UEA-Prodem

2026-06-10 12:38:42 -03:00
parent 3f33154e16
commit c41625e542
9352 changed files with 1128292 additions and 14752 deletions
@@ -0,0 +1,116 @@
+/**
+ * Flat lookup table: maps Unicode code point → ASCII digit (0-9).
+ * Only decimal digit characters (Unicode category Nd) are included.
+ *
+ * Strategy: Int32Array of size (maxCodePoint - minCodePoint + 1).
+ * Value 0xFF means "not a digit". Value 0-9 is the ASCII digit value.
+ * This gives O(1) lookup with no branching, no bisect, no loop.
+ *
+ * Memory: range is 0x0660 to 0x1FBF0 → ~129,936 entries × 1 byte = ~127 KB.
+ * Acceptable for a one-time init; lookup is a single array index.
+ */
+
+// All known Unicode Nd (decimal digit) script zero code points.
+// Each script has exactly 10 consecutive digits: zero+0 .. zero+9.
+const SCRIPT_ZEROS = [
+  // Basic Latin (ASCII) — included for completeness / pass-through
+  0x0030, // 0-9
+
+  // Arabic scripts
+  0x0660, // Arabic-Indic ٠١٢٣٤٥٦٧٨٩
+  0x06F0, // Extended Arabic-Indic (Urdu/Persian/Sindhi) ۰۱۲۳
+
+  // Indic scripts
+  0x0966, // Devanagari ०१२३४५६७८९
+  0x09E6, // Bengali ০১২৩৪৫৬৭৮৯
+  0x0A66, // Gurmukhi ੦੧੨੩੪੫੬੭੮੯
+  0x0AE6, // Gujarati ૦૧૨૩૪૫૬૭૮૯
+  0x0B66, // Odia ୦୧୨୩୪୫୬୭୮୯
+  0x0BE6, // Tamil ௦௧௨௩௪௫௬௭௮௯
+  0x0C66, // Telugu ౦౧౨౩౪౫౬౭౮౯
+  0x0CE6, // Kannada ೦೧೨೩೪೫೬೭೮೯
+  0x0D66, // Malayalam ൦൧൨൩൪൫൬൭൮൯
+  0x0DE6, // Sinhala Archaic ෦෧෨෩෪෫෬෭෮෯
+
+  // Southeast Asian scripts
+  0x0E50, // Thai ๐๑๒๓๔๕๖๗๘๙
+  0x0ED0, // Lao ໐໑໒໓໔໕໖໗໘໙
+  0x0F20, // Tibetan ༠༡༢༣༤༥༦༧༨༩
+  0x1040, // Myanmar ၀၁၂၃၄၅၆၇၈၉
+  0x1090, // Myanmar Shan ႐႑႒႓႔႕႖႗႘႙
+  0x17E0, // Khmer ០១២៣៤៥៦៧៨៩
+  0x1810, // Mongolian ᠐᠑᠒᠓᠔᠕᠖᠗᠘᠙
+  0x1946, // Limbu ᥆᥇᥈᥉᥊᥋᥌᥍᥎᥏
+  0x19D0, // New Tai Lue ᧐᧑᧒᧓᧔᧕᧖᧗᧘᧙
+  0x1A80, // Tai Tham Hora ᪀᪁᪂᪃᪄᪅᪆᪇᪈᪉
+  0x1A90, // Tai Tham Tham ᪐᪑᪒᪓᪔᪕᪖᪗᪘᪙
+  0x1B50, // Balinese ᭐᭑᭒᭓᭔᭕᭖᭗᭘᭙
+  0x1BB0, // Sundanese ᮰᮱᮲᮳᮴᮵᮶᮷᮸᮹
