captcha-ocr 1.0.0

package/dist/rules/arithmetic.js

@@ -0,0 +1,145 @@
+// src/rules/arithmetic.ts
+var DEFAULT_OPERATORS = {
+  "+": {
+    evaluate: (left, right) => left + right
+  },
+  "-": {
+    evaluate: (left, right) => left - right
+  },
+  "*": {
+    aliases: ["x", "X", "\xD7", "\xB7"],
+    evaluate: (left, right) => left * right
+  },
+  "/": {
+    aliases: ["\xF7"],
+    evaluate: (left, right) => left / right,
+    validate: (left, right) => right === 0 ? {
+      code: "DIVISION_BY_ZERO",
+      message: "\u9664\u6570\u4E0D\u80FD\u4E3A 0"
+    } : null
+  }
+};
+function createArithmeticRule(options = {}) {
+  const operators = options.operators ?? DEFAULT_OPERATORS;
+  const aliasToOperator = /* @__PURE__ */ new Map();
+  for (const [symbol, definition] of Object.entries(operators)) {
+    aliasToOperator.set(symbol, symbol);
+    for (const alias of definition.aliases ?? []) {
+      aliasToOperator.set(alias, symbol);
+    }
+  }
+  const operatorPattern = Object.keys(operators).sort((left, right) => right.length - left.length).map(escapeRegExp).join("|");
+  const expressionPattern = new RegExp(
+    `^(\\d{1,6})(${operatorPattern})(\\d{1,6})(?:=)?\\??$`
+  );
+  const normalize = (text) => {
+    let normalized = text.normalize("NFKC").replace(/\s+/g, "");
+    for (const [alias, symbol] of [...aliasToOperator.entries()].sort(
+      ([left], [right]) => right.length - left.length
+    )) {
+      normalized = normalized.split(alias).join(symbol);
+    }
+    normalized = normalized.replace(/[Oo]/g, "0").replace(/[Il|]/g, "1");
+    return normalized;
+  };
+  const parse = (text) => {
+    const normalized = normalize(text);
+    const match = normalized.match(expressionPattern);
+    if (!match) {
+      return {
+        ok: false,
+        normalized,
+        error: {
+          code: "EXPRESSION_NOT_MATCHED",
+          message: "\u65E0\u6CD5\u8BC6\u522B\u4E3A\u201C\u6570\u5B57 \u8FD0\u7B97\u7B26 \u6570\u5B57 =?\u201D\u683C\u5F0F"
+        }
+      };
+    }
+    const parsed = {
+      left: Number(match[1]),
+      operator: match[2],
+      right: Number(match[3])
+    };
+    const validationError = operators[parsed.operator].validate?.(parsed.left, parsed.right) ?? null;
+    if (validationError) {
+      return {
+        ok: false,
+        normalized,
+        error: validationError
+      };
+    }
+    return {
+      ok: true,
+      normalized,
+      expression: `${parsed.left}${parsed.operator}${parsed.right}`,
+      parsed
+    };
+  };
+  return {
+    id: options.id ?? "arithmetic-v1",
+    normalize,
+    parse,
+    evaluate: (parsed) => operators[parsed.operator].evaluate(parsed.left, parsed.right)
+  };
+}
+var arithmeticRule = createArithmeticRule();
+function normalizeOcrText(text) {
+  return arithmeticRule.normalize(text);
+}
+function parseExpression(text) {
+  const parsed = arithmeticRule.parse(text);
+  if (!parsed.ok) {
+    return {
+      ok: false,
