compromise-speech 0.1.0 → 0.2.0

package/builds/compromise-speech.min.js

@@ -1 +1 @@
-var e,t;e=this,t=function(){var e={dedup:e=>e.replace(/([^c])\1/g,"$1"),dropInitialLetters:e=>e.match(/^(kn|gn|pn|ae|wr)/)?e.substring(1,e.length-1):e,dropBafterMAtEnd:e=>e.replace(/mb$/,"m"),cchange:e=>(e=(e=(e=e.replace(/([^s]|^)(c)(h)/g,"$1x$3").trim()).replace(/cia/g,"xia")).replace(/c([iey])/g,"s$1")).replace(/c/g,"k"),dchange:e=>(e=e.replace(/d(ge|gy|gi)/g,"j$1")).replace(/d/g,"t"),dropG:e=>(e=e.replace(/gh(^$|[^aeiou])/g,"h$1")).replace(/g(n|ned)$/g,"$1"),changeG:e=>(e=(e=(e=e.replace(/gh/g,"f")).replace(/([^g]|^)(g)([iey])/g,"$1j$3")).replace(/gg/g,"g")).replace(/g/g,"k"),dropH:e=>e.replace(/([aeiou])h([^aeiou]|$)/g,"$1$2"),changeCK:e=>e.replace(/ck/g,"k"),changePH:e=>e.replace(/ph/g,"f"),changeQ:e=>e.replace(/q/g,"k"),changeS:e=>e.replace(/s(h|io|ia)/g,"x$1"),changeT:e=>e=e.replace(/t(ia[^n]|io)/g,"x$1"),dropT:e=>e.replace(/tch/g,"ch"),changeV:e=>e.replace(/v/g,"f"),changeWH:e=>e.replace(/^wh/,"w"),dropW:e=>e.replace(/w([^aeiou]|$)/g,"$1"),changeX:e=>(e=e.replace(/^x/,"s")).replace(/x/g,"ks"),dropY:e=>e.replace(/y([^aeiou]|$)/g,"$1"),changeZ:e=>e.replace(/z/,"s"),dropVowels:e=>e},t=function(t){return t=e.dedup(t),t=e.dropInitialLetters(t),t=e.dropBafterMAtEnd(t),t=e.changeCK(t),t=e.cchange(t),t=e.dchange(t),t=e.dropG(t),t=e.changeG(t),t=e.dropH(t),t=e.changePH(t),t=e.changeQ(t),t=e.changeS(t),t=e.changeX(t),t=e.changeT(t),t=e.dropT(t),t=e.changeV(t),t=e.changeWH(t),t=e.dropW(t),t=e.dropY(t),t=e.changeZ(t),(t=e.dropVowels(t)).trim()};const a=/^(eu)/i,c=/^[^aeiou]e([^d]|$)/,n=/^([^aeiouy])[aeiouy]\1[aeiouy]/,l=/^[^aeiouy]([aeiouy])[^aeiouy]\1/,h=/^([tg][aeiouy]){2}/,o=/^[^aeiouy]+$/,i=/[aeiouy]$/,r=/^[^aeiouy]h?[aeiouy]/,g=[/^[^aeiou]?ion/,/^[^aeiou]?ised/,/^[^aeiou]?iled/,/[aeiou]n[gt]$/,/\wa[gt]e$/];var p=function(e){let t=(e=(e=e.map((function(e){return e.trim()}))).filter((function(e){return""!==e}))).length;if(t>1){let a=e[t-2]+e[t-1];for(let c=0;c<g.length;c++)a.match(g[c])&&(e[t-2]=e[t-2]+e[t-1],e.pop())}if(e.length>1){let t=(1===e[0].length||e[0].match(r))&&e[0].match(i),a=e[1].match(c);if(t&&a){let t=e[0]+e[1];t.match(n)||t.match(l)||t.match(h)||(e[0]=e[0]+e[1],e.splice(1,1))}}if(e.length>1){let t=e[e.length-2].match(r)&&e[e.length-2].match(i),a=e[e.length-1].match(c)&&g.every(t=>!e[e.length-1].match(t));if(t&&a){let t=e[e.length-2]+e[e.length-1];t.match(n)||t.match(l)||t.match(h)||(e[e.length-2]=e[e.length-2]+e[e.length-1],e.splice(e.length-1,1))}}if(e.length>1){let t=e[0]+e[1];t.match(a)&&(e[0]=t,e.splice(1,1))}return e.length>1&&e[e.length-1].match(o)&&(e[e.length-2]=e[e.length-2]+e[e.length-1],e.splice(e.length-1,1)),e};const u=/ +/g,s=/[aeiouy]$/,f=/^[^aeiouy]h?[aeiouy]/,d=/^e[sm]/,m=/^e/,y=/(eo|eu|ia|oa|ua|ui)$/i,$=/[aiouy]/,k=/ee$/,b=function(e){let t=[],a=e.split(""),c="",n="",l="";for(let e=0;e<a.length;e++){c=a.slice(0,e).join(""),l=a[e],n=a.slice(e+1,a.length).join("");let h=c+a[e];if(c.match(s)&&!l.match(s))return n.match(d)&&(h+="e",n=n.replace(m,"")),t.push(h),t.concat(b(n));if(h.match(y))return t.push(c),t.push(l),t.concat(b(n));if(h.match(s)&&n.match(f))return t.push(h),t.concat(b(n))}if(e.match($)||e.match(k))t.push(e);else if(e){let a=t.length-1;a<0&&(a=0),t[a]=(t[a]||"")+e}return t};var x=function(e){let t=[];return e?((e=e.replace(/[.,?]/g,"")).split(u).map(e=>{t=t.concat(b(e))}),t=p(t),0===t.length&&(t=[e]),t=t.filter(e=>e),t):t};return{api:function(e){e.prototype.syllables=function(){this.compute("syllables");let e=[];return this.docs.forEach(t=>{let a=[];t.forEach(e=>{a=a.concat(e.syllables)}),a.length>0&&e.push(a)}),e},e.prototype.soundsLike=function(){this.compute("soundsLike");let e=[];return this.docs.forEach(t=>{let a=[];t.forEach(e=>{a=a.concat(e.soundsLike)}),a.length>0&&e.push(a)}),e}},compute:{soundsLike:function(e){e.docs.forEach(e=>{e.forEach(e=>{e.soundsLike=t(e.normal||e.text)})})},syllables:function(e){e.docs.forEach(e=>{e.forEach(e=>{e.syllables=x(e.normal||e.text)})})}}}},"object"==typeof exports&&"undefined"!=typeof module?module.exports=t():"function"==typeof define&&define.amd?define(t):(e="undefined"!=typeof globalThis?globalThis:e||self).compromiseSpeech=t();