+  0x1C40, // Lepcha ᱀᱁᱂᱃᱄᱅᱆᱇᱈᱉
+  0x1C50, // Ol Chiki ᱐᱑᱒᱓᱔᱕᱖᱗᱘᱙
+
+  // Fullwidth (CJK context)
+  0xFF10, // Fullwidth ０１２３４５６７８９
+
+  // Mathematical digit variants (Unicode math block)
+  0x1D7CE, // Mathematical Bold
+  0x1D7D8, // Mathematical Double-Struck
+  0x1D7E2, // Mathematical Sans-Serif
+  0x1D7EC, // Mathematical Sans-Serif Bold
+  0x1D7F6, // Mathematical Monospace
+
+  // Other scripts
+  0x104A0, // Osmanya 𐒠𐒡𐒢𐒣𐒤𐒥𐒦𐒧𐒨𐒩
+  0x10D30, // Hanifi Rohingya 𐴰𐴱𐴲𐴳𐴴𐴵𐴶𐴷𐴸𐴹
+  0x11066, // Brahmi 𑁦𑁧𑁨𑁩𑁪𑁫𑁬𑁭𑁮𑁯
+  0x110F0, // Sora Sompeng 𑃰𑃱𑃲𑃳𑃴𑃵𑃶𑃷𑃸𑃹
+  0x11136, // Chakma 𑄶𑄷𑄸𑄹𑄺𑄻𑄼𑄽𑄾𑄿
+  0x111D0, // Sharada 𑇐𑇑𑇒𑇓𑇔𑇕𑇖𑇗𑇘𑇙
+  0x112F0, // Khudawadi 𑋰𑋱𑋲𑋳𑋴𑋵𑋶𑋷𑋸𑋹
+  0x11450, // Newa 𑑐𑑑𑑒𑑓𑑔𑑕𑑖𑑗𑑘𑑙
+  0x114D0, // Tirhuta 𑓐𑓑𑓒𑓓𑓔𑓕𑓖𑓗𑓘𑓙
+  0x11650, // Modi 𑙐𑙑𑙒𑙓𑙔𑙕𑙖𑙗𑙘𑙙
+  0x116C0, // Takri 𑛀𑛁𑛂𑛃𑛄𑛅𑛆𑛇𑛈𑛉
+  0x11730, // Ahom 𑜰𑜱𑜲𑜳𑜴𑜵𑜶𑜷𑜸𑜹
+  0x118E0, // Warang Citi 𑣠𑣡𑣢𑣣𑣤𑣥𑣦𑣧𑣨𑣩
+  0x11950, // Dives Akuru 𑥐𑥑𑥒𑥓𑥔𑥕𑥖𑥗𑥘𑥙
+  0x11BF0, // Khitan Small Script 𑯰𑯱𑯲𑯳𑯴𑯵𑯶𑯷𑯸𑯹
+  0x11C50, // Bhaiksuki 𑱐𑱑𑱒𑱓𑱔𑱕𑱖𑱗𑱘𑱙
+  0x11D50, // Masaram Gondi 𑵐𑵑𑵒𑵓𑵔𑵕𑵖𑵗𑵘𑵙
+  0x11DA0, // Gunjala Gondi 𑶠𑶡𑶢𑶣𑶤𑶥𑶦𑶧𑶨𑶩
+  0x11F50, // Kawi 𑽐𑽑𑽒𑽓𑽔𑽕𑽖𑽗𑽘𑽙
+  0x16A60, // Mro 𖩠𖩡𖩢𖩣𖩤𖩥𖩦𖩧𖩨𖩩
+  0x16AC0, // Tangsa 𖫀𖫁𖫂𖫃𖫄𖫅𖫆𖫇𖫈𖫉
+  0x16B50, // Pahawh Hmong 𖭐𖭑𖭒𖭓𖭔𖭕𖭖𖭗𖭘𖭙
+  0x1E140, // Nyiakeng Puachue Hmong 𞅀𞅁𞅂𞅃𞅄𞅅𞅆𞅇𞅈𞅉
+  0x1E2F0, // Wancho 𞋰𞋱𞋲𞋳𞋴𞋵𞋶𞋷𞋸𞋹
+  0x1E4F0, // Nag Mundari 𞓰𞓱𞓲𞓳𞓴𞓵𞓶𞓷𞓸𞓹
+  0x1E950, // Adlam 𞥐𞥑𞥒𞥓𞥔𞥕𞥖𞥗𞥘𞥙
+  0x1FBF0, // Segmented digit symbols 🯰🯱🯲🯳🯴🯵🯶🯷🯸🯹
+];
+
+// Build a sparse Map for scripts above 0xFFFF (surrogate-pair range).
+// These can't go into a flat Uint8Array indexed by code point efficiently.
+const NOT_DIGIT = 0xFF;
+const HIGH_MAP = new Map(); // codePoint → digit value (0-9)
+
+const LOW_MAX = 0xFFFF;
+const LOW_MIN = 0x0660; // first non-ASCII digit script
+
+// Flat Uint8Array covering 0x0660 .. 0xFFFF
+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);
+    }
+  }
+}
+
+export { TABLE, TABLE_OFFSET, HIGH_MAP, NOT_DIGIT };