+      normalized: parsed.normalized,
+      expression: null,
+      answer: null,
+      error: parsed.error
+    };
+  }
+  return {
+    ok: true,
+    normalized: parsed.normalized,
+    expression: parsed.expression,
+    answer: arithmeticRule.evaluate(parsed.parsed),
+    error: null
+  };
+}
+function calculate(left, operator, right) {
+  const definition = DEFAULT_OPERATORS[operator];
+  if (!definition) {
+    throw new Error(`\u4E0D\u652F\u6301\u7684\u8FD0\u7B97\u7B26: ${operator}`);
+  }
+  const validationError = definition.validate?.(left, right);
+  if (validationError) {
+    throw new Error(validationError.message);
+  }
+  return definition.evaluate(left, right);
+}
+var arithmeticProfile = {
+  id: "arithmetic-v1",
+  ocr: {
+    language: "eng",
+    whitelist: "0123456789+-*/=?xX\xD7\xF7",
+    pageSegMode: 7,
+    preserveInterwordSpaces: false
+  },
+  preprocess: {
+    scale: 4,
+    thresholds: [160, 120, 180, 200]
+  },
+  rule: arithmeticRule
+};
+function escapeRegExp(value) {
+  return value.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
+}
+export {
+  DEFAULT_OPERATORS,
+  arithmeticProfile,
+  arithmeticRule,
+  calculate,
+  createArithmeticRule,
+  normalizeOcrText,
+  parseExpression
+};
+//# sourceMappingURL=arithmetic.js.map

package/dist/rules/arithmetic.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../src/rules/arithmetic.ts"],"sourcesContent":["import type {\n  CaptchaRule,\n  CaptchaProfile,\n  RecognitionError,\n  RuleParseResult\n} from '../types.js';\n\nexport interface ArithmeticExpression {\n  left: number;\n  operator: string;\n  right: number;\n}\n\nexport interface ArithmeticOperator {\n  aliases?: string[];\n  evaluate(left: number, right: number): number;\n  validate?(left: number, right: number): RecognitionError | null;\n}\n\nexport const DEFAULT_OPERATORS: Record<string, ArithmeticOperator> = {\n  '+': {\n    evaluate: (left, right) => left + right\n  },\n  '-': {\n    evaluate: (left, right) => left - right\n  },\n  '*': {\n    aliases: ['x', 'X', '×', '·'],\n    evaluate: (left, right) => left * right\n  },\n  '/': {\n    aliases: ['÷'],\n    evaluate: (left, right) => left / right,\n    validate: (left, right) =>\n      right === 0\n        ? {\n            code: 'DIVISION_BY_ZERO',\n            message: '除数不能为 0'\n          }\n        : null\n  }\n};\n\nexport interface ArithmeticRuleOptions {\n  id?: string;\n  whitelist?: string;\n  operators?: Record<string, ArithmeticOperator>;\n}\n\nexport function createArithmeticRule(\n  options: ArithmeticRuleOptions = {}\n): CaptchaRule<ArithmeticExpression, number> {\n  const operators = options.operators ?? DEFAULT_OPERATORS;\n  const aliasToOperator = new Map<string, string>();\n\n  for (const [symbol, definition] of Object.entries(operators)) {\n    aliasToOperator.set(symbol, symbol);\n    for (const alias of definition.aliases ?? []) {\n      aliasToOperator.set(alias, symbol);\n    }\n  }\n\n  const operatorPattern = Object.keys(operators)\n    .sort((left, right) => right.length - left.length)\n    .map(escapeRegExp)\n    .join('|');\n  const expressionPattern = new RegExp(\n    `^(\\\\d{1,6})(${operatorPattern})(\\\\d{1,6})(?:=)?