Correctly handle unicode ranges

This should correctly generate UTF-16 ranges for
Peggy from any valid ABNF range.
Fixes #25.

.gitignore

@@ -5,3 +5,4 @@ examples/*.peggy
 examples/*.pest
 node_modules
 t.abnf
+.cert/

bin/abnf_gen.js

@@ -112,6 +112,6 @@ program
   })
   .parseAsync()
   .catch(er => {
-    console.error(er.message);
+    console.error(er);
     process.exit(1);
   });

lib/ast.js

@@ -42,15 +42,28 @@ function slug(s) {
   return s.replace(/-/g, "_");
 }
 
-function str(s) {
-  return `"${s.replace(/[\\"\x00-\x19\x7f-\xff]/g, c => `\\${{
+function str(s, opts) {
+  s = `"${s.replace(/[\\"\x00-\x19\x7f-\xff]/g, c => `\\${{
     "\r": "r",
     "\n": "n",
     '"': '"',
     "\t": "t",
     "\v": "x0B",
     "\\": "\\",
   }[c] || `x${c.charCodeAt(0).toString(16).padStart(2, "0")}`}`)}"`;
+
+  if (opts?.format === "peggy") {
+    s = s.replace(
+      /[\u0100-\uffff]/,
+      c => `\\u${c.codePointAt(0).toString(16).padStart(4, 0)}`
+    );
+    s = s.replace(
+      /[\u{10000}-\u{10ffff}]/u,
+      c => `\\u{${c.codePointAt(0).toString(16)}}`
+    );
+  }
+
+  return s;
 }
 
 function fromArray(opts, a, joiner, needed, parent) {
@@ -132,7 +145,7 @@ export class CaseInsensitiveString extends Base {
   }
 
   toFormat(opts) {
-    const s = str(this.str);
+    const s = str(this.str, opts);
     if (this.str.match(/[a-z]/i)) {
       switch (opts.format) {
         case "peggy":
@@ -155,8 +168,8 @@ export class CaseSensitiveString extends Base {
     this.base = base;
   }
 
-  toFormat(_opts) {
-    return str(this.str);
+  toFormat(opts) {
+    return str(this.str, opts);
   }
 }
 
@@ -299,21 +312,92 @@ export class Range extends Base {
   }
 
   static create(base, first, last, loc, utf16 = true) {
-    if (utf16 && (first <= 0xffff) && (last === 0x10ffff)) {
-      // Special case "all high Unicode" since it shows up a lot
-      // This should be generalized.
+    if (first > last) {
+      throw new Error(`Range out of order ${first.toString(16)}-${last.toString(16)}`);
+    }
+    if (first === last) {
+      return new CaseSensitiveString(String.fromCodePoint(first), base, loc);
+    }
+    if (utf16) {
+      if (last < 0xd800) {
+        return new Range(base, first, last, loc);
+      }
+      if ((first > 0xd7ff) && (first < 0xe000) && (last < 0xe000)) {
+        throw new Error(`Range consists of all surrogates ${first.toString(16)}-${last.toString(16)}`);
+      }
+
+      // Remove the range 0xd800-0xdfff.  These possible ranges remain:
+      // - first-0xd7ff
+      // - 0xe000-0xffff
+      // - 0x10000-last
       const alts = [];
+
       if (first < 0xd800) {
-        alts.push(new Range(base, first, 0xd7ff, loc));
-        alts.push(new Range(base, 0xe000, 0xffff, loc));
-      } else {
-        alts.push(new Range(base, first, 0xffff, loc));
+        alts.push(Range.create(base, first, 0xd7ff, loc, false));
       }
-      alts.push(new Concatenation([
-        new Range(base, 0xd800, 0xdbff, loc),
-        new Range(base, 0xdc00, 0xdfff, loc),
-      ], loc));
+      if (last < 0x10000) {
+        alts.push(
+          Range.create(base, Math.max(0xe000, first), last, loc, false)
+        );
+      } else {
+        if (first < 0x10000) {
+          alts.push(
+            Range.create(base, Math.max(0xe000, first), 0xffff, loc, false)
+          );
+          first = 0x10000;
+        }
 
