punctilio 3.0.7 → 3.0.9

package/src/quotes.ts

@@ -3,7 +3,7 @@
  */
 
 import escapeStringRegexp from "escape-string-regexp"
-import { UNICODE_SYMBOLS, DEFAULT_SEPARATOR, LATIN_LETTERS, TERMINAL_PUNCTUATION } from "./constants.js"
+import { UNICODE_SYMBOLS, DEFAULT_SEPARATOR, LATIN_LETTERS, TERMINAL_PUNCTUATION, cachedRegExp } from "./constants.js"
 
 const {
   EM_DASH,
@@ -34,8 +34,8 @@ function convertSingleQuotes(text: string, sep: string): string {
   // Only match straight quotes, not already-converted curly quotes
   const singleQuoteChars = `'${LEFT_SINGLE_QUOTE}${RIGHT_SINGLE_QUOTE}${MODIFIER_LETTER_APOSTROPHE}`
   const singleQuoteOrWord = `[${singleQuoteChars}\\w]`
-  text = text.replace(new RegExp(`(?<!${singleQuoteOrWord})''(?!${singleQuoteOrWord})`, "g"), `${LEFT_SINGLE_QUOTE}${RIGHT_SINGLE_QUOTE}`)
-  text = text.replace(new RegExp(`(?<!${singleQuoteOrWord})'(\\s+)'(?!${singleQuoteOrWord})`, "g"), `${LEFT_SINGLE_QUOTE}$1${RIGHT_SINGLE_QUOTE}`)
+  text = text.replace(cachedRegExp(`(?<!${singleQuoteOrWord})''(?!${singleQuoteOrWord})`, "g"), `${LEFT_SINGLE_QUOTE}${RIGHT_SINGLE_QUOTE}`)
+  text = text.replace(cachedRegExp(`(?<!${singleQuoteOrWord})'(\\s+)'(?!${singleQuoteOrWord})`, "g"), `${LEFT_SINGLE_QUOTE}$1${RIGHT_SINGLE_QUOTE}`)
 
   const afterEndingSinglePatterns = `\\s\\.!?;,\\)${EM_DASH}\\-\\]"`
   // Full pattern with optional 's' for lookahead detection in apostropheRegex
@@ -45,32 +45,32 @@ function convertSingleQuotes(text: string, sep: string): string {
   // the general rules produced LSQ+n+RSQ which was fine. But MLA requires both
   // quotes to be MLA (semantic apostrophes), and the general rules can't achieve
   // that: neither quote is in a contraction context (no Latin letter on both sides).
-  text = text.replace(new RegExp(`(?<=\\w${escapedSep}? )['${RIGHT_SINGLE_QUOTE}]n['${RIGHT_SINGLE_QUOTE}](?= ${escapedSep}?\\w)`, "gm"), `${MODIFIER_LETTER_APOSTROPHE}n${MODIFIER_LETTER_APOSTROPHE}`)
+  text = text.replace(cachedRegExp(`(?<=\\w${escapedSep}? )['${RIGHT_SINGLE_QUOTE}]n['${RIGHT_SINGLE_QUOTE}](?= ${escapedSep}?\\w)`, "gm"), `${MODIFIER_LETTER_APOSTROPHE}n${MODIFIER_LETTER_APOSTROPHE}`)
 
   // Possessive: 's followed by ending context (e.g., dog's) → U+02BC
   const afterPossessive = `(?=${escapedSep}?s${escapedSep}?(?:[${afterEndingSinglePatterns}]|$))`
   const possessiveSingle = `(?<=[^\\s${LEFT_DOUBLE_QUOTE}'])['${RIGHT_SINGLE_QUOTE}]${afterPossessive}`
-  text = text.replace(new RegExp(possessiveSingle, "gm"), MODIFIER_LETTER_APOSTROPHE)
+  text = text.replace(cachedRegExp(possessiveSingle, "gm"), MODIFIER_LETTER_APOSTROPHE)
 
