site-metadata-extractor
Version:
web(site) resource metadata extractor
180 lines • 22.9 kB
JavaScript
;
var __importDefault = (this && this.__importDefault) || function (mod) {
return (mod && mod.__esModule) ? mod : { "default": mod };
};
Object.defineProperty(exports, "__esModule", { value: true });
exports.ulToText = exports.replaceWithText = exports.replaceCharacters = exports.removeNegativescoresNodes = exports.removeFewWordsParagraphs = exports.linksToText = exports.convertToText = exports.cleanParagraphText = exports.addNewlineToBr = void 0;
const xregexp_1 = __importDefault(require("xregexp"));
const stopwords_1 = __importDefault(require("./stopwords"));
const addNewlineToBr = (doc, topNode) => {
const brs = topNode.find("br");
brs.each((_index, element) => {
const br = doc(element);
br.replaceWith("\n\n");
});
return doc;
};
exports.addNewlineToBr = addNewlineToBr;
const cleanParagraphText = (rawText) => {
const text = rawText.trim();
text.replace(/[\s\t]+/g, " ");
return text;
};
exports.cleanParagraphText = cleanParagraphText;
const convertToText = (doc, topNode) => {
let texts = [];
const nodes = topNode.contents();
let hangingText = "";
nodes.each((_index, element) => {
const node = doc(element);
const nodeType = node.get(0) ? node.get(0).type : null;
const nodeName = node.get(0) ? node.get(0).tagName : null;
if (nodeType === "text") {
hangingText += node.text();
return true;
}
else if (nodeName === "ul") {
hangingText += (0, exports.ulToText)(doc, node);
return true;
}
if (hangingText.length > 0) {
const txt = (0, exports.cleanParagraphText)(hangingText);
texts = texts.concat(txt.split(/\r?\n/));
hangingText = "";
}
let text = (0, exports.cleanParagraphText)(node.text());
text = text.replace(/(\w+\.)([A-Z]+)/, "$1 $2");
texts = texts.concat(text.split(/\r?\n/));
});
if (hangingText.length > 0) {
const text = (0, exports.cleanParagraphText)(hangingText);
texts = texts = texts.concat(text.split(/\r?\n/));
}
texts = texts.map((txt) => {
return txt.trim();
});
const regex = (0, xregexp_1.default)("[\\p{Number}\\p{Letter}]");
texts = texts.filter((txt) => {
return regex.test(txt);
});
return texts.join("\n\n");
};
exports.convertToText = convertToText;
const linksToText = (doc, topNode) => {
const nodes = topNode.find("a");
nodes.each((_index, element) => {
const htmlEl = doc(element).html();
if (htmlEl) {
doc(element).replaceWith(htmlEl);
}
});
return doc;
};
exports.linksToText = linksToText;
const removeFewWordsParagraphs = (doc, topNode, lang) => {
const allNodes = topNode.find("*");
allNodes.each((_index, element) => {
const el = doc(element);
const tag = el.get(0).tagName;
const text = el.text();
const stopWords = (0, stopwords_1.default)(text, lang);
if ((tag !== "br" || text !== "\\r") &&
stopWords.stopWordCount < 3 &&
el.find("object").length === 0 &&
el.find("embed").length === 0) {
doc(el).remove();
}
else {
const trimmed = text.trim();
if (trimmed[0] === "(" && trimmed[trimmed.length - 1] === ")") {
doc(el).remove();
}
}
});
return doc;
};
exports.removeFewWordsParagraphs = removeFewWordsParagraphs;
