compromise-speech 0.0.1 → 0.0.2

package/builds/compromise-speech.cjs

@@ -21,7 +21,7 @@
     cchange: (s) => {
       s = s.replace(/([^s]|^)(c)(h)/g, '$1x$3').trim();
       s = s.replace(/cia/g, 'xia');
-      s = s.replace(/c(i|e|y)/g, 's$1');
+      s = s.replace(/c([iey])/g, 's$1');
       return s.replace(/c/g, 'k')
     },
     dchange: (s) => {
@@ -29,12 +29,14 @@
       return s.replace(/d/g, 't')
     },
     dropG: (s) => {
-      s = s.replace(/gh(^$|[^aeiou])/g, 'h$1');
+      // Drop 'G' if followed by 'H' and 'H' is not at the end or before a vowel. 
+      s = s.replace(/gh(^$|[^aeiou])/g, 'h$1');//eslint-disable-line
+      // Drop 'G' if followed by 'N' or 'NED' and is at the end.
       return s.replace(/g(n|ned)$/g, '$1')
     },
     changeG: (s) => {
       s = s.replace(/gh/g, 'f');
-      s = s.replace(/([^g]|^)(g)(i|e|y)/g, '$1j$3');
+      s = s.replace(/([^g]|^)(g)([iey])/g, '$1j$3');
       s = s.replace(/gg/g, 'g');
       return s.replace(/g/g, 'k')
     },
@@ -84,51 +86,56 @@
       return s //.charAt(0) + s.substr(1, s.length).replace(/[aeiou]/g, '');
     },
   };
+  var m = transformations;
 
   //a js version of the metaphone (#1) algorithm
 
   const metaphone = function (s) {
-    s = transformations.dedup(s);
-    s = transformations.dropInitialLetters(s);
-    s = transformations.dropBafterMAtEnd(s);
-    s = transformations.changeCK(s);
-    s = transformations.cchange(s);
-    s = transformations.dchange(s);
-    s = transformations.dropG(s);
-    s = transformations.changeG(s);
-    s = transformations.dropH(s);
-    s = transformations.changePH(s);
-    s = transformations.changeQ(s);
-    s = transformations.changeS(s);
-    s = transformations.changeX(s);
-    s = transformations.changeT(s);
-    s = transformations.dropT(s);
-    s = transformations.changeV(s);
-    s = transformations.changeWH(s);
-    s = transformations.dropW(s);
-    s = transformations.dropY(s);
-    s = transformations.changeZ(s);
-    s = transformations.dropVowels(s);
+    s = m.dedup(s);
+    s = m.dropInitialLetters(s);
+    s = m.dropBafterMAtEnd(s);
+    s = m.changeCK(s);
+    s = m.cchange(s);
+    s = m.dchange(s);
+    s = m.dropG(s);
+    s = m.changeG(s);
+    s = m.dropH(s);
+    s = m.changePH(s);
+    s = m.changeQ(s);
+    s = m.changeS(s);
+    s = m.changeX(s);
+    s = m.changeT(s);
+    s = m.dropT(s);
+    s = m.changeV(s);
+    s = m.changeWH(s);
+    s = m.dropW(s);
+    s = m.dropY(s);
+    s = m.changeZ(s);
+    s = m.dropVowels(s);
     return s.trim()
   };
 
+  var metaphone$1 = metaphone;
+
   const soundsLike = function (view) {
     view.docs.forEach(terms => {
       terms.forEach(term => {
-        term.soundsLike = metaphone(term.normal || term.text);
+        term.soundsLike = metaphone$1(term.normal || term.text);
       });
     });
   };
 
+  var soundsLike$1 = soundsLike;
+
   const starts_with_single_vowel_combos = /^(eu)/i;
-  const joining_consonant_vowel = /^[^aeiou][e]([^d]|$)/;
+  const joining_consonant_vowel = /^[^aeiou]e([^d]|$)/;
   const cvcv_same_consonant = /^([^aeiouy])[aeiouy]\1[aeiouy]/;
   const cvcv_same_vowel = /^[^aeiouy]([aeiouy])[^aeiouy]\1/;
   const cvcv_known_consonants = /^([tg][aeiouy]){2}/;
   const only_one_or_more_c = /^[^aeiouy]+$/;
 
   const ends_with_vowel$1 = /[aeiouy]$/;
-  const starts_with_consonant_vowel$1 = /^[^aeiouy][h]?[aeiouy]/;
+  const starts_with_consonant_vowel$1 = /^[^aeiouy]h?[aeiouy]/;
 
   const ones = [
     /^[^aeiou]?ion/,
@@ -136,7 +143,7 @@
     /^[^aeiou]?iled/,
 
     // -ing, -ent
-    /[aeiou][n][gt]$/,
+    /[aeiou]n[gt]$/,
 
     // -ate, -age
     /\wa[gt]e$/,
@@ -226,12 +233,13 @@
 
     return arr
   };
+  var postProcess = postprocess;
 