+        // This code follows the logic in regenerate:
+        // https://github.com/mathiasbynens/regenerate/blob/11567339f40fd262435934d544885bc047cb4220/regenerate.js#L996
+        // I didn't use regenerate directly because:
+        // a) I only needed a small part of it
+        // b) Regenerate will only generate a string, which I then would have
+        //    to parse to get the info I needed out.
+        // I believe this use is within the spirit of the MIT license.
+        const firstH = Math.floor((first - 0x10000) / 0x400) + 0xd800;
+        const firstL = ((first - 0x10000) % 0x400) + 0xdc00;
+        const lastH = Math.floor((last - 0x10000) / 0x400) + 0xd800;
+        const lastL = ((last - 0x10000) % 0x400) + 0xdc00;
+        let complete = false;
+        if (
+          (firstH === lastH)
+          || ((firstL === 0xdc00) && (lastL === 0xdfff))
+        ) {
+          alts.push(new Concatenation([
+            Range.create(base, firstH, lastH, loc, false),
+            Range.create(base, firstL, lastL, loc, false),
+          ], loc));
+          complete = true;
+        } else {
+          alts.push(new Concatenation([
+            Range.create(base, firstH, firstH, loc, false),
+            Range.create(base, firstL, 0xdfff, loc, false),
+          ], loc));
+        }
+        if (!complete && (firstH + 1 < lastH)) {
+          if (lastL === 0xdfff) {
+            alts.push(new Concatenation([
+              Range.create(base, firstH + 1, lastH + 1, loc, false),
+              Range.create(base, 0xdc00, lastL, loc, false),
+            ], loc));
+            complete = true;
+          } else {
+            alts.push(new Concatenation([
+              Range.create(base, firstH + 1, lastH - 1, loc, false),
+              Range.create(base, 0xdc00, 0xdfff, loc, false),
+            ], loc));
+          }
+        }
+        if (!complete) {
+          alts.push(new Concatenation([
+            Range.create(base, lastH, lastH, loc, false),
+            Range.create(base, 0xdc00, lastL, loc, false),
+          ], loc));
+        }
+      }
+      if (alts.length === 1) {
+        return alts[0];
+      }
       return new Alternation(alts, loc);
     }
     return new Range(base, first, last, loc);
@@ -339,6 +423,9 @@ export class Range extends Base {
     const { rangeBefore, rangeAfter, rangeSep } = delims(opts);
     const first = Range.escape(opts, this.first);
     const last = Range.escape(opts, this.last);
+    if (this.first + 1 === this.last) {
+      return `${rangeBefore}${first}${last}${rangeAfter}`;
+    }
     return `${rangeBefore}${first}${rangeSep}${last}${rangeAfter}`;
   }
 }

package.json

@@ -50,26 +50,20 @@
     "ci": "npm run coverage && npm run lint"
   },
   "dependencies": {
-    "commander": "^12.1.0",
+    "commander": "^13.0.0",
     "peggy": "^4.2.0"
   },
   "devDependencies": {
     "@peggyjs/coverage": "1.3.2",
-    "@peggyjs/eslint-config": "^5.0.1",
-    "@typescript-eslint/eslint-plugin": "^8.18.0",
-    "@typescript-eslint/parser": "^8.18.0",
+    "@peggyjs/eslint-config": "^5.0.3",
+    "@typescript-eslint/eslint-plugin": "^8.19.0",
+    "@typescript-eslint/parser": "^8.19.0",
     "ava": "6.2.0",
     "c8": "10.1.3",
-    "eslint": "^9.16.0",
+    "eslint": "^9.17.0",
     "typescript": "^5.7.2"
   },
-  "packageManager": "pnpm@9.15.0",
-  "pnpm": {
-    "overrides": {
-      "cross-spawn": "^7.0.6",
-      "@eslint/plugin-kit": "^0.2.4"
-    }
-  },
+  "packageManager": "pnpm@9.15.2",
   "engines": {
     "node": ">=18"
   }