compromise/src/data/people/female.js

natural language processing in the browser

'use strict';
const fns = require('../fns');
//names with a distinctive signal that they identify as a female, internationally

//compressed by frequent suffixes
//comprssed with
//https://github.com/nlp-compromise/thumb/blob/master/src/compress/compress.js
const compressed = {
  stine: 'chri,erne,ju,kri',
  rlene: 'a,cha,da,ma',
  eline: 'ad,ang,jacqu,mad',
  nette: 'an,antoi,jean,ly',
  elia: 'ad,am,ang,cec,c,corn,d,of,sh',
  anne: ',di,je,jo,le,mari,rox,sus,suz',
  elle: 'dani,est,gabri,isab,jan,mich,rach,roch',
  ella: 'd,est,isab,lu,marc,st',
  rina: 'kata,kat,ma,sab,t',
  icia: 'al,fel,let,patr,tr',
  ette: 'bernad,b,claud,paul,yv',
  leen: 'ai,cath,col,ei,kath',
  ndra: 'alexa,cassa,ke,sa,so',
  elma: ',s,th,v',
  anda: 'am,mir,w,yol',
  etta: ',henri,lor,ros',
  isha: 'al,ke,lat,tr',
  tina: 'cris,mar,,valen',
  inda: 'bel,l,luc,mel',
  arla: 'c,d,k,m',
  lena: 'e,je,,magda',
  ine: 'carol,cather,cel,ela,franc,gerald,jan,jasm,jeann,joseph,kathar,kather,lorra,max,nad,paul',
  ice: 'al,beatr,bern,cand,clar,eun,jan,patr',
  ela: 'andj,ang,carm,gabri,graci,l,manu,pam',
  ara: 'barb,c,cl,k,l,tam,t,z',
  ora: 'c,d,fl,isid,len,l,n,teod',
  ina: 'am,catal,d,georg,g,josef,n',
  ita: 'an,arp,bon,juan,kav,margar,r',
  nna: 'dea,do,gle,je,joha,lado,sha',
  lyn: 'caro,eve,gwendo,jac,jacque,joce,mari',
  ica: 'angel,er,jess,mil,mon,patr,veron',
  ene: 'adri,hel,imog,ir,jol,lor',
  ana: 'adri,d,jov,ju,l,sus',
  nda: 'bre,gle,ly,rho,ro',
  nia: 'anto,euge,so,to,virgi',
  ley: 'ash,kel,kimber,les,shir',
  sha: 'lata,mar,nata,ta',
  ian: 'jill,lill,mar,viv',
  isa: 'al,el,l,lu',
  ann: ',jo,le,mary',
  ise: 'den,el,elo,lou',
  ida: 'a,,rach,sa',
  nya: 'lato,so,ta,to',
  ssa: 'aly,mari,meli,vane',
  tha: 'ber,mar,saman,tabi',
  ia: 'cecil,claud,cynth,dam,georg,glor,jul,luc,lyd,marc,mar,nad,oliv,silv,sof,soph,sylv,victor',
  la: 'eu,kay,lei,leo,li,lo,pau,priscil,shei,ursu,vio,wil',
  na: 'de,ed,leo,lor,mo,myr,ramo,re,shau,shaw,shee,ver',
  le: 'ade,camil,caro,ceci,ga,gay,lucil,mab,myrt,nicho,nico',
  en: 'carm,dore,ell,gretch,gw,hel,kar,kirst,krist,laur,maure',
  ra: 'aud,barb,deb,elvi,javie,lau,may,my,pet,ve',
  ma: 'al,em,er,fati,ir,kari,nai,nor,wil',
  el: 'eth,isab,laur,mab,marib,muri,racha,rach,raqu',
  ta: 'alber,al,chris,ek,kris,mandakran,mar,rober',
  ey: 'audr,brittn,courtn,linds,stac,trac,whitn',
  ri: 'je,kanyakuma,ka,ker,sha,she,ter',
  ne: 'corin,daph,ja,laver,lyn,simo,yvon',
  th: 'be,edi,elisabe,elizabe,judi,meredi,ru',
  ah: 'aish,beul,debor,hann,le,rebek,sar',
  is: 'delor,dor,jan,lo,mav,phyll',
  da: 'a,fre,frie,hil,matil,priyamva',
  ce: 'canda,constan,floren,gra,joy',
  es: 'agn,delor,dolor,franc,merced',
  er: 'amb,est,esth,heath,jennif',
  et: 'bridg,harri,jan,margar,margr',
  ca: 'bian,blan,francis,rebec',
  ja: 'an,khadi,mari,son',
  sa: 'el,ro,tere,there',
  ee: 'aim,d,desir,ren',
  va: 'a,el,e,i',
  in: 'caitl,er,kar,krist',
  on: 'alis,man,shann,shar',
  an: 'meag,meg,megh,sus'
};

