers win STATUS_RESOLVED_NO: int = 3 # chain definitively broken — NO stakers win STATUS_CANCELLED: int = 4 # cancelled before activity; stakes refundable STATUS_EXPIRED: int = 5 # deadline passed without full chain — NO wins STATUS_NAMES: dict[int, str] = { STATUS_OPEN: "OPEN", STATUS_RESOLVING: "RESOLVING", STATUS_RESOLVED_YES: "RESOLVED_YES", STATUS_RESOLVED_NO: "RESOLVED_NO", STATUS_CANCELLED: "CANCELLED", STATUS_EXPIRED: "EXPIRED", } # Per-step verification states. STEP_PENDING: int = 0 # not yet evaluated, or evidence inconclusive so far STEP_FULFILLED: int = 1 # verified to have occurred STEP_FAILED: int = 2 # verified NOT to have occurred / can no longer occur STEP_NAMES: dict[int, str] = { STEP_PENDING: "PENDING", STEP_FULFILLED: "FULFILLED", STEP_FAILED: "FAILED", } SIDE_YES: int = 1 SIDE_NO: int = 2 # Hard limits — sanity rails, generous enough not to constrain real use. MAX_STEPS_PER_MARKET: int = 12 MIN_STEPS_PER_MARKET: int = 2 MAX_SOURCES_PER_STEP: int = 5 MAX_TITLE_LEN: int = 200 MAX_DESCRIPTION_LEN: int = 2000 MAX_STEP_TEXT_LEN: int = 600 MAX_URL_LEN: int = 500 MAX_CATEGORY_LEN: int = 40 MAX_EVIDENCE_EXCERPT: int = 1200 # chars of rendered page fed to the LLM per source MAX_REASONING_STORED: int = 2000 # chars of reasoning persisted per step # Confidence is bucketed to a coarse 0..100 int; the DEFAULT floor a market # needs for a step to flip from PENDING to FULFILLED/FAILED. Coarse buckets # keep validators convergent. DEFAULT_CONFIDENCE_FLOOR: int = 70 MIN_CONFIDENCE_FLOOR: int = 55 MAX_CONFIDENCE_FLOOR: int = 95 # Fees are expressed in basis points (1/10000). DEFAULT_PROTOCOL_FEE_BPS: int = 100 # 1.0% of the losing pool DEFAULT_CREATOR_FEE_BPS: int = 50 # 0.5% of the losing pool MAX_TOTAL_FEE_BPS: int = 1000 # combined fees can never exceed 10% BPS_DENOMINATOR: int = 10000 # Error prefixes — deterministic, machine-parseable failure classes. ERR_EXPECTED = "EXPECTED: " # caller mistake (bad input, wrong state) ERR_EXTERNAL = "EXTERNAL: " # upstream/web failure ERR_TRANSIENT = "TRANSIENT: " # retryable condition ERR_LLM = "LLM_ERROR: " # model output unusable after sanitation DEFAULT_CATEGORIES: list[str] = [ "Technology", "Finance", "Geopolitics", "Science", "Sports", "Crypto", "Entertainment", "Climate", "Other", ] # ============================================================================ # Storage dataclasses # ============================================================================ @allow_storage @dataclass class ChainStep: """One condition in a market's causal chain. A step describes a single verifiable real-world condition, the public web sources validators should consult, and the current verification state plus an audit trail of the last adjudication. """ description: str # natural-language condition sources: DynArray[str] # public URLs to consult as evidence state: u8 # STEP_PENDING / STEP_FULFILLED / STEP_FAILED confidence: u8 # 0..100 bucketed confidence of last verdict reasoning: str # persisted transparent reasoning (truncated) evidence_summary: str # short summary of what evidence supported the verdict checked_at_ts: u64 # unix ts (caller-supplied clock) of last check check_count: u32 # how many adjudication passes touched this step @allow_storage @dataclass class Market: """A causal-chain prediction market.""" id: u64 creator: Address title: str description: str category: str created_ts: u64 deadline_ts: u64 # after this, an unresolved chain expires to NO status: u8 confidence_floor: u8 # per-market floor for step verdicts yes_pool: u256 # native units staked on YES no_pool: u256 # native units staked on NO creation_bond: u256 # bond posted by creator, returned on clean resolution resolved_ts: u64 resolution_summary: str # overall transparent reasoning of the final verdict fee_paid: bool # protocol+creator fee already carved from losing pool stake_count: u32 # number of stake actions (activity metric) steps_fulfilled: u32 # cached count of FULFILLED steps steps_failed: u32 # cached count of FAILED steps @allow_storage @dataclass class Position: """A user's stake in one market.""" yes_amount: u256 no_amount: u256 claimed: bool @allow_storage @dataclass class ActivityEvent: """Append-only activity log entry for a market (staking / lifecycle).""" kind: str # "CREATE" | "STAKE_YES" | "STAKE_NO" | "RESOLVE" | "CLAIM" | "CANCEL" | "EXPIRE" | "CHECK" actor: Address amount: u256 ts: u64 note: str # ============================================================================ # Pure helpers (deterministic — safe anywhere) # ============================================================================ def _require(cond: bool, message: str) -> None: """Deterministic guard that raises a user-facing EXPECTED error.""" if not cond: raise gl.vm.UserError(ERR_EXPECTED + message) def _clamp_int(value: int, low: int, high: int) -> int: """Clamp an int into [low, high].""" if value < low: return low if value > high: return high return value def _truncate(text: str, limit: int) -> str: """Trim text to a storage-safe length without exploding mid-codepoint.""" if len(text) <= limit: return text return text[: limit - 1] + "…" def _normalize_url(url: str) -> str: """Light URL validation/normalization. Rejects obviously invalid input.""" u = url.strip() _require(0 < len(u) <= MAX_URL_LEN, f"source URL must be 1..