const removeNegativescoresNodes = (doc, topNode) => {
const gravityItems = topNode.find("*[gravityScore]");
gravityItems.each((_index, element) => {
let score = 0;
const item = doc(element);
const gravityScore = item.attr("gravityScore");
if (gravityScore) {
score = parseInt(gravityScore, 10) || 0;
}
if (score < 1) {
doc(item).remove();
}
});
return doc;
};
exports.removeNegativescoresNodes = removeNegativescoresNodes;
const replaceCharacters = (text, html, chars) => {
let processedText = text;
// if element does not match any in map and starts with & and ends with ;, replace with empty string
const htmlEntities = {
"&": "&",
"'": "'",
"¢": "¢",
"©": "©",
"€": "€",
">": ">",
"<": "<",
" ": " ",
"£": "£",
""": '"',
"®": "®",
"¥": "¥",
};
const escapeChars = {
'"': '"',
"'": "'",
// tslint:disable-next-line: object-literal-sort-keys
"\n": " ",
"\r": " ",
};
if (html) {
for (const key of Object.keys(htmlEntities)) {
const htmlregex = new RegExp(key, "g");
processedText = processedText.replace(htmlregex, htmlEntities[key]);
}
}
if (chars) {
for (const key of Object.keys(escapeChars)) {
const escapeCharsRegex = new RegExp(key, "g");
processedText = processedText.replace(escapeCharsRegex, escapeChars[key]);
}
}
return processedText;
};
exports.replaceCharacters = replaceCharacters;
const replaceWithText = (doc, topNode) => {
const nodes = topNode.find("b, strong, i, br, sup");
nodes.each((_index, element) => {
doc(element).replaceWith(doc(element).text());
});
return doc;
};
exports.replaceWithText = replaceWithText;
const ulToText = (doc, node) => {
const nodes = node.find("li");
let text = "";
nodes.each((_index, element) => {
text = `${text}\n * ${doc(element).text()}`;
});
text = `${text}\n`;
return text;
};
exports.ulToText = ulToText;
const formatter = (doc, topNode, lang) => {
(0, exports.removeNegativescoresNodes)(doc, topNode);
(0, exports.linksToText)(doc, topNode);
(0, exports.addNewlineToBr)(doc, topNode);
(0, exports.replaceWithText)(doc, topNode);
(0, exports.removeFewWordsParagraphs)(doc, topNode, lang);
const convertedText = (0, exports.convertToText)(doc, topNode).replace(/\n/g, " ");
return (0, exports.replaceCharacters)(convertedText, false, true);
};
exports.default = formatter;
//# sourceMappingURL=data:application/json;base64,{"version":3,"file":"formatter.js","sourceRoot":"","sources":["../src/formatter.ts"],"names":[],"mappings":";;;;;;AACA,sDAA8B;AAC9B,4DAAoC;AAE7B,MAAM,cAAc,GAAG,CAC5B,GAAiB,EACjB,OAAwB,EACV,EAAE;IAChB,MAAM,GAAG,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAE/B,GAAG,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAwB,EAAE,EAAE;QACpD,MAAM,EAAE,GAAG,GAAG,CAAC,OAAO,CAAC,CAAC;QACxB,EAAE,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC;IACzB,CAAC,CAAC,CAAC;IAEH,OAAO,GAAG,CAAC;AACb,CAAC,CAAC;AAZW,QAAA,cAAc,kBAYzB;AAEK,MAAM,kBAAkB,GAAG,CAAC,OAAe,EAAU,EAAE;IAC5D,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,EAAE,CAAC;IAC5B,IAAI,CAAC,OAAO,CAAC,UAAU,EAAE,GAAG,CAAC,CAAC;IAC9B,OAAO,IAAI,CAAC;AACd,CAAC,CAAC;AAJW,QAAA,kBAAkB,sBAI7B;AAEK,MAAM,aAAa,GAAG