   //chop a string into pronounced syllables
 
   const all_spaces = / +/g;
   const ends_with_vowel = /[aeiouy]$/;
-  const starts_with_consonant_vowel = /^[^aeiouy][h]?[aeiouy]/;
+  const starts_with_consonant_vowel = /^[^aeiouy]h?[aeiouy]/;
   const starts_with_e_then_specials = /^e[sm]/;
   const starts_with_e = /^e/;
   const ends_with_noisy_vowel_combos = /(eo|eu|ia|oa|ua|ui)$/i;
@@ -290,7 +298,7 @@
     return all
   };
 
-  let syllables$1 = function (str) {
+  let syllables$2 = function (str) {
     let all = [];
     if (!str) {
       return all
@@ -301,7 +309,7 @@
     });
 
     // str.split(whitespace_dash).forEach(doWord)
-    all = postprocess(all);
+    all = postProcess(all);
 
     //for words like 'tree' and 'free'
     if (all.length === 0) {
@@ -313,17 +321,25 @@
     return all
   };
 
+  // console.log(syllables('civilised'))
+
+  var getSyllables = syllables$2;
+
+  // const defaultObj = { normal: true, text: true, terms: false }
+
   const syllables = function (view) {
     view.docs.forEach(terms => {
       terms.forEach(term => {
-        term.syllables = syllables$1(term.normal || term.text);
+        term.syllables = getSyllables(term.normal || term.text);
       });
     });
   };
 
+  var syllables$1 = syllables;
+
   var compute = {
-    soundsLike,
-    syllables
+    soundsLike: soundsLike$1,
+    syllables: syllables$1
   };
 
   const api = function (View) {
@@ -332,16 +348,36 @@
       this.compute('syllables');
       let all = [];
       this.docs.forEach(terms => {
+        let some = [];
+        terms.forEach(term => {
+          some = some.concat(term.syllables);
+        });
+        if (some.length > 0) {
+          all.push(some);
+        }
+      });
+      return all
+    };
+    /** */
+    View.prototype.soundsLike = function () {
+      this.compute('soundsLike');
+      let all = [];
+      this.docs.forEach(terms => {
+        let some = [];
         terms.forEach(term => {
-          all = all.concat(term.syllables);
+          some = some.concat(term.soundsLike);
         });
+        if (some.length > 0) {
+          all.push(some);
+        }
       });
       return all
     };
   };
+  var api$1 = api;
 
   var plugin = {
-    api,
+    api: api$1,
     compute
   };
 

