tla-ts 0.0.1 → 0.0.2

package/src/parser.ts

@@ -1,24 +1,29 @@
 import type {
+  AssumeProve,
   Assumption,
   Bound,
   BoundPattern,
   CaseArm,
   Declaration,
+  DefineStep,
   Definition,
   ExceptPathStep,
-  ExceptUpdate,
   Expression,
+  Fact,
   Module,
   Instance,
-  OuterScope,
-  InstanceSubstitution,
+  NewSymbol,
   OperatorParam,
-  RecordField,
+  Proof,
+  ProofStep,
   Selector,
   SourceLocation,
+  SourceRange,
+  StepBody,
+  UseOrHide,
 } from "./ast";
 import { TlaSyntaxError } from "./errors";
-import { tokenize, type Token } from "./lexer";
+import { getNumeralValue, tokenize, type Token } from "./lexer";
 
 interface InfixOperator {
   low: number;
@@ -26,7 +31,7 @@ interface InfixOperator {
   isAssociative: boolean;
 }
 
-const createInfix = (low: number, high: number, isAssociative: boolean): InfixOperator => ({
+const createInfix = (low: number, high: number, isAssociative: boolean) => ({
   low,
   high,
   isAssociative,
@@ -134,6 +139,20 @@ const POSTFIX_OPERATORS = new Set(["^+", "^*", "^#"]);
 
 const PREFIX_OPERATOR_REFERENCES = new Set(["-", "~", "-."]);
 
+const EXPRESSION_PREFIXES = new Set([
+  "-",
+  "~",
+  "/\\",
+  "\\/",
+  "SUBSET",
+  "UNION",
+  "DOMAIN",
+  "ENABLED",
+  "UNCHANGED",
+  "[]",
+  "<>",
+]);
+
 const OPERATOR_ARGUMENTS = new Set([
   "SUBSET",
   "UNION",
@@ -146,6 +165,9 @@ const OPERATOR_ARGUMENTS = new Set([
   ...POSTFIX_OPERATORS,
 ]);
 
+const isOperatorName = (text: string) =>
+  INFIX_OPERATORS.has(text) || POSTFIX_OPERATORS.has(text) || text === "-.";
+
 const getOperatorApplication = (
   operator: string,
   argument: Expression,
@@ -187,27 +209,111 @@ const UNIT_KEYWORDS = new Set([
 
 const THEOREM_KEYWORDS = new Set(["THEOREM", "LEMMA", "COROLLARY", "PROPOSITION"]);
 
-const PROOF_WORDS = new Set(["BY", "OBVIOUS", "OMITTED", "QED", "USE", "HIDE"]);
+const PROOF_STARTS = new Set(["BY", "PROOF", "OBVIOUS", "OMITTED"]);
+
+const PROVABLE_STEPS = new Set(["Qed", "Pick", "Case", "Assert"]);
+
+const UNNAMED_STEP_LABELS = new Set(["", "*", "-"]);
+
+const FACT_KEYWORDS = new Set([
+  "CASE",
+  "CHOOSE",
+  "DOMAIN",
+  "ENABLED",
+  "IF",
+  "LAMBDA",
+  "LET",
+  "SF_",
+  "SUBSET",
+  "UNCHANGED",
+  "UNION",
+  "WF_",
+]);
+
+const NEW_SYMBOL_LEVELS = new Map<string, NewSymbol["level"]>([
+  ["CONSTANT", "Constant"],
+  ["CONSTANTS", "Constant"],
+  ["VARIABLE", "Variable"],
+  ["VARIABLES", "Variable"],
+  ["STATE", "State"],
+  ["ACTION", "Action"],
+  ["ACTIONS", "Action"],
+  ["TEMPORAL", "Temporal"],
+  ["TEMPORALS", "Temporal"],
+]);
+
+const NEW_SYMBOL_STARTS = new Set(["NEW", ...NEW_SYMBOL_LEVELS.keys()]);
+
+const OPENING_BRACKETS = new Set(["(", "[", "{", "<<"]);
+
+const CLOSING_BRACKETS = new Set([")", "]", "}", ">>", "]_", ">>_"]);
+
+const BUILTIN_CONSTANTS = new Set(["TRUE", "FALSE", "BOOLEAN", "STRING"]);
 
 const NO_OFFSIDE = -1;
 
+const getModuleHeader = (name: string) => new RegExp(`^-{4,}\\s*MODULE\\s+${name}\\s*-{3,}$`);
+
+// The text of a module that starts on its own line in a file, as TLC's monolith extractor reads
+// modules that follow the root module.
+export const extractModule = (source: string, name: string) => {
+  const lines = source.split(/\r?\n/);
+  const start = lines.findIndex((line) => getModuleHeader(name).test(line));
+  if (start === -1) return undefined;
+  const end = lines.findIndex((line, index) => index > start && line.startsWith("===="));
+  return lines.slice(start, end === -1 ? undefined : end + 1).join("\n");
+};
+
 export const parseModule = (source: string): Module => {
   const tokens = tokenize(source);
-  const cursor = { index: 0, limit: tokens.length, offsideColumns: [] as number[] };
-  const recursiveNames = new Set<string>();
+  const cursor = { index: 0, offsideColumns: [] as number[] };
+  // A RECURSIVE declaration covers the definitions of its own LET or module.
+  const recursiveScopes: Set<string>[] = [];
+  const factNames = new Set<string>();
+  // The NEW, TAKE, and PICK names in scope, which a DEF clause cannot cite.
+  const proofScopes: Set<string>[] = [];
+  // A fact may cite an operator of a parameterized instance without the instance's arguments.
+  const factStart = { index: -1 };
+  const openModules: Module[] = [];
+
+  const isRecursiveName = (name: string) => recursiveScopes[recursiveScopes.length - 1].has(name);
 
   const getLocation = (token: Token): SourceLocation => ({
     line: token.line,
     column: token.column,
+    range: undefined,
   });
 
