parseman 0.50.6 → 0.50.7

package/dist/cli/index.js

@@ -7,14 +7,14 @@ import { pathToFileURL } from "node:url";
 import { register } from "node:module";
 
 // src/combinators/first-set.ts
-function classifyFinalChoice(arms, resolve2) {
+function classifyFinalChoice(arms, resolve2, activeTrivia) {
   const firstSets = [];
   const nullable2 = [];
   let exclusive = true;
   for (const arm of arms) {
     const empty2 = matchesEmpty(arm, /* @__PURE__ */ new Set(), resolve2);
     nullable2.push(empty2);
-    firstSets.push(empty2 ? { kind: "any" } : firstSetOf(arm, /* @__PURE__ */ new Set(), resolve2));
+    firstSets.push(empty2 ? { kind: "any" } : firstSetOf(arm, /* @__PURE__ */ new Set(), resolve2, activeTrivia));
     if (empty2) exclusive = false;
   }
   if (exclusive) {
@@ -75,7 +75,7 @@ function intersects(a, b) {
   return false;
 }
 function fromChar(code) {
-  return { kind: "ranges", ranges: [{ lo: code, hi: code }] };
+  return fromRange(code, code);
 }
 function startsFirstSet(combinator, input, pos) {
   const fs = combinator._meta.firstSet;
@@ -168,11 +168,10 @@ function matchesEmptyBody(p, seen, resolve2) {
   }
 }
 function isZeroWidthAssertion(p) {
-  const tag = p._def.tag;
-  return tag === "not" || tag === "adjacency";
+  return p._tag === "not" || p._tag === "adjacency";
 }
 function isPositiveLookahead(p) {
-  return p._def.tag === "peek";
+  return p._tag === "peek";
 }
 function narrowBy(acc, constraint) {
   if (constraint.kind === "any") return acc;
@@ -188,34 +187,42 @@ function narrowBy(acc, constraint) {
   return ranges.length === 0 ? { kind: "empty" } : { kind: "ranges", ranges };
 }
 function applyAssertion(fs, assertion) {
-  if (assertion === null) return fs;
-  if (fs.kind === "empty") return assertion;
-  return narrowBy(fs, assertion) ?? fs;
+  return assertion === null ? fs : fs.kind === "empty" ? assertion : narrowBy(fs, assertion);
 }
-function sequenceFirstSet(parsers) {
-  let fs = empty();
-  let assertion = null;
-  for (const p of parsers) {
+function sequenceSet(parsers, fs, empties, trivia2) {
+  const afterBoundary = (i) => {
+    const rest = at(i);
+    return trivia2 ? union(rest, fs(trivia2)) : rest;
+  };
+  const at = (i) => {
+    const p = parsers[i];
     if (isPositiveLookahead(p)) {
-      assertion = narrowBy(assertion, p._meta.firstSet);
-      continue;
+      const constraint = fs(p);
+      return i + 1 === parsers.length ? constraint : applyAssertion(afterBoundary(i + 1), constraint);
     }
-    if (!isZeroWidthAssertion(p)) fs = union(fs, p._meta.firstSet);
-    if (!matchesEmpty(p)) return applyAssertion(fs, assertion);
-  }
-  return applyAssertion(fs, assertion);
+    if (isZeroWidthAssertion(p)) {
+      return i + 1 === parsers.length ? empty() : afterBoundary(i + 1);
+    }
+    const first = fs(p);
+    if (!empties(p) || i + 1 === parsers.length) return first;
+    return union(first, afterBoundary(i + 1));
+  };
+  return at(0);
+}
+function sequenceFirstSet(parsers) {
+  return sequenceSet(parsers, (p) => p._meta.firstSet, matchesEmpty);
 }
-function firstSetOf(p, seen = /* @__PURE__ */ new Set(), resolve2) {
+function firstSetOf(p, seen = /* @__PURE__ */ new Set(), resolve2, activeTrivia) {
   if (seen.has(p)) return any();
   seen.add(p);
   try {
-    return firstSetBody(p, seen, resolve2);
+    return firstSetBody(p, seen, resolve2, activeTrivia);
   } finally {
     seen.delete(p);
   }
 }
-function firstSetBody(p, seen, resolve2) {
-  const fs = (c) => firstSetOf(c, seen, resolve2);
+function firstSetBody(p, seen, resolve2, activeTrivia) {
+  const fs = (c) => firstSetOf(c, seen, resolve2, activeTrivia);
   const empties = (c) => matchesEmpty(c, /* @__PURE__ */ new Set(), resolve2);
   const d = p._def;
   switch (d.tag) {
@@ -241,17 +248,7 @@ function firstSetBody(p, seen, resolve2) {
       return out;
     }
     case "sequence": {
-      let out = empty();
-      let assertion = null;
-      for (const term of d.parsers) {
-        if (isPositiveLookahead(term)) {
-          assertion = narrowBy(assertion, fs(term));
-          continue;
-        }
-        if (!isZeroWidthAssertion(term)) out = union(out, fs(term));
-        if (!empties(term)) return applyAssertion(out, assertion);
-      }
-      return applyAssertion(out, assertion);
+      return sequenceSet(d.parsers, fs, empties, activeTrivia);
     }
     case "peek": {
       const inner = d.parser;
@@ -266,12 +263,16 @@ function firstSetBody(p, seen, resolve2) {
     case "token":
     case "leaf":
     case "node":
-    case "grammar":
-    case "expect":
-      return fs(d.parser);
     case "sepBy":
+    case "expect":
       return fs(d.parser);
-    // both min 0 and min 1 start with the item