\\\\??$`\n  );\n\n  const normalize = (text: string): string => {\n    let normalized = text.normalize('NFKC').replace(/\\s+/g, '');\n    for (const [alias, symbol] of [...aliasToOperator.entries()].sort(\n      ([left], [right]) => right.length - left.length\n    )) {\n      normalized = normalized.split(alias).join(symbol);\n    }\n    normalized = normalized.replace(/[Oo]/g, '0').replace(/[Il|]/g, '1');\n    return normalized;\n  };\n\n  const parse = (text: string): RuleParseResult<ArithmeticExpression> => {\n    const normalized = normalize(text);\n    const match = normalized.match(expressionPattern);\n\n    if (!match) {\n      return {\n        ok: false,\n        normalized,\n        error: {\n          code: 'EXPRESSION_NOT_MATCHED',\n          message: '无法识别为“数字 运算符 数字 =?”格式'\n        }\n      };\n    }\n\n    const parsed: ArithmeticExpression = {\n      left: Number(match[1]),\n      operator: match[2],\n      right: Number(match[3])\n    };\n    const validationError =\n      operators[parsed.operator].validate?.(parsed.left, parsed.right) ?? null;\n\n    if (validationError) {\n      return {\n        ok: false,\n        normalized,\n        error: validationError\n      };\n    }\n\n    return {\n      ok: true,\n      normalized,\n      expression: `${parsed.left}${parsed.operator}${parsed.right}`,\n      parsed\n    };\n  };\n\n  return {\n    id: options.id ?? 'arithmetic-v1',\n    normalize,\n    parse,\n    evaluate: (parsed) =>\n      operators[parsed.operator].evaluate(parsed.left, parsed.right)\n  };\n}\n\nexport const arithmeticRule = createArithmeticRule();\n\nexport interface LegacyExpressionSuccess {\n  ok: true;\n  normalized: string;\n  expression: string;\n  answer: number;\n  error: null;\n}\n\nexport interface LegacyExpressionFailure {\n  ok: false;\n  normalized: string;\n  expression: null;\n  answer: null;\n  error: RecognitionError;\n}\n\nexport type LegacyExpressionResult =\n  | LegacyExpressionSuccess\n  | LegacyExpressionFailure;\n\nexport function normalizeOcrText(text: string): string {\n  return arithmeticRule.normalize(text);\n}\n\nexport function parseExpression(text: string): LegacyExpressionResult {\n  const parsed = arithmeticRule.parse(text);\n  if (!parsed.ok) {\n    return {\n      ok: false,\n      normalized: parsed.normalized,\n      expression: null,\n      answer: null,\n      error: parsed.error\n    };\n  }\n\n  return {\n    ok: true,\n    normalized: parsed.normalized,\n    expression: parsed.expression,\n    answer: arithmeticRule.evaluate(parsed.parsed),\n    error: null\n  };\n}\n\nexport function calculate(\n  left: number,\n  operator: string,\n  right: number\n): number {\n  const definition = DEFAULT_OPERATORS[operator];\n  if (!definition) {\n    throw new Error(`不支持的运算符: ${operator}`);\n  }\n  const validationError = definition.validate?.