   // Closing single quote: ending context without 's' → U+2019
   const afterClosingSingle = `(?=${escapedSep}?(?:[${afterEndingSinglePatterns}]|$))`
   const closingSingle = `(?<=[^\\s${LEFT_DOUBLE_QUOTE}'])[']${afterClosingSingle}`
-  text = text.replace(new RegExp(closingSingle, "gm"), RIGHT_SINGLE_QUOTE)
+  text = text.replace(cachedRegExp(closingSingle, "gm"), RIGHT_SINGLE_QUOTE)
 
   const contraction = `(?<=[${LATIN_LETTERS}])['${RIGHT_SINGLE_QUOTE}${MODIFIER_LETTER_APOSTROPHE}](?=${escapedSep}?[${LATIN_LETTERS}])`
-  text = text.replace(new RegExp(contraction, "gm"), MODIFIER_LETTER_APOSTROPHE)
+  text = text.replace(cachedRegExp(contraction, "gm"), MODIFIER_LETTER_APOSTROPHE)
 
   const apostropheWhitelist = `(?=n${MODIFIER_LETTER_APOSTROPHE} )`
   const endQuoteNotContraction = `(?!${contraction})[${RIGHT_SINGLE_QUOTE}${MODIFIER_LETTER_APOSTROPHE}]${afterEndingSingle}`
   // Limit lookahead scan to 1000 chars to prevent catastrophic backtracking on pathological inputs
-  const apostropheRegex = new RegExp(
+  const apostropheRegex = cachedRegExp(
     `(?<=^|[^\\w])['](${apostropheWhitelist}|(?![^${LEFT_SINGLE_QUOTE}'\\n]{0,1000}${endQuoteNotContraction}))`,
     "gm"
   )
   text = text.replace(apostropheRegex, MODIFIER_LETTER_APOSTROPHE)
 
   const beginningSingle = `(?<beforeContext>(?:^|[\\s${LEFT_DOUBLE_QUOTE}${RIGHT_DOUBLE_QUOTE}${EM_DASH}\\-\\(])${escapedSep}?)['](?=${escapedSep}?\\S)`
-  text = text.replace(new RegExp(beginningSingle, "gm"), `$<beforeContext>${LEFT_SINGLE_QUOTE}`)
+  text = text.replace(cachedRegExp(beginningSingle, "gm"), `$<beforeContext>${LEFT_SINGLE_QUOTE}`)
 
   text = convertUnmatchedPluralPossessives(text, sep)
 
@@ -85,7 +85,7 @@ function convertSingleQuotes(text: string, sep: string): string {
 function convertUnmatchedPluralPossessives(text: string, sep: string): string {
   let singleQuoteBalance = 0
   return text.replace(
-    new RegExp(`[${LEFT_SINGLE_QUOTE}${RIGHT_SINGLE_QUOTE}]`, "g"),
+    cachedRegExp(`[${LEFT_SINGLE_QUOTE}${RIGHT_SINGLE_QUOTE}]`, "g"),
     (match, offset) => {
       if (match === LEFT_SINGLE_QUOTE) {
         singleQuoteBalance++
@@ -115,19 +115,19 @@ function convertDoubleQuotes(text: string, sep: string): string {
   // Handle whitespace-only quotes " " - require non-quote chars on both sides
   text = text.replace(/(?<=^|[\s([{])"(?<whitespace>\s+)"(?=$|[\s)\]}.!?,;:])/g, `${LEFT_DOUBLE_QUOTE}$<whitespace>${RIGHT_DOUBLE_QUOTE}`)
 
-  const beginningDouble = new RegExp(
+  const beginningDouble = cachedRegExp(
     `(?<=^|[\\s\\(\\/\\[\\{\\-${EM_DASH}${escapedSep}])(?<beforeChr>${escapedSep}?)["](?<afterChr>(?<sepWithPunct>${escapedSep}[ .,])|(?=${escapedSep}?\\.{3}|${escapedSep}?[^\\s\\)\\${EM_DASH},!?${escapedSep};:.\\}]))`,
     "gm"
   )
   text = text.replace(beginningDouble, `$<beforeChr>${LEFT_DOUBLE_QUOTE}$<afterChr>`)
 
