culled words

This commit is contained in:
lingdocs 2021-10-04 21:15:31 -04:00
parent eff4335f91
commit ddf9f33fb2
9 changed files with 7 additions and 17 deletions

View File

@ -64,12 +64,17 @@ function getVerbsFromTsS(entries) {
} }
function getNounsAdjsFromTsS(entries) { function getNounsAdjsFromTsS(entries) {
return allNounAdjTsS.map(item => { const b = allNounAdjTsS.map(item => {
const entry = entries.find(x => item.ts === x.ts); const entry = entries.find(x => item.ts === x.ts);
if (!entry) { if (!entry) {
console.log("couldn't find ts", item); console.log("couldn't find ts", item);
return undefined; return undefined;
} }
const firstWord = entry.e.split(",")[0].split(";")[0].split("(")[0].trim();
console.log(firstWord, entry.f, entry.ts);
// if (firstWord.contains(" ")) console.log("SPACE PRESENT");
return { entry, def: item.e, category: item.category }; return { entry, def: item.e, category: item.category };
}).filter(x => x); }).filter(x => x);
return b;
// console.log(b.length, "number of nouns/adjs");
} }

View File

@ -24,7 +24,6 @@ module.exports = [
{ ts: 1578009902092, e: `veil, burka` }, // برقه - bUrqá { ts: 1578009902092, e: `veil, burka` }, // برقه - bUrqá
{ ts: 1527816994, e: `program` }, // برنامه - barnaama { ts: 1527816994, e: `program` }, // برنامه - barnaama
{ ts: 1579294091093, e: `balcony, veranda, porch` }, // برنډه - baranDá { ts: 1579294091093, e: `balcony, veranda, porch` }, // برنډه - baranDá
{ ts: 1527817837, e: `whorehouse, brothel` }, // بړوا خانه - baRwaakhaana
{ ts: 1527823617, e: `crime, offense, sin, guilt, fault` }, // بزه - bazá { ts: 1527823617, e: `crime, offense, sin, guilt, fault` }, // بزه - bazá
{ ts: 1527823619, e: `moth` }, // بزه - bUzá { ts: 1527823619, e: `moth` }, // بزه - bUzá
{ ts: 1527823620, e: `patch (in a garment)` }, // بزه - bza { ts: 1527823620, e: `patch (in a garment)` }, // بزه - bza
@ -85,7 +84,6 @@ module.exports = [
{ ts: 1527823717, e: `cloth, fabric, material, clothing, garment` }, // کپړه - kapRá { ts: 1527823717, e: `cloth, fabric, material, clothing, garment` }, // کپړه - kapRá
{ ts: 1527816257, e: `notebook, little book` }, // کتابچه - kitaabcha { ts: 1527816257, e: `notebook, little book` }, // کتابچه - kitaabcha
{ ts: 1527820050, e: `bag, purse` }, // کڅوړه - katsóRa { ts: 1527820050, e: `bag, purse` }, // کڅوړه - katsóRa
{ ts: 1594834749012, e: `flirting, wink, flirtatious wink` }, // کرشمه - karishmá, karashmá
{ ts: 1527813252, e: `line, trace` }, // کرښه - kurxa { ts: 1527813252, e: `line, trace` }, // کرښه - kurxa
{ ts: 1527823590, e: `sphere, globe` }, // کره - kará { ts: 1527823590, e: `sphere, globe` }, // کره - kará
{ ts: 1527823591, e: `shovel, scraper, scoop` }, // کره - kára { ts: 1527823591, e: `shovel, scraper, scoop` }, // کره - kára
@ -99,5 +97,4 @@ module.exports = [
{ ts: 1591034128816, e: `mistake, error, blunder, fault` }, // سهوه - sáhwa { ts: 1591034128816, e: `mistake, error, blunder, fault` }, // سهوه - sáhwa
{ ts: 1527814370, e: `nostril` }, // سوږمه - soGma { ts: 1527814370, e: `nostril` }, // سوږمه - soGma
{ ts: 1527817498, e: `famine, starvation, serious hunger/lack of food, drought, crop failure` }, // سوکړه - sookRá { ts: 1527817498, e: `famine, starvation, serious hunger/lack of food, drought, crop failure` }, // سوکړه - sookRá
{ ts: 1527814144, e: `charred log` }, // سوکه - swaka
]; ];

