@nyaomaru/divider 2.0.3 → 2.0.5

package/dist/index.cjs

@@ -122,7 +122,10 @@ function isNoneMode(value) {
 function normalizeSeparators(separators) {
   return Array.from(new Set(separators)).filter(
     (separator) => !isEmptyString(separator)
-  );
+  ).sort((left, right) => right.length - left.length);
+}
+function createCacheKey(normalizedSeparators) {
+  return normalizedSeparators.join(CACHE_KEY_SEPARATOR);
 }
 var RegexCache = class {
   cache = /* @__PURE__ */ new Map();
@@ -138,6 +141,15 @@ var RegexCache = class {
    */
   get(separators) {
     const key = this.createKey(separators);
+    return this.getByKey(key);
+  }
+  /**
+   * Retrieves a cached RegExp for a precomputed cache key.
+   *
+   * @param key - Cache key string
+   * @returns Cached RegExp or null if not found
+   */
+  getByKey(key) {
     const regex = this.cache.get(key);
     if (regex) {
       this.cache.delete(key);
@@ -153,6 +165,15 @@ var RegexCache = class {
    */
   set(separators, regex) {
     const key = this.createKey(separators);
+    this.setByKey(key, regex);
+  }
+  /**
+   * Stores a RegExp in the cache for a precomputed cache key.
+   *
+   * @param key - Cache key string
+   * @param regex - Compiled RegExp to cache
+   */
+  setByKey(key, regex) {
     if (this.cache.has(key)) {
       this.cache.delete(key);
     }
@@ -173,7 +194,7 @@ var RegexCache = class {
    */
   createKey(separators) {
     const normalizedSeparators = normalizeSeparators(separators);
-    return normalizedSeparators.join(CACHE_KEY_SEPARATOR);
+    return createCacheKey(normalizedSeparators);
   }
   /**
    * Gets current cache size for debugging/monitoring.
@@ -194,15 +215,16 @@ var RegexCache = class {
 var regexCache = new RegexCache();
 function getRegex(separators) {
   if (isEmptyArray(separators)) return null;
-  const cached = regexCache.get(separators);
-  if (cached) return cached;
   const uniqueSeparators = normalizeSeparators(separators);
   if (uniqueSeparators.length === 0) {
     return null;
   }
+  const cacheKey = createCacheKey(uniqueSeparators);
+  const cached = regexCache.getByKey(cacheKey);
+  if (cached) return cached;
   const pattern = uniqueSeparators.map(escapeRegExp).join("|");
   const regex = new RegExp(`(?:${pattern})`, "g");
-  regexCache.set(separators, regex);
+  regexCache.setByKey(cacheKey, regex);
   return regex;
 }
 function escapeRegExp(str) {
@@ -422,13 +444,41 @@ function dividePreserve(input, separator) {
   if (isEmptyString(input)) return [""];
   return divider(input, separator, { preserveEmpty: true });
 }
-function countUnescaped(text, quote) {
-  const pair = quote + quote;
+var countUnescapedSingleChar = (text, quote) => {
   let count = 0;
-  for (const chunk of dividePreserve(text, pair)) {
-    count += dividePreserve(chunk, quote).length - 1;
+  for (let index = 0; index < text.length; index++) {
+    if (text[index] !== quote) continue;
+    if (text[index + 1] === quote) {
+      index++;
+      continue;
+    }
+    count++;
   }
   return count;
+};
+var countUnescapedMultiChar = (text, quote) => {
+  const escapedPair = quote + quote;
+  const quoteLength = quote.length;
+  const escapedPairLength = escapedPair.length;
+  let count = 0;
+  for (let index = 0; index < text.length; ) {
+    if (text.startsWith(escapedPair, index)) {
+      index += escapedPairLength;
+      continue;
+    }
+    if (text.startsWith(quote, index)) {
+      count++;
+      index += quoteLength;
+      continue;
+    }
+    index++;
+  }
+  return count;
+};
+function countUnescaped(text, quote) {
+  if (isEmptyString(quote)) return 0;
+  if (quote.length === 1) return countUnescapedSingleChar(text, quote);
+  return countUnescapedMultiChar(text, quote);
 }
 function stripOuterQuotes(text, quoteChar, { lenient = true } = {}) {
   const escapedPair = quoteChar + quoteChar;
@@ -480,7 +530,8 @@ var buildQuotedFields = (line, delimiter, quote, trim, lenient) => {
   const pieces = dividePreserve(line, delimiter);
   const state = {
     fields: [],
-    current: ""
+    current: "",
+    insideQuotes: false
   };
   for (const piece of pieces) {
     appendPiece(state, piece, delimiter, quote, trim, lenient);
@@ -490,10 +541,18 @@ var buildQuotedFields = (line, delimiter, quote, trim, lenient) => {
   }
   return state.fields;
 };
+var advanceQuoteState = (insideQuotes, segment, quote) => {
+  let nextInsideQuotes = insideQuotes;
+  for (const char of segment) {
+    if (char === quote) nextInsideQuotes = !nextInsideQuotes;
+  }
+  return nextInsideQuotes;
+};
 var appendPiece = (state, piece, delimiter, quote, trim, lenient) => {
-  state.current = isEmptyString(state.current) ? piece : state.current + delimiter + piece;
-  const insideQuotes = countUnescaped(state.current, quote) % 2 === 1;
-  if (!insideQuotes) {
+  const segment = isEmptyString(state.current) ? piece : delimiter + piece;
+  state.current += segment;
+  state.insideQuotes = quote.length === 1 ? advanceQuoteState(state.insideQuotes, segment, quote) : countUnescaped(state.current, quote) % 2 === 1;
+  if (!state.insideQuotes) {
     flushField(state, quote, trim, lenient);
   }
 };
@@ -502,6 +561,7 @@ var flushField = (state, quote, trim, lenient) => {
   if (trim) fieldValue = fieldValue.trim();
   state.fields.push(fieldValue);
   state.current = "";
+  state.insideQuotes = false;
 };
 