+  const tokenStarts: SourceLocation[] = [];
+  const tokenEnds: SourceLocation[] = [];
+  const tokenRanges: SourceRange[] = [];
+
+  const getRange = (startIndex: number) => {
+    const endIndex = cursor.index - 1;
+    const cached = startIndex === endIndex ? tokenRanges[startIndex] : undefined;
+    if (cached) return cached;
+    const last = tokens[endIndex];
+    tokenStarts[startIndex] ??= getLocation(tokens[startIndex]);
+    tokenEnds[endIndex] ??= { line: last.line, column: last.endColumn };
+    const range = {
+      module: openModules[openModules.length - 1].name,
+      start: tokenStarts[startIndex],
+      end: tokenEnds[endIndex],
+    };
+    if (startIndex === endIndex) tokenRanges[startIndex] = range;
+    return range;
+  };
+
+  // Each node gets its own location object from getLocation, so the range can be set in place.
+  const withRange = <Node extends { location: SourceLocation }>(node: Node, startIndex: number) => {
+    node.location.range = getRange(startIndex);
+    return node;
+  };
+
   const peekRaw = (offset = 0) => tokens[Math.min(cursor.index + offset, tokens.length - 1)];
 
   const peek = (offset = 0) => {
-    const token =
-      cursor.index + offset >= cursor.limit
-        ? { ...peekRaw(offset), kind: "End" as const, text: "" }
-        : peekRaw(offset);
+    const token = peekRaw(offset);
     const offsideColumn = cursor.offsideColumns[cursor.offsideColumns.length - 1];
     if (offsideColumn === undefined || offsideColumn === NO_OFFSIDE || token.kind === "End") {
       return token;
@@ -216,9 +322,9 @@ export const parseModule = (source: string): Module => {
   };
 
   const describe = (token: Token) =>
-    token.kind === "End" ? "end of input" : `${JSON.stringify(token.text)}`;
+    token.kind === "End" ? "end of input" : JSON.stringify(token.text);
 
-  const createError = (message: string, token = peekRaw()) =>
+  const createError = (message: string, token: { line: number; column: number } = peekRaw()) =>
     new TlaSyntaxError(`${message} at line ${token.line}, column ${token.column}`);
 
   const isFieldName = (token: Token) => token.kind === "Identifier" || token.kind === "Keyword";
@@ -304,7 +410,7 @@ export const parseModule = (source: string): Module => {
     return { kind: "Tuple", names };
   };
 
-  const parseBounds = (): Bound[] =>
+  const parseBounds = () =>
     parseSeparated(() => {
       const patterns = parseSeparated(parsePattern);
       expect("\\in");
@@ -315,26 +421,64 @@ export const parseModule = (source: string): Module => {
   const parseSubscript = (): Expression => {
     const token = peek();
     if (token.kind === "Identifier") {
-      cursor.index++;
-      return { kind: "Identifier", name: token.text, location: getLocation(token) };
+      const start = cursor.index;
+      const names = [expectIdentifier()];
+      while (isAt("!") && peek(1).kind === "Identifier") {
+        advance();
+        names.push(expectIdentifier());
+      }
+      return withRange(
+        { kind: "Identifier", name: names.join("!"), location: getLocation(token) },
+        start,
+      );
     }
     if (isAt("<<") || isAt("(")) return parsePrimary();
     throw createError(`Expected a subscript but found ${describe(token)}`);
   };
 
+  const isDefinitionAhead = (operator: Token) => {
+    const token = peek(1);
+    const next = peek(2);
+    if (token.kind === "Identifier" && next.text === "==") return true;
+    if (token.line === operator.line) return false;
+    if (token.text === "-.") return true;
+    return (
+      token.kind === "Identifier" &&
+      (next.text === "(" ||
+        next.text === "[" ||
+        (next.kind === "Operator" && isAt("==", POSTFIX_OPERATORS.has(next.text) ? 3 : 4)))
+    );
+  };
+
+  const isStandalone = (operator: Token) => {
+    if (!EXPRESSION_PREFIXES.has(operator.text)) return true;
+    const token = peek(1);
+    return (
+      token.kind === "End" ||
+      token.kind === "ModuleEnd" ||
+      token.kind === "Separator" ||
+      isAt(",", 1) ||
+      isAt(")", 1) ||
+      (token.kind === "Keyword" && UNIT_KEYWORDS.has(token.text)) ||
+      isDefinitionAhead(operator)
+    );
+  };
+
   const parseArgument = (): Expression => {
     const token = peek();
-    const next = peek(1);
     const isOperatorAlone =
       token.kind === "Operator" &&
       (INFIX_OPERATORS.has(token.text) || PREFIX_OPERATOR_REFERENCES.has(token.text)) &&
-      (next.text === "," || next.text === ")");
+      isStandalone(token);
     if (isOperatorAlone) {
       cursor.index++;
-      return { kind: "OperatorReference", operator: token.text, location: getLocation(token) };
+      return withRange(
+        { kind: "OperatorReference", operator: token.text, location: getLocation(token) },
+        cursor.index - 1,
+      );
     }
     const isPrefixAlone =
-      OPERATOR_ARGUMENTS.has(token.text) && (next.text === "," || next.text === ")");
+      token.kind !== "String" && OPERATOR_ARGUMENTS.has(token.text) && isStandalone(token);
     if (isPrefixAlone) {
       cursor.index++;
       const location = getLocation(token);
@@ -350,10 +494,12 @@ export const parseModule = (source: string): Module => {
       };
     }
     if (isAt("LAMBDA")) {
+      const start = cursor.index;
       advance();
       const params = parseSeparated(expectIdentifier);
       expect(":");
-      return { kind: "Lambda", params, body: parseExpression(), location: getLocation(token) };
+      const body = parseExpression();
+      return withRange({ kind: "Lambda", params, body, location: getLocation(token) }, start);
     }
     return parseExpression();
   };
@@ -387,7 +533,7 @@ export const parseModule = (source: string): Module => {
       const location = getLocation(start);
       if (peek().kind === "Identifier" && (isAt("|->", 1) || isAt(":", 1))) {
         const kind = isAt("|->", 1) ? "Record" : "RecordSet";
-        const fields = parseSeparated((): RecordField => {
+        const fields = parseSeparated(() => {
           const name = expectIdentifier();
           expect(kind === "Record" ? "|->" : ":");
           return { name, value: parseExpression() };
@@ -409,7 +555,7 @@ export const parseModule = (source: string): Module => {
         return { kind: "FunctionSet", domain: expression, range, location };
       }
       if (accept("EXCEPT")) {
-        const updates = parseSeparated((): ExceptUpdate => {
+        const updates = parseSeparated(() => {
           expect("!");
           const path: ExceptPathStep[] = [];
           while (isAt(".") || isAt("[")) {
@@ -468,10 +614,9 @@ export const parseModule = (source: string): Module => {
 