,CAC3B,GAAiB,EACjB,OAAwB,EAChB,EAAE;IACV,IAAI,KAAK,GAAa,EAAE,CAAC;IACzB,MAAM,KAAK,GAAG,OAAO,CAAC,QAAQ,EAAE,CAAC;IAEjC,IAAI,WAAW,GAAG,EAAE,CAAC;IAErB,KAAK,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAwB,EAAE,EAAE;QACtD,MAAM,IAAI,GAAG,GAAG,CAAC,OAAO,CAAC,CAAC;QAC1B,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC;QACvD,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC;QAE1D,IAAI,QAAQ,KAAK,MAAM,EAAE;YACvB,WAAW,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC;YAC3B,OAAO,IAAI,CAAC;SACb;aAAM,IAAI,QAAQ,KAAK,IAAI,EAAE;YAC5B,WAAW,IAAI,IAAA,gBAAQ,EAAC,GAAG,EAAE,IAAI,CAAC,CAAC;YACnC,OAAO,IAAI,CAAC;SACb;QAED,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE;YAC1B,MAAM,GAAG,GAAG,IAAA,0BAAkB,EAAC,WAAW,CAAC,CAAC;YAC5C,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC;YACzC,WAAW,GAAG,EAAE,CAAC;SAClB;QAED,IAAI,IAAI,GAAG,IAAA,0BAAkB,EAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC;QAC3C,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,iBAAiB,EAAE,OAAO,CAAC,CAAC;QAChD,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC;IAC5C,CAAC,CAAC,CAAC;IAEH,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE;QAC1B,MAAM,IAAI,GAAG,IAAA,0BAAkB,EAAC,WAAW,CAAC,CAAC;QAC7C,KAAK,GAAG,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC;KACnD;IAED,KAAK,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE;QACxB,OAAO,GAAG,CAAC,IAAI,EAAE,CAAC;IACpB,CAAC,CAAC,CAAC;IAEH,MAAM,KAAK,GAAG,IAAA,iBAAO,EAAC,0BAA0B,CAAC,CAAC;IAClD,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE;QAC3B,OAAO,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IACzB,CAAC,CAAC,CAAC;IAEH,OAAO,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;AAC5B,CAAC,CAAC;AAhDW,QAAA,aAAa,iBAgDxB;AAEK,MAAM,WAAW,GAAG,CACzB,GAAiB,EACjB,OAAwB,EACV,EAAE;IAChB,MAAM,KAAK,GAAG,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IAChC,KAAK,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAwB,EAAE,EAAE;QACtD,MAAM,MAAM,GAAG,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,CAAC;QACnC,IAAI,MAAM,EAAE;YACV,GAAG,CAAC,OAAO,CAAC,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC;SAClC;IACH,CAAC,CAAC,CAAC;IACH,OAAO,GAAG,CAAC;AACb,CAAC,CAAC;AAZW,QAAA,WAAW,eAYtB;AAEK,MAAM,wBAAwB,GAAG,CACtC,GAAiB,EACjB,OAAwB,EACxB,IAAY,EACE,EAAE;IAChB,MAAM,QAAQ,GAAG,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IAEnC,QAAQ,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAwB,EAAE,EAAE;QACzD,MAAM,EAAE,GAAG,GAAG,CAAC,OAAO,CAAC,CAAC;QACxB,MAAM,GAAG,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC;QAC9B,MAAM,IAAI,GAAG,EAAE,CAAC,IAAI,EAAE,CAAC;QAEvB,MAAM,SAAS,GAAG,IAAA,mBAAS,EAAC,IAAI,EAAE,IAAI,CAAC,CAAC;QACxC,IACE,CAAC,GAAG,KAAK,IAAI,IAAI,IAAI,KAAK,KAAK,CAAC;YAChC,SAAS,CAAC,aAAa,GAAG,CAAC;YAC3B,EAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,MAAM,KAAK,CAAC;YAC9B,EAAE,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,KAAK,CAAC,EAC7B;YACA,GAAG,CAAC,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC;SAClB;aAAM;YACL,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;YAC5B,IAAI,OAAO,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