View File

@ -6,7 +6,6 @@ module.exports = [
{ ts: 1527811859, e: `union, alliance` }, // اتحاد - itihaad { ts: 1527811859, e: `union, alliance` }, // اتحاد - itihaad
{ ts: 1527822033, e: `joining, connection, contiguity, junction` }, // اتصال - ittisáal { ts: 1527822033, e: `joining, connection, contiguity, junction` }, // اتصال - ittisáal
{ ts: 1527811858, e: `unity, alliance, agreement, understanding, consent; coincidence` }, // اتفاق - itifaaq { ts: 1527811858, e: `unity, alliance, agreement, understanding, consent; coincidence` }, // اتفاق - itifaaq
{ ts: 1527820147, e: `the Atan, the Pashtun/Afghan national dance` }, // اتڼ - atáN
{ ts: 1527813560, e: `accusation, charge, indictment` }, // اتهام - itihaam { ts: 1527813560, e: `accusation, charge, indictment` }, // اتهام - itihaam
{ ts: 1527812105, e: `respect, honor, esteem, deference` }, // احترام - ihtiraam { ts: 1527812105, e: `respect, honor, esteem, deference` }, // احترام - ihtiraam
{ ts: 1527819653, e: `possibility, probability, likelihood` }, // احتمال - ihtimaal { ts: 1527819653, e: `possibility, probability, likelihood` }, // احتمال - ihtimaal
@ -86,7 +85,6 @@ module.exports = [
{ ts: 1527812176, e: `evening` }, // ماښام - maaxaam { ts: 1527812176, e: `evening` }, // ماښام - maaxaam
{ ts: 1527813601, e: `death` }, // مرګ - marg { ts: 1527813601, e: `death` }, // مرګ - marg
{ ts: 1527817691, e: `future` }, // مستقبل - mUstaqbil { ts: 1527817691, e: `future` }, // مستقبل - mUstaqbil
{ ts: 1527817963, e: `victim-hood, oppression` }, // مظلومیت - mazloomiyát
{ ts: 1527811866, e: `damage, defect, loss` }, // نقصان - nUqsaan { ts: 1527811866, e: `damage, defect, loss` }, // نقصان - nUqsaan
{ ts: 1527815122, e: `name` }, // نوم - noom { ts: 1527815122, e: `name` }, // نوم - noom
{ ts: 1527812661, e: `boy, young lad` }, // هلک - halík, halúk { ts: 1527812661, e: `boy, young lad` }, // هلک - halík, halúk

View File

@ -61,7 +61,6 @@ module.exports = [
{ ts: 1527819672, e: "crafty" }, // چالاک - chaaláak { ts: 1527819672, e: "crafty" }, // چالاک - chaaláak
{ ts: 1527811230, e: "quick, fast" }, // چټک - chaTak { ts: 1527811230, e: "quick, fast" }, // چټک - chaTak
{ ts: 1527812524, e: "started, in motion" }, // چلان - chalaan { ts: 1527812524, e: "started, in motion" }, // چلان - chalaan
{ ts: 1527822858, e: "curly, wavy" }, // څپول - tsapól
{ ts: 1527815370, e: "clear, apparent" }, // څرګند - tsărgund { ts: 1527815370, e: "clear, apparent" }, // څرګند - tsărgund
{ ts: 1576366107077, e: "straight, upright" }, // څک - tsak { ts: 1576366107077, e: "straight, upright" }, // څک - tsak
{ ts: 1527812113, e: "present, on hand, ready" }, // حاضر - haazir, haazur { ts: 1527812113, e: "present, on hand, ready" }, // حاضر - haazir, haazur
@ -73,13 +72,11 @@ module.exports = [
{ ts: 1527814219, e: "relative, one's own" }, // خپل - khpul { ts: 1527814219, e: "relative, one's own" }, // خپل - khpul
{ ts: 1527812795, e: "relative" }, // خپلوان - khpulwaan { ts: 1527812795, e: "relative" }, // خپلوان - khpulwaan
{ ts: 1527812808, e: "poor, miserable" }, // خوار - khwaar { ts: 1527812808, e: "poor, miserable" }, // خوار - khwaar
{ ts: 1611065752352, e: "moldy" }, // دربجن - drabjan
{ ts: 1527814880, e: "tall" }, // دنګ - dung { ts: 1527814880, e: "tall" }, // دنګ - dung
{ ts: 1527812537, e: "assured" }, // ډاډمن - DaaDmun { ts: 1527812537, e: "assured" }, // ډاډمن - DaaDmun
{ ts: 1527812583, e: "full" }, // ډک - Duk { ts: 1527812583, e: "full" }, // ډک - Duk
{ ts: 1527822674, e: "gaunt" }, // ډنګر - Dungár, Dangár { ts: 1527822674, e: "gaunt" }, // ډنګر - Dungár, Dangár
{ ts: 1527817256, e: "sunk" }, // ډوب - Doob { ts: 1527817256, e: "sunk" }, // ډوب - Doob
{ ts: 1527812559, e: "very, lots" }, // ډېر - Der
{ ts: 1527814277, e: "healthy" }, // روغ - rogh { ts: 1527814277, e: "healthy" }, // روغ - rogh
{ ts: 1609780006604, e: "fruitful" }, // زرخېز - zarkhéz { ts: 1609780006604, e: "fruitful" }, // زرخېز - zarkhéz
{ ts: 1527817116, e: "green, flourishing" }, // زرغون - zarghoon { ts: 1527817116, e: "green, flourishing" }, // زرغون - zarghoon