   const parseTuple = (start: Token): Expression => {
     const location = getLocation(start);
-    const elements = withoutOffside(() => {
-      if (isAt(">>") || isAt(">>_")) return [];
-      return parseSeparated(() => parseExpression());
-    });
+    const elements = withoutOffside(() =>
+      isAt(">>") || isAt(">>_") ? [] : parseSeparated(() => parseExpression()),
+    );
     if (accept(">>_")) {
       if (elements.length !== 1)
         throw createError("Expected a single action inside <<A>>_v", start);
@@ -497,18 +642,16 @@ export const parseModule = (source: string): Module => {
 
   const parseCase = (start: Token): Expression => {
     const arms: CaseArm[] = [];
-    const other = { value: undefined as Expression | undefined };
     do {
       if (accept("OTHER")) {
         expect("->");
-        other.value = parseExpression();
-        break;
+        return { kind: "Case", arms, other: parseExpression(), location: getLocation(start) };
       }
       const condition = parseExpression();
       expect("->");
       arms.push({ condition, body: parseExpression() });
     } while (accept("[]"));
-    return { kind: "Case", arms, other: other.value, location: getLocation(start) };
+    return { kind: "Case", arms, other: undefined, location: getLocation(start) };
   };
 
   const isSelectorAhead = () => {
@@ -527,7 +670,7 @@ export const parseModule = (source: string): Module => {
 
   const parseSelector = (): Selector => {
     const token = advance();
-    if (token.kind === "Number") return { kind: "Operand", operand: Number(token.text) };
+    if (token.kind === "Number") return { kind: "Operand", operand: getNumeralValue(token.text) };
     switch (token.text) {
       case "<<":
       case ">>":
@@ -545,25 +688,46 @@ export const parseModule = (source: string): Module => {
   };
 
   const parsePrimary = (): Expression => {
+    const isFactStart = cursor.index === factStart.index;
     const token = advance();
     const location = getLocation(token);
-    if (token.kind === "Number") return { kind: "Number", value: Number(token.text), location };
+    if (token.kind === "Number") {
+      return { kind: "Number", value: getNumeralValue(token.text), text: token.text, location };
+    }
+    if (token.kind === "Decimal") return { kind: "Decimal", text: token.text, location };
     if (token.kind === "String") return { kind: "String", value: token.text, location };
-    if (token.kind === "Identifier") {
+    const isStep = token.kind === "Step" && isStepReference(token);
+    if (isStep && getProofLevel() < 0 && proofLevels.length <= 1) {
+      throw createError("Step number used outside proof", token);
+    }
+    if (isStep && token.text[1] === "+") {
+      throw createError("<+> step number used in an expression", token);
+    }
+    if (token.kind === "Identifier" || isStep) {
+      const name = isStep ? (getStepName(token) ?? token.text) : token.text;
       if (isLabelAhead()) {
         const params = accept("(") ? parseSeparated(expectIdentifier) : [];
         if (params.length > 0) expect(")");
         expect("::");
-        return { kind: "Label", name: token.text, params, body: parseExpression(), location };
+        return { kind: "Label", name, params, body: parseExpression(), location };
       }
       const hasArguments = isAt("(");
       const args = hasArguments ? parseArguments() : [];
       if (!isSelectorAhead()) {
         return hasArguments
-          ? { kind: "Apply", operator: token.text, args, location }
-          : { kind: "Identifier", name: token.text, location };
+          ? { kind: "Apply", operator: name, args, location }
+          : { kind: "Identifier", name, location };
+      }
+      const instance = openModules
+        .flatMap((module) => module.instances)
+        .find((candidate) => candidate.name === name);
+      if (instance && !isFactStart && instance.params.length !== args.length) {
+        throw createError(
+          `The operator ${name} requires ${instance.params.length} arguments`,
+          token,
+        );
       }
-      const selectors: Selector[] = [{ kind: "Name", name: token.text, args }];
+      const selectors: Selector[] = [{ kind: "Name", name, args }];
       while (isSelectorAhead()) {
         advance();
         selectors.push(parseSelector());
@@ -580,9 +744,7 @@ export const parseModule = (source: string): Module => {
         ? { kind: "Apply", operator, args: qualifiedArgs, location }
         : { kind: "Identifier", name: operator, location };
     }
-    const isOperatorName =
-      INFIX_OPERATORS.has(token.text) || POSTFIX_OPERATORS.has(token.text) || token.text === "-.";
-    if (token.kind === "Operator" && isOperatorName && isAt("("))
+    if (token.kind === "Operator" && isOperatorName(token.text) && isAt("("))
       return { kind: "Apply", operator: token.text, args: parseArguments(), location };
     switch (token.text) {
       case "(": {
@@ -615,15 +777,17 @@ export const parseModule = (source: string): Module => {
       case "LET": {
         const definitions: Definition[] = [];
         const instances: Instance[] = [];
+        recursiveScopes.push(new Set());
         while (!accept("IN")) {
           if (accept("RECURSIVE")) {
-            parseDeclarations().forEach((declaration) => recursiveNames.add(declaration.name));
+            declareRecursive();
             continue;
           }
           const unit = parseDefinition(false);
           if (unit.kind === "Instance") instances.push(unit);
           else definitions.push(unit);
         }
+        recursiveScopes.pop();
         const body = parseExpression();
         return instances.length === 0
           ? { kind: "Let", definitions, body, location }
@@ -692,8 +856,10 @@ export const parseModule = (source: string): Module => {
   };
 