,OAAO,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,KAAK,GAAG,EAAE;gBAC7D,GAAG,CAAC,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC;aAClB;SACF;IACH,CAAC,CAAC,CAAC;IAEH,OAAO,GAAG,CAAC;AACb,CAAC,CAAC;AA7BW,QAAA,wBAAwB,4BA6BnC;AAEK,MAAM,yBAAyB,GAAG,CACvC,GAAiB,EACjB,OAAwB,EACV,EAAE;IAChB,MAAM,YAAY,GAAG,OAAO,CAAC,IAAI,CAAC,iBAAiB,CAAC,CAAC;IAErD,YAAY,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAwB,EAAE,EAAE;QAC7D,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,MAAM,IAAI,GAAG,GAAG,CAAC,OAAO,CAAC,CAAC;QAC1B,MAAM,YAAY,GAAG,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;QAC/C,IAAI,YAAY,EAAE;YAChB,KAAK,GAAG,QAAQ,CAAC,YAAY,EAAE,EAAE,CAAC,IAAI,CAAC,CAAC;SACzC;QAED,IAAI,KAAK,GAAG,CAAC,EAAE;YACb,GAAG,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,CAAC;SACpB;IACH,CAAC,CAAC,CAAC;IAEH,OAAO,GAAG,CAAC;AACb,CAAC,CAAC;AApBW,QAAA,yBAAyB,6BAoBpC;AAEK,MAAM,iBAAiB,GAAG,CAC/B,IAAY,EACZ,IAAa,EACb,KAAc,EACN,EAAE;IACV,IAAI,aAAa,GAAG,IAAI,CAAC;IACzB,oGAAoG;IACpG,MAAM,YAAY,GAA2B;QAC3C,OAAO,EAAE,GAAG;QACZ,QAAQ,EAAE,GAAG;QACb,QAAQ,EAAE,GAAG;QACb,QAAQ,EAAE,GAAG;QACb,QAAQ,EAAE,GAAG;QACb,MAAM,EAAE,GAAG;QACX,MAAM,EAAE,GAAG;QACX,QAAQ,EAAE,GAAG;QACb,SAAS,EAAE,GAAG;QACd,QAAQ,EAAE,GAAG;QACb,OAAO,EAAE,GAAG;QACZ,OAAO,EAAE,GAAG;KACb,CAAC;IAEF,MAAM,WAAW,GAA2B;QAC1C,GAAG,EAAE,GAAG;QACR,GAAG,EAAE,GAAG;QACR,qDAAqD;QACrD,IAAI,EAAE,GAAG;QACT,IAAI,EAAE,GAAG;KACV,CAAC;IAEF,IAAI,IAAI,EAAE;QACR,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,CAAC,YAAY,CAAC,EAAE;YAC3C,MAAM,SAAS,GAAG,IAAI,MAAM,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC;YACvC,aAAa,GAAG,aAAa,CAAC,OAAO,CAAC,SAAS,EAAE,YAAY,CAAC,GAAG,CAAC,CAAC,CAAC;SACrE;KACF;IAED,IAAI,KAAK,EAAE;QACT,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,CAAC,WAAW,CAAC,EAAE;YAC1C,MAAM,gBAAgB,GAAG,IAAI,MAAM,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC;YAC9C,aAAa,GAAG,aAAa,CAAC,OAAO,CAAC,gBAAgB,EAAE,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC;SAC3E;KACF;IAED,OAAO,aAAa,CAAC;AACvB,CAAC,CAAC;AA7CW,QAAA,iBAAiB,qBA6C5B;AAEK,MAAM,eAAe,GAAG,CAC7B,GAAiB,EACjB,OAAwB,EACV,EAAE;IAChB,MAAM,KAAK,GAAG,OAAO,CAAC,IAAI,CAAC,uBAAuB,CAAC,CAAC;IACpD,KAAK,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAwB,EAAE,EAAE;QACtD,GAAG,CAAC,OAAO,CAAC,CAAC,WAAW,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;IAChD,CAAC,CAAC,CAAC;IACH,OAAO,GAAG,CAAC;AACb,CAAC,CAAC;AATW,QAAA,eAAe,mBAS1B;AAEK,MAAM,QAAQ,GAAG,CAAC,GAAiB,EAAE,IAAqB,EAAU,EAAE;IAC3E,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC9B,IAAI,IAAI,GAAG,EAAE,CAAC;IAEd,KAAK,CAAC,IAAI,CAAC,CAAC,MAAc,EAAE,OAAwB,EAAE,EAAE;QACtD,IAAI,GAAG,GAAG,IAAI,QAAQ,GAAG,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC;IAC9C,CAAC,CAAC,CAAC;IACH,IAAI,GAAG,GAAG,IAAI,IAAI,CAAC;IACnB,OAAO,IAAI,CAAC;AACd,CAAC,CAAC;AATW,QAAA,QAAQ,YASnB;AAEF,MAAM,SAAS,GAAG,CAChB,GAAiB,EACjB,OAAwB,EACxB,IAAY,EACJ,EAAE;IACV,IAAA,iCAAyB,EAAC,GAAG,EAAE,OAAO,CAAC,CAAC;IACxC,IAAA,mBAAW,EAAC,GAAG,EAAE,OAAO,CAAC,CAAC;IAC1B,IAAA,sBAAc,EAAC,GAAG,EAAE,OAAO,CAAC,CAAC;IAC7B,IAAA,uBAAe,EAAC,GAAG,EAAE,OAAO,CAAC,CAAC;IAC9B,IAAA,gCAAwB,EAAC,GAAG,EAAE,OAAO,EAAE,IAAI,CAAC,CAAC;IAC7C,MAAM,aAAa,GAAG,IAAA,qBAAa,EAAC,GAAG,EAAE,OAAO,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC;IACtE,OAAO,IAAA,yBAAiB,EAAC,aAAa,EAAE,KAAK,EAAE,IAAI,CAAC,CAAC;AACvD,CAAC,CAAC;AAEF,kBAAe,SAAS,CAAC","sourcesContent":["import