-  text = text.replace(new RegExp(`(?<=\\{)(?<sepSpace>${escapedSep}? )?["]`, "g"), `$<sepSpace>${LEFT_DOUBLE_QUOTE}`)
+  text = text.replace(cachedRegExp(`(?<=\\{)(?<sepSpace>${escapedSep}? )?["]`, "g"), `$<sepSpace>${LEFT_DOUBLE_QUOTE}`)
 
   const endingDouble = `(?<beforeQuote>[^\\s\\(])["]((?<sepAfter>${escapedSep})?)(?=${escapedSep}|[\\s/\\).,;${EM_DASH}:\\-\\}!?s]|$)`
-  text = text.replace(new RegExp(endingDouble, "g"), `$<beforeQuote>${RIGHT_DOUBLE_QUOTE}$<sepAfter>`)
+  text = text.replace(cachedRegExp(endingDouble, "g"), `$<beforeQuote>${RIGHT_DOUBLE_QUOTE}$<sepAfter>`)
 
-  text = text.replace(new RegExp(`["](?<sepEnd>${escapedSep}?)$`, "g"), `${RIGHT_DOUBLE_QUOTE}$<sepEnd>`)
-  text = text.replace(new RegExp(`'(?=${RIGHT_DOUBLE_QUOTE})`, "gu"), RIGHT_SINGLE_QUOTE)
+  text = text.replace(cachedRegExp(`["](?<sepEnd>${escapedSep}?)$`, "g"), `${RIGHT_DOUBLE_QUOTE}$<sepEnd>`)
+  text = text.replace(cachedRegExp(`'(?=${RIGHT_DOUBLE_QUOTE})`, "gu"), RIGHT_SINGLE_QUOTE)
 
   return text
 }
@@ -138,14 +138,14 @@ function applyPunctuationStyle(text: string, sep: string, style: PunctuationStyl
 
   if (style === "american") {
     // Period outside → inside: "Hello". → "Hello."
-    const periodOutsideRegex = new RegExp(
+    const periodOutsideRegex = cachedRegExp(
       `(?<![${TERMINAL_PUNCTUATION_CLASS}])(?<sepBefore>${escapedSep}?)(?<quote>[${RIGHT_SINGLE_QUOTE}${RIGHT_DOUBLE_QUOTE}])(?<sepAfter>${escapedSep}?)(?!\\.\\.\\.)\\.`,
       "g"
     )
     text = text.replace(periodOutsideRegex, "$<sepBefore>.$<quote>$<sepAfter>")
 
     // Comma outside → inside: "Hello", → "Hello,"
-    const commaOutsideRegex = new RegExp(
+    const commaOutsideRegex = cachedRegExp(
       `(?<![${TERMINAL_PUNCTUATION_CLASS}])(?<sepBefore>${escapedSep}?)(?<quote>[${RIGHT_SINGLE_QUOTE}${RIGHT_DOUBLE_QUOTE}])(?<sepAfter>${escapedSep}?),`,
       "g"
     )
@@ -153,14 +153,14 @@ function applyPunctuationStyle(text: string, sep: string, style: PunctuationStyl
   } else if (style === "british") {
     // Period inside → outside: "Hello." → "Hello".
     // No terminal punctuation guard — "Stop!." inside is always wrong; move the period out.
-    const periodInsideRegex = new RegExp(
+    const periodInsideRegex = cachedRegExp(
       `(?<sepBefore>${escapedSep}?)\\.(?<sepMiddle>${escapedSep}?)(?<quote>[${RIGHT_SINGLE_QUOTE}${RIGHT_DOUBLE_QUOTE}])`,
       "g"
     )
     text = text.replace(periodInsideRegex, "$<sepBefore>$<sepMiddle>$<quote>.")
 