{MAX_URL_LEN} chars") _require( u.startswith("https://") or u.startswith("http://"), f"source URL must start with http(s):// — got '{u[:40]}'", ) return u def _confidence_bucket(raw: typing.Any) -> int: """Coerce arbitrary LLM confidence output into a coarse 0..100 bucket. Accepts ints, floats, numeric strings, and percentages expressed as 0..1 floats. Buckets to the nearest 5 so leader and validators that reason to slightly different confidences still converge. """ try: if isinstance(raw, bool): value = 100.0 if raw else 0.0 elif isinstance(raw, (int, float)): value = float(raw) elif isinstance(raw, str): cleaned = raw.strip().rstrip("%").strip() value = float(cleaned) else: value = 0.0 except (ValueError, TypeError): value = 0.0 if 0.0 < value <= 1.0: value *= 100.0 value = max(0.0, min(100.0, value)) return int(round(value / 5.0) * 5) def _coerce_bool(raw: typing.Any) -> typing.Optional[bool]: """Interpret common LLM spellings of a boolean. None if unparseable.""" if isinstance(raw, bool): return raw if isinstance(raw, (int, float)): if raw == 1: return True if raw == 0: return False return None if isinstance(raw, str): lowered = raw.strip().lower() if lowered in ("true", "yes", "y", "1", "fulfilled", "occurred", "happened", "confirmed"): return True if lowered in ("false", "no", "n", "0", "failed", "not occurred", "did not happen", "refuted"): return False return None def _first_present(payload: dict, keys: list[str]) -> typing.Any: """Return the first present key from a list of aliases, else None.""" for key in keys: if key in payload: return payload[key] return None def _sanitize_json_text(text: str) -> str: """Strip markdown fences and leading chatter around a JSON object.""" stripped = text.strip() if stripped.startswith("```"): # remove the first fence line and any trailing fence first_newline = stripped.find("\n") if first_newline != -1: stripped = stripped[first_newline + 1 :] if stripped.rstrip().endswith("```"): stripped = stripped.rstrip()[:-3] # fall back to the outermost braces if there is chatter around the object start = stripped.find("{") end = stripped.rfind("}") if start != -1 and end != -1 and end > start: stripped = stripped[start : end + 1] return stripped.strip() def _parse_verdict_payload(raw: typing.Any) -> dict: """Normalize an LLM verdict into {occurred, can_still_occur, confidence, reasoning, evidence_summary}. Tolerant of alias keys and stringly-typed values; raises LLM_ERROR only when the outcome itself is unrecoverable. """ payload: typing.Any = raw if isinstance(payload, str): try: payload = json.loads(_sanitize_json_text(payload)) except (json.JSONDecodeError, ValueError): raise gl.vm.UserError(ERR_LLM + "verdict was not parseable JSON") if not isinstance(payload, dict): raise gl.vm.UserError(ERR_LLM + "verdict JSON was not an object") occurred_raw = _first_present( payload, ["occurred", "happened", "fulfilled", "result", "verdict", "outcome", "answer"], ) occurred = _coerce_bool(occurred_raw) can_still_raw = _first_present( payload, ["can_still_occur", "still_possible", "possible", "pending_possible", "may_still_occur"], ) can_still = _coerce_bool(can_still_raw) if can_still is None: # Conservative default: if the model didn't say, assume the event # can still occur so we stay PENDING rather than flipping to FAILED. can_still = True confidence = _confidence_bucket( _first_present(payload, ["confidence", "confidence_pct", "certainty", "score"]) ) reasoning_raw = _first_present(payload, ["reasoning", "rationale", "explanation", "analysis"]) reasoning = str(reasoning_raw) if reasoning_raw is not None else "" evidence_raw = _first_present(payload, ["evidence_summary", "evidence", "sources_summary"]) evidence_summary = str(evidence_raw) if evidence_raw is not None else "" if occurred is None: # Unclear verdict — treat as low-confidence "not yet", never crash. occurred = False can_still = True confidence = 0 return { "occurred": occurred, "can_still_occur": can_still, "confidence": confidence, "reasoning": reasoning, "evidence_summary": evidence_summary, } # ============================================================================ # Native transfer target — payouts go to MetaMask wallets (EOAs), not other # Intelligent Contracts. gl.get_contract_at(...).emit_transfer(...) is the # IC-to-IC path and does not settle real balance for an externally-owned # account. The EOA/EVM path requires this contract-interface stub instead. # ============================================================================ @gl.evm.contract_interface class _Recipient: class View: pass class Write: pass def _send_gen(to_address: Address, amount: int) -> None: """Single