cheerio from \"cheerio\";\nimport xregexp from \"xregexp\";\nimport stopwords from \"./stopwords\";\n\nexport const addNewlineToBr = (\n  doc: cheerio.Root,\n  topNode: cheerio.Cheerio,\n): cheerio.Root => {\n  const brs = topNode.find(\"br\");\n\n  brs.each((_index: number, element: cheerio.Element) => {\n    const br = doc(element);\n    br.replaceWith(\"\\n\\n\");\n  });\n\n  return doc;\n};\n\nexport const cleanParagraphText = (rawText: string): string => {\n  const text = rawText.trim();\n  text.replace(/[\\s\\t]+/g, \" \");\n  return text;\n};\n\nexport const convertToText = (\n  doc: cheerio.Root,\n  topNode: cheerio.Cheerio,\n): string => {\n  let texts: string[] = [];\n  const nodes = topNode.contents();\n\n  let hangingText = \"\";\n\n  nodes.each((_index: number, element: cheerio.Element) => {\n    const node = doc(element);\n    const nodeType = node.get(0) ? node.get(0).type : null;\n    const nodeName = node.get(0) ? node.get(0).tagName : null;\n\n    if (nodeType === \"text\") {\n      hangingText += node.text();\n      return true;\n    } else if (nodeName === \"ul\") {\n      hangingText += ulToText(doc, node);\n      return true;\n    }\n\n    if (hangingText.length > 0) {\n      const txt = cleanParagraphText(hangingText);\n      texts = texts.concat(txt.split(/\\r?\\n/));\n      hangingText = \"\";\n    }\n\n    let text = cleanParagraphText(node.text());\n    text = text.replace(/(\\w+\\.)([A-Z]+)/, \"$1 $2\");\n    texts = texts.concat(text.split(/\\r?\\n/));\n  });\n\n  if (hangingText.length > 0) {\n    const text = cleanParagraphText(hangingText);\n    texts = texts = texts.concat(text.split(/\\r?\\n/));\n  }\n\n  texts = texts.map((txt) => {\n    return txt.trim();\n  });\n\n  const regex = xregexp(\"[\\\\p{Number}\\\\p{Letter}]\");\n  texts = texts.filter((txt) => {\n    return regex.test(txt);\n  });\n\n  return texts.join(\"\\n\\n\");\n};\n\nexport const linksToText = (\n  doc: cheerio.Root,\n  topNode: cheerio.Cheerio,\n): cheerio.Root => {\n  const nodes = topNode.find(\"a\");\n  nodes.each((_index: number, element: cheerio.Element) => {\n    const htmlEl = doc(element).html();\n    if (htmlEl) {\n      doc(element).replaceWith(htmlEl);\n    }\n  });\n  return doc;\n};\n\nexport const removeFewWordsParagraphs = (\n  doc: cheerio.Root,\n  topNode: cheerio.Cheerio,\n  lang: string,\n): cheerio.Root => {\n  const allNodes = topNode.find(\"*\");\n\n  allNodes.each((_index: number, element: cheerio.Element) => {\n    const el = doc(element);\n    const tag = el.get(0).tagName;\n    const text = el.text();\n\n    const stopWords = stopwords(text, lang);\n    if (\n      (tag !== \"br\" || text !