     // Comma inside → outside: "Hello," → "Hello",
-    const commaInsideRegex = new RegExp(
+    const commaInsideRegex = cachedRegExp(
       `,(?<sepAndQuote>${escapedSep}?[${RIGHT_DOUBLE_QUOTE}${RIGHT_SINGLE_QUOTE}])`,
       "g"
     )

package/src/rehype.ts

@@ -389,7 +389,8 @@ export function collectTransformableElements(
     // Otherwise, recurse into children
     for (const child of node.children) {
       if (child.type === "element") {
-        results.push(...collectTransformableElements(child, shouldSkip, depth + 1))
+        const childResults = collectTransformableElements(child, shouldSkip, depth + 1)
+        for (const r of childResults) results.push(r)
       }
     }
   }

package/src/symbols.ts

@@ -2,8 +2,7 @@
  * Symbol transformations: ellipses, multiplication, math symbols, arrows.
  */
 
-import escapeStringRegexp from "escape-string-regexp"
-import { UNICODE_SYMBOLS, ESCAPED_DEFAULT_SEPARATOR, LATIN_LETTERS, wordBoundaryEnd, SPACE_CHARS, spaceBoundaryStart, spaceBoundaryEnd, cachedRegExp } from "./constants.js"
+import { UNICODE_SYMBOLS, LATIN_LETTERS, wordBoundaryEnd, SPACE_CHARS, spaceBoundaryStart, spaceBoundaryEnd, cachedRegExp, getEscapedSeparator } from "./constants.js"
 
 export interface SymbolOptions {
   /** Boundary marker for HTML element boundaries. Default: "\uE000" */
@@ -41,9 +40,7 @@ const {
 
 /** Convert "..." or ". . ." to "…". */
 export function ellipsis(text: string, options: SymbolOptions = {}): string {
-  const chr = options.separator
-    ? escapeStringRegexp(options.separator)
-    : ESCAPED_DEFAULT_SEPARATOR
+  const chr = getEscapedSeparator(options)
 
   // Convert consecutive or spaced dots: ... or . . . → …
   // Captures preserve any separators between dots
@@ -59,9 +56,7 @@ export function ellipsis(text: string, options: SymbolOptions = {}): string {
 
 /** Convert "5x5" to "5×5". Skips hex (0x5F). */
 export function multiplication(text: string, options: SymbolOptions = {}): string {
-  const chr = options.separator
-    ? escapeStringRegexp(options.separator)
-    : ESCAPED_DEFAULT_SEPARATOR
+  const chr = getEscapedSeparator(options)
 
   // Match entire multiplication chains in one pass: "5 x 5 x 5" or "5x5x5"
   // Pattern matches: digit(s), then one or more (operator, digit(s)) groups
@@ -113,9 +108,7 @@ const MATH_SYMBOL_MAP: [string, string, string, string][] = [
 
 /** Convert !=, <=, >=, +/-, ~= to Unicode equivalents. */
 export function mathSymbols(text: string, options: SymbolOptions = {}): string {
-  const chr = options.separator
-    ? escapeStringRegexp(options.separator)
-    : ESCAPED_DEFAULT_SEPARATOR
+  const chr = getEscapedSeparator(options)
 
   for (const [left, right, lookahead, replacement] of MATH_SYMBOL_MAP) {
     const pattern = cachedRegExp(`${left}${chr}?${right}${lookahead}`, "g")
@@ -168,9 +161,7 @@ const ARROW_MAP: [string, string][] = [
 
 /** Convert -> and <-> to arrows. */
 export function arrows(text: string, options: SymbolOptions = {}): string {
-  const chr = options.separator
-    ? escapeStringRegexp(options.separator)
-    : ESCAPED_DEFAULT_SEPARATOR
+  const chr = getEscapedSeparator(options)
 
   const start = spaceBoundaryStart(chr)
   const end = spaceBoundaryEnd(chr)
@@ -185,9 +176,7 @@ export function arrows(text: string, options: SymbolOptions = {}): string {
 
 /** Convert "20 C" or "20 F" to "20 °C" or "20 °F". Only matches uppercase C/F. */
 export function degrees(text: string, options: SymbolOptions = {}): string {
-  const chr = options.separator
-    ? escapeStringRegexp(options.separator)
-    : ESCAPED_DEFAULT_SEPARATOR
+  const chr = getEscapedSeparator(options)
 
   // Temperature with optional space before C or F (uppercase only)
   // Handles separator between digit and unit
@@ -288,9 +277,7 @@ function balancedPrimeReplacer(primeChar: string, escapedSeparator: string) {
 