(left, right);\n  if (validationError) {\n    throw new Error(validationError.message);\n  }\n  return definition.evaluate(left, right);\n}\n\nexport const arithmeticProfile: CaptchaProfile<\n  ArithmeticExpression,\n  number\n> = {\n  id: 'arithmetic-v1',\n  ocr: {\n    language: 'eng',\n    whitelist: '0123456789+-*/=?xX×÷',\n    pageSegMode: 7,\n    preserveInterwordSpaces: false\n  },\n  preprocess: {\n    scale: 4,\n    thresholds: [160, 120, 180, 200]\n  },\n  rule: arithmeticRule\n};\n\nfunction escapeRegExp(value: string): string {\n  return value.replace(/[.*+?^${}()|[\\]\\\\]/g, '\\\\$&');\n}\n"],"mappings":";AAmBO,IAAM,oBAAwD;AAAA,EACnE,KAAK;AAAA,IACH,UAAU,CAAC,MAAM,UAAU,OAAO;AAAA,EACpC;AAAA,EACA,KAAK;AAAA,IACH,UAAU,CAAC,MAAM,UAAU,OAAO;AAAA,EACpC;AAAA,EACA,KAAK;AAAA,IACH,SAAS,CAAC,KAAK,KAAK,QAAK,MAAG;AAAA,IAC5B,UAAU,CAAC,MAAM,UAAU,OAAO;AAAA,EACpC;AAAA,EACA,KAAK;AAAA,IACH,SAAS,CAAC,MAAG;AAAA,IACb,UAAU,CAAC,MAAM,UAAU,OAAO;AAAA,IAClC,UAAU,CAAC,MAAM,UACf,UAAU,IACN;AAAA,MACE,MAAM;AAAA,MACN,SAAS;AAAA,IACX,IACA;AAAA,EACR;AACF;AAQO,SAAS,qBACd,UAAiC,CAAC,GACS;AAC3C,QAAM,YAAY,QAAQ,aAAa;AACvC,QAAM,kBAAkB,oBAAI,IAAoB;AAEhD,aAAW,CAAC,QAAQ,UAAU,KAAK,OAAO,QAAQ,SAAS,GAAG;AAC5D,oBAAgB,IAAI,QAAQ,MAAM;AAClC,eAAW,SAAS,WAAW,WAAW,CAAC,GAAG;AAC5C,sBAAgB,IAAI,OAAO,MAAM;AAAA,IACnC;AAAA,EACF;AAEA,QAAM,kBAAkB,OAAO,KAAK,SAAS,EAC1C,KAAK,CAAC,MAAM,UAAU,MAAM,SAAS,KAAK,MAAM,EAChD,IAAI,YAAY,EAChB,KAAK,GAAG;AACX,QAAM,oBAAoB,IAAI;AAAA,IAC5B,eAAe,eAAe;AAAA,EAChC;AAEA,QAAM,YAAY,CAAC,SAAyB;AAC1C,QAAI,aAAa,KAAK,UAAU,MAAM,EAAE,QAAQ,QAAQ,EAAE;AAC1D,eAAW,CAAC,OAAO,MAAM,KAAK,CAAC,GAAG,gBAAgB,QAAQ,CAAC,EAAE;AAAA,MAC3D,CAAC,CAAC,IAAI,GAAG,CAAC,KAAK,MAAM,MAAM,SAAS,KAAK;AAAA,IAC3C,GAAG;AACD,mBAAa,WAAW,MAAM,KAAK,EAAE,KAAK,MAAM;AAAA,IAClD;AACA,iBAAa,WAAW,QAAQ,SAAS,GAAG,EAAE,QAAQ,UAAU,GAAG;AACnE,WAAO;AAAA,EACT;AAEA,QAAM,QAAQ,CAAC,SAAwD;AACrE,UAAM,aAAa,UAAU,IAAI;AACjC,UAAM,QAAQ,WAAW,MAAM,iBAAiB;AAEhD,QAAI,CAAC,OAAO;AACV,aAAO;AAAA,QACL,IAAI;AAAA,QACJ;AAAA,QACA,OAAO;AAAA,UACL,MAAM;AAAA,UACN,SAAS;AAAA,QACX;AAAA,MACF;AAAA,IACF;AAEA,UAAM,SAA+B;AAAA,MACnC,MAAM,OAAO,MAAM,CAAC,CAAC;AAAA,MACrB,UAAU,MAAM,CAAC;AAAA,MACjB,OAAO,OAAO,MAAM,CAAC,CAAC;AAAA,IACxB;AACA,UAAM,kBACJ,UAAU,OAAO,QAAQ,EAAE,WAAW,OAAO,MAAM,OAAO,KAAK,KAAK;AAEtE,QAAI,iBAAiB;AACnB,aAAO;AAAA,QACL,IAAI;AAAA,QACJ;AAAA,QACA,OAAO;AAAA,MACT;AAAA,IACF;AAEA,WAAO;AAAA,MACL,IAAI;AAAA,MACJ;AAAA,MACA,YAAY,GAAG,OAAO,IAAI,GAAG,OAAO,QAAQ,GAAG,OAAO,KAAK;AAAA,MAC3D;AAAA,IACF;AAAA,EACF;AAEA,SAAO;AAAA,IACL,IAAI,QAAQ,MAAM;AAAA,IAClB;AAAA,IACA;AAAA,IACA,UAAU,CAAC,WACT,UAAU,OAAO,QAAQ,EAAE,SAAS,OAAO,MAAM,OAAO,KAAK;AAAA,EACjE;AACF;AAEO,IAAM,iBAAiB,qBAAqB;AAsB5C,SAAS,iBAAiB,MAAsB;AACrD,SAAO,eAAe,UAAU,IAAI;AACtC;AAEO,SAAS,gBAAgB,MAAsC;AACpE,QAAM,SAAS,eAAe,MAAM,IAAI;AACxC,MAAI,CAAC,OAAO,IAAI;AACd,WAAO;AAAA,MACL,IAAI;AAAA,MACJ,YAAY,OAAO;AAAA,MACnB,YAAY;AAAA,MACZ,QAAQ;AAAA,MACR,OAAO,OAAO;AAAA,IAChB;AAAA,EACF;AAEA,SAAO;AAAA,IACL,IAAI;AAAA,IACJ,YAAY,OAAO;AAAA,IACnB,YAAY,OAAO;AAAA,IACnB,QAAQ,eAAe,SAAS,OAAO,MAAM;AAAA,IAC7C,OAAO;AAAA,EACT;AACF;AAEO,SAAS,UACd,MACA,UACA,OACQ;AACR,QAAM,aAAa,kBAAkB,QAAQ;AAC7C,MAAI,CAAC,YAAY;AACf,UAAM,IAAI,MAAM,+CAAY,QAAQ,EAAE;AAAA,EACxC;AACA,QAAM,kBAAkB,WAAW,WAAW,MAAM,KAAK;AACzD,MAAI,iBAAiB;AACnB,UAAM,IAAI,MAAM,gBAAgB,OAAO;AAAA,EACzC;AACA,SAAO,WAAW,SAAS,MAAM,KAAK;AACxC;AAEO,IAAM,oBAGT;AAAA,EACF,IAAI;AAAA,EACJ,KAAK;AAAA,IACH,UAAU;AAAA,IACV,WAAW;AAAA,IACX,aAAa;AAAA,IACb,yBAAyB;AAAA,EAC3B;AAAA,EACA,YAAY;AAAA,IACV,OAAO;AAAA,IACP,YAAY,CAAC,KAAK,KAAK,KAAK,GAAG;AAAA,EACjC;AAAA,EACA,MAAM;AACR;AAEA,SAAS,aAAa,OAAuB;AAC3C,SAAO,MAAM,QAAQ,uBAAuB,MAAM;AACpD;","names":[]}

package/package.json

@@ -0,0 +1,66 @@
+{
+  "name": "captcha-ocr",
+  "version": "1.0.0",
+  "private": false,
+  "description": "TypeScript captcha OCR recognizer with configurable profiles and rules",
+  "type": "module",
+  "main": "./dist/index.cjs",
+  "module": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "author": "wisen",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js",
+      "require": "./dist/index.cjs"
+    },
+    "./rules/arithmetic": {
+      "types": "./dist/rules/arithmetic.d.ts",
+      "import": "./dist/rules/arithmetic.js",
+      "require": "./dist/rules/arithmetic.cjs"
+    }
+  },
+  "files": [
+    "dist/index.*",
+    "dist/arithmetic-*.d.ts",
+    "dist/arithmetic-*.d.cts",
+    "dist/rules/*",
+    "README.md",
+    "README.zh-CN.md",
+    "CHANGELOG.md"
+  ],
+  "publishConfig": {
+    "registry": "https://registry.npmjs.org/",
+    "access": "public"
+  },
+  "engines": {
+    "node": ">=20"
+  },
+  "dependencies": {
+    "sharp": "^0.35.3",
+    "tesseract.js": "^5.1.1"
+  },
+  "devDependencies": {
+    "@types/node": "^22.10.2",
+    "changelogen": "^0.6.2",
+    "tsup": "^8.3.5",
+    "tsx": "^4.19.2",
+    "typescript": "^5.7.2"
+  },
+  "scripts": {
+    "build": "tsup",
+    "start": "pnpm build && node dist/bin/server.cjs",
+    "dev": "tsx bin/server.ts",
+    "cli": "pnpm build && node dist/bin/cli.cjs",
+    "test": "tsx --test test/*.test.ts",
+    "pack:check": "pnpm build && pnpm pack --pack-destination ./tmp",
+    "release:check": "pnpm test && pnpm exec tsc --noEmit && pnpm run pack:check",
+    "release": "pnpm run release:auto",
+    "release:auto": "pnpm run release:check && changelogen --release --clean --hideAuthorEmail && pnpm run release:publish",
+    "release:patch": "pnpm run release:check && changelogen --release --clean --patch --hideAuthorEmail && pnpm run release:publish",
+    "release:base": "pnpm run release:check && changelogen --release --clean --minor --hideAuthorEmail && pnpm run release:publish",
+    "release:minor": "pnpm run release:base",
+    "release:major": "pnpm run release:check && changelogen --release --clean --major --hideAuthorEmail && pnpm run release:publish",
+    "release:publish": "pnpm publish --registry https://registry.npmjs.org/ --access public"
+  }
+}