@happyvertical/smrt-content 0.37.0 → 0.37.1

package/dist/index.js

@@ -7017,7 +7017,6 @@ const CURRENCY = {
   yen: "¥",
   euro: "€",
   dollar: "$",
-  euro: "€",
   fnof: "ƒ",
   inr: "₹",
   af: "؋",
@@ -7232,6 +7231,7 @@ class EntityDecoder {
    */
   decode(str) {
     if (typeof str !== "string" || str.length === 0) return str;
+    if (str.indexOf("&") === -1) return str;
     const original = str;
     const chunks = [];
     const len = str.length;
@@ -7862,17 +7862,244 @@ function validateEntityName(name) {
   else
     throw new Error(`Invalid entity name ${name}`);
 }
+const SCRIPT_ZEROS = [
+  // Basic Latin (ASCII) — included for completeness / pass-through
+  48,
+  // 0-9
+  // Arabic scripts
+  1632,
+  // Arabic-Indic ٠١٢٣٤٥٦٧٨٩
+  1776,
+  // Extended Arabic-Indic (Urdu/Persian/Sindhi) ۰۱۲۳
+  // Indic scripts
+  2406,
+  // Devanagari ०१२३४५६७८९
+  2534,
+  // Bengali ০১২৩৪৫৬৭৮৯
+  2662,
+  // Gurmukhi ੦੧੨੩੪੫੬੭੮੯
+  2790,
+  // Gujarati ૦૧૨૩૪૫૬૭૮૯
+  2918,
+  // Odia ୦୧୨୩୪୫୬୭୮୯
+  3046,
+  // Tamil ௦௧௨௩௪௫௬௭௮௯
+  3174,
+  // Telugu ౦౧౨౩౪౫౬౭౮౯
+  3302,
+  // Kannada ೦೧೨೩೪೫೬೭೮೯
+  3430,
+  // Malayalam ൦൧൨൩൪൫൬൭൮൯
+  3558,
+  // Sinhala Archaic ෦෧෨෩෪෫෬෭෮෯
+  // Southeast Asian scripts
+  3664,
+  // Thai ๐๑๒๓๔๕๖๗๘๙
+  3792,
+  // Lao ໐໑໒໓໔໕໖໗໘໙
+  3872,
+  // Tibetan ༠༡༢༣༤༥༦༧༨༩
+  4160,
+  // Myanmar ၀၁၂၃၄၅၆၇၈၉
+  4240,
+  // Myanmar Shan ႐႑႒႓႔႕႖႗႘႙
+  6112,
+  // Khmer ០១២៣៤៥៦៧៨៩
+  6160,
+  // Mongolian ᠐᠑᠒᠓᠔᠕᠖᠗᠘᠙
+  6470,
+  // Limbu ᥆᥇᥈᥉᥊᥋᥌᥍᥎᥏
+  6608,
+  // New Tai Lue ᧐᧑᧒᧓᧔᧕᧖᧗᧘᧙
+  6784,
+  // Tai Tham Hora ᪀᪁᪂᪃᪄᪅᪆᪇᪈᪉
+  6800,
+  // Tai Tham Tham ᪐᪑᪒᪓᪔᪕᪖᪗᪘᪙
+  6992,
+  // Balinese ᭐᭑᭒᭓᭔᭕᭖᭗᭘᭙
+  7088,
+  // Sundanese ᮰᮱᮲᮳᮴᮵᮶᮷᮸᮹
+  7232,
+  // Lepcha ᱀᱁᱂᱃᱄᱅᱆᱇᱈᱉
+  7248,
+  // Ol Chiki ᱐᱑᱒᱓᱔᱕᱖᱗᱘᱙
+  // Fullwidth (CJK context)
+  65296,
+  // Fullwidth ０１２３４５６７８９
+  // Mathematical digit variants (Unicode math block)
+  120782,
+  // Mathematical Bold
+  120792,
+  // Mathematical Double-Struck
+  120802,
+  // Mathematical Sans-Serif
+  120812,
+  // Mathematical Sans-Serif Bold
+  120822,
+  // Mathematical Monospace
+  // Other scripts
+  66720,
+  // Osmanya 𐒠𐒡𐒢𐒣𐒤𐒥𐒦𐒧𐒨𐒩
+  68912,
+  // Hanifi Rohingya 𐴰𐴱𐴲𐴳𐴴𐴵𐴶𐴷𐴸𐴹
+  69734,
+  // Brahmi 𑁦𑁧𑁨𑁩𑁪𑁫𑁬𑁭𑁮𑁯
+  69872,
+  // Sora Sompeng 𑃰𑃱𑃲𑃳𑃴𑃵𑃶𑃷𑃸𑃹
+  69942,
+  // Chakma 𑄶𑄷𑄸𑄹𑄺𑄻𑄼𑄽𑄾𑄿
+  70096,
+  // Sharada 𑇐𑇑𑇒𑇓𑇔𑇕𑇖𑇗𑇘𑇙
+  70384,
+  // Khudawadi 𑋰𑋱𑋲𑋳𑋴𑋵𑋶𑋷𑋸𑋹
+  70736,
+  // Newa 𑑐𑑑𑑒𑑓𑑔𑑕𑑖𑑗𑑘𑑙
+  70864,
+  // Tirhuta 𑓐𑓑𑓒𑓓𑓔𑓕𑓖𑓗𑓘𑓙
+  71248,
+  // Modi 𑙐𑙑𑙒𑙓𑙔𑙕𑙖𑙗𑙘𑙙
+  71360,
+  // Takri 𑛀𑛁𑛂𑛃𑛄𑛅𑛆𑛇𑛈𑛉
+  71472,
+  // Ahom 𑜰𑜱𑜲𑜳𑜴𑜵𑜶𑜷𑜸𑜹
+  71904,
+  // Warang Citi 𑣠𑣡𑣢𑣣𑣤𑣥𑣦𑣧𑣨𑣩
+  72016,
+  // Dives Akuru 𑥐𑥑𑥒𑥓𑥔𑥕𑥖𑥗𑥘𑥙
+  72688,
+  // Khitan Small Script 𑯰𑯱𑯲𑯳𑯴𑯵𑯶𑯷𑯸𑯹
+  72784,
+  // Bhaiksuki 𑱐𑱑𑱒𑱓𑱔𑱕𑱖𑱗𑱘𑱙
+  73040,
+  // Masaram Gondi 𑵐𑵑𑵒𑵓𑵔𑵕𑵖𑵗𑵘𑵙
+  73120,
+  // Gunjala Gondi 𑶠𑶡𑶢𑶣𑶤𑶥𑶦𑶧𑶨𑶩
+  73552,
+  // Kawi 𑽐𑽑𑽒𑽓𑽔𑽕𑽖𑽗𑽘𑽙
+  92768,
+  // Mro 𖩠𖩡𖩢𖩣𖩤𖩥𖩦𖩧𖩨𖩩
+  92864,
+  // Tangsa 𖫀𖫁𖫂𖫃𖫄𖫅𖫆𖫇𖫈𖫉
+  93008,