+!function(e,t){"object"==typeof exports&&"undefined"!=typeof module?module.exports=t():"function"==typeof define&&define.amd?define(t):(e="undefined"!=typeof globalThis?globalThis:e||self).compromiseSpeech=t()}(this,function(){"use strict";const e={acre:[1],cafe:[2],abalone:[2,3,5],aborigine:[2,3,6,7],comanche:[2,5],dolce:[3],acacia:[1,3],fuchsia:[4],concierge:[3,5],concierges:[3,5],comedienne:[2,4,6],massacre:[3,5],massacred:[3,5],massacring:[3,5],aisle:[],isle:[],wednesday:[6],buoyant:[4],buoyancy:[4,6],acquiesce:[2,5],acquiesced:[2,5],acquiescent:[2,5,7],acquiescence:[2,5,7],people:[3],queue:[],business:[4],quiet:[3],science:[3],poet:[2],poetry:[2,4],every:[3],everything:[3,5],everybody:[3,5,7],ilyich:[2],juan:[],wyatt:[2],batesville:[5],belleville:[5],bridgeport:[6],cambridgeport:[3,9],cambridgeside:[3,9],janesville:[5],roseberry:[4,7],roseville:[4],belge:[3],canadienne:[2,4,6],mildred:[3],nietzsche:[5],odea:[1,3],zoologist:[2,4,5],zoology:[2,4,5]},t=t=>{if(Object.hasOwn(e,t))return e[t];const a=t.replace(/s$/,"");return Object.hasOwn(e,a)?/[csxz]e$/.test(a)?e[a].concat(a.length-2):e[a]:null},a=[/^note(?=book)/,/^side(?=kick|line|show|step|stream|track|walk|water|way|winder|wise)/,/^some(?=bod|one|time|where)/,/^none(?=such)/,/^goose(?=berr)/,/^above(?=board|ground)/,/^a?fore(?=clos|taste|said|mentioned)/,/^ware(?=house|housing)/,/^space(?=port|ship)/,/^race(?=horse)/,/^lake(?=side)/,/^white(?=side)/,/^state(?=side)/,/^juke(?=box)/,/^police(?=woman|women|man|men)/,/^(?:sale|spoke)s?(?=woman|women|man|men)/],s=/[aeiouy]/,o=[{match:/[^aeiou]e(?=['’](?:s|d|ll|re|ve)$|$)/g,offset:1,except:/[^aeiouyl]le(?:['’](?:s|d|ll|re|ve))?$/},{match:/[^td]ed(?=['’](?:s|d|ll|re|ve)$|$)/g,offset:1,except:/(?:[^aeiouyl]led|^(?:hundred|kindred|sacred|wicked|rugged|ragged|wretched|naked|crooked|(?:a|em|death|river|water|flower|hot|sick|sun|sea)bed|(?:in|inter|thorough|cross|pure)bred|infrared|newlywed|premed))(?:['’](?:s|d|ll|re|ve))?$/},{match:/es(?=['’](?:s|d|ll|re|ve)$|$)/g,offset:0,except:/([sxzcg]|ch|sh|[^aeiouyl]l)es(?:['’](?:s|d|ll|re|ve))?$/},{match:/e(?=ments?$|thing$|s?man$)/g,offset:0,except:/(?:[^aeiouyl]le(?:ments?|thing|s?man)|elements?)$/},{match:/[^aeiou]e(?=ness(?:es)?$|l(?:y|iness)$|ful(?:ly)?$|less$|ships?$|some$)/g,offset:1,except:/[^aeiouyl]le(?:ness(?:es)?|l(?:y|iness)|ful(?:ly)?|less|ships?|some)$/},{match:/['’][rv]e$/g,offset:2},{match:/qu(?=[aeiouy])/g,offset:1},{match:/que(?=s?$|ly$|ness$)/g,offset:2},{match:/^y(?=[aeiou])/g,offset:0},{match:/[aeiouwy]y(?=[aeiou])/g,offset:1},{match:/[aeiou]ny(?=[aeiou])/g,offset:2,except:/^any/},{match:/gu(?=a)/g,offset:1},{match:/busi(?=ness)/g,offset:3},{match:/ever(?=y(?:body|thing|where|one|day|time|$))/g,offset:2}],n=/^mc(?=[a-z])/,i=[/(?:[sxz]e?|[cs]h|[cg]e)['’](s)$/,/[ai]s(m)(?:s|['’]s)?$/,/[iy]th(m)(?:s|['’]s)?$/,/[^aeiouyn](n)['’]t$/,/^(?:could|should|would|might|must)['’](v)e$/],c=[/^(?:re)?cre(?=at[eio])/,/^re(?=act)/,/^re(?=absorb|app|elect|emerg|enter|establish|evalu|impos|invent|issue|us[ei])/,/^pre(?=adolesc|ambl|arrang|eminen|empt|exist|ignition|invas)/,/^(?:micro|macro)(?=[aeiou])/,/^co(?=operat|ordinat|alesc|erc)/,/^pro(?=act)/,/^flu(?=id)/,/^ru(?=in)/,/di(?=et|en[ct])/,/^ide(?=a)/,/^sovi(?=et)/,/^soci(?=et)/,/i(?=ell[aoi])/,/i(?=et(?:te)?s?$)/,/^qui(?=et)/,/^acre(?=age)/,/^sci(?=on)/,/^intra(?=[aeiou])/,/re(?=al(?!ly|m))/,/[^q]u(?=en[ctdz])/,/[^g]u(?=ing)/,/argu(?=ing)/,/^lou(?=is)/,/[eo]u(?=illes?$)/,/ou(?=ett)/,/o(?=as?$)/],r=[{before:1,after:1,match:/^(ai|au|ay|ea|ee|ei|eu|ey|ie|oa|oe|oi|oo|ou|oy|ue|ui)$/},{before:3,after:2,match:/^llion/},{before:2,after:3,match:/^[cst]i(?:on|an|al|ous)/},{before:2,after:4,match:/^cia(?:s|['’]s)?$/},{before:2,after:1,match:/^eau$/},{before:1,after:3,match:/^ae(?:r|st)/},{before:2,after:3,match:/^[cg]eous/},{before:3,after:3,match:/^hteous/}],l=new Set(["bl","br","ch","cl","cr","dr","fl","fr","gl","gr","ph","pl","pr","qu","sh","sk","sl","sm","sn","sp","st","sw","th","tr","tw","wh","scr","shr","spl","spr","str","thr"]),h=[{minGap:1,maxGap:1,gap:/^x$/,left:1},{minGap:2,gap:/^(.)