   const parseExpression = (minimumPrecedence = 0): Expression => {
+    const start = cursor.index;
+    const prefix = parsePrefix();
     const state = {
-      left: parsePrefix(),
+      left: prefix.location.range ? prefix : withRange(prefix, start),
       previous: undefined as (InfixOperator & { text: string }) | undefined,
     };
     while (true) {
@@ -702,24 +868,31 @@ export const parseModule = (source: string): Module => {
       const location = getLocation(token);
       if (token.text === "'" && minimumPrecedence <= 15) {
         advance();
-        state.left = { kind: "Prime", expression: state.left, location };
+        state.left = withRange({ kind: "Prime", expression: state.left, location }, start);
         continue;
       }
       if (POSTFIX_OPERATORS.has(token.text) && minimumPrecedence <= 15) {
         advance();
-        state.left = { kind: "Apply", operator: token.text, args: [state.left], location };
+        state.left = withRange(
+          { kind: "Apply", operator: token.text, args: [state.left], location },
+          start,
+        );
         continue;
       }
       if (token.text === "[") {
         advance();
         const args = withoutOffside(() => parseSeparated(() => parseExpression()));
         withoutOffside(() => expect("]"));
-        state.left = { kind: "FunctionApplication", function: state.left, args, location };
+        state.left = withRange(
+          { kind: "FunctionApplication", function: state.left, args, location },
+          start,
+        );
         continue;
       }
       if (token.text === "." && isFieldName(peek(1))) {
         advance();
-        state.left = { kind: "FieldAccess", record: state.left, field: advance().text, location };
+        const field = advance().text;
+        state.left = withRange({ kind: "FieldAccess", record: state.left, field, location }, start);
         continue;
       }
       const infix = INFIX_OPERATORS.get(token.text);
@@ -752,44 +925,59 @@ export const parseModule = (source: string): Module => {
       } else {
         state.left = { kind: "Apply", operator: token.text, args: [left, right], location };
       }
+      state.left = withRange(state.left, start);
       state.previous = { ...infix, text: token.text };
     }
     return state.left;
   };
 
-  const parseParams = (): OperatorParam[] => {
+  const parseParams = () => {
     if (!accept("(")) return [];
     const params = parseDeclarations();
     expect(")");
     return params;
   };
 
+  const parseBody = () => {
+    const start = getLocation(peek());
+    const body = parseExpression();
+    const last = tokens[cursor.index - 1];
+    const end = { line: last.line, column: last.endColumn };
+    return { body, bodyRange: { module: openModules[openModules.length - 1].name, start, end } };
+  };
+
   const parseDefinition = (isLocal: boolean): Definition | Instance => {
+    const start = cursor.index;
+    return withRange(parseDefinitionUnit(isLocal), start);
+  };
+
+  const parseDefinitionUnit = (isLocal: boolean): Definition | Instance => {
     const token = peek();
     const location = getLocation(token);
+    const parseOperator = (name: string, params: OperatorParam[]): Definition => ({
+      kind: "Operator",
+      name,
+      params,
+      ...parseBody(),
+      isLocal,
+      isRecursive: isRecursiveName(name),
+      location,
+    });
     if (
       token.kind === "Identifier" &&
       peek(1).kind === "Operator" &&
-      INFIX_OPERATORS.has(peek(1).text)
+      INFIX_OPERATORS.has(peek(1).text) &&
+      peek(2).kind === "Identifier" &&
+      isAt("==", 3)
     ) {
-      if (peek(2).kind === "Identifier" && isAt("==", 3)) {
-        const left = expectIdentifier();
-        const operator = advance().text;
-        const right = expectIdentifier();
-        expect("==");
-        return {
-          kind: "Operator",
-          name: operator,
-          params: [
-            { name: left, arity: 0 },
-            { name: right, arity: 0 },
-          ],
-          body: parseExpression(),
-          isLocal,
-          isRecursive: recursiveNames.has(operator),
-          location,
-        };
-      }
+      const left = expectIdentifier();
+      const operator = advance().text;
+      const right = expectIdentifier();
+      expect("==");
+      return parseOperator(operator, [
+        { name: left, arity: 0 },
+        { name: right, arity: 0 },
+      ]);
     }
     const isPrefixDefinition = token.kind === "Operator" && token.text === "-.";
     const isPostfixDefinition =
@@ -799,15 +987,7 @@ export const parseModule = (source: string): Module => {
       cursor.index += isPrefixDefinition ? 1 : 2;
       const param = isPrefixDefinition ? expectIdentifier() : token.text;
       expect("==");
-      return {
-        kind: "Operator",
-        name: operator,
-        params: [{ name: param, arity: 0 }],
-        body: parseExpression(),
-        isLocal,
-        isRecursive: recursiveNames.has(operator),
-        location,
-      };
+      return parseOperator(operator, [{ name: param, arity: 0 }]);
     }
     const name = expectIdentifier();
     if (accept("[")) {
@@ -820,22 +1000,14 @@ export const parseModule = (source: string): Module => {
         bounds,
         body: parseExpression(),
         isLocal,
-        isRecursive: recursiveNames.has(name),
+        isRecursive: isRecursiveName(name),
         location,
       };
     }
     const params = parseParams();
     expect("==");
     if (isAt("INSTANCE")) return parseInstance(name, params, isLocal, location);
-    return {
-      kind: "Operator",
-      name,
-      params,
-      body: parseExpression(),
-      isLocal,
-      isRecursive: recursiveNames.has(name),
-      location,
-    };
+    return parseOperator(name, params);
   };
 
   const parseInstance = (
@@ -847,7 +1019,7 @@ export const parseModule = (source: string): Module => {
     expect("INSTANCE");
     const module = expectIdentifier();
     const substitutions = accept("WITH")
-      ? parseSeparated((): InstanceSubstitution => {
+      ? parseSeparated(() => {
           const target = advance().text;
           expect("<-");
           return { name: target, expression: parseArgument() };
@@ -856,101 +1028,444 @@ export const parseModule = (source: string): Module => {
     return { kind: "Instance", name, params, module, substitutions, isLocal, location };
   };
 