emission choke point for every native-token payout. Zero the ledger field and persist state BEFORE calling this — never after — so a reentrant call always finds the balance already zeroed.""" if amount <= 0: return _Recipient(to_address).emit_transfer(value=u256(int(amount))) # ============================================================================ # The Contract # ============================================================================ class TrueChain(gl.Contract): """Causal-chain prediction markets with trustless web-evidence resolution and a real native-token value-transfer path.""" # ---- ownership / config ------------------------------------------------- owner: Address paused: bool protocol_fee_bps: u32 creator_fee_bps: u32 accrued_protocol_fees: u256 min_creation_bond: u256 min_stake: u256 # ---- market storage ----------------------------------------------------- market_count: u64 markets: TreeMap[u32, Market] market_steps: TreeMap[u32, DynArray[ChainStep]] # positions keyed by "marketId:0xaddress" positions: TreeMap[str, Position] # per-user list of market ids they have staked in user_markets: TreeMap[Address, DynArray[u64]] # per-market activity log activity: TreeMap[u32, DynArray[ActivityEvent]] # internal withdrawable native balances (credited by deposits, claims, # refunds, creator fees); withdraw() turns credits into real transfers. balances: TreeMap[Address, u256] # ---- platform metrics --------------------------------------------------- total_volume: u256 total_stakes: u64 total_resolved: u64 total_payouts: u256 # ------------------------------------------------------------------------ # Construction # ------------------------------------------------------------------------ def __init__( self, min_creation_bond: int = 0, min_stake: int = 0, ): """Deploy the platform. Args: min_creation_bond: native units a creator must attach to create_market (anti-spam). 0 disables the bond requirement. min_stake: minimum native units per stake action. 0 disables. """ self.owner = gl.message.sender_address self.paused = False self.protocol_fee_bps = u32(DEFAULT_PROTOCOL_FEE_BPS) self.creator_fee_bps = u32(DEFAULT_CREATOR_FEE_BPS) self.accrued_protocol_fees = u256(0) self.min_creation_bond = u256(max(0, min_creation_bond)) self.min_stake = u256(max(0, min_stake)) self.market_count = u64(0) self.total_volume = u256(0) self.total_stakes = u64(0) self.total_resolved = u64(0) self.total_payouts = u256(0) # ------------------------------------------------------------------------ # Internal utilities # ------------------------------------------------------------------------ def _not_paused(self) -> None: if self.paused: raise gl.vm.UserError(ERR_EXPECTED + "platform is paused") def _only_owner(self) -> None: if gl.message.sender_address != self.owner: raise gl.vm.UserError(ERR_EXPECTED + "only the owner may call this") def _now_ts(self) -> int: """Authenticated, consensus-agreed clock. GenVM patches datetime.now() to the network's block time, which every validator computes identically — it is never read from caller-supplied arguments or calldata, so it cannot be spoofed by a transaction sender to fabricate a future or stale time.""" return int(datetime.datetime.now(datetime.timezone.utc).timestamp()) def _require_adjudication_rights(self, market: Market, now_ts: int) -> None: """Adjudication policy: before the market deadline only the market creator or the platform owner may trigger step checks (progressive verification of long chains). After the deadline adjudication is fully permissionless — anyone (including the platform's automatic resolver) may trigger it, and validators decide trustlessly.""" if now_ts > int(market.deadline_ts): return sender = gl.message.sender_address _require( sender == market.creator or sender == self.owner, "before the deadline only the market creator or owner may trigger adjudication", ) def _get_market(self, market_id: int) -> Market: mid = u32(market_id) market = self.markets.get(mid) if market is None: raise gl.vm.UserError(ERR_EXPECTED + f"market {market_id} does not exist") return market def _position_key(self, market_id: int, addr: Address) -> str: return f"{market_id}:{addr.as_hex}" def _get_or_create_position(self, market_id: int, addr: Address) -> Position: key = self._position_key(market_id, addr) pos = self.positions.get(key) if pos is None: self.positions[key] = Position( yes_amount=u256(0), no_amount=u256(0), claimed=False ) pos = self.positions[key] return pos def _credit_balance(self, addr: Address, amount: int) -> None: """Credit an internal withdrawable balance (value stays in-contract until withdraw() emits the real native transfer).""" if amount <= 0: return current = self.balances.get(addr) base = int(current) if current is not None else 0 self.balances[addr] = u256(base + int(amount)) def _log(self, market_id: int, kind: str, actor: Address, amount: int, ts: int, note: str) -> None: """Append to the market's activity log.""" mid = u32(market_id) if