parseman 0.23.0 → 0.24.0

package/dist/index.js

@@ -343,24 +343,35 @@ function literal(value, opts = {}) {
       { lo: firstUpper, hi: firstUpper }
     ] } : firstSet2;
   }
+  const expected = [JSON.stringify(value)];
+  const parse2 = !caseInsensitive && value.length === 1 ? (() => {
+    const code = value.charCodeAt(0);
+    return function parse3(input, pos, ctx) {
+      if (input.charCodeAt(pos) === code) {
+        const span = { start: pos, end: pos + 1 };
+        if (cstCaptureActive(ctx)) pushCstLeaf(ctx, { _tag: "leaf", value, span });
+        return { ok: true, value, span };
+      }
+      return failAt(ctx, expected, pos);
+    };
+  })() : function parse3(input, pos, ctx) {
+    const end = pos + value.length;
+    if (end > input.length) {
+      return failAt(ctx, expected, pos);
+    }
+    const matchedValue = caseInsensitive ? input.slice(pos, end) : value;
+    if (caseInsensitive ? asciiFoldEq(matchedValue, value) : input.startsWith(value, pos)) {
+      const span = { start: pos, end };
+      if (cstCaptureActive(ctx)) pushCstLeaf(ctx, { _tag: "leaf", value: matchedValue, span });
+      return { ok: true, value: matchedValue, span };
+    }
+    return failAt(ctx, expected, pos);
+  };
   return {
     _tag: "literal",
     _meta: meta,
     _def: { tag: "literal", value, caseInsensitive },
-    parse(input, pos, ctx) {
-      const end = pos + value.length;
-      if (end > input.length) {
-        return failAt(ctx, [JSON.stringify(value)], pos);
-      }
-      const matchedValue = caseInsensitive ? input.slice(pos, end) : value;
-      if (caseInsensitive ? asciiFoldEq(matchedValue, value) : input.startsWith(value, pos)) {
-        const span = { start: pos, end };
-        const leaf = { _tag: "leaf", value: matchedValue, span };
-        if (cstCaptureActive(ctx)) pushCstLeaf(ctx, leaf);
-        return { ok: true, value: matchedValue, span };
-      }
-      return failAt(ctx, [JSON.stringify(value)], pos);
-    }
+    parse: parse2
   };
 }
 