-  const parseDeclarations = (): Declaration[] =>
-    parseSeparated(() => {
-      if (isAt("_") && peek(1).kind === "Operator") {
-        advance();
-        const operator = advance().text;
-        return accept("_") ? { name: operator, arity: 2 } : { name: operator, arity: 1 };
-      }
-      if (peek().kind === "Operator" && isAt("_", 1)) {
-        const operator = advance().text;
-        advance();
-        return { name: operator === "-" ? "-." : operator, arity: 1 };
-      }
-      const name = expectIdentifier();
-      if (!accept("(")) return { name, arity: 0 };
-      const placeholders = parseSeparated(() => expect("_"));
-      expect(")");
-      return { name, arity: placeholders.length };
-    });
+  const parseDeclaration = (): Declaration => {
+    const location = getLocation(peek());
+    if (isAt("_") && peek(1).kind === "Operator") {
+      advance();
+      const name = advance().text;
+      return { name, arity: accept("_") ? 2 : 1, location };
+    }
+    if (peek().kind === "Operator" && isAt("_", 1)) {
+      const operator = advance().text;
+      advance();
+      return { name: operator === "-" ? "-." : operator, arity: 1, location };
+    }
+    const name = expectIdentifier();
+    if (!accept("(")) return { name, arity: 0, location };
+    const placeholders = parseSeparated(() => expect("_"));
+    expect(")");
+    return { name, arity: placeholders.length, location };
+  };
 
-  const isUnitStart = (token: Token, next: Token) => {
-    if (token.column !== 1) return false;
-    if (token.kind === "Separator" || token.kind === "ModuleEnd" || token.kind === "End")
-      return true;
-    if (token.kind === "Keyword") return UNIT_KEYWORDS.has(token.text);
-    return (
-      token.kind === "Identifier" && (next.text === "==" || next.text === "(" || next.text === "[")
+  const parseDeclarations = () => parseSeparated(parseDeclaration);
+
+  const declareRecursive = () =>
+    parseDeclarations().forEach((declaration) =>
+      recursiveScopes[recursiveScopes.length - 1].add(declaration.name),
     );
+
+  const getClosingOffset = (offset: number) => {
+    const depth = { value: 0, offset };
+    do {
+      const token = peek(depth.offset);
+      if (token.kind === "End") return depth.offset;
+      if (token.kind === "Operator" && OPENING_BRACKETS.has(token.text)) depth.value++;
+      if (token.kind === "Operator" && CLOSING_BRACKETS.has(token.text)) depth.value--;
+      depth.offset++;
+    } while (depth.value > 0);
+    return depth.offset;
+  };
+
+  // SANY inserts a break before each definition, which ends the fact list or step before it.
+  const isDefinitionStart = () => {
+    const token = peek();
+    const next = peek(1);
+    if (token.kind !== "Identifier") {
+      return isAt("-.") && next.kind === "Identifier" && isAt("==", 2);
+    }
+    if (isAt("(", 1) || isAt("[", 1)) return isAt("==", getClosingOffset(1));
+    if (next.kind === "Operator" && POSTFIX_OPERATORS.has(next.text)) return isAt("==", 2);
+    if (next.kind === "Operator" && INFIX_OPERATORS.has(next.text)) {
+      return peek(2).kind === "Identifier" && isAt("==", 3);
+    }
+    return isAt("==", 1);
   };
 
-  const skipUnit = (start: Token) => {
-    while (peekRaw().kind !== "End") {
-      const token = peekRaw();
-      if (token.line > start.line && isUnitStart(token, peekRaw(1))) return;
+  // SANY's proof levels, one for each open proof, which stay -1 until a step sets them.
+  const proofLevels: number[] = [];
+
+  const getProofLevel = () => proofLevels[proofLevels.length - 1] ?? -1;
+
+  const getStepLevel = (step: Token) => {
+    if (step.text[1] === "*") return -1;
+    if (step.text[1] === "+") return -2;
+    return Number(step.text.slice(1, step.text.indexOf(">")));
+  };
+
+  const getStepLabel = (step: Token) => {
+    const label = step.text.slice(step.text.indexOf(">") + 1);
+    const end = { value: label.length };
+    while (label[end.value - 1] === ".") end.value--;
+    return label.slice(0, end.value);
+  };
+
+  const isStepReference = (step: Token) =>
+    !step.text.endsWith(".") && !UNNAMED_STEP_LABELS.has(getStepLabel(step));
+
+  const getStepName = (step: Token) => {
+    const label = getStepLabel(step);
+    if (UNNAMED_STEP_LABELS.has(label)) return undefined;
+    const level = getStepLevel(step);
+    if (level >= 0) return `<${level}>${label}`;
+    const current = getProofLevel();
+    const outer = proofLevels[proofLevels.length - 2];
+    return `<${current < 0 && outer !== undefined ? outer : current}>${label}`;
+  };
+
+  const isProofStart = (token: Token) => {
+    if (token.kind === "Keyword") return PROOF_STARTS.has(token.text);
+    if (token.kind !== "Step") return false;
+    if (token.text[1] === "*") return proofLevels.length === 0;
+    if (token.text[1] === "+" || proofLevels.length === 0) return true;
+    const level = getProofLevel();
+    return level >= 0 && getStepLevel(token) > level;
+  };
+
+  const isCorrectLevel = (step: Token) => {
+    const level = getStepLevel(step);
+    const top = proofLevels.length - 1;
+    const outer = proofLevels[top - 1] ?? -1;
+    const isFirst = proofLevels[top] < 0;
+    if (isFirst) proofLevels[top] = level < 0 ? outer + 1 : level;
+    if (level === -1) return true;
+    if (level === -2) return isFirst;
+    return isFirst ? level > outer : proofLevels[top] === level;
+  };
+
+  const isDeclared = (name: string) =>
+    !factNames.has(name) &&
+    openModules.some(
+      (module) =>
+        module.constants.some((constant) => constant.name === name) ||
+        module.variables.some((variable) => variable.name === name) ||
+        module.definitions.some((definition) => definition.name === name),
+    );
+
+  const isHideFact = (fact: Fact) => {
+    if (fact.kind === "ModuleFact" || fact.kind === "Subexpression") return true;
+    if (fact.kind === "Apply") return fact.operator.includes("!");
+    if (fact.kind !== "Identifier") return false;
+    if (fact.name.startsWith("<") || fact.name.includes("!")) return true;
+    return !BUILTIN_CONSTANTS.has(fact.name) && !isDeclared(fact.name);
+  };
+
+  const isFactAhead = () => {
+    const token = peek();
+    if (token.kind === "Step") return isStepReference(token);
+    if (token.kind === "Keyword") return FACT_KEYWORDS.has(token.text);
+    if (token.kind === "End" || token.kind === "ModuleEnd" || token.kind === "Separator") {
+      return false;
+    }
+    return !isDefinitionStart();
+  };
+
+  // SANY reads an infix or postfix operator alone as its name, but a prefix operator such as - as the
+  // start of an expression.
+  const parseFact = (): Fact => {
+    const start = cursor.index;
+    factStart.index = start;
+    const token = peek();
+    if (
+      token.kind === "Operator" &&
+      !EXPRESSION_PREFIXES.has(token.text) &&
+      isOperatorName(token.text)
+    ) {
       cursor.index++;
+      return withRange(
+        { kind: "OperatorReference", operator: token.text, location: getLocation(token) },
+        start,
+      );
     }
+    if (!accept("MODULE")) return parseExpression();
+    return withRange(
+      { kind: "ModuleFact", name: expectIdentifier(), location: getLocation(token) },
+      start,
+    );
   };
 