View File

@ -20,7 +20,6 @@ module.exports = [
{ ts: 1527818084, e: `drinker, drunkard, alcoholic, wine-bibber` }, // شرابي - sharaabee { ts: 1527818084, e: `drinker, drunkard, alcoholic, wine-bibber` }, // شرابي - sharaabee
{ ts: 1527821950, e: `prince` }, // شهزاده - shahzaadá { ts: 1527821950, e: `prince` }, // شهزاده - shahzaadá
{ ts: 1588158828142, e: `hunter` }, // ښکاري - xkaaree { ts: 1588158828142, e: `hunter` }, // ښکاري - xkaaree
{ ts: 1527811789, e: `religious warrior for the faith (in Islam)` }, // غازي - ghaazee
{ ts: 1527815206, e: `judge, religious authority/judge` }, // قاضي - qaazee { ts: 1527815206, e: `judge, religious authority/judge` }, // قاضي - qaazee
{ ts: 1527818500, e: `contractor, supplier` }, // قراردادي - qaraardaadee { ts: 1527818500, e: `contractor, supplier` }, // قراردادي - qaraardaadee
{ ts: 1527816446, e: `paternal uncle, term of address for elderly man` }, // کاکا - kaakaa { ts: 1527816446, e: `paternal uncle, term of address for elderly man` }, // کاکا - kaakaa

View File

@ -77,7 +77,6 @@ module.exports = [
{ ts: 1588161660483, e: `crutch, walking-stick, cane` }, // ډانګوری - Daangooréy { ts: 1588161660483, e: `crutch, walking-stick, cane` }, // ډانګوری - Daangooréy
{ ts: 1527813493, e: `majority; heap, pile` }, // ډېری - Derey { ts: 1527813493, e: `majority; heap, pile` }, // ډېری - Derey
{ ts: 1527823700, e: `avalanche, flood, shower` }, // راشی - raashey { ts: 1527823700, e: `avalanche, flood, shower` }, // راشی - raashey
{ ts: 1527819525, e: `diminutive form of زړه - little heart; precious, dear` }, // زړګی - zuRgéy
{ ts: 1527819732, e: `young, youth, young lad` }, // زلمی - zalméy { ts: 1527819732, e: `young, youth, young lad` }, // زلمی - zalméy
{ ts: 1527813708, e: `good news, gospel` }, // زېری - zerey { ts: 1527813708, e: `good news, gospel` }, // زېری - zerey
{ ts: 1588758498458, e: `jaundice` }, // زېړی - zeRéy { ts: 1588758498458, e: `jaundice` }, // زېړی - zeRéy
@ -110,7 +109,6 @@ module.exports = [
{ ts: 1527818483, e: `plate` }, // غوری - ghorey { ts: 1527818483, e: `plate` }, // غوری - ghorey
{ ts: 1527816181, e: `vomit, nausea (Arabic)` }, // قی - qey { ts: 1527816181, e: `vomit, nausea (Arabic)` }, // قی - qey
{ ts: 1527814715, e: `user` }, // کاروونکی - kaarawoonkey { ts: 1527814715, e: `user` }, // کاروونکی - kaarawoonkey
{ ts: 1578706351012, e: `dress, clothing, underwear, ornament (usually plural)` }, // کالی - kaaley
{ ts: 1527823295, e: `stone, rock` }, // کاڼی - káaNey { ts: 1527823295, e: `stone, rock` }, // کاڼی - káaNey
{ ts: 1527818563, e: `muscle` }, // کبوړی - kabóoRey { ts: 1527818563, e: `muscle` }, // کبوړی - kabóoRey
{ ts: 1527822824, e: `booklet, notebook` }, // کتاب ګوټی - kitaabgóTey { ts: 1527822824, e: `booklet, notebook` }, // کتاب ګوټی - kitaabgóTey
@ -130,7 +128,6 @@ module.exports = [
{ ts: 1527812416, e: `brook, rivulet, small irrigation` }, // لښتی - laxtey { ts: 1527812416, e: `brook, rivulet, small irrigation` }, // لښتی - laxtey
{ ts: 1527814401, e: `toy` }, // لوبونی - lobawuney { ts: 1527814401, e: `toy` }, // لوبونی - lobawuney
{ ts: 1527814519, e: `side, direction` }, // لوری - lorey { ts: 1527814519, e: `side, direction` }, // لوری - lorey
{ ts: 1527813846, e: `smoke` }, // لوګی - loogéy
{ ts: 1527823103, e: `perspective, viewpoint` }, // لیدلوری - leedlorey { ts: 1527823103, e: `perspective, viewpoint` }, // لیدلوری - leedlorey
{ ts: 1527819920, e: `mosquito, midge` }, // ماشی - maashey { ts: 1527819920, e: `mosquito, midge` }, // ماشی - maashey
{ ts: 1527820224, e: `fly swatter` }, // مچوژی - muchwajzéy { ts: 1527820224, e: `fly swatter` }, // مچوژی - muchwajzéy
@ -144,7 +141,6 @@ module.exports = [
{ ts: 1527823577, e: `sapling, seedling, sprout, young tree` }, // نیالګی - niyaalgey { ts: 1527823577, e: `sapling, seedling, sprout, young tree` }, // نیالګی - niyaalgey
{ ts: 1527812073, e: `bone` }, // هډوکی - haDookey { ts: 1527812073, e: `bone` }, // هډوکی - haDookey
{ ts: 1527812668, e: `welcome` }, // هرکلی - hărkáley { ts: 1527812668, e: `welcome` }, // هرکلی - hărkáley
{ ts: 1527820928, e: `awkward/silly laugh, unmotivated laugh` }, // هرکی - hurkéy
{ ts: 1588153218244, e: `height, elevation, tallness` }, // هسکوالی - haskwáaley { ts: 1588153218244, e: `height, elevation, tallness` }, // هسکوالی - haskwáaley
{ ts: 1585309922022, e: `flu, respiratory illness, influenza, cold` }, // والګی - waalgéy { ts: 1585309922022, e: `flu, respiratory illness, influenza, cold` }, // والګی - waalgéy
{ ts: 1527813014, e: `vein, nerve` }, // وژی - wajzey { ts: 1527813014, e: `vein, nerve` }, // وژی - wajzey