 // src/presets/csv-divider.ts

package/dist/index.js

@@ -88,7 +88,10 @@ function isNoneMode(value) {
 function normalizeSeparators(separators) {
   return Array.from(new Set(separators)).filter(
     (separator) => !isEmptyString(separator)
-  );
+  ).sort((left, right) => right.length - left.length);
+}
+function createCacheKey(normalizedSeparators) {
+  return normalizedSeparators.join(CACHE_KEY_SEPARATOR);
 }
 var RegexCache = class {
   cache = /* @__PURE__ */ new Map();
@@ -104,6 +107,15 @@ var RegexCache = class {
    */
   get(separators) {
     const key = this.createKey(separators);
+    return this.getByKey(key);
+  }
+  /**
+   * Retrieves a cached RegExp for a precomputed cache key.
+   *
+   * @param key - Cache key string
+   * @returns Cached RegExp or null if not found
+   */
+  getByKey(key) {
     const regex = this.cache.get(key);
     if (regex) {
       this.cache.delete(key);
@@ -119,6 +131,15 @@ var RegexCache = class {
    */
   set(separators, regex) {
     const key = this.createKey(separators);
+    this.setByKey(key, regex);
+  }
+  /**
+   * Stores a RegExp in the cache for a precomputed cache key.
+   *
+   * @param key - Cache key string
+   * @param regex - Compiled RegExp to cache
+   */
+  setByKey(key, regex) {
     if (this.cache.has(key)) {
       this.cache.delete(key);
     }
@@ -139,7 +160,7 @@ var RegexCache = class {
    */
   createKey(separators) {
     const normalizedSeparators = normalizeSeparators(separators);
-    return normalizedSeparators.join(CACHE_KEY_SEPARATOR);
+    return createCacheKey(normalizedSeparators);
   }
   /**
    * Gets current cache size for debugging/monitoring.
@@ -160,15 +181,16 @@ var RegexCache = class {
 var regexCache = new RegexCache();
 function getRegex(separators) {
   if (isEmptyArray(separators)) return null;
-  const cached = regexCache.get(separators);
-  if (cached) return cached;
   const uniqueSeparators = normalizeSeparators(separators);
   if (uniqueSeparators.length === 0) {
     return null;
   }
+  const cacheKey = createCacheKey(uniqueSeparators);
+  const cached = regexCache.getByKey(cacheKey);
+  if (cached) return cached;
   const pattern = uniqueSeparators.map(escapeRegExp).join("|");
   const regex = new RegExp(`(?:${pattern})`, "g");
-  regexCache.set(separators, regex);
+  regexCache.setByKey(cacheKey, regex);
   return regex;
 }
 function escapeRegExp(str) {
@@ -388,13 +410,41 @@ function dividePreserve(input, separator) {
   if (isEmptyString(input)) return [""];
   return divider(input, separator, { preserveEmpty: true });
 }
-function countUnescaped(text, quote) {
-  const pair = quote + quote;
+var countUnescapedSingleChar = (text, quote) => {
   let count = 0;
-  for (const chunk of dividePreserve(text, pair)) {
-    count += dividePreserve(chunk, quote).length - 1;