-  const isProofStart = (index: number, start: Token) => {
-    const token = tokens[index];
-    if (token.kind === "End") return true;
-    if (token.line > start.line && isUnitStart(token, tokens[index + 1])) return true;
-    if (token.text === "PROOF" || PROOF_WORDS.has(token.text)) return true;
-    return (
-      token.text === "<" && tokens[index + 1]?.kind === "Number" && tokens[index + 2]?.text === ">"
+  const declareProofSymbol = (name: string) => proofScopes[proofScopes.length - 1]?.add(name);
+
+  const withProofScope = <Result>(parse: () => Result) => {
+    proofScopes.push(new Set());
+    const result = parse();
+    proofScopes.pop();
+    return result;
+  };
+
+  const isDeclaredSymbol = (name: string) =>
+    proofScopes.some((scope) => scope.has(name)) ||
+    openModules.some(
+      (module) =>
+        module.constants.some((constant) => constant.name === name) ||
+        module.variables.some((variable) => variable.name === name),
     );
+
+  const parseDefs = () => {
+    const defs = parseSeparated(parseFact);
+    const invalid = defs.find((def) => def.kind === "Identifier" && isDeclaredSymbol(def.name));
+    if (invalid) {
+      throw createError("DEF clause entry should describe a defined operator", invalid.location);
+    }
+    return defs;
+  };
+
+  const parseFacts = () => ({
+    facts: isFactAhead() ? parseSeparated(parseFact) : [],
+    defs: accept("DEF") || accept("DEFS") ? parseDefs() : [],
+  });
+
+  const parseUseOrHide = (): UseOrHide => {
+    const start = cursor.index;
+    const keyword = advance();
+    const isOnly = keyword.text === "USE" && accept("ONLY");
+    const use: UseOrHide = {
+      kind: keyword.text === "USE" ? "Use" : "Hide",
+      isOnly,
+      ...parseFacts(),
+      location: getLocation(keyword),
+    };
+    const invalid = use.kind === "Hide" ? use.facts.find((fact) => !isHideFact(fact)) : undefined;
+    if (invalid) {
+      throw createError(
+        "The only expression allowed as a fact in a HIDE is the name of a theorem, assumption, or step",
+        invalid.location,
+      );
+    }
+    return withRange(use, start);
   };
 