\1/,minLeft:1},{minGap:2,tail:/^ed$/,right:-1},{minGap:2,gap:/^sh$/,tail:/^(es|ing)$/,right:0},{minGap:2,gap:/l$/,tail:/^e(?:[sd]|['’](?:s|d|ll|re|ve))?$/,right:-2},{minGap:2,gap:/l$/,tail:/^e(?:ments?|ships?|ness(?:es)?|less|ful(?:ly)?|ly)(?:['’]s)?$/,right:-2},{minGap:1,gap:/l$/,tail:/^ess(?:ly|ness(?:es)?)?$/,right:-1},{minGap:3,gap:/^ssl$/,tail:/^y$/,right:-1},{minGap:1,gap:/^(?:s|th)$/,tail:/^m(?:s|['’]s)?$/,right:0},{minGap:1,tail:/^n['’]t$/,right:0},{minGap:1,gap:/^[-‐‑–—]+$/,right:0}],u=e=>{const u=t(e);if(null!==u)return u;const g=((e,t)=>{const s=[],o=a.flatMap(t=>{const a=t.exec(e);return a?[a[0].length]:[]});for(let a=1;a<t.length;a+=1){const n=t[a-1].end,i=t[a].start,c=e.slice(n,i),r=e.slice(i);let u=n;if(c.length>1){u=i-1;for(let e=Math.min(3,c.length);e>1;e-=1)if(l.has(c.slice(-e))){u=i-e;break}}h.forEach(e=>{c.length<e.minGap||void 0!==e.maxGap&&c.length>e.maxGap||e.gap&&!e.gap.test(c)||e.tail&&!e.tail.test(r)||(u=void 0!==e.left?n+e.left:void 0!==e.right?i+e.right:Math.max(u,n+e.minLeft))}),o.forEach(e=>{e>=n&&e<=i&&(u=e)}),s.push(u)}return s})(e,(e=>{const t=new Set;a.forEach(a=>{const s=a.exec(e);s&&t.add(s[0].lastIndexOf("e"))}),o.forEach(a=>{if(!a.except||!a.except.test(e))for(const s of e.matchAll(a.match))t.add(s.index+a.offset)});const l=new Set;c.forEach(t=>{const a=t.exec(e);a&&l.add(a.index+a[0].length)});const h=[],u=n.exec(e);u&&h.push({start:0,end:u[0].length});const g=new Set;i.forEach(t=>{const a=t.exec(e);a&&g.add(a.index+a[0].lastIndexOf(a[1]))});for(let a=0;a<e.length;a+=1){if(!s.test(e[a])&&!g.has(a)||t.has(a))continue;const o=h[h.length-1];o&&o.end===a&&!l.has(a)&&r.some(t=>t.match.test(e.slice(Math.max(0,a-t.before),a+t.after)))?o.end=a+1:h.push({start:a,end:a+1})}return h})(e)),p=e.replace(/['’](?:s|d|ll|re|ve)$/,"");if(p===e)return g;const f=t(p);return null!==f?f.concat(g.filter(e=>e>=p.length)):g},g=/(?<!\p{Punctuation})\p{Punctuation}+$/u,p=/^[\p{Punctuation}$]+/u,f=(e,t={})=>{if(!e.text)return[];let a=e.text;t.normalize&&(a=e.normal||e.text);const{clean:s,prefix:o}=(e=>{let t=e.toLowerCase();t=t.replace(g,"");const a=t.match(p)?.[0]||"";return{clean:t.slice(a.length),prefix:a}})(a);let n=u(s);return o&&o.length>0&&(n=((e,t="")=>e.map(e=>e+t.length))(n,o)),((e,t)=>{let a=0;const s=t.map(t=>{const s=e.slice(a,t);return a=t,s});return s.push(e.slice(a)),s})(a,n)},d=e=>e.replace(/([^c])\1/g,"$1"),m=e=>/^(kn|gn|pn|ae|wr)/.test(e)?e.substring(1,e.length-1):e,$=e=>e.replace(/mb$/,"m"),y=e=>(e=(e=(e=e.replace(/([^s]|^)(c)(h)/g,"$1x$3").trim()).replace(/cia/g,"xia")).replace(/c([iey])/g,"s$1")).replace(/c/g,"k"),b=e=>(e=e.replace(/d(ge|gy|gi)/g,"j$1")).replace(/d/g,"t"),x=e=>(e=e.replace(/gh(^$|[^aeiou])/g,"h$1")).replace(/g(n|ned)$/g,"$1"),w=e=>(e=(e=(e=e.replace(/gh/g,"f")).replace(/([^g]|^)(g)([iey])/g,"$1j$3")).replace(/gg/g,"g")).replace(/g/g,"k"),v=e=>e.replace(/([aeiou])h([^aeiou]|$)/g,"$1$2"),k=e=>e.replace(/ck/g,"k"),G=e=>e.replace(/ph/g,"f"),E=e=>e.replace(/q/g,"k"),q=e=>e.replace(/s(h|io|ia)/g,"x$1"),z=e=>e=e.replace(/t(ia[^n]|io)/g,"x$1"),j=e=>e.replace(/tch/g,"ch"),L=e=>e.replace(/v/g,"f"),O=e=>e.replace(/^wh/,"w"),S=e=>e.replace(/w([^aeiou]|$)/g,"$1"),M=e=>(e=e.replace(/^x/,"s")).replace(/x/g,"ks"),P=e=>e.replace(/y([^aeiou]|$)/g,"$1"),I=e=>e.replace(/z/,"s"),T=e=>e;return{version:"0.2.0",api:function(e){e.prototype.syllables=function(e={}){this.compute("syllables",e);const t=[];return this.docs.forEach(e=>{let a=[];e.forEach(e=>{a=a.concat(e.syllables)}),a.length>0&&t.push(a)}),t},e.prototype.soundsLike=function(){this.compute("soundsLike");const e=[];return this.docs.forEach(t=>{let a=[];t.forEach(e=>{a=a.concat(e.soundsLike)}),a.length>0&&e.push(a)}),e}},compute:{syllables:function(e,t={}){e.docs.forEach(e=>{e.forEach(e=>{e.syllables=f(e,t)})})},soundsLike:function(e){e.docs.forEach(e=>{e.forEach(e=>{var t;e.soundsLike=(t=e.normal||e.text,t=d(t),t=m(t),t=$(t),t=k(t),t=y(t),t=b(t),t=x(t),t=w(t),t=v(t),t=G(t),t=E(t),t=q(t),t=M(t),t=z(t),t=j(t),t=L(t),t=O(t),t=S(t),t=P(t),t=I(t),(t=T(t)).trim())})})}}}});