+  for (let index = 0; index < text.length; index++) {
+    if (text[index] !== quote) continue;
+    if (text[index + 1] === quote) {
+      index++;
+      continue;
+    }
+    count++;
   }
   return count;
+};
+var countUnescapedMultiChar = (text, quote) => {
+  const escapedPair = quote + quote;
+  const quoteLength = quote.length;
+  const escapedPairLength = escapedPair.length;
+  let count = 0;
+  for (let index = 0; index < text.length; ) {
+    if (text.startsWith(escapedPair, index)) {
+      index += escapedPairLength;
+      continue;
+    }
+    if (text.startsWith(quote, index)) {
+      count++;
+      index += quoteLength;
+      continue;
+    }
+    index++;
+  }
+  return count;
+};
+function countUnescaped(text, quote) {
+  if (isEmptyString(quote)) return 0;
+  if (quote.length === 1) return countUnescapedSingleChar(text, quote);
+  return countUnescapedMultiChar(text, quote);
 }
 function stripOuterQuotes(text, quoteChar, { lenient = true } = {}) {
   const escapedPair = quoteChar + quoteChar;
@@ -446,7 +496,8 @@ var buildQuotedFields = (line, delimiter, quote, trim, lenient) => {
   const pieces = dividePreserve(line, delimiter);
   const state = {
     fields: [],
-    current: ""
+    current: "",
+    insideQuotes: false
   };
   for (const piece of pieces) {
     appendPiece(state, piece, delimiter, quote, trim, lenient);
@@ -456,10 +507,18 @@ var buildQuotedFields = (line, delimiter, quote, trim, lenient) => {
   }
   return state.fields;
 };
+var advanceQuoteState = (insideQuotes, segment, quote) => {
+  let nextInsideQuotes = insideQuotes;
+  for (const char of segment) {
+    if (char === quote) nextInsideQuotes = !nextInsideQuotes;
+  }
+  return nextInsideQuotes;
+};
 var appendPiece = (state, piece, delimiter, quote, trim, lenient) => {
-  state.current = isEmptyString(state.current) ? piece : state.current + delimiter + piece;
-  const insideQuotes = countUnescaped(state.current, quote) % 2 === 1;
-  if (!insideQuotes) {
+  const segment = isEmptyString(state.current) ? piece : delimiter + piece;
+  state.current += segment;
+  state.insideQuotes = quote.length === 1 ? advanceQuoteState(state.insideQuotes, segment, quote) : countUnescaped(state.current, quote) % 2 === 1;
+  if (!state.insideQuotes) {
     flushField(state, quote, trim, lenient);
   }
 };
@@ -468,6 +527,7 @@ var flushField = (state, quote, trim, lenient) => {
   if (trim) fieldValue = fieldValue.trim();
   state.fields.push(fieldValue);
   state.current = "";
+  state.insideQuotes = false;
 };
 
 // src/presets/csv-divider.ts

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@nyaomaru/divider",
   "type": "module",
-  "version": "2.0.3",
+  "version": "2.0.5",
   "description": "To divide string or string[] with a given separator",
   "main": "./dist/index.cjs",
   "module": "./dist/index.js",