-  const parseTheorem = (): Definition | undefined => {
-    const start = advance();
-    const isNamed = peek().kind === "Identifier" && isAt("==", 1) && !isAt("ASSUME", 2);
-    if (!isNamed) {
-      skipUnit(start);
-      return undefined;
+  const parseNewSymbol = (): NewSymbol => {
+    const start = cursor.index;
+    const location = getLocation(peek());
+    const isNew = accept("NEW");
+    const level = peek().kind === "Keyword" ? NEW_SYMBOL_LEVELS.get(peek().text) : undefined;
+    if (!isNew && level === undefined) {
+      throw createError(`Expected a declaration but found ${describe(peek())}`);
     }
-    const name = expectIdentifier();
-    advance();
-    const end = { value: cursor.index };
-    while (!isProofStart(end.value, start)) end.value++;
-    cursor.limit = end.value;
-    const body = parseExpression();
-    const isComplete = cursor.index === end.value;
-    cursor.limit = tokens.length;
-    skipUnit(start);
-    if (!isComplete) return undefined;
+    if (level !== undefined) advance();
+    if (level === "Variable") {
+      const name = expectIdentifier();
+      declareProofSymbol(name);
+      return withRange(
+        { kind: "NewSymbol", name, arity: 0, level, set: undefined, location },
+        start,
+      );
+    }
+    const declaration = parseDeclaration();
+    declareProofSymbol(declaration.name);
+    const isBounded = (level ?? "Constant") === "Constant" && isAt("\\in");
+    if (isBounded && declaration.arity > 0) {
+      throw createError("declared symbol with arguments before \\in");
+    }
+    if (isBounded) advance();
+    return withRange(
+      {
+        kind: "NewSymbol",
+        name: declaration.name,
+        arity: declaration.arity,
+        level: level ?? "Constant",
+        set: isBounded ? parseExpression() : undefined,
+        location,
+      },
+      start,
+    );
+  };
+
+  const parseAssumeProve = (): AssumeProve => {
+    const start = cursor.index;
+    const location = getLocation(expect("ASSUME"));
+    const assumptions = parseSeparated(() => {
+      if (isAt("ASSUME")) return withProofScope(parseAssumeProve);
+      if (peek().kind === "Keyword" && NEW_SYMBOL_STARTS.has(peek().text)) return parseNewSymbol();
+      return parseExpression();
+    });
+    expect("PROVE");
+    return withRange(
+      { kind: "AssumeProve", assumptions, conclusion: parseExpression(), location },
+      start,
+    );
+  };
+
+  const isNameListAhead = (terminator: string) => {
+    const offset = { value: 0 };
+    while (peek(offset.value).kind === "Identifier" && isAt(",", offset.value + 1)) {
+      offset.value += 2;
+    }
+    return peek(offset.value).kind === "Identifier" && isAt(terminator, offset.value + 1);
+  };
+
+  const declareProofSymbols = (names: string[], bounds: Bound[]) => {
+    names.forEach(declareProofSymbol);
+    bounds.forEach((bound) =>
+      (bound.pattern.kind === "Name" ? [bound.pattern.name] : bound.pattern.names).forEach(
+        declareProofSymbol,
+      ),
+    );
+  };
+
+  const parseStepBody = (): StepBody => {
+    const token = peek();
+    if (isAt("USE") || isAt("HIDE")) return parseUseOrHide();
+    if (isAt("INSTANCE")) {
+      return withRange(parseInstance(undefined, [], false, getLocation(token)), cursor.index);
+    }
+    if (isAt("DEFINE") || isDefinitionStart()) {
+      const start = cursor.index;
+      accept("DEFINE");
+      const definitions = [parseDefinition(false)];
+      while (isDefinitionStart()) definitions.push(parseDefinition(false));
+      const define: DefineStep = { kind: "Define", definitions, location: getLocation(token) };
+      return withRange(define, start);
+    }
+    if (accept("HAVE")) return { kind: "Have", expression: parseExpression() };
+    if (accept("TAKE")) {
+      const isBounded = isBoundAhead();
+      const names = isBounded ? [] : parseSeparated(expectIdentifier);
+      const bounds = isBounded ? parseBounds() : [];
+      declareProofSymbols(names, bounds);
+      return { kind: "Take", names, bounds };
+    }
+    if (accept("WITNESS")) {
+      return { kind: "Witness", expressions: parseSeparated(() => parseExpression()) };
+    }
+    if (accept("PICK")) {
+      const isNames = isNameListAhead(":");
+      const names = isNames ? parseSeparated(expectIdentifier) : [];
+      const bounds = isNames ? [] : parseBounds();
+      declareProofSymbols(names, bounds);
+      expect(":");
+      return { kind: "Pick", names, bounds, body: parseExpression() };
+    }
+    if (accept("CASE")) return { kind: "Case", expression: parseExpression() };
     return {
-      kind: "Operator",
-      name,
-      params: [],
-      body,
-      isLocal: false,
-      isRecursive: false,
-      location: getLocation(start),
+      kind: "Assert",
+      isSuffices: accept("SUFFICES"),
+      statement: isAt("ASSUME") ? parseAssumeProve() : parseExpression(),
     };
   };
 
+  const parseStepStart = (isQed: boolean) => {
+    const step = peek();
+    if (step.kind !== "Step") {
+      throw createError(`Expected a step number but found ${describe(step)}`);
+    }
+    if (!isCorrectLevel(step)) {
+      throw createError(`${isQed ? "QED step" : "step"}'s number has bad level`, step);
+    }
+    if (getStepLevel(step) < 0 && getStepName(step) !== undefined) {
+      throw createError("<*> and <+> cannot be used for a named step", step);
+    }
+    cursor.index++;
+    return { name: getStepName(step), location: getLocation(step) };
+  };
+
+  // The names an assertion assumes stay in its own proof, and other steps declare theirs for the rest
+  // of the enclosing proof.
+  const parseStep = (): ProofStep => {
+    const start = cursor.index;
+    const step = parseStepStart(isAt("QED", 1));
+    const scope = new Set<string>();
+    proofScopes.push(scope);
+    const body: StepBody = accept("QED") ? { kind: "Qed" } : parseStepBody();
+    const hasProof = isProofStart(peek());
+    if (hasProof && !PROVABLE_STEPS.has(body.kind)) {
+      throw createError("Proof of a step that does not take a proof");
+    }
+    const proof = hasProof ? parseProof() : undefined;
+    proofScopes.pop();
+    if (body.kind !== "Assert" || body.isSuffices) scope.forEach(declareProofSymbol);
+    return withRange({ ...step, body, proof }, start);
+  };
+
+  const parseProofBody = (): Proof => {
+    const location = getLocation(peek());
+    if (isAt("BY") || (isAt("PROOF") && isAt("BY", 1))) {
+      accept("PROOF");
+      advance();
+      return { kind: "By", isOnly: accept("ONLY"), ...parseFacts(), location };
+    }
+    const terminal = peek(isAt("PROOF") ? 1 : 0);
+    if (
+      terminal.kind === "Keyword" &&
+      (terminal.text === "OBVIOUS" || terminal.text === "OMITTED")
+    ) {
+      accept("PROOF");
+      advance();
+      return { kind: terminal.text === "OBVIOUS" ? "Obvious" : "Omitted", location };
+    }
+    accept("PROOF");
+    const steps = [parseStep()];
+    while (steps[steps.length - 1].body.kind !== "Qed") steps.push(parseStep());
+    return { kind: "Steps", level: getProofLevel(), steps, location };
+  };
+
+  const parseProof = (): Proof => {
+    const start = cursor.index;
+    proofLevels.push(-1);
+    const proof = withProofScope(parseProofBody);
+    proofLevels.pop();
+    return withRange(proof, start);
+  };
+
+  const parseTheorem = (module: Module) => {
+    const start = cursor.index;
+    const keyword = advance();
+    const isNamed = peek().kind === "Identifier" && isAt("==", 1);
+    const name = isNamed ? expectIdentifier() : undefined;
+    if (name !== undefined) {
+      factNames.add(name);
+      advance();
+    }
+    const parsed = withProofScope(() => {
+      const statement = isAt("ASSUME") ? parseAssumeProve() : parseExpression();
+      return { statement, proof: isProofStart(peek()) ? parseProof() : undefined };
+    });
+    const statement = parsed.statement;
+    module.theorems.push(
+      withRange({ name, statement, proof: parsed.proof, location: getLocation(keyword) }, start),
+    );
+    if (name === undefined || statement.kind === "AssumeProve") return;
+    module.definitions.push(
+      withRange(
+        {
+          kind: "Operator",
+          name,
+          params: [],
+          body: statement,
+          isLocal: false,
+          isRecursive: false,
+          location: getLocation(keyword),
+        },
+        start,
+      ),
+    );
+  };
+
   const parseAssumption = (): Assumption => {
+    const start = cursor.index;
+    return withRange(parseAssumptionUnit(), start);
+  };
+
+  const parseAssumptionUnit = (): Assumption => {
     const keyword = advance();
     const location = getLocation(keyword);
     const isAxiom = keyword.text === "AXIOM";
     if (peek().kind === "Identifier" && isAt("==", 1)) {
       const name = expectIdentifier();
+      factNames.add(name);
       advance();
       return { name, expression: parseExpression(), isAxiom, location };
     }
     return { name: undefined, expression: parseExpression(), isAxiom, location };
   };
 