package/builds/compromise-speech.mjs

@@ -1 +1 @@
-var e={dedup:e=>e.replace(/([^c])\1/g,"$1"),dropInitialLetters:e=>e.match(/^(kn|gn|pn|ae|wr)/)?e.substring(1,e.length-1):e,dropBafterMAtEnd:e=>e.replace(/mb$/,"m"),cchange:e=>(e=(e=(e=e.replace(/([^s]|^)(c)(h)/g,"$1x$3").trim()).replace(/cia/g,"xia")).replace(/c([iey])/g,"s$1")).replace(/c/g,"k"),dchange:e=>(e=e.replace(/d(ge|gy|gi)/g,"j$1")).replace(/d/g,"t"),dropG:e=>(e=e.replace(/gh(^$|[^aeiou])/g,"h$1")).replace(/g(n|ned)$/g,"$1"),changeG:e=>(e=(e=(e=e.replace(/gh/g,"f")).replace(/([^g]|^)(g)([iey])/g,"$1j$3")).replace(/gg/g,"g")).replace(/g/g,"k"),dropH:e=>e.replace(/([aeiou])h([^aeiou]|$)/g,"$1$2"),changeCK:e=>e.replace(/ck/g,"k"),changePH:e=>e.replace(/ph/g,"f"),changeQ:e=>e.replace(/q/g,"k"),changeS:e=>e.replace(/s(h|io|ia)/g,"x$1"),changeT:e=>e=e.replace(/t(ia[^n]|io)/g,"x$1"),dropT:e=>e.replace(/tch/g,"ch"),changeV:e=>e.replace(/v/g,"f"),changeWH:e=>e.replace(/^wh/,"w"),dropW:e=>e.replace(/w([^aeiou]|$)/g,"$1"),changeX:e=>(e=e.replace(/^x/,"s")).replace(/x/g,"ks"),dropY:e=>e.replace(/y([^aeiou]|$)/g,"$1"),changeZ:e=>e.replace(/z/,"s"),dropVowels:e=>e};var a=function(a){return a=e.dedup(a),a=e.dropInitialLetters(a),a=e.dropBafterMAtEnd(a),a=e.changeCK(a),a=e.cchange(a),a=e.dchange(a),a=e.dropG(a),a=e.changeG(a),a=e.dropH(a),a=e.changePH(a),a=e.changeQ(a),a=e.changeS(a),a=e.changeX(a),a=e.changeT(a),a=e.dropT(a),a=e.changeV(a),a=e.changeWH(a),a=e.dropW(a),a=e.dropY(a),a=e.changeZ(a),(a=e.dropVowels(a)).trim()};const t=/^(eu)/i,c=/^[^aeiou]e([^d]|$)/,n=/^([^aeiouy])[aeiouy]\1[aeiouy]/,l=/^[^aeiouy]([aeiouy])[^aeiouy]\1/,h=/^([tg][aeiouy]){2}/,r=/^[^aeiouy]+$/,o=/[aeiouy]$/,g=/^[^aeiouy]h?[aeiouy]/,i=[/^[^aeiou]?ion/,/^[^aeiou]?ised/,/^[^aeiou]?iled/,/[aeiou]n[gt]$/,/\wa[gt]e$/];var p=function(e){let a=(e=(e=e.map((function(e){return e.trim()}))).filter((function(e){return""!==e}))).length;if(a>1){let t=e[a-2]+e[a-1];for(let c=0;c<i.length;c++)t.match(i[c])&&(e[a-2]=e[a-2]+e[a-1],e.pop())}if(e.length>1){let a=(1===e[0].length||e[0].match(g))&&e[0].match(o),t=e[1].match(c);if(a&&t){let a=e[0]+e[1];a.match(n)||a.match(l)||a.match(h)||(e[0]=e[0]+e[1],e.splice(1,1))}}if(e.length>1){let a=e[e.length-2].match(g)&&e[e.length-2].match(o),t=e[e.length-1].match(c)&&i.every(a=>!e[e.length-1].match(a));if(a&&t){let a=e[e.length-2]+e[e.length-1];a.match(n)||a.match(l)||a.match(h)||(e[e.length-2]=e[e.length-2]+e[e.length-1],e.splice(e.length-1,1))}}if(e.length>1){let a=e[0]+e[1];a.match(t)&&(e[0]=a,e.splice(1,1))}return e.length>1&&e[e.length-1].match(r)&&(e[e.length-2]=e[e.length-2]+e[e.length-1],e.splice(e.length-1,1)),e};const u=/ +/g,s=/[aeiouy]$/,d=/^[^aeiouy]h?[aeiouy]/,f=/^e[sm]/,m=/^e/,$=/(eo|eu|ia|oa|ua|ui)$/i,y=/[aiouy]/,k=/ee$/,E=function(e){let a=[],t=e.split(""),c="",n="",l="";for(let e=0;e<t.length;e++){c=t.slice(0,e).join(""),l=t[e],n=t.slice(e+1,t.length).join("");let h=c+t[e];if(c.match(s)&&!l.match(s))return n.match(f)&&(h+="e",n=n.replace(m,"")),a.push(h),a.concat(E(n));if(h.match($))return a.push(c),a.push(l),a.concat(E(n));if(h.match(s)&&n.match(d))return a.push(h),a.concat(E(n))}if(e.match(y)||e.match(k))a.push(e);else if(e){let t=a.length-1;t<0&&(t=0),a[t]=(a[t]||"")+e}return a};var x=function(e){let a=[];return e?((e=e.replace(/[.,?]/g,"")).split(u).map(e=>{a=a.concat(E(e))}),a=p(a),0===a.length&&(a=[e]),a=a.filter(e=>e),a):a};var b={api:function(e){e.prototype.syllables=function(){this.compute("syllables");let e=[];return this.docs.forEach(a=>{let t=[];a.forEach(e=>{t=t.concat(e.syllables)}),t.length>0&&e.push(t)}),e},e.prototype.soundsLike=function(){this.compute("soundsLike");let e=[];return this.docs.forEach(a=>{let t=[];a.forEach(e=>{t=t.concat(e.soundsLike)}),t.length>0&&e.push(t)}),e}},compute:{soundsLike:function(e){e.docs.forEach(e=>{e.forEach(e=>{e.soundsLike=a(e.normal||e.text)})})},syllables:function(e){e.docs.forEach(e=>{e.forEach(e=>{e.syllables=x(e.normal||e.text)})})}}};export{b as default};
+const e={acre:[1],cafe:[2],abalone:[2,3,5],aborigine:[2,3,6,7],comanche:[2,5],dolce:[3],acacia:[1,3],fuchsia:[4],concierge:[3,5],concierges:[3,5],comedienne:[2,4,6],massacre:[3,5],massacred:[3,5],massacring:[3,5],aisle:[],isle:[],wednesday:[6],buoyant:[4],buoyancy:[4,6],acquiesce:[2,5],acquiesced:[2,5],acquiescent:[2,5,7],acquiescence:[2,5,7],people:[3],queue:[],business:[4],quiet:[3],science:[3],poet:[2],poetry:[2,4],every:[3],everything:[3,5],everybody:[3,5,7],ilyich:[2],juan:[],wyatt:[2],batesville:[5],belleville:[5],bridgeport:[6],cambridgeport:[3,9],cambridgeside:[3,9],janesville:[5],roseberry:[4,7],roseville:[4],belge:[3],canadienne:[2,4,6],mildred:[3],nietzsche:[5],odea:[1,3],zoologist:[2,4,5],zoology:[2,4,5]},t=t=>{if(Object.hasOwn(e,t))return e[t];const a=t.replace(/s$/,"");return Object.hasOwn(e,a)?/[csxz]e$/.test(a)?e[a].concat(a.length-2):e[a]:null},a=[/^note(?=book)/,/^side(?=kick|line|show|step|stream|track|walk|water|way|winder|wise)/,/^some(?=bod|one|time|where)/,/^none(?=such)/,/^goose(?=berr)/,/^above(?=board|ground)/,/^a?fore(?=clos|taste|said|mentioned)/,/^ware(?=house|housing)/,/^space(?=port|ship)/,/^race(?=horse)/,/^lake(?=side)/,/^white(?=side)/,/^state(?=side)/,/^juke(?=box)/,/^police(?=woman|women|man|men)/,/^(?:sale|spoke)s?(?=woman|women|man|men)/],s=/[aeiouy]/,r=[{match:/[^aeiou]e(?=['’](?:s|d|ll|re|ve)$|$)/g,offset:1,except:/[^aeiouyl]le(?:['’](?:s|d|ll|re|ve))?$/},{match:/[^td]ed(?=['’](?:s|d|ll|re|ve)$|$)/g,offset:1,except:/(?:[^aeiouyl]led|^(?:hundred|kindred|sacred|wicked|rugged|ragged|wretched|naked|crooked|(?:a|em|death|river|water|flower|hot|sick|sun|sea)bed|(?:in|inter|thorough|cross|pure)bred|infrared|newlywed|premed))(?:['’](?:s|d|ll|re|ve))?$/},{match:/es(?=['’](?:s|d|ll|re|ve)$|$)/g,offset:0,except:/([sxzcg]|ch|sh|[^aeiouyl]l)es(?:['’](?:s|d|ll|re|ve))?$/},{match:/e(?=ments?$|thing$|s?man$)/g,offset:0,except:/(?:[^aeiouyl]le(?:ments?|thing|s?man)|elements?)$/},{match:/[^aeiou]e(?