@@ -495,8 +506,7 @@ function regex(pattern, flags = "") {
         if (scanEnd === null) return failAt(ctx, [`/${source}/`], pos);
         const value = input.slice(pos, scanEnd);
         const span2 = { start: pos, end: scanEnd };
-        const leaf2 = { _tag: "leaf", value, span: span2 };
-        if (cstCaptureActive(ctx)) pushCstLeaf(ctx, leaf2);
+        if (cstCaptureActive(ctx)) pushCstLeaf(ctx, { _tag: "leaf", value, span: span2 });
         return { ok: true, value, span: span2 };
       }
       anchored.lastIndex = pos;
@@ -505,8 +515,7 @@ function regex(pattern, flags = "") {
         return failAt(ctx, [`/${source}/`], pos);
       }
       const span = { start: pos, end: pos + m[0].length };
-      const leaf = { _tag: "leaf", value: m[0], span };
-      if (cstCaptureActive(ctx)) pushCstLeaf(ctx, leaf);
+      if (cstCaptureActive(ctx)) pushCstLeaf(ctx, { _tag: "leaf", value: m[0], span });
       return { ok: true, value: m[0], span };
     }
   };
@@ -814,8 +823,7 @@ function keywords(words, opts = {}) {
       }
       const value = m[0];
       const span = { start: pos, end: pos + value.length };
-      const leaf = { _tag: "leaf", value, span };
-      if (cstCaptureActive(ctx)) pushCstLeaf(ctx, leaf);
+      if (cstCaptureActive(ctx)) pushCstLeaf(ctx, { _tag: "leaf", value, span });
       return { ok: true, value, span };
     }
   };
@@ -1305,33 +1313,136 @@ function loopScanner(arms) {
 }
 function regexTriviaScanner(parser2) {
   if (parser2._def.tag !== "regex" || parser2._def.flags) return null;
-  const ws = whitespaceSource(parser2._def.source);
-  if (ws) return ws;
-  return blockCommentSource(parser2._def.source) ? scanBlockComment : null;
+  const source = parser2._def.source;
+  return classRunSource(source) ?? altStarSource(source) ?? (blockCommentSource(source) ? scanBlockComment : null);
 }
-function whitespaceSource(source) {
-  const m = /^\[([ \\tnrf]+)\][*+]$/.exec(source);
-  if (!m) return null;
-  const chars = /* @__PURE__ */ new Set();
-  for (let i = 0; i < m[1].length; i++) {
-    const ch = m[1][i];
-    if (ch !== "\\") {
-      chars.add(ch.charCodeAt(0));
-      continue;
+function inRanges2(cp, ranges) {
+  for (let i = 0; i < ranges.length; i++) {
+    const r = ranges[i];
+    if (cp >= r[0] && cp <= r[1]) return true;
+  }
+  return false;
+}
+function classScanner(classBody) {
+  const ranges = parseClassRanges(classBody);
+  if (!ranges) return null;
+  return (input, cur) => {
+    let pos = cur;
+    while (pos < input.length && inRanges2(input.charCodeAt(pos), ranges)) pos++;
+    return pos;
+  };
+}
+function lineCommentScanner(commentCode) {
+  return (input, cur) => {
+    if (input.charCodeAt(cur) !== commentCode) return cur;
+    let pos = cur + 1;
+    while (pos < input.length) {
+      const cc = input.charCodeAt(pos);
+      if (cc === 10 || cc === 13) break;
+      pos++;
     }
-    const esc = m[1][++i];
-    if (esc === "t") chars.add(9);
-    else if (esc === "n") chars.add(10);
-    else if (esc === "r") chars.add(13);
-    else if (esc === "f") chars.add(12);
-    else return null;
+    return pos;
+  };
+}
+function classRunSource(source) {
+  const m = /^\[([^\]^](?:[^\]]|\\.)*)\][*+]$/.exec(source);
+  return m ? classScanner(m[1]) : null;
+}
+function classifyTriviaArm(arm) {
+  const cls = /^\[([^\]^](?:[^\]]|\\.)*)\][*+]?$/.exec(arm);
+  if (cls) {
+    const ranges = parseClassRanges(cls[1]);
+    return ranges ? { kind: "class", ranges } : null;
+  }
+  const lc = /^(\\?.)\[\^\\n\\r\]\*$/.exec(arm);
+  if (lc) {
+    const marker = lc[1];
+    return { kind: "comment", code: (marker.length === 2 ? marker[1] : marker[0]).charCodeAt(0) };
   }
+  return null;
+}
+function armScanner(arm) {
+  if (arm.kind === "comment") return lineCommentScanner(arm.code);
+  const ranges = arm.ranges;
+  return (input, cur) => {
+    let pos = cur;
+    while (pos < input.length && inRanges2(input.charCodeAt(pos), ranges)) pos++;
+    return pos;
+  };
+}
+function fusedTriviaScanner(ranges, commentCodes) {
+  const c0 = commentCodes[0];
+  const single = commentCodes.length === 1;
   return (input, cur) => {
     let pos = cur;
-    while (pos < input.length && chars.has(input.charCodeAt(pos))) pos++;
+    const len = input.length;
+    for (; ; ) {
+      const c = input.charCodeAt(pos);
+      if (pos < len && inRanges2(c, ranges)) {
+        pos++;
+        continue;
+      }
+      if (single ? c === c0 : commentCodes.includes(c)) {
+        pos++;
+        while (pos < len) {
+          const cc = input.charCodeAt(pos);
+          if (cc === 10 || cc === 13) break;
+          pos++;
+        }
+        continue;
+      }
+      break;
+    }
     return pos;
   };
 }
+function splitTopLevelAlts(body) {
+  const arms = [];
+  let inClass = false;
+  let start = 0;
+  for (let i = 0; i < body.length; i++) {
+    const c = body[i];
+    if (c === "\\") {
+      i++;
+      continue;
+    }
+    if (inClass) {
+      if (c === "]") inClass = false;
+      continue;
+    }
+    if (c === "[") inClass = true;
+    else if (c === "(") return null;
+    else if (c === "|") {
+      arms.push(body.slice(start, i));
+      start = i + 1;
+    }
+  }
+  arms.push(body.slice(start));
+  return arms;
+}
+function altStarSource(source) {
+  const m = /^\(\?:(.*)\)[*+]$/.exec(source);
+  if (!m) return null;
+  const armSrcs = splitTopLevelAlts(m[1]);
+  if (!armSrcs || armSrcs.length < 2) return null;
+  const arms = [];
+  for (const src of armSrcs) {
+    const arm = classifyTriviaArm(src);
+    if (!arm) return null;
+    arms.push(arm);
+  }
+  const ranges = [];
+  const commentCodes = [];
+  for (const arm of arms) {
+    if (arm.kind === "class") ranges.push(...arm.ranges);
+    else commentCodes.push(arm.code);
+  }
+  if (commentCodes.some((code) => inRanges2(code, ranges))) {
+    return loopScanner(arms.map(armScanner));
+  }
+  if (commentCodes.length === 0) return armScanner({ kind: "class", ranges });
+  return fusedTriviaScanner(ranges, commentCodes);
+}
 function blockCommentSource(source) {
   return source === "\\/\\*(?:[^*]|\\*(?!\\/))*\\*\\/" || source === "\\/\\*[^]*?\\*\\/";
 }
@@ -1685,6 +1796,7 @@ function markUnusedValues(root) {
 }
 
 // src/combinators/parser.ts
+var RULE_ORDER = "__parsemanRuleOrder";
 function tagRule(r, key) {
   ;
   r._ruleName = key;
@@ -1729,6 +1841,11 @@ function rules(a, b) {
   for (const key of Object.keys(definitions)) {
     markUnusedValues(cache[key]);
   }
+  Object.defineProperty(cache, RULE_ORDER, {
+    value: Object.keys(definitions),
+    enumerable: false,
+    configurable: true
+  });
   return cache;
 }