View File

@ -25,7 +25,6 @@ module.exports = [
{ ts: 1591871233587, e: "well-spoken" }, // خوږژبی { ts: 1591871233587, e: "well-spoken" }, // خوږژبی
{ ts: 1527814118, e: "painful, agonizing" }, // دردوونکی { ts: 1527814118, e: "painful, agonizing" }, // دردوونکی
{ ts: 1527820657, e: "mediator, arbitrator" }, // درېیمګړی { ts: 1527820657, e: "mediator, arbitrator" }, // درېیمګړی
{ ts: 1527818179, e: "cross-eyed, blind in one eye, one eyed" }, // دوه ګوری
{ ts: 1527815713, e: "coming, future" }, // راتلونکی { ts: 1527815713, e: "coming, future" }, // راتلونکی
{ ts: 1527812142, e: "true, truthful" }, // رښتنی { ts: 1527812142, e: "true, truthful" }, // رښتنی
{ ts: 1527812161, e: "true, truthful" }, // رښتونی { ts: 1527812161, e: "true, truthful" }, // رښتونی

View File

@ -45,7 +45,6 @@ module.exports = [
{ ts: 1527818632, e: `twin` }, // غبرګونی - ghbargoney { ts: 1527818632, e: `twin` }, // غبرګونی - ghbargoney
{ ts: 1527812624, e: `thief` }, // غل - ghul { ts: 1527812624, e: `thief` }, // غل - ghul
{ ts: 1613561408232, e: `guilty, at fault` }, // قصوروار - qUsoorwáar { ts: 1613561408232, e: `guilty, at fault` }, // قصوروار - qUsoorwáar
{ ts: 1527812724, e: `of Kabul (a person)` }, // کابلی - kaabuley
{ ts: 1527814715, e: `user` }, // کاروونکی - kaarawoonkey { ts: 1527814715, e: `user` }, // کاروونکی - kaarawoonkey
{ ts: 1527816256, e: `great-grandson` }, // کړوسی - kaRwaséy { ts: 1527816256, e: `great-grandson` }, // کړوسی - kaRwaséy
{ ts: 1594906790729, e: `child` }, // ککی - kakéy { ts: 1594906790729, e: `child` }, // ککی - kakéy

File diff suppressed because one or more lines are too long