=ness(?:es)?$|l(?:y|iness)$|ful(?:ly)?$|less$|ships?$|some$)/g,offset:1,except:/[^aeiouyl]le(?:ness(?:es)?|l(?:y|iness)|ful(?:ly)?|less|ships?|some)$/},{match:/['’][rv]e$/g,offset:2},{match:/qu(?=[aeiouy])/g,offset:1},{match:/que(?=s?$|ly$|ness$)/g,offset:2},{match:/^y(?=[aeiou])/g,offset:0},{match:/[aeiouwy]y(?=[aeiou])/g,offset:1},{match:/[aeiou]ny(?=[aeiou])/g,offset:2,except:/^any/},{match:/gu(?=a)/g,offset:1},{match:/busi(?=ness)/g,offset:3},{match:/ever(?=y(?:body|thing|where|one|day|time|$))/g,offset:2}],o=/^mc(?=[a-z])/,c=[/(?:[sxz]e?|[cs]h|[cg]e)['’](s)$/,/[ai]s(m)(?:s|['’]s)?$/,/[iy]th(m)(?:s|['’]s)?$/,/[^aeiouyn](n)['’]t$/,/^(?:could|should|would|might|must)['’](v)e$/],n=[/^(?:re)?cre(?=at[eio])/,/^re(?=act)/,/^re(?=absorb|app|elect|emerg|enter|establish|evalu|impos|invent|issue|us[ei])/,/^pre(?=adolesc|ambl|arrang|eminen|empt|exist|ignition|invas)/,/^(?:micro|macro)(?=[aeiou])/,/^co(?=operat|ordinat|alesc|erc)/,/^pro(?=act)/,/^flu(?=id)/,/^ru(?=in)/,/di(?=et|en[ct])/,/^ide(?=a)/,/^sovi(?=et)/,/^soci(?=et)/,/i(?=ell[aoi])/,/i(?=et(?:te)?s?$)/,/^qui(?=et)/,/^acre(?=age)/,/^sci(?=on)/,/^intra(?=[aeiou])/,/re(?=al(?!ly|m))/,/[^q]u(?=en[ctdz])/,/[^g]u(?=ing)/,/argu(?=ing)/,/^lou(?=is)/,/[eo]u(?=illes?$)/,/ou(?=ett)/,/o(?=as?$)/],i=[{before:1,after:1,match:/^(ai|au|ay|ea|ee|ei|eu|ey|ie|oa|oe|oi|oo|ou|oy|ue|ui)$/},{before:3,after:2,match:/^llion/},{before:2,after:3,match:/^[cst]i(?:on|an|al|ous)/},{before:2,after:4,match:/^cia(?:s|['’]s)?$/},{before:2,after:1,match:/^eau$/},{before:1,after:3,match:/^ae(?:r|st)/},{before:2,after:3,match:/^[cg]eous/},{before:3,after:3,match:/^hteous/}],l=new Set(["bl","br","ch","cl","cr","dr","fl","fr","gl","gr","ph","pl","pr","qu","sh","sk","sl","sm","sn","sp","st","sw","th","tr","tw","wh","scr","shr","spl","spr","str","thr"]),h=[{minGap:1,maxGap:1,gap:/^x$/,left:1},{minGap:2,gap:/^(.)\1/,minLeft:1},{minGap:2,tail:/^ed$/,right:-1},{minGap:2,gap:/^sh$/,tail:/^(es|ing)$/,right:0},{minGap:2,gap:/l$/,tail:/^e(?:[sd]|['’](?:s|d|ll|re|ve))?$/,right:-2},{minGap:2,gap:/l$/,tail:/^e(?:ments?|ships?|ness(?:es)?|less|ful(?:ly)?|ly)(?:['’]s)?$/,right:-2},{minGap:1,gap:/l$/,tail:/^ess(?:ly|ness(?:es)?)?$/,right:-1},{minGap:3,gap:/^ssl$/,tail:/^y$/,right:-1},{minGap:1,gap:/^(?:s|th)$/,tail:/^m(?:s|['’]s)?$/,right:0},{minGap:1,tail:/^n['’]t$/,right:0},{minGap:1,gap:/^[-‐‑–—]+$/,right:0}],g=e=>{const g=t(e);if(null!==g)return g;const u=((e,t)=>{const s=[],r=a.flatMap(t=>{const a=t.exec(e);return a?[a[0].length]:[]});for(let a=1;a<t.length;a+=1){const o=t[a-1].end,c=t[a].start,n=e.slice(o,c),i=e.slice(c);let g=o;if(n.length>1){g=c-1;for(let e=Math.min(3,n.length);e>1;e-=1)if(l.has(n.slice(-e))){g=c-e;break}}h.forEach(e=>{n.length<e.minGap||void 0!==e.maxGap&&n.length>e.maxGap||e.gap&&!e.gap.test(n)||e.tail&&!e.tail.test(i)||(g=void 0!==e.left?o+e.left:void 0!==e.right?c+e.right:Math.max(g,o+e.minLeft))}),r.forEach(e=>{e>=o&&e<=c&&(g=e)}),s.push(g)}return s})(e,(e=>{const t=new Set;a.forEach(a=>{const s=a.exec(e);s&&t.add(s[0].lastIndexOf("e"))}),r.forEach(a=>{if(!a.except||!a.except.test(e))for(const s of e.matchAll(a.match))t.add(s.index+a.offset)});const l=new Set;n.forEach(t=>{const a=t.exec(e);a&&l.add(a.index+a[0].length)});const h=[],g=o.exec(e);g&&h.push({start:0,end:g[0].length});const u=new Set;c.forEach(t=>{const a=t.exec(e);a&&u.add(a.index+a[0].lastIndexOf(a[1]))});for(let a=0;a<e.length;a+=1){if(!s.test(e[a])&&!u.has(a)||t.has(a))continue;const r=h[h.length-1];r&&r.end===a&&!l.has(a)&&i.some(t=>t.match.test(e.slice(Math.max(0,a-t.before),a+t.after)))?r.end=a+1:h.push({start:a,end:a+1})}return h})(e)),p=e.replace(/['’](?:s|d|ll|re|ve)$/,"");if(p===e)return u;const f=t(p);return null!==f?f.concat(u.filter(e=>e>=p.length)):u},u=/(?<!\p{Punctuation})\p{Punctuation}+$/u,p=/^[\p{Punctuation}$]+/u,f=(e,t={})=>{if(!e.text)return[];let a=e.text;t.normalize&&(a=e.normal||e.text);const{clean:s,prefix:r}=(e=>{let t=e.toLowerCase();t=t.replace(u,"");const a=t.match(p)?.[0]||"";return{clean:t.slice(a.length),prefix:a}})(a);let o=g(s);return r&&r.length>0&&(o=((e,t="")=>e.map(e=>e+t.length))(o,r)),((e,t)=>{let a=0;const s=t.map(t=>{const s=e.slice(a,t);return a=t,s});return s.push(e.slice(a)),s})(a,o)},m=e=>e.replace(/([^c])\1/g,"$1"),d=e=>/^(kn|gn|pn|ae|wr)/.test(e)?e.substring(1,e.length-1):e,$=e=>e.replace(/mb$/,"m"),y=e=>(e=(e=(e=e.replace(/([^s]|^)(c)(h)/g,"$1x$3").trim()).replace(/cia/g,"xia")).replace(/c([iey])/g,"s$1")).replace(/c/g,"k"),b=e=>(e=e.replace(/d(ge|gy|gi)/g,"j$1")).replace(/d/g,"t"),x=e=>(e=e.replace(/gh(^$|[^aeiou])/g,"h$1")).replace(/g(n|ned)$/g,"$1"),w=e=>(e=(e=(e=e.replace(/gh/g,"f")).replace(/([^g]|^)(g)([iey])/g,"$1j$3")).replace(/gg/g,"g")).replace(/g/g,"k"),v=e=>e.replace(/([aeiou])h([^aeiou]|$)/g,"$1$2"),k=e=>e.replace(/ck/g,"k"),G=e=>e.replace(/ph/g,"f"),E=e=>e.replace(/q/g,"k"),q=e=>e.replace(/s(h|io|ia)/g,"x$1"),z=e=>e=e.replace(/t(ia[^n]|io)/g,"x$1"),L=e=>e.replace(/tch/g,"ch"),j=e=>e.replace(/v/g,"f"),O=e=>e.replace(/^wh/,"w"),M=e=>e.replace(/w([^aeiou]|$)/g,"$1"),S=e=>(e=e.replace(/^x/,"s")).replace(/x/g,"ks"),P=e=>e.replace(/y([^aeiou]|$)/g,"$1"),I=e=>e.replace(/z/,"s"),A=e=>e;var C={version:"0.2.0",api:function(e){e.prototype.syllables=function(e={}){this.compute("syllables",e);const t=[];return this.docs.forEach(e=>{let a=[];e.forEach(e=>{a=a.concat(e.syllables)}),a.length>0&&t.push(a)}),t},e.prototype.soundsLike=function(){this.compute("soundsLike");const e=[];return this.docs.forEach(t=>{let a=[];t.forEach(e=>{a=a.concat(e.soundsLike)}),a.length>0&&e.push(a)}),e}},compute:{syllables:function(e,t={}){e.docs.forEach(e=>{e.forEach(e=>{e.syllables=f(e,t)})})},soundsLike:function(e){e.docs.forEach(e=>{e.forEach(e=>{var t;e.soundsLike=(t=e.normal||e.text,t=m(t),t=d(t),t=$(t),t=k(t),t=y(t),t=b(t),t=x(t),t=w(t),t=v(t),t=G(t),t=E(t),t=q(t),t=S(t),t=z(t),t=L(t),t=j(t),t=O(t),t=M(t),t=P(t),t=I(t),(t=A(t)).trim())})})}}};export{C as default};