let list = [
  'abigail',
  'aicha',
  'alya',
  'andrea',
  'beatriz',
  'bettye',
  'brandi',
  'brooke',
  'carol',
  'celeste',
  'chelsea',
  'cheryl',
  'chloe',
  'claire',
  'cleo',
  'constanza',
  'consuelo',
  'crystal',
  'dominique',
  'dorothea',
  'eleanor',
  'eliza',
  'erika',
  'fay',
  'faye',
  'fern',
  'gail',
  'genevieve',
  'gertrude',
  'gladys',
  'heidi',
  'ingrid',
  'jade',
  'jill',
  'jo',
  'joni',
  'kate',
  'katie',
  'kathryn',
  'kay',
  'kim',
  'krystal',
  'latoya',
  'laxmi',
  'leigh',
  'lindsay',
  'lupe',
  'lynn',
  'mae',
  'malika',
  'margo',
  'marguerite',
  'marisol',
  'maritza',
  'maude',
  'maya',
  'mildred',
  'miriam',
  'monique',
  'mrignayani',
  'naomi',
  'nell',
  'nikki',
  'olga',
  'paige',
  'pam',
  'parvati',
  'pearl',
  'reba',
  'robyn',
  'rosalind',
  'sheryl',
  'sue',
  'sybil',
  'tami',
  'tamika',
  'therese',
  'toni',
  'gisele'
];
list = fns.uncompress_suffixes(list, compressed);

for (let i = 0; i < list.length; i++) {
  let str = list[i];
  if (str.match(/[^ea]y$/)) {
    list.push(str.replace(/y$/, 'i'));
  }
  if (str.match(/ll/)) {
    list.push(str.replace(/ll/, 'l'));
  }
  if (str.match(/nn/)) {
    list.push(str.replace(/nn/, 'n'));
  }
  if (str.match(/ah/)) {
    list.push(str.replace(/ah/, 'a'));
  }
  if (str.match(/t$/)) {
    list.push(str.replace(/t$/, 'tte'));
  }
  if (str.match(/ey$/)) {
    list.push(str.replace(/ey$/, 'y'));
  }
  if (str.match(/ie$/)) {
    list.push(str.replace(/ie$/, 'y'));
  }
  if (str.match(/ne$/)) {
    list.push(str.replace(/ne$/, 'na'));
  }
  if (str.match(/ss/)) {
    list.push(str.replace(/ss/, 's'));
  }
  if (str.match(/rr/)) {
    list.push(str.replace(/rr/, 'r'));
  }
}
const no_change = [
  'amy',
  'becky',
  'betty',
  'beverly',
  'cathy',
  'dolly',
  'dorothy',
  'hilary',
  'hillary',
  'kimberly',
  'rosemary',
  'sally',
  'shelly',
  'trudy',
  'tammy',
  'wendy',
  'ruby',
  'susi'
];
list = list.concat(no_change);
module.exports = list;
// console.log(list.indexOf('kelley'));