 /** Convert 5'10" to 5′10″ (prime marks). Call before smart quotes. */
 export function primeMarks(text: string, options: SymbolOptions = {}): string {
-  const escapedSeparator = options.separator
-    ? escapeStringRegexp(options.separator)
-    : ESCAPED_DEFAULT_SEPARATOR
+  const escapedSeparator = getEscapedSeparator(options)
 
   const quotePrimePairs: [string, string][] = [
     ["'", PRIME],
@@ -298,8 +285,8 @@ export function primeMarks(text: string, options: SymbolOptions = {}): string {
   ]
 
   for (const [quote, primeChar] of quotePrimePairs) {
-    const escapedQuote = escapeStringRegexp(quote)
-    const pattern = buildQuoteClassificationPattern(escapedQuote, escapedSeparator)
+    // ' and " are not regex-special, so they can be used directly as patterns
+    const pattern = buildQuoteClassificationPattern(quote, escapedSeparator)
     text = text.replace(pattern, balancedPrimeReplacer(primeChar, escapedSeparator))
   }
 
@@ -326,9 +313,7 @@ const FRACTION_MAP: Record<string, string> = {
 
 /** Convert 1/2, 1/4, etc. to ½, ¼, etc. */
 export function fractions(text: string, options: SymbolOptions = {}): string {
-  const chr = options.separator
-    ? escapeStringRegexp(options.separator)
-    : ESCAPED_DEFAULT_SEPARATOR
+  const chr = getEscapedSeparator(options)
 
   for (const [ascii, unicode] of Object.entries(FRACTION_MAP)) {
     // Negative lookbehind/lookahead: ensures fraction is not part of a larger number or path
@@ -356,9 +341,7 @@ const ORDINAL_MAP: Record<string, string> = {
 
 /** Convert 1st, 2nd, 3rd, 4th to superscript ordinals. */
 export function superscriptOrdinal(text: string, options: SymbolOptions = {}): string {
-  const chr = options.separator
-    ? escapeStringRegexp(options.separator)
-    : ESCAPED_DEFAULT_SEPARATOR
+  const chr = getEscapedSeparator(options)
 
   // Match number + optional separator + ordinal suffix at word boundary
   // Use case-insensitive matching for the suffix
@@ -397,9 +380,7 @@ const PUNCTUATION_LIGATURE_MAP: [string, string, string][] = [
 
 /** Convert ?? to ⁇, ?! to ⁈, !? to ⁉. Poor font support, disabled by default. */
 export function punctuationLigatures(text: string, options: SymbolOptions = {}): string {
-  const chr = options.separator
-    ? escapeStringRegexp(options.separator)
-    : ESCAPED_DEFAULT_SEPARATOR
+  const chr = getEscapedSeparator(options)
 
   for (const [first, repeated, replacement] of PUNCTUATION_LIGATURE_MAP) {
     const pattern = cachedRegExp(`${first}(${chr})?${repeated}(?:${chr}?${repeated})*`, "g")

package/src/tests/utils.test.ts

@@ -1,5 +1,5 @@
 import { assertSeparatorAbsent, countSeparators, assertSeparatorCountPreserved, formatErrorString } from "../utils.js"
-import { DEFAULT_SEPARATOR } from "../constants.js"
+import { DEFAULT_SEPARATOR, cachedRegExp } from "../constants.js"
 
 describe("assertSeparatorAbsent", () => {
   it.each([
@@ -75,6 +75,19 @@ describe("assertSeparatorCountPreserved", () => {
   })
 })
 
+describe("cachedRegExp", () => {
+  it("evicts oldest entry when cache exceeds max size", () => {
+    // Fill the cache well beyond 1000 entries with unique patterns
+    for (let i = 0; i < 1010; i++) {
+      const re = cachedRegExp(`unique-pattern-${i}`, "g")
+      expect(re).toBeInstanceOf(RegExp)
+    }
+    // Verify the cache still works correctly after eviction
+    const re = cachedRegExp("unique-pattern-1009", "g")
+    expect(re.source).toBe("unique-pattern-1009")
+  })
+})
+
 describe("formatErrorString", () => {
   let originalWrite: typeof process.stderr.write
   let stderrOutput: string[]