package/package.json

@@ -1,38 +1,59 @@
 {
   "name": "compromise-speech",
-  "description": "plugin for nlp-compromise",
-  "version": "0.1.0",
-  "author": "Spencer Kelly <spencermountain@gmail.com> (http://spencermounta.in)",
-  "main": "./src/plugin.js",
-  "unpkg": "./builds/compromise-speech.min.js",
-  "module": "./builds/compromise-speech.mjs",
+  "version": "0.2.0",
+  "description": "Syllable counting and pronunciation estimates for compromise",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/spencermountain/compromise.git",
+    "directory": "plugins/speech"
+  },
+  "homepage": "https://github.com/spencermountain/compromise/tree/master/plugins/speech",
+  "author": "Spencer Kelly <spencermountain@gmail.com> (https://spencermountain.dev)",
   "type": "module",
-  "types": "index.d.ts",
-  "sideEffects": false,
   "exports": {
     ".": {
-      "import": "./src/plugin.js",
-      "require": "./builds/compromise-speech.cjs"
+      "import": {
+        "types": "./types/index.d.ts",
+        "default": "./src/plugin.js"
+      },
+      "require": {
+        "types": "./types/index.d.cts",
+        "default": "./builds/compromise-speech.cjs"
+      }
     }
   },
-  "repository": {
-    "type": "git",
-    "url": "git://github.com/spencermountain/compromise.git"
-  },
-  "homepage": "https://github.com/spencermountain/compromise/tree/master/plugins/speech",
-  "scripts": {
-    "test": "tape \"./tests/**/*.test.js\" | tap-dancer  --color always",
-    "testb": "TESTENV=prod tape \"./tests/**/*.test.js\" | tap-dancer  --color always",
-    "watch": "amble ./scratch.js",
-    "build": "rollup -c --silent"
+  "main": "./builds/compromise-speech.cjs",
+  "module": "./builds/compromise-speech.mjs",
+  "types": "./types/index.d.ts",
+  "sideEffects": false,
+  "engines": {
+    "node": ">=18"
   },
-  "eslintIgnore": [
-    "builds/*.js"
+  "files": [
+    "types/",
+    "builds/",
+    "src/"
   ],
+  "devDependencies": {
+    "@rollup/plugin-node-resolve": "16.0.3",
+    "@rollup/plugin-terser": "1.0.0",
+    "rollup": "4.64.8",
+    "rollup-plugin-filesize-check": "2.0.1",
+    "tape": "5.10.3",
+    "cross-env": "10.1.0",
+    "tap-dancer": "0.4.4"
+  },
   "peerDependencies": {
-    "compromise": ">=14.0.0"
+    "compromise": "^14.19.0"
   },
-  "dependencies": {},
-  "devDependencies": {},
-  "license": "MIT"
+  "unpkg": "./builds/compromise-speech.min.js",
+  "scripts": {
+    "test": "tape \"./tests/**/*.test.js\" | tap-dancer --color always",
+    "testb": "cross-env TESTENV=prod tape \"./tests/**/*.test.js\" | tap-dancer --color always",
+    "test:hmm": "tape \"./tests/**/*.hmm.js\" | tap-dancer --color always",
+    "watch": "node --watch ./scratch.js",
+    "version": "node ./scripts/version.js",
+    "build": "node ./scripts/version.js && rollup -c --silent"
+  }
 }

package/src/_version.js

@@ -0,0 +1 @@
+export default '0.2.0'

package/src/api.js

@@ -1,8 +1,8 @@
 const api = function (View) {
   /** */
-  View.prototype.syllables = function () {
-    this.compute('syllables')
-    let all = []
+  View.prototype.syllables = function (options = {}) {
+    this.compute('syllables', options)
+    const all = []
     this.docs.forEach(terms => {
       let some = []
       terms.forEach(term => {
@@ -17,7 +17,7 @@ const api = function (View) {
   /** */
   View.prototype.soundsLike = function () {
     this.compute('soundsLike')
-    let all = []
+    const all = []
     this.docs.forEach(terms => {
       let some = []
       terms.forEach(term => {
@@ -30,4 +30,4 @@ const api = function (View) {
     return all
   }
 }
-export default api
+export default api

package/src/plugin.js

@@ -1,7 +1,13 @@
-import compute from './compute/index.js'
+import version from './_version.js'
 import api from './api.js'
+import syllables from './syllables/index.js'
+import soundsLike from './sounds-like/index.js'
 
 export default {
+  version,
   api,
-  compute
-}
+  compute: {
+    syllables,
+    soundsLike,
+  },
+}

package/src/{compute/soundsLike → sounds-like}/transformations.js

@@ -4,7 +4,7 @@ const transformations = {
     return s.replace(/([^c])\1/g, '$1')
   },
   dropInitialLetters: (s) => {
-    if (s.match(/^(kn|gn|pn|ae|wr)/)) {
+    if (/^(kn|gn|pn|ae|wr)/.test(s)) {
       return s.substring(1, s.length - 1)
     }
     return s

package/src/syllables/index.js

@@ -0,0 +1,12 @@
+import splitTerm from './split/index.js'
+
+// Store syllables on each term; the public API collects them by sentence.
+const syllables = function (view, options = {}) {
+  view.docs.forEach(terms => {
+    terms.forEach(term => {
+      term.syllables = splitTerm(term, options)
+    })
+  })
+}
+
+export default syllables

package/src/syllables/split/_lib.js

@@ -0,0 +1,32 @@
+const trailingPunct = /(?<!\p{Punctuation})\p{Punctuation}+$/u
+const leadingPunct = /^[\p{Punctuation}$]+/u
+
+// Prepare spelling for rule matching; cuts are still applied to the original text.
+const normalize = input => {
+  let txt = input.toLowerCase()
+  // Ignore trailing punctuation runs for analysis; internal apostrophes stay intact.
+  txt = txt.replace(trailingPunct, '')
+  // Keep the prefix length so cuts still align with the original spelling.
+  const prefix = txt.match(leadingPunct)?.[0] || ''
+  const clean = txt.slice(prefix.length)
+  return { clean, prefix }
+}
+
+const pushCutsBack = (cuts, prefix = '') => {
+  return cuts.map(cut => cut + prefix.length)
+}
+
+// split a string into chunks based on the given cuts
+const applyCuts = (text, cuts) => {
+  let start = 0
+  const parts = cuts.map(end => {
+    const part = text.slice(start, end)
+    start = end
+    return part
+  })
+  // Cuts are interior boundaries; the final chunk always runs to the text's end.
+  parts.push(text.slice(start))
+  return parts
+}
+
+export { applyCuts, normalize, pushCutsBack }

package/src/syllables/split/find-cuts/00-exceptions.js

@@ -0,0 +1,81 @@
+// Whole-word splits that override the vowel-span and boundary rules.
+// Keys are lowercase; values are interior cut offsets, excluding the word's end.
+const exceptions = {
+  acre: [1], // a-cre: the final re carries a syllable
+  // Borrowed words whose final e is pronounced.
+  cafe: [2], // ca-fe
+  abalone: [2, 3, 5], // ab-a-lo-ne
+  aborigine: [2, 3, 6, 7], // ab-o-rig-i-ne
+  comanche: [2, 5], // co-man-che
+  dolce: [3], // dol-ce
+  // Borrowed vowel combinations do not follow the usual English pair rules.
+  acacia: [1, 3], // a-ca-cia
+  fuchsia: [4], // fuch-sia
+  concierge: [3, 5], // con-ci-erge
+  concierges: [3, 5], // con-ci-erges
+  comedienne: [2, 4, 6], // co-me-di-enne
+  // The final re in massacre carries a syllable, including its inflected forms.
+  massacre: [3, 5], // mas-sa-cre
+  massacred: [3, 5], // mas-sa-cred
+  massacring: [3, 5], // mas-sa-cring
+  // Silent consonants make these look like syllabic -le endings.
+  aisle: [],
+  isle: [],
+  wednesday: [6], // wednes-day
+  // Keep the spoken vowel groups together or apart explicitly.
+  buoyant: [4], // buoy-ant
+  buoyancy: [4, 6], // buoy-an-cy
+  acquiesce: [2, 5], // ac-qui-esce
+  acquiesced: [2, 5], // ac-qui-esced
+  acquiescent: [2, 5, 7], // ac-qui-es-cent
+  acquiescence: [2, 5, 7], // ac-qui-es-cence
+  people: [3], //'peo ple',
+  queue: [], //'queue',
+  business: [4], //'busi ness',
+  quiet: [3], //'qui et',
+  science: [3], //'sci ence',
+  poet: [2], //'po et',
+  poetry: [2, 4], //'po et ry',
+  every: [3], //'eve ry',
+  everything: [3, 5], // 'eve ry thing',
+  everybody: [3, 5, 7], //'eve ry bo dy',
+  ilyich: [2], // 'il yich': consonantal y in this name
+  juan: [], // 'juan':
+  wyatt: [2], // 'wy att'
+  // Place names retain their recognizable compound boundaries.
+  batesville: [5], // bates-ville
+  belleville: [5], // belle-ville
+  bridgeport: [6], // bridge-port
+  cambridgeport: [3, 9], // cam-bridge-port
+  cambridgeside: [3, 9], // cam-bridge-side
+  janesville: [5], // janes-ville
+  roseberry: [4, 7], // rose-ber-ry
+  roseville: [4], // rose-ville
+  // These names retain vowels that the general spelling rules suppress or blend.
+  belge: [3], // bel-ge
+  canadienne: [2, 4, 6], // ca-na-di-enne
+  mildred: [3], // mil-dred
+  nietzsche: [5], // nietz-sche
+  odea: [1, 3], // o-de-a
+  // The two written o's belong to separate syllables in these words.
+  zoologist: [2, 4, 5], // zo-ol-o-gist
+  zoology: [2, 4, 5], // zo-ol-o-gy
+}
+
+const checkExceptions = str => {
+  if (Object.hasOwn(exceptions, str)) {
+    return exceptions[str]
+  }
+  // Reuse singular cuts for a plain -s ending; exact entries above take priority.
+  const singular = str.replace(/s$/, '')
+  if (Object.hasOwn(exceptions, singular)) {
+    // A sibilant before silent e adds a syllable: sci-ence becomes sci-en-ces.
+    if (/[csxz]e$/.test(singular)) {
+      return exceptions[singular].concat(singular.length - 2)
+    }
+    return exceptions[singular]
+  }
+  return null
+}
+
+export default checkExceptions