+  // Pahawh Hmong 𖭐𖭑𖭒𖭓𖭔𖭕𖭖𖭗𖭘𖭙
+  123200,
+  // Nyiakeng Puachue Hmong 𞅀𞅁𞅂𞅃𞅄𞅅𞅆𞅇𞅈𞅉
+  123632,
+  // Wancho 𞋰𞋱𞋲𞋳𞋴𞋵𞋶𞋷𞋸𞋹
+  124144,
+  // Nag Mundari 𞓰𞓱𞓲𞓳𞓴𞓵𞓶𞓷𞓸𞓹
+  125264,
+  // Adlam 𞥐𞥑𞥒𞥓𞥔𞥕𞥖𞥗𞥘𞥙
+  130032
+  // Segmented digit symbols 🯰🯱🯲🯳🯴🯵🯶🯷🯸🯹
+];
+const NOT_DIGIT = 255;
+const HIGH_MAP = /* @__PURE__ */ new Map();
+const LOW_MAX = 65535;
+const LOW_MIN = 1632;
+const TABLE_OFFSET = LOW_MIN;
+const TABLE_SIZE = LOW_MAX - LOW_MIN + 1;
+const TABLE = new Uint8Array(TABLE_SIZE).fill(NOT_DIGIT);
+for (const zero of SCRIPT_ZEROS) {
+  for (let d = 0; d < 10; d++) {
+    const cp = zero + d;
+    if (cp <= LOW_MAX) {
+      TABLE[cp - TABLE_OFFSET] = d;
+    } else {
+      HIGH_MAP.set(cp, d);
+    }
+  }
+}
+const CHAR_0 = 48;
+const CHAR_9 = 57;
+const CHAR_MINUS = 45;
+const MINUS_SET = /* @__PURE__ */ new Set([8722, 65293, 65123]);
+function anynum(str) {
+  if (typeof str !== "string") return str;
+  const len = str.length;
+  if (len === 0) return str;
+  let firstHit = -1;
+  for (let i = 0; i < len; i++) {
+    const cc = str.charCodeAt(i);
+    if (cc >= CHAR_0 && cc <= CHAR_9 || cc === CHAR_MINUS) continue;
+    if (cc < TABLE_OFFSET) {
+      if (MINUS_SET.has(cc)) {
+        firstHit = i;
+        break;
+      }
+      continue;
+    }
+    if (cc >= 55296 && cc <= 56319) {
+      if (i + 1 < len) {
+        const low = str.charCodeAt(i + 1);
+        if (low >= 56320 && low <= 57343) {
+          const cp = 65536 + (cc - 55296 << 10) + (low - 56320);
+          if (HIGH_MAP.has(cp)) {
+            firstHit = i;
+            break;
+          }
+        }
+      }
+      continue;
+    }
+    if (TABLE[cc - TABLE_OFFSET] !== NOT_DIGIT || MINUS_SET.has(cc)) {
+      firstHit = i;
+      break;
+    }
+  }
+  if (firstHit === -1) return str;
+  const chars = [];
+  if (firstHit > 0) chars.push(str.slice(0, firstHit));
+  for (let i = firstHit; i < len; i++) {
+    const cc = str.charCodeAt(i);
+    if (cc >= CHAR_0 && cc <= CHAR_9 || cc === CHAR_MINUS) {
+      chars.push(str[i]);
+      continue;
+    }
+    if (cc < TABLE_OFFSET) {
+      chars.push(MINUS_SET.has(cc) ? "-" : str[i]);
+      continue;
+    }
+    if (cc >= 55296 && cc <= 56319) {
+      if (i + 1 < len) {
+        const low = str.charCodeAt(i + 1);
+        if (low >= 56320 && low <= 57343) {
+          const cp = 65536 + (cc - 55296 << 10) + (low - 56320);
+          const d2 = HIGH_MAP.get(cp);
+          if (d2 !== void 0) {
+            chars.push(String.fromCharCode(d2 + 48));
+            i++;
+            continue;
+          }
+        }
+      }
+      chars.push(str[i]);
+      continue;
+    }
+    if (MINUS_SET.has(cc)) {
+      chars.push("-");
+      continue;
+    }
+    const d = TABLE[cc - TABLE_OFFSET];
+    chars.push(d !== NOT_DIGIT ? String.fromCharCode(d + 48) : str[i]);
+  }
+  return chars.join("");
+}
 const hexRegex = /^[-+]?0x[a-fA-F0-9]+$/;
