fix: throw when lookaheads contain anchors

Author: gruhn

benchmark/toStdRegex_output_length-result.txt

@@ -1,12 +1,12 @@
 
 failed instances:
-- parseError         : 17
-- cacheOverflow      : 94
-- veryLargeSyntaTree : 29
-- stackOverflow      : 11
+- parseError         : 45
+- cacheOverflow      : 89
+- veryLargeSyntaTree : 24
+- stackOverflow      : 12
 - regexSyntaxError   : 0
 
 size multipliers:
-- mean   : 324.6198162588445
-- median : 1.5333333333333334
+- mean   : 323.9035686487585
+- median : 1.2692307692307692
 - max    : 178441.7894736842

benchmark/toStdRegex_output_length.ts

@@ -1,5 +1,6 @@
 import fs from 'fs'
 import * as RE from '../src/regex'
+import * as AST from '../src/ast'
 import { UnsupportedSyntaxError, ParseError } from '../src/index'
 import { parseRegExp } from '../src/regex-parser'
 import { toStdRegex } from '../src/dfa.js'
@@ -16,7 +17,7 @@ const mults: number[] = []
 function run(inputRegExp: RegExp, index: number) {
   console.log('#' + index, inputRegExp)
 
-  const inputRegex = RE.fromRegExpAST(parseRegExp(inputRegExp))
+  const inputRegex = AST.toExtRegex(parseRegExp(inputRegExp))
   const outputRegex = toStdRegex(inputRegex)
   const outputRegExp = RE.toRegExp(outputRegex)
 

src/ast.ts

@@ -1,5 +1,6 @@
 import * as CharSet from './char-set'
 import * as RE from './regex'
+import { UnsupportedSyntaxError } from './regex-parser'
 import { assert, checkedAllCases, isOneOf } from './utils'
 
 /**
@@ -243,16 +244,22 @@ function pullUpStartAnchor(ast: RegExpAST): RegExpAST {
       }
     }
     case "positive-lookahead": {
+      const inner = pullUpStartAnchor(ast.inner)
       const right = pullUpStartAnchor(ast.right)
-      if (right.type === 'start-anchor') {
+      if (inner.type === 'start-anchor') {
+        throw new UnsupportedSyntaxError('start anchors (^) inside lookaheads are not supported')
+      } else if (right.type === 'start-anchor') {
         return startAnchor(undefined, positiveLookahead(ast.inner, right.right))
       } else {
         return positiveLookahead(ast.inner, right)
       }
     }
     case "negative-lookahead": {
+      const inner = pullUpStartAnchor(ast.inner)
       const right = pullUpStartAnchor(ast.right)
-      if (right.type === 'start-anchor') {
+      if (inner.type === 'start-anchor') {
+        throw new UnsupportedSyntaxError('start anchors (^) inside lookaheads are not supported')
+      } else if (right.type === 'start-anchor') {
         return startAnchor(undefined, negativeLookahead(ast.inner, right.right))
       } else {
         return negativeLookahead(ast.inner, right)
@@ -386,16 +393,22 @@ function pullUpEndAnchor(ast: RegExpAST): RegExpAST {
       }
     }
     case "positive-lookahead": {
+      const inner = pullUpEndAnchor(ast.inner)
       const right = pullUpEndAnchor(ast.right)
-      if (right.type === 'end-anchor') {
+      if (inner.type === 'end-anchor') {
+        throw new UnsupportedSyntaxError('end anchors ($) inside lookaheads are not supported')
+      } else if (right.type === 'end-anchor') {
         return endAnchor(positiveLookahead(ast.inner, right.left), undefined)
       } else {
         return positiveLookahead(ast.inner, right)
       }
     }
     case "negative-lookahead": {
+      const inner = pullUpEndAnchor(ast.inner)
       const right = pullUpEndAnchor(ast.right)
-      if (right.type === 'end-anchor') {
+      if (inner.type === 'end-anchor') {
+        throw new UnsupportedSyntaxError('end anchors ($) inside lookaheads are not supported')
+      } else if (right.type === 'end-anchor') {
         return endAnchor(negativeLookahead(ast.inner, right.right), undefined)
       } else {
         return negativeLookahead(ast.inner, right)

test/ast.spec.ts

@@ -77,6 +77,11 @@ describe('toExtRegex', () => {
         assert.equal(actual.hash, expected.hash)
       })
     }
+
+    it('fixme', {todo:true}, () => {
+      const re = /^(?!a)(?!.*a$)(?!.*a{2})a+$/
+      AST.toExtRegex(parseRegExp(re))     
+    })
   })
 
   describe('lookahead elimination', () => {