== \"\\\\r\") &&\n      stopWords.stopWordCount < 3 &&\n      el.find(\"object\").length === 0 &&\n      el.find(\"embed\").length === 0\n    ) {\n      doc(el).remove();\n    } else {\n      const trimmed = text.trim();\n      if (trimmed[0] === \"(\" && trimmed[trimmed.length - 1] === \")\") {\n        doc(el).remove();\n      }\n    }\n  });\n\n  return doc;\n};\n\nexport const removeNegativescoresNodes = (\n  doc: cheerio.Root,\n  topNode: cheerio.Cheerio,\n): cheerio.Root => {\n  const gravityItems = topNode.find(\"*[gravityScore]\");\n\n  gravityItems.each((_index: number, element: cheerio.Element) => {\n    let score = 0;\n    const item = doc(element);\n    const gravityScore = item.attr(\"gravityScore\");\n    if (gravityScore) {\n      score = parseInt(gravityScore, 10) || 0;\n    }\n\n    if (score < 1) {\n      doc(item).remove();\n    }\n  });\n\n  return doc;\n};\n\nexport const replaceCharacters = (\n  text: string,\n  html: boolean,\n  chars: boolean,\n): string => {\n  let processedText = text;\n  // if element does not match any in map and starts with & and ends with ;, replace with empty string\n  const htmlEntities: Record<string, string> = {\n    \"&amp;\": \"&\",\n    \"&apos;\": \"'\",\n    \"&cent;\": \"¢\",\n    \"&copy;\": \"©\",\n    \"&euro;\": \"€\",\n    \"&gt;\": \">\",\n    \"&lt;\": \"<\",\n    \"&nbsp;\": \" \",\n    \"&pound;\": \"£\",\n    \"&quot;\": '\"',\n    \"&reg;\": \"®\",\n    \"&yen;\": \"¥\",\n  };\n\n  const escapeChars: Record<string, string> = {\n    '\"': '\"',\n    \"'\": \"'\",\n    // tslint:disable-next-line: object-literal-sort-keys\n    \"\\n\": \" \",\n    \"\\r\": \" \",\n  };\n\n  if (html) {\n    for (const key of Object.keys(htmlEntities)) {\n      const htmlregex = new RegExp(key, \"g\");\n      processedText = processedText.replace(htmlregex, htmlEntities[key]);\n    }\n  }\n\n  if (chars) {\n    for (const key of Object.keys(escapeChars)) {\n      const escapeCharsRegex = new RegExp(key, \"g\");\n      processedText = processedText.replace(escapeCharsRegex, escapeChars[key]);\n    }\n  }\n\n  return processedText;\n};\n\nexport const replaceWithText = (\n  doc: cheerio.Root,\n  topNode: cheerio.Cheerio,\n): cheerio.Root => {\n  const nodes = topNode.find(\"b, strong, i, br, sup\");\n  nodes.each((_index: number, element: cheerio.Element) => {\n    doc(element).replaceWith(doc(element).text());\n  });\n  return doc;\n};\n\nexport const ulToText = (doc: cheerio.Root, node: cheerio.Cheerio): string => {\n  const nodes = node.find(\"li\");\n  let text = \"\";\n\n  nodes.each((_index: number, element: cheerio.Element) => {\n    text = `${text}\\n * ${doc(element).text()}`;\n  });\n  text = `${text}\\n`;\n  return text;\n};\n\nconst formatter = (\n  doc: cheerio.Root,\n  topNode: cheerio.Cheerio,\n  lang: string,\n): string => {\n  removeNegativescoresNodes(doc, topNode);\n  linksToText(doc, topNode);\n  addNewlineToBr(doc, topNode);\n  replaceWithText(doc, topNode);\n  removeFewWordsParagraphs(doc, topNode, lang);\n  const convertedText = convertToText(doc, topNode).replace(/\\n/g, \" \");\n  return replaceCharacters(convertedText, false, true);\n};\n\nexport default formatter;\n"]}