self.activity.get(mid) is None: self.activity[mid] = [] self.activity[mid].append( ActivityEvent( kind=kind, actor=actor, amount=u256(max(0, amount)), ts=u64(max(0, ts)), note=_truncate(note, 200), ) ) def _record_user_market(self, addr: Address, market_id: int) -> None: """Track which markets a user has touched (idempotent-ish; we accept duplicates being filtered at the view layer to keep writes cheap).""" if self.user_markets.get(addr) is None: self.user_markets[addr] = [] arr = self.user_markets[addr] mid = u32(market_id) # small-scan dedupe; user market lists are short in practice for existing in arr: if existing == mid: return arr.append(mid) def _send_native(self, recipient: Address, amount: int) -> None: """Emit an actual native-token transfer from this contract to `recipient` (a wallet/EOA). Outbound half of the value-transfer path. Delegates to the module-level `_send_gen` choke point — callers must zero and persist the ledger field(s) BEFORE invoking this.""" _send_gen(recipient, amount) # ------------------------------------------------------------------------ # Market serialization for views (schema-safe primitives only) # ------------------------------------------------------------------------ def _step_dict(self, step: ChainStep, index: int) -> dict: return { "index": index, "description": step.description, "sources": [s for s in step.sources], "state": STEP_NAMES.get(int(step.state), "PENDING"), "confidence": int(step.confidence), "reasoning": step.reasoning, "evidence_summary": step.evidence_summary, "checked_at_ts": int(step.checked_at_ts), "check_count": int(step.check_count), } def _market_dict(self, market: Market, include_steps: bool) -> dict: yes_pool = int(market.yes_pool) no_pool = int(market.no_pool) total = yes_pool + no_pool # implied probability from pool ratio, in basis points implied_yes_bps = (yes_pool * BPS_DENOMINATOR) // total if total > 0 else 5000 result = { "id": int(market.id), "creator": market.creator.as_hex, "title": market.title, "description": market.description, "category": market.category, "created_ts": int(market.created_ts), "deadline_ts": int(market.deadline_ts), "status": STATUS_NAMES.get(int(market.status), "OPEN"), "confidence_floor": int(market.confidence_floor), "yes_pool": yes_pool, "no_pool": no_pool, "total_pool": total, "implied_yes_bps": implied_yes_bps, "creation_bond": int(market.creation_bond), "resolved_ts": int(market.resolved_ts), "resolution_summary": market.resolution_summary, "stake_count": int(market.stake_count), "steps_fulfilled": int(market.steps_fulfilled), "steps_failed": int(market.steps_failed), "step_count": len(self.market_steps[u32(int(market.id))]) if self.market_steps.get(u32(int(market.id))) is not None else 0, } if include_steps: steps = self.market_steps.get(u32(int(market.id))) result["steps"] = ( [self._step_dict(step, i) for i, step in enumerate(steps)] if steps is not None else [] ) return result # ------------------------------------------------------------------------ # Non-deterministic adjudication core # ------------------------------------------------------------------------ def _fetch_evidence(self, sources: list[str]) -> list[dict]: """Fetch each evidence URL defensively inside the nondet block. Runs INSIDE a leader/validator function — never call from deterministic code. A dead or slow source yields an error record instead of aborting adjudication. """ evidence: list[dict] = [] for url in sources[:MAX_SOURCES_PER_STEP]: try: text = gl.nondet.web.render(url, mode="text") excerpt = str(text)[:MAX_EVIDENCE_EXCERPT] evidence.append({"url": url, "ok": True, "excerpt": excerpt}) except Exception as exc: # noqa: BLE001 — degrade per-source evidence.append({"url": url, "ok": False, "excerpt": f"[fetch failed: {str(exc)[:120]}]"}) return evidence def _build_step_prompt( self, market_title: str, step_description: str, step_index: int, prior_context: str, deadline_ts: int, now_ts: int, evidence: list[dict], ) -> str: """Compose the adjudication prompt for one chain step.""" evidence_blocks = [] for item in evidence: status = "OK" if item["ok"] else "FETCH_FAILED" evidence_blocks.append(f"--- SOURCE ({status}): {item['url']}\n{item['excerpt']}") evidence_text = "\n\n".join(evidence_blocks) if evidence_blocks else "(no sources provided)" return f"""You are a neutral adjudicator for a causal-chain prediction market called "{market_title}". You must decide whether ONE specific condition in the chain has occurred in the real world, based ONLY on the evidence excerpts below plus widely-known public facts. Do not speculate beyond the evidence. CONDITION (step {step_index + 1} of the chain): "{step_description}" CONTEXT FROM EARLIER CHAIN STEPS: {prior_context if prior_context else "(this is the first step)"} TIMING: current unix time is {now_ts}; the market deadline is unix time {deadline_ts}. EVIDENCE: {evidence_text} Respond with ONLY a JSON object, no markdown, with exactly these keys: {{ "occurred": true or false — has the condition verifiably occurred? "can_still_occur": true or false — if it has not occurred, is it still possible before the deadline? "confidence": integer 0-100 — your confidence in the "occurred" answer, "reasoning": one short paragraph explaining the verdict, "evidence_summary": one sentence naming which evidence supported the verdict }} Rules: - "occurred" must be true ONLY with concrete supporting evidence. - If evidence is missing or ambiguous, set occurred=false, can_still_occur=true, confidence below 50. - If the deadline has passed and it did not occur, set can_still_occur=false. - Keep reasoning under 120 words.""" def _adjudicate_step_nondet( self, market_title: str, step_description: str, step_index: int, sources: list[str], prior_context: str, deadline_ts: int, now_ts: int, confidence_floor: int, ) -> dict: """Run the full web-fetch + LLM verdict for one step under a comparative equivalence principle. The principle is deliberately outcome-focused and tolerant: the leader's result stands as long as validators agree on the verdict booleans and rough confidence band, regardless of prose wording. This prevents leader rotation / Undetermined results caused by immaterial differences in fetched page snapshots. """ def leader() -> str: evidence = self._fetch_evidence(sources) prompt = self._build_step_prompt( market_title, step_description, step_index, prior_context, deadline_ts, now_ts, evidence, ) raw = gl.nondet.exec_prompt(prompt, response_format="json") verdict = _parse_verdict_payload(raw) # Return a compact canonical JSON string — comparative NLP # equivalence then compares meaning, not bytes. return json.dumps( { "occurred": verdict["occurred"], "can_still_occur": verdict["can_still_occur"], "confidence": verdict["confidence"], "reasoning": _truncate(verdict["reasoning"], 800), "evidence_summary": _truncate(verdict["evidence_summary"], 300), }, sort_keys=True, ) principle = ( "Both results are JSON verdicts about whether the same real-world " "condition occurred. Treat them as equivalent if they agree on the " "boolean value of 'occurred' AND on the boolean value of " "'can_still_occur'. " f"CRITICALLY IMPORTANT: Both confidences must either BOTH be >= {confidence_floor} " f"or BOTH be < {confidence_floor}. If one is above the floor and one is below, " "they are NOT equivalent. Differences in wording of 'reasoning' or " "'evidence_summary' are irrelevant." ) raw_result = gl.eq_principle.prompt_comparative(leader, principle) return _parse_verdict_payload(raw_result) def _apply_verdict_to_step(self, market: Market, step: ChainStep, verdict: dict, now_ts: int) -> None: """Fold an adjudication verdict into a step's stored state.""" floor = int(market.confidence_floor) confidence = int(verdict["confidence"]) occurred = bool(verdict["occurred"]) can_still = bool(verdict["can_still_occur"]) previous_state = int(step.state) new_state = previous_state if occurred and confidence >= floor: new_state = STEP_FULFILLED elif (not occurred) and (not can_still) and confidence >= floor: new_state = STEP_FAILED # Otherwise: stay PENDING — inconclusive evidence never flips a step. # A fulfilled step is sticky: real-world events don't un-happen. A # later low-confidence pass cannot revert it. if previous_state == STEP_FULFILLED: new_state = STEP_FULFILLED if new_state != previous_state: if new_state == STEP_FULFILLED: market.steps_fulfilled = u32(int(market.steps_fulfilled) + 1) elif new_state == STEP_FAILED: market.steps_failed = u32(int(market.steps_failed) + 1) step.state = u8(new_state) step.confidence = u8(_clamp_int(confidence, 0, 100)) step.reasoning = _truncate(str(verdict["reasoning"]), MAX_REASONING_STORED) step.evidence_summary = _truncate(str(verdict["evidence_summary"]), 400) step.checked_at_ts = u64(max(0, now_ts)) step.check_count = u32(int(step.check_count) + 1) def _maybe_finalize(self, market: Market, now_ts: int) -> None: """Flip a RESOLVING market to a terminal status when the chain is decided: all steps fulfilled → YES; any step failed → NO; deadline passed with unfulfilled steps → EXPIRED (NO side wins).""" steps = self.market_steps[u32(int(market.id))] total_steps = len(steps) fulfilled = int(market.steps_fulfilled) failed = int(market.steps_failed) if failed > 0: market.status = u8(STATUS_RESOLVED_NO) elif fulfilled == total_steps and total_steps > 0: market.status = u8(STATUS_RESOLVED_YES) elif now_ts > int(market.deadline_ts): market.status = u8(STATUS_EXPIRED) else: return # still undecided — stays RESOLVING market.resolved_ts = u64(max(0, now_ts)) self.total_resolved = u64(int(self.total_resolved) + 1) outcome = STATUS_NAMES[int(market.status)] parts = [f"Chain verdict: {outcome}."] for i, step in enumerate(steps): parts.append( f"Step {i + 1} [{STEP_NAMES[int(step.state)]}, conf {int(step.confidence)}%]: " f"{_truncate(step.reasoning, 240)}" ) market.resolution_summary = _truncate(" ".join(parts), 4000) self._log(int(market.id), "RESOLVE", gl.message.sender_address, 