-  const parseModuleBody = (outer?: OuterScope): Module => {
-    expect("----");
+  const parseModuleBody = (): Module => {
+    const location = getLocation(expect("----"));
     expect("MODULE");
     const name = expectIdentifier();
     expect("----");
     const module: Module = {
       name,
-      outer,
       modules: [],
       extends: accept("EXTENDS") ? parseSeparated(expectIdentifier) : [],
       constants: [],
@@ -958,13 +1473,27 @@ export const parseModule = (source: string): Module => {
       definitions: [],
       instances: [],
       assumptions: [],
+      location,
+      theorems: [],
+      uses: [],
     };
+    openModules.push(module);
+    recursiveScopes.push(new Set());
     while (true) {
+      const tokenIndex = cursor.index;
       const token = peek();
-      if (token.kind === "ModuleEnd" || token.kind === "End") return module;
+      if (token.kind === "ModuleEnd" || token.kind === "End") {
+        openModules.pop();
+        recursiveScopes.pop();
+        return module;
+      }
       if (token.kind === "Separator") {
-        if (isAt("MODULE", 1)) module.modules.push(parseNestedModule(getOuterScope(module)));
-        else advance();
+        if (isAt("MODULE", 1)) {
+          module.modules.push(parseModuleBody());
+          expect("====");
+          continue;
+        }
+        advance();
         continue;
       }
       switch (token.text) {
@@ -976,7 +1505,13 @@ export const parseModule = (source: string): Module => {
         case "VARIABLE":
         case "VARIABLES":
           advance();
-          module.variables.push(...parseSeparated(expectIdentifier));
+          module.variables.push(
+            ...parseSeparated(() => {
+              const start = cursor.index;
+              const location = getLocation(peek());
+              return withRange({ name: expectIdentifier(), arity: 0, location }, start);
+            }),
+          );
           continue;
         case "ASSUME":
         case "ASSUMPTION":
@@ -996,17 +1531,25 @@ export const parseModule = (source: string): Module => {
           }
           continue;
         }
+        case "USE":
+        case "HIDE":
+          module.uses.push(parseUseOrHide());
+          continue;
         case "RECURSIVE":
           advance();
-          parseDeclarations().forEach((declaration) => recursiveNames.add(declaration.name));
+          declareRecursive();
           continue;
         case "INSTANCE":
-          module.instances.push(parseInstance(undefined, [], false, getLocation(token)));
+          module.instances.push(
+            withRange(parseInstance(undefined, [], false, getLocation(token)), tokenIndex),
+          );
           continue;
         case "LOCAL": {
           advance();
           if (isAt("INSTANCE")) {
-            module.instances.push(parseInstance(undefined, [], true, getLocation(token)));
+            module.instances.push(
+              withRange(parseInstance(undefined, [], true, getLocation(token)), tokenIndex),
+            );
             continue;
           }
           const unit = parseDefinition(true);
@@ -1016,8 +1559,7 @@ export const parseModule = (source: string): Module => {
         }
       }
       if (token.kind === "Keyword" && THEOREM_KEYWORDS.has(token.text)) {
-        const theorem = parseTheorem();
-        if (theorem) module.definitions.push(theorem);
+        parseTheorem(module);
         continue;
       }
       const unit = parseDefinition(false);
@@ -1026,27 +1568,5 @@ export const parseModule = (source: string): Module => {
     }
   };
 
-  const getOuterScope = (module: Module): OuterScope => ({
-    extends: [...(module.outer?.extends ?? []), ...module.extends],
-    symbols: [
-      ...(module.outer?.symbols ?? []),
-      ...module.constants,
-      ...module.variables.map((variable) => ({ name: variable, arity: 0 })),
-      ...module.definitions.map((definition) => ({
-        name: definition.name,
-        arity: definition.kind === "Operator" ? definition.params.length : 0,
-      })),
-    ],
-  });
-
-  const parseNestedModule = (outer?: OuterScope) => {
-    const nested = parseModuleBody(outer);
-    accept("====");
-    return nested;
-  };
-
-  const root = parseModuleBody();
-  accept("====");
-  while (peekRaw().kind === "Separator") root.modules.push(parseNestedModule());
-  return root;
+  return parseModuleBody();
 };