package/builds/compromise-speech.min.js

@@ -1 +1 @@
-var e,t;e=this,t=function(){const e=e=>e.replace(/([^c])\1/g,"$1"),t=e=>e.match(/^(kn|gn|pn|ae|wr)/)?e.substr(1,e.length-1):e,a=e=>e.replace(/mb$/,"m"),l=e=>(e=(e=(e=e.replace(/([^s]|^)(c)(h)/g,"$1x$3").trim()).replace(/cia/g,"xia")).replace(/c(i|e|y)/g,"s$1")).replace(/c/g,"k"),c=e=>(e=e.replace(/d(ge|gy|gi)/g,"j$1")).replace(/d/g,"t"),i=e=>(e=e.replace(/gh(^$|[^aeiou])/g,"h$1")).replace(/g(n|ned)$/g,"$1"),n=e=>(e=(e=(e=e.replace(/gh/g,"f")).replace(/([^g]|^)(g)(i|e|y)/g,"$1j$3")).replace(/gg/g,"g")).replace(/g/g,"k"),o=e=>e.replace(/([aeiou])h([^aeiou]|$)/g,"$1$2"),h=e=>e.replace(/ck/g,"k"),r=e=>e.replace(/ph/g,"f"),g=e=>e.replace(/q/g,"k"),u=e=>e.replace(/s(h|io|ia)/g,"x$1"),p=e=>e=e.replace(/t(ia[^n]|io)/g,"x$1"),s=e=>e.replace(/tch/g,"ch"),f=e=>e.replace(/v/g,"f"),m=e=>e.replace(/^wh/,"w"),y=e=>e.replace(/w([^aeiou]|$)/g,"$1"),$=e=>(e=e.replace(/^x/,"s")).replace(/x/g,"ks"),d=e=>e.replace(/y([^aeiou]|$)/g,"$1"),b=e=>e.replace(/z/,"s"),x=e=>e,k=/^(eu)/i,E=/^[^aeiou][e]([^d]|$)/,j=/^([^aeiouy])[aeiouy]\1[aeiouy]/,w=/^[^aeiouy]([aeiouy])[^aeiouy]\1/,v=/^([tg][aeiouy]){2}/,L=/^[^aeiouy]+$/,T=/[aeiouy]$/,q=/^[^aeiouy][h]?[aeiouy]/,z=[/^[^aeiou]?ion/,/^[^aeiou]?ised/,/^[^aeiou]?iled/,/[aeiou][n][gt]$/,/\wa[gt]e$/],S=/ +/g,A=/[aeiouy]$/,B=/^[^aeiouy][h]?[aeiouy]/,C=/^e[sm]/,D=/^e/,F=/(eo|eu|ia|oa|ua|ui)$/i,G=/[aiouy]/,H=/ee$/,I=function(e){let t=[],a=e.split(""),l="",c="",i="";for(let e=0;e<a.length;e++){l=a.slice(0,e).join(""),i=a[e],c=a.slice(e+1,a.length).join("");let n=l+a[e];if(l.match(A)&&!i.match(A))return c.match(C)&&(n+="e",c=c.replace(D,"")),t.push(n),t.concat(I(c));if(n.match(F))return t.push(l),t.push(i),t.concat(I(c));if(n.match(A)&&c.match(B))return t.push(n),t.concat(I(c))}if(e.match(G)||e.match(H))t.push(e);else if(e){let a=t.length-1;a<0&&(a=0),t[a]=(t[a]||"")+e}return t};let J=function(e){let t=[];return e?((e=e.replace(/[.,?]/g,"")).split(S).map(e=>{t=t.concat(I(e))}),t=function(e){let t=(e=(e=e.map((function(e){return e.trim()}))).filter((function(e){return""!==e}))).length;if(t>1){let a=e[t-2]+e[t-1];for(let l=0;l<z.length;l++)a.match(z[l])&&(e[t-2]=e[t-2]+e[t-1],e.pop())}if(e.length>1){let t=(1===e[0].length||e[0].match(q))&&e[0].match(T),a=e[1].match(E);if(t&&a){let t=e[0]+e[1];t.match(j)||t.match(w)||t.match(v)||(e[0]=e[0]+e[1],e.splice(1,1))}}if(e.length>1){let t=e[e.length-2].match(q)&&e[e.length-2].match(T),a=e[e.length-1].match(E)&&z.every(t=>!e[e.length-1].match(t));if(t&&a){let t=e[e.length-2]+e[e.length-1];t.match(j)||t.match(w)||t.match(v)||(e[e.length-2]=e[e.length-2]+e[e.length-1],e.splice(e.length-1,1))}}if(e.length>1){let t=e[0]+e[1];t.match(k)&&(e[0]=t,e.splice(1,1))}return e.length>1&&e[e.length-1].match(L)&&(e[e.length-2]=e[e.length-2]+e[e.length-1],e.splice(e.length-1,1)),e}(t),0===t.length&&(t=[e]),t=t.filter(e=>e),t):t};return{api:function(e){e.prototype.syllables=function(){this.compute("syllables");let e=[];return this.docs.forEach(t=>{t.forEach(t=>{e=e.concat(t.syllables)})}),e}},compute:{soundsLike:function(k){k.docs.forEach(k=>{k.forEach(k=>{var E;k.soundsLike=(E=k.normal||k.text,E=e(E),E=t(E),E=a(E),E=h(E),E=l(E),E=c(E),E=i(E),E=n(E),E=o(E),E=r(E),E=g(E),E=u(E),E=$(E),E=p(E),E=s(E),E=f(E),E=m(E),E=y(E),E=d(E),E=b(E),(E=x(E)).trim())})})},syllables:function(e){e.docs.forEach(e=>{e.forEach(e=>{e.syllables=J(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();
+var e,t;e=this,t=function(){var e={dedup:e=>e.replace(/([^c])\1/g,"$1"),dropInitialLetters:e=>e.match(/^(kn|gn|pn|ae|wr)/)?e.substr(1,e.length-1):e,dropBafterMAtEnd:e=>e.replace(/mb$/,"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();

package/builds/compromise-speech.mjs

@@ -1 +1 @@
-const e=e=>e