0, now_ts, outcome) self._settle_fees(market) def _settle_fees(self, market: Market) -> None: """Carve protocol + creator fees out of the losing pool once, and return the creator's bond. Credits flow through internal balances and become real native transfers on withdraw().""" if market.fee_paid: return status = int(market.status) if status == STATUS_RESOLVED_YES: losing_pool = int(market.no_pool) elif status in (STATUS_RESOLVED_NO, STATUS_EXPIRED): losing_pool = int(market.yes_pool) else: return protocol_fee = (losing_pool * int(self.protocol_fee_bps)) // BPS_DENOMINATOR creator_fee = (losing_pool * int(self.creator_fee_bps)) // BPS_DENOMINATOR self.accrued_protocol_fees = u256(int(self.accrued_protocol_fees) + protocol_fee) self._credit_balance(market.creator, creator_fee) # bond returned on any clean terminal resolution self._credit_balance(market.creator, int(market.creation_bond)) market.fee_paid = True def _losing_pool_after_fees(self, market: Market) -> tuple[int, int]: """Return (winning_pool, distributable_losing_pool_after_fees).""" status = int(market.status) if status == STATUS_RESOLVED_YES: winning, losing = int(market.yes_pool), int(market.no_pool) else: winning, losing = int(market.no_pool), int(market.yes_pool) total_fee_bps = int(self.protocol_fee_bps) + int(self.creator_fee_bps) distributable = losing - (losing * total_fee_bps) // BPS_DENOMINATOR return winning, max(0, distributable) # ======================================================================== # PUBLIC WRITES — market lifecycle # ======================================================================== @gl.public.write.payable def create_market( self, title: str, description: str, category: str, steps_json: str, deadline_ts: int, confidence_floor: int = DEFAULT_CONFIDENCE_FLOOR, ) -> int: """Create a causal-chain market. Attach at least `min_creation_bond` native units as an anti-spam bond (returned on resolution). Args: title / description / category: market metadata. steps_json: JSON array of {"description": str, "sources": [url,...]} objects, ordered — the causal chain. 2..12 steps. deadline_ts: unix time after which an unresolved chain expires. confidence_floor: 55..95, per-step confidence needed to flip a step's state. Returns: the new market id. """ self._not_paused() sender = gl.message.sender_address bond = int(gl.message.value) now_ts = self._now_ts() _require(bond >= int(self.min_creation_bond), "creation bond below minimum") _require(0 < len(title.strip()) <= MAX_TITLE_LEN, f"title must be 1..{MAX_TITLE_LEN} chars") _require(len(description) <= MAX_DESCRIPTION_LEN, f"description exceeds {MAX_DESCRIPTION_LEN} chars") _require(0 < len(category.strip()) <= MAX_CATEGORY_LEN, "category required") _require(deadline_ts > now_ts, "deadline must be in the future") floor = _clamp_int(int(confidence_floor), MIN_CONFIDENCE_FLOOR, MAX_CONFIDENCE_FLOOR) # Parse and validate the chain definition. if isinstance(steps_json, (list, tuple)): raw_steps = list(steps_json) else: try: raw_steps = json.loads(steps_json) except (json.JSONDecodeError, ValueError, TypeError): raise gl.vm.UserError(ERR_EXPECTED + "steps_json is not valid JSON") _require(isinstance(raw_steps, list), "steps_json must be a JSON array") _require( MIN_STEPS_PER_MARKET <= len(raw_steps) <= MAX_STEPS_PER_MARKET, f"chain must have {MIN_STEPS_PER_MARKET}..{MAX_STEPS_PER_MARKET} steps", ) parsed_steps: list[tuple[str, list[str]]] = [] for i, raw in enumerate(raw_steps): _require(isinstance(raw, dict), f"step {i} must be an object") desc = str(raw.get("description", "")).strip() _require( 0 < len(desc) <= MAX_STEP_TEXT_LEN, f"step {i} description must be 1..{MAX_STEP_TEXT_LEN} chars", ) raw_sources = raw.get("sources", []) _require(isinstance(raw_sources, list), f"step {i} sources must be an array") _require( 1 <= len(raw_sources) <= MAX_SOURCES_PER_STEP, f"step {i} needs 1..{MAX_SOURCES_PER_STEP} evidence source URLs", ) sources = [_normalize_url(str(u)) for u in raw_sources] parsed_steps.append((desc, sources)) market_id = int(self.market_count) self.market_count = u64(market_id + 1) mid = u32(market_id) self.markets[mid] = Market( id=mid, creator=sender, title=title.strip(), description=description.strip(), category=category.strip(), created_ts=u64(now_ts), deadline_ts=u64(deadline_ts), status=u8(STATUS_OPEN), confidence_floor=u8(floor), yes_pool=u256(0), no_pool=u256(0), creation_bond=u256(bond), resolved_ts=u64(0), resolution_summary="", fee_paid=False, stake_count=u32(0), steps_fulfilled=u32(0), steps_failed=u32(0), ) self.market_steps[mid] = [] steps_arr = self.market_steps[mid] for desc, sources in parsed_steps: steps_arr.append( ChainStep( description=desc, sources=sources, state=u8(STEP_PENDING), confidence=u8(0), reasoning="", evidence_summary="", checked_at_ts=u64(0), check_count=u32(0), ) ) self._record_user_market(sender, market_id) self._log(market_id, "CREATE", sender, bond, now_ts, title.strip()[:100]) return market_id def _stake(self, market_id: int, side: int, amount: int) -> None: """Shared staking core for payable and balance-funded stakes.""" self._not_paused() now_ts = self._now_ts() market = self._get_market(market_id) _require( int(market.status) in (STATUS_OPEN, STATUS_RESOLVING), "market is not open for staking", ) _require(now_ts <= int(market.deadline_ts), "market deadline has passed") _require(amount > 0, "stake amount must be positive") _require(amount >= int(self.min_stake), "stake below minimum") sender = gl.message.sender_address pos = self._get_or_create_position(market_id, sender) if side == SIDE_YES: market.yes_pool = u256(int(market.yes_pool) + amount) pos.yes_amount = u256(int(pos.yes_amount) + amount) kind = "STAKE_YES" else: market.no_pool = u256(int(market.no_pool) + amount) pos.no_amount = u256(int(pos.no_amount) + amount) kind = "STAKE_NO" market.stake_count = u32(int(market.stake_count) + 1) self.total_volume = u256(int(self.total_volume) + amount) self.total_stakes = u64(int(self.total_stakes) + 1) self._record_user_market(sender, market_id) self._log(market_id, kind, sender, amount, now_ts, "") @gl.public.write.payable def stake_yes(self, market_id: int) -> None: """Stake attached native value on the full chain occurring. This IS a real value transfer: the chain moves gl.message.value from the caller into the contract before this body runs.""" self._stake(market_id, SIDE_YES, int(gl.message.value)) @gl.public.write.payable def stake_no(self, market_id: int) -> None: """Stake attached native value on the chain NOT fully occurring.""" self._stake(market_id, SIDE_NO, int(gl.message.value)) @gl.public.write def stake_from_balance(self, market_id: int, side: str, amount: int) -> None: """Stake using previously deposited / won internal balance instead of attaching new value.""" sender = gl.message.sender_address side_normalized = side.strip().upper() _require(side_normalized in ("YES", "NO"), "side must be 'YES' or 'NO'") try: amount = int(amount) except (ValueError, TypeError): raise gl.vm.UserError(ERR_EXPECTED + "amount must be an integer") current = self.balances.get(sender) available = int(current) if current is not None else 0 _require(amount > 0 and amount <= available, "insufficient internal balance") self.balances[sender] = u256(available - amount) self._stake( market_id, SIDE_YES if side_normalized == "YES" else SIDE_NO, amount, ) # ======================================================================== # PUBLIC WRITES — value transfer path (deposit / withdraw / claim) # ======================================================================== @gl.public.write.payable def deposit(self) -> None: """Deposit native tokens into an internal withdrawable balance. Inbound half of the value-transfer path.""" amount = int(gl.message.value) _require(amount > 0, "attach a positive value to deposit") self._credit_balance(gl.message.sender_address, amount) @gl.public.write def withdraw(self, amount: int) -> None: """Withdraw internal balance as a REAL native-token transfer from the contract to the caller, emitted on finalization. Outbound half of the value-transfer path.""" sender = gl.message.sender_address try: amount = int(amount) # tolerate stringly-typed calldata amounts except (ValueError, TypeError): raise gl.vm.UserError(ERR_EXPECTED + "amount must be an integer") current = self.balances.get(sender) available = int(current) if current is not None else 0 _require(amount > 0, "withdraw amount must be positive") _require(amount <= available, f"insufficient balance: have {available}") self.balances[sender] = u256(available - amount) self.total_payouts = u256(int(self.total_payouts) + amount) self._send_native(sender, amount) @gl.public.write def claim(self, market_id: int) -> int: """After terminal resolution, claim winnings: the caller's winning stake back plus a pro-rata share of the losing pool (after fees). Credits the internal balance; call withdraw() to move tokens out. Returns the credited amount. """ now_ts = self._now_ts() market = self._get_market(market_id) status = int(market.status) _require( status in (STATUS_RESOLVED_YES, STATUS_RESOLVED_NO, STATUS_EXPIRED), "market is not resolved yet", ) sender = gl.message.sender_address key = self._position_key(market_id, sender) pos = self.positions.get(key) _require(pos is not None, "no position in this market") _require(not pos.claimed, "already claimed") winning_pool, distributable = self._losing_pool_after_fees(market) my_winning_stake = ( int(pos.yes_amount) if status == STATUS_RESOLVED_YES else int(pos.no_amount) ) _require(my_winning_stake > 0, "no winning stake to claim") share = (distributable * my_winning_stake) // winning_pool if winning_pool > 0 else 0 payout = my_winning_stake + share pos.claimed = True self._credit_balance(sender, payout) self._log(market_id, "CLAIM", sender, payout, now_ts, "") return payout @gl.public.write def refund_cancelled(self, market_id: int) -> int: """Reclaim full stakes from a cancelled market. Returns amount credited.""" now_ts = self._now_ts() market = self._get_market(market_id) _require(int(market.status) == STATUS_CANCELLED, "market is not cancelled") sender = gl.message.sender_address key = self._position_key(market_id, sender) pos = self.positions.get(key) _require(pos is not None, "no position in this market") _require(not pos.claimed, "already refunded") refund = int(pos.yes_amount) + int(pos.no_amount) _require(refund > 0, "nothing to refund") pos.claimed = True self._credit_balance(sender, refund) self._log(market_id, "CLAIM", sender, refund, now_ts, "refund") return refund @gl.public.write def cancel_market(self, market_id: int) -> None: """Cancel an open market. Creator may cancel only while it has no stakes; the owner may cancel any open/resolving market (stakes become refundable, bond returned).""" now_ts = self._now_ts() market = self._get_market(market_id) sender = gl.message.sender_address status = int(market.status) _require(status in (STATUS_OPEN, STATUS_RESOLVING), "market cannot be cancelled") if sender == market.creator and sender != self.owner: _require(int(market.stake_count) == 0, "creator can only cancel before any stakes") else: self._only_owner() market.status = u8(STATUS_CANCELLED) market.resolved_ts = u64(max(0, now_ts)) market.resolution_summary = "Cancelled; all stakes refundable via refund_cancelled()." self._credit_balance(market.creator, int(market.creation_bond)) market.fee_paid = True # no fees on cancellation self._log(market_id, "CANCEL", sender, 0, now_ts, "") # ======================================================================== # PUBLIC WRITES — trustless resolution # ======================================================================== @gl.public.write def check_step(self, market_id: int, step_index: int) -> dict: """Adjudicate a single chain step against live web evidence. Anyone may call this — resolution is permissionless. The step's verdict is decided by validator consensus over web evidence and LLM reasoning, then folded into market state. Finalizes the market when the chain becomes decided. Returns the step's post-adjudication view dict. """ self._not_paused() now_ts = self._now_ts() market = self._get_market(market_id) status = int(market.status) _require(status in (STATUS_OPEN, STATUS_RESOLVING), "market is not resolvable") self._require_adjudication_rights(market, now_ts) steps = self.market_steps[u32(market_id)] _require(0 <= step_index < len(steps), "step index out of range") step = steps[step_index] # Steps must be verified in order: everything before must be fulfilled. for i in range(step_index): _require( int(steps[i].state) == STEP_FULFILLED, f"step {i + 1} must be fulfilled before checking step {step_index + 1}", ) if status == STATUS_OPEN: market.status = u8(STATUS_RESOLVING) if int(step.state) == STEP_FULFILLED: # Nothing to do — sticky fulfilled state. self._maybe_finalize(market, now_ts) return self._step_dict(step, step_index) prior_context = " ".join( f"Step {i + 1} ('{_truncate(steps[i].description, 120)}') was verified FULFILLED." for i in range(step_index) ) verdict = self._adjudicate_step_nondet( market.title, step.description, step_index, [s for s in step.sources], prior_context, int(market.deadline_ts), now_ts, int(market.confidence_floor), ) self._apply_verdict_to_step(market, step, verdict, now_ts) self._log( market_id, "CHECK", gl.message.sender_address, 0, now_ts, f"step {step_index + 1}: {STEP_NAMES[int(step.state)]} @{int(step.confidence)}%", ) self._maybe_finalize(market, now_ts) return self._step_dict(step, step_index) @gl.public.write def request_resolution(self, market_id: int) -> dict: """Run a full adjudication pass over the chain: walks steps in order, adjudicating each pending step until one fails to fulfill (chain order matters). Finalizes if the chain becomes decided. Returns the market view dict after the pass. """ self._not_paused() now_ts = self._now_ts() market = self._get_market(market_id) status = int(market.status) _require(status in (STATUS_OPEN, STATUS_RESOLVING), "market is not resolvable") self._require_adjudication_rights(market, now_ts) if status == STATUS_OPEN: market.status = u8(STATUS_RESOLVING) steps = self.market_steps[u32(market_id)] prior_context = "" for index in range(len(steps)): step = steps[index] state = int(step.state) if state == STEP_FULFILLED: prior_context += ( f"Step {index + 1} ('{_truncate(step.description, 120)}') was verified FULFILLED. " ) continue if state == STEP_FAILED: break verdict = self._adjudicate_step_nondet( market.title, step.description, index, [s for s in step.sources], prior_context, int(market.deadline_ts), now_ts, int(market.confidence_floor), ) self._apply_verdict_to_step(market, step, verdict, now_ts) self._log( market_id, "CHECK", gl.message.sender_address, 0, now_ts, f"step {index + 1}: {STEP_NAMES[int(step.state)]} @{int(step.confidence)}%", ) if int(step.state) != STEP_FULFILLED: break self._maybe_finalize(market, now_ts) return self._market_dict(market, include_steps=True)