package/src/syllables/split/find-cuts/01-vowel-span.js

@@ -0,0 +1,155 @@
+import compoundPrefixes from './_compound-prefixes.js'
+
+// Roots ending in ed keep their vowel; these are not regular past-tense suffixes.
+const voicedEd = /(?:[^aeiouyl]led|^(?:hundred|kindred|sacred|wicked|rugged|ragged|wretched|naked|crooked|(?:a|em|death|river|water|flower|hot|sick|sun|sea)bed|(?:in|inter|thorough|cross|pure)bred|infrared|newlywed|premed))(?:['’](?:s|d|ll|re|ve))?$/
+
+// Find vowel spans in lowercase spelling; return character offsets, not syllables.
+// Candidate vowel letters; ignored rules handle consonantal uses of u and y.
+const isVowel = /[aeiouy]/
+
+// Ignore the character at match.index + offset. Rules may veto a whole word.
+// Global patterns visit every occurrence; except prevents that rule from applying.
+const ignored = [
+  { match: /[^aeiou]e(?=['’](?:s|d|ll|re|ve)$|$)/g, offset: 1, except: /[^aeiouyl]le(?:['’](?:s|d|ll|re|ve))?$/ }, // silent e
+  { match: /[^td]ed(?=['’](?:s|d|ll|re|ve)$|$)/g, offset: 1, except: voicedEd }, // walked vs hundred/bubbled
+  { match: /es(?=['’](?:s|d|ll|re|ve)$|$)/g, offset: 0, except: /([sxzcg]|ch|sh|[^aeiouyl]l)es(?:['’](?:s|d|ll|re|ve))?$/ }, // cakes vs boxes/tables
+  {
+    match: /e(?=ments?$|thing$|s?man$)/g,
+    offset: 0,
+    except: /(?:[^aeiouyl]le(?:ments?|thing|s?man)|elements?)$/,
+  }, // movement(s), but settlement(s) and element(s)
+  // Silent e stays silent before consonant suffixes, but syllabic -le survives.
+  {
+    match: /[^aeiou]e(?=ness(?:es)?$|l(?:y|iness)$|ful(?:ly)?$|less$|ships?$|some$)/g,
+    offset: 1,
+    except: /[^aeiouyl]le(?:ness(?:es)?|l(?:y|iness)|ful(?:ly)?|less|ships?|some)$/,
+  },
+  { match: /['’][rv]e$/g, offset: 2 }, // we're, they've: suffix e adds no syllable
+  { match: /qu(?=[aeiouy])/g, offset: 1 }, // consonantal u
+  { match: /que(?=s?$|ly$|ness$)/g, offset: 2 }, // opaque, opaques, opaquely
+  { match: /^y(?=[aeiou])/g, offset: 0 }, // yard, yellow
+  { match: /[aeiouwy]y(?=[aeiou])/g, offset: 1 }, // beyond, lawyer, doubled y
+  { match: /[aeiou]ny(?=[aeiou])/g, offset: 2, except: /^any/ }, // canyon, but anyone
+  { match: /gu(?=a)/g, offset: 1 }, // guard, language
+  { match: /busi(?=ness)/g, offset: 3 }, // businesses, agribusiness: the i stays silent
+  { match: /ever(?=y(?:body|thing|where|one|day|time|$))/g, offset: 2 }, // everywhere, everyone
+]
+
+// Mc contributes a spoken syllable despite having no written vowel.
+const syllabicPrefix = /^mc(?=[a-z])/
+
+// These consonants carry a syllable despite having no written vowel.
+// The captured consonant starts the span; surrounding letters provide context.
+const syllabicConsonants = [
+  /(?:[sxz]e?|[cs]h|[cg]e)['’](s)$/, // horse's, church's
+  /[ai]s(m)(?:s|['’]s)?$/, // chasm, criticism, and plurals/possessives
+  /[iy]th(m)(?:s|['’]s)?$/, // algorithm, rhythms
+  /[^aeiouyn](n)['’]t$/, // couldn't, wasn't; not can't or hasn't
+  /^(?:could|should|would|might|must)['’](v)e$/, // modal + have: would-'ve, unlike we've
+]
+
+// Keep a boundary after these matched prefixes, even inside a vowel group.
+const separate = [
+  /^(?:re)?cre(?=at[eio])/, // create, recreate, recreation (but not creature)
+  /^re(?=act)/, // re-act and re-action
+  /^re(?=absorb|app|elect|emerg|enter|establish|evalu|impos|invent|issue|us[ei])/, // reuse, reenter; not reed
+  /^pre(?=adolesc|ambl|arrang|eminen|empt|exist|ignition|invas)/, // preempt, preexisting; not preen
+  /^(?:micro|macro)(?=[aeiou])/, // micro-organism, macro-economic
+  /^co(?=operat|ordinat|alesc|erc)/, // cooperate, coercion; not cooperage
+  /^pro(?=act)/, // proactive; not approach
+  /^flu(?=id)/, // flu-id, unlike fruit
+  /^ru(?=in)/, // ru-in and ru-ined
+  /di(?=et|en[ct])/, // diet, audience, obedient; not died or soldier
+  /^ide(?=a)/, // idea, ideal; not deal
+  /^sovi(?=et)/, // Soviet and its derivatives retain separate i and e sounds
+  /^soci(?=et)/, // society, societies, societal; not social
+  /i(?=ell[aoi])/, // Gabriella, Gabrielli, -iello names; not field
+  /i(?=et(?:te)?s?$)/, // Juliet, Henriette
+  /^qui(?=et)/, // quiet keeps its separate vowels in quietly and quieter
+  /^acre(?=age)/, // acreage retains the vowel of acre before -age
+  /^sci(?=on)/, // scion has two vowels, unlike the ending in vision
+  /^intra(?=[aeiou])/, // intrauterine, intraocular
+  /re(?=al(?!ly|m))/, // cereal, reality; not really or realm
+  /[^q]u(?=en[ctdz])/, // fluent, innuendo, influenza; not subsequent
+  /[^g]u(?=ing)/, // suing, continuing; not plaguing or intriguing
+  /argu(?=ing)/, // arguing retains its pronounced u
+  /^lou(?=is)/, // Louis, Louisville
+  /[eo]u(?=illes?$)/, // separate the vowel groups in -euille/-ouille
+  /ou(?=ett)/, // silhouette, pirouette: ou and e are separate vowels
+  /o(?=as?$)/, // boa, boas; not boat
+]
+
+// Match a bounded window around the current vowel: [i - before, i + after).
+const joined = [
+  // Common vowel pairs share a span unless a separate rule overrides them.
+  { before: 1, after: 1, match: /^(ai|au|ay|ea|ee|ei|eu|ey|ie|oa|oe|oi|oo|ou|oy|ue|ui)$/ },
+  { before: 3, after: 2, match: /^llion/ }, // million, unlike lion
+  { before: 2, after: 3, match: /^[cst]i(?:on|an|al|ous)/ }, // nation, special, musician
+  { before: 2, after: 4, match: /^cia(?:s|['’]s)?$/ }, // final -cia blends; association and glaciation keep separate vowels
+  { before: 2, after: 1, match: /^eau$/ }, // extend ea to eau in beautiful
+  { before: 1, after: 3, match: /^ae(?:r|st)/ }, // aerospace, aesthetic
+  { before: 2, after: 3, match: /^[cg]eous/ }, // gorgeous, herbaceous; not courteous
+  { before: 3, after: 3, match: /^hteous/ }, // righteous and its suffixes
+]
+
+// Mark pronounced vowel groups before choosing any consonant boundaries.
+const vowelSpan = word => {
+  // Record ignored positions without deleting letters or shifting offsets.
+  const silent = new Set()
+  // A silent e remains silent inside a recognized compound.
+  compoundPrefixes.forEach(pattern => {
+    const match = pattern.exec(word)
+    if (match) {
+      silent.add(match[0].lastIndexOf('e'))
+    }
+  })
+  ignored.forEach(rule => {
+    if (!rule.except || !rule.except.test(word)) {
+      for (const match of word.matchAll(rule.match)) {
+        silent.add(match.index + rule.offset)
+      }
+    }
+  })
+  // Forced separations take precedence over the joined-vowel rules below.
+  const breaks = new Set()
+  separate.forEach(pattern => {
+    const match = pattern.exec(word)
+    if (match) {
+      breaks.add(match.index + match[0].length)
+    }
+  })
+  // Spans use slice-compatible offsets: start is inclusive, end is exclusive.
+  const groups = []
+  const prefix = syllabicPrefix.exec(word)
+  if (prefix) {
+    groups.push({ start: 0, end: prefix[0].length })
+  }
+  // Mark the captured consonants without inserting letters or shifting offsets.
+  const voiced = new Set()
+  syllabicConsonants.forEach(pattern => {
+    const match = pattern.exec(word)
+    if (match) {
+      voiced.add(match.index + match[0].lastIndexOf(match[1]))
+    }
+  })
+  for (let i = 0; i < word.length; i += 1) {
+    if ((!isVowel.test(word[i]) && !voiced.has(i)) || silent.has(i)) {
+      continue
+    }
+    const previous = groups[groups.length - 1]
+    // Only adjacent pronounced vowels may extend an existing span.
+    const blended =
+      previous &&
+      previous.end === i &&
+      !breaks.has(i) &&
+      joined.some(rule => rule.match.test(word.slice(Math.max(0, i - rule.before), i + rule.after)))
+    if (blended) {
+      previous.end = i + 1
+    } else {
+      groups.push({ start: i, end: i + 1 })
+    }
+  }
+  return groups
+}
+
+export default vowelSpan