+    case "grammar":
+      return firstSetOf(
+        d.parser,
+        seen,
+        resolve2,
+        d.clearTrivia ? null : d.triviaParser ?? activeTrivia
+      );
     default:
       return p._meta.firstSet;
   }
@@ -7659,7 +7660,9 @@ var Encoder = class {
   followClass(parsers, i) {
     let fs = { kind: "empty" };
     const rr = this.refResolver();
-    for (let j = i + 1; j < parsers.length; j++) fs = union(fs, firstSetOf(parsers[j], /* @__PURE__ */ new Set(), rr));
+    for (let j = i + 1; j < parsers.length; j++) {
+      fs = union(fs, firstSetOf(parsers[j], /* @__PURE__ */ new Set(), rr, this.activeTrivia));
+    }
     return this.charClass(fs);
   }
   constant(v) {
@@ -7876,7 +7879,7 @@ var Encoder = class {
    */
   subtree(c) {
     const ip = this.node(c).ip;
-    const fs = firstSetOf(c, /* @__PURE__ */ new Set(), this.refResolver());
+    const fs = firstSetOf(c, /* @__PURE__ */ new Set(), this.refResolver(), this.activeTrivia);
     return [ip, fs.kind === "empty" ? -2 : this.charClass(fs)];
   }
   /** Pool one scan spec. One per combinator: `node()` memoizes by identity. */
@@ -8095,12 +8098,12 @@ var Encoder = class {
           return this.covWrap(gated, armCovIds?.[src], 1);
         });
         const rr = this.refResolver();
-        const finalChoice = classifyFinalChoice(arms, rr);
+        const finalChoice = classifyFinalChoice(arms, rr, this.activeTrivia);
         const transactionalFirst = arms.map((arm, i) => {
           if (!finalChoice.nullable[i]) return void 0;
           const def = arm._def;
           if (def.tag !== "attempt" || matchesEmpty(def.parser, /* @__PURE__ */ new Set(), rr)) return void 0;
-          return firstSetOf(def.parser, /* @__PURE__ */ new Set(), rr);
+          return firstSetOf(def.parser, /* @__PURE__ */ new Set(), rr, this.activeTrivia);
         });
         const gateFirst = finalChoice.firstSets.map((first, i) => transactionalFirst[i] ?? first);
         const classes = gateFirst.map((first, i) => finalChoice.nullable[i] && transactionalFirst[i] === void 0 ? -1 : this.charClass(first));
@@ -8147,7 +8150,7 @@ var Encoder = class {
       case "oneOrMore": {
         const child = this.node(d.parser).ip;
         const rr = this.refResolver();
-        const itemClass = matchesEmpty(d.parser, /* @__PURE__ */ new Set(), rr) ? -1 : this.existingCharClass(firstSetOf(d.parser, /* @__PURE__ */ new Set(), rr));
+        const itemClass = matchesEmpty(d.parser, /* @__PURE__ */ new Set(), rr) ? -1 : this.existingCharClass(firstSetOf(d.parser, /* @__PURE__ */ new Set(), rr, this.activeTrivia));
         const op = d.valueUnused ? OP_REPV : OP_REP;
         const min = d.tag === "many" ? 0 : d.min;
         return this.rec ? this.emit(op, child, min, d.max ?? -1, -1, 0, this.expected(deriveExpected(d.parser)), itemClass) : this.emit(op, child, min, d.max ?? -1, -1, 0);
@@ -8165,7 +8168,7 @@ var Encoder = class {
             sep,
             flags,
             this.expected(deriveExpected(d.parser)),
-            this.charClass(firstSetOf(d.separator, /* @__PURE__ */ new Set(), this.refResolver()))
+            this.charClass(firstSetOf(d.separator, /* @__PURE__ */ new Set(), this.refResolver(), this.activeTrivia))
           );
         }
         return d.min >= 2 ? this.emit(OP_REP, child, d.min, d.max ?? -1, sep, flags, this.expected(deriveExpected(d.parser))) : this.emit(OP_REP, child, d.min, d.max ?? -1, sep, flags);
@@ -8220,7 +8223,7 @@ var Encoder = class {
           d.tags === void 0 ? -1 : this.constant(d.tags)
         );
         if (matchesEmpty(d.parser)) return body;
-        const fs = d.parser._def.tag === "lazy" ? firstSetOf(d.parser, /* @__PURE__ */ new Set(), this.refResolver()) : d.parser._meta.firstSet;
+        const fs = d.parser._def.tag === "lazy" ? firstSetOf(d.parser, /* @__PURE__ */ new Set(), this.refResolver(), this.activeTrivia) : d.parser._meta.firstSet;
         if (fs.kind === "any") return body;
         const cls = this.charClass(fs);
         if (cls < 0) return body;
@@ -8334,7 +8337,7 @@ var Encoder = class {
       case "attempt": {
         const inner = this.emit(OP_ATTEMPT, this.node(d.parser).ip);
         if (!this.failureNeedsRollback(d.parser, void 0)) this.failureRollbackCleanSites.add(inner);
-        const fs = d.parser._def.tag === "lazy" ? firstSetOf(d.parser, /* @__PURE__ */ new Set(), this.refResolver()) : d.parser._meta.firstSet;
+        const fs = d.parser._def.tag === "lazy" ? firstSetOf(d.parser, /* @__PURE__ */ new Set(), this.refResolver(), this.activeTrivia) : d.parser._meta.firstSet;
         if (fs.kind === "any" || matchesEmpty(d.parser)) return inner;
         const cls = this.charClass(fs);
         if (cls < 0) return inner;