+const binRegex = /^0b[01]+$/;
+const octRegex = /^0o[0-7]+$/;
 const numRegex = /^([\-\+])?(0*)([0-9]*(\.[0-9]*)?)$/;
 const consider = {
   hex: true,
-  // oct: false,
+  binary: false,
+  octal: false,
   leadingZeros: true,
   decimalPoint: ".",
   eNotation: true,
   //skipLike: /regex/,
-  infinity: "original"
+  infinity: "original",
   // "null", "infinity" (Infinity type), "string" ("Infinity" (the string literal))
+  unicode: false
 };
 function toNumber(str, options = {}) {
   options = Object.assign({}, consider, options);
@@ -7881,8 +8108,16 @@ function toNumber(str, options = {}) {
   if (trimmedStr.length === 0) return str;
   else if (options.skipLike !== void 0 && options.skipLike.test(trimmedStr)) return str;
   else if (trimmedStr === "0") return 0;
-  else if (options.hex && hexRegex.test(trimmedStr)) {
+  if (options.unicode) {
+    trimmedStr = anynum(trimmedStr);
+    if (trimmedStr === "0") return 0;
+  }
+  if (options.hex && hexRegex.test(trimmedStr)) {
     return parse_int(trimmedStr, 16);
+  } else if (options.binary && binRegex.test(trimmedStr)) {
+    return parse_int(trimmedStr, 2);
+  } else if (options.octal && octRegex.test(trimmedStr)) {
+    return parse_int(trimmedStr, 8);
   } else if (!isFinite(trimmedStr)) {
     return handleInfinity(str, Number(trimmedStr), options);
   } else if (trimmedStr.includes("e") || trimmedStr.includes("E")) {
@@ -7962,6 +8197,8 @@ function trimZeros(numStr) {
   return numStr;
 }
 function parse_int(numStr, base) {
+  const str = numStr.trim();
+  if (base === 2 || base === 8) numStr = str.substring(2);
   if (parseInt) return parseInt(numStr, base);
   else if (Number.parseInt) return Number.parseInt(numStr, base);
   else if (window && window.parseInt) return window.parseInt(numStr, base);
@@ -9294,7 +9531,7 @@ function readStopNodeData(xmlData, tagName, i) {
         const closeIndex = findClosingIndex(xmlData, "]]>", i, "StopNode is not closed.") - 2;
         i = closeIndex;
       } else {
-        const tagData = readTagExp(xmlData, i, ">");
+        const tagData = readTagExp(xmlData, i, false);
         if (tagData) {
           const openTagName = tagData && tagData.tagName;
           if (openTagName === tagName && tagData.tagExp[tagData.tagExp.length - 1] !== "/") {
@@ -9378,6 +9615,9 @@ function compress(arr, options, matcher, readonlyMatcher) {
     } else if (tagObj[property]) {
       let val = compress(tagObj[property], options, matcher, readonlyMatcher);
       const isLeaf = isLeafTag(val, options);
+      if (Object.keys(val).length === 0 && options.alwaysCreateTextNode) {
+        val[options.textNodeName] = "";
+      }
       if (tagObj[":@"]) {
         assignAttributes(val, tagObj[":@"], readonlyMatcher, options);
       } else if (Object.keys(val).length === 1 && val[options.textNodeName] !== void 0 && !options.alwaysCreateTextNode) {