package/src/syllables/split/find-cuts/02-boundaries.js

@@ -0,0 +1,86 @@
+import compoundPrefixes from './_compound-prefixes.js'
+
+// Turn vowel spans into cut positions by assigning the consonants between them.
+// Consonant clusters allowed to start a chunk; the picker checks longest first.
+const onset = new Set([
+  'bl', 'br', 'ch', 'cl', 'cr', 'dr', 'fl', 'fr', 'gl', 'gr',
+  'ph', 'pl', 'pr', 'qu', 'sh', 'sk', 'sl', 'sm', 'sn', 'sp',
+  'st', 'sw', 'th', 'tr', 'tw', 'wh', 'scr', 'shr', 'spl', 'spr', 'str', 'thr',
+])
+
+// Applied in order after onset placement; later rules take precedence.
+// left/right are offsets from the surrounding vowel groups' edges.
+// minLeft only moves a boundary forward, preserving a longer consonant cluster.
+// gap matches intervening letters; tail matches the complete remaining spelling.
+const adjustments = [
+  { minGap: 1, maxGap: 1, gap: /^x$/, left: 1 }, // box-es
+  { minGap: 2, gap: /^(.)\1/, minLeft: 1 }, // rab-bit
+  { minGap: 2, tail: /^ed$/, right: -1 }, // res-ted
+  { minGap: 2, gap: /^sh$/, tail: /^(es|ing)$/, right: 0 }, // wish-ing
+  { minGap: 2, gap: /l$/, tail: /^e(?:[sd]|['’](?:s|d|ll|re|ve))?$/, right: -2 }, // bot-tle, ta-bles, bub-bled
+  { minGap: 2, gap: /l$/, tail: /^e(?:ments?|ships?|ness(?:es)?|less|ful(?:ly)?|ly)(?:['’]s)?$/, right: -2 }, // set-tle-ment, bat-tle-ships
+  { minGap: 1, gap: /l$/, tail: /^ess(?:ly|ness(?:es)?)?$/, right: -1 }, // help-less: keep the suffix intact
+  { minGap: 3, gap: /^ssl$/, tail: /^y$/, right: -1 }, // help-less-ly, not help-les-sly
+  { minGap: 1, gap: /^(?:s|th)$/, tail: /^m(?:s|['’]s)?$/, right: 0 }, // chas-m, rhyth-m
+  { minGap: 1, tail: /^n['’]t$/, right: 0 }, // could-n't: keep the base consonants together
+  { minGap: 1, gap: /^[-‐‑–—]+$/, right: 0 }, // re-enter: keep the hyphen on the left
+]
+
+// Choose one cut between each pair of vowel spans, preserving their count.
+const pickBoundaries = (word, vowelSpans) => {
+  const cuts = []
+  // Keep recognized compound components intact, such as side-walk and some-one.
+  const compoundCuts = compoundPrefixes.flatMap(pattern => {
+    const match = pattern.exec(word)
+    return match ? [match[0].length] : []
+  })
+  for (let i = 1; i < vowelSpans.length; i += 1) {
+    // The gap contains the letters not assigned to either vowel span.
+    const left = vowelSpans[i - 1].end
+    const right = vowelSpans[i].start
+    const gap = word.slice(left, right)
+    // Match full suffixes so long endings are not mistaken for shorter ones.
+    const tail = word.slice(right)
+    // Adjacent vowels split directly; a single consonant starts the next chunk.
+    let cut = left
+    if (gap.length > 1) {
+      cut = right - 1
+      // Prefer the longest recognized consonant cluster at the next chunk's start.
+      for (let size = Math.min(3, gap.length); size > 1; size -= 1) {
+        if (onset.has(gap.slice(-size))) {
+          cut = right - size
+          break
+        }
+      }
+    }
+    // Apply every matching adjustment in order, allowing specific rules to override.
+    adjustments.forEach(rule => {
+      if (
+        gap.length < rule.minGap ||
+        (rule.maxGap !== undefined && gap.length > rule.maxGap) ||
+        (rule.gap && !rule.gap.test(gap)) ||
+        (rule.tail && !rule.tail.test(tail))
+      ) {
+        return
+      }
+      // Exact offsets replace the cut; minLeft only advances an earlier cut.
+      if (rule.left !== undefined) {
+        cut = left + rule.left
+      } else if (rule.right !== undefined) {
+        cut = right + rule.right
+      } else {
+        cut = Math.max(cut, left + rule.minLeft)
+      }
+    })
+    // A compound boundary overrides ordinary consonant placement within this gap.
+    compoundCuts.forEach(boundary => {
+      if (boundary >= left && boundary <= right) {
+        cut = boundary
+      }
+    })
+    cuts.push(cut)
+  }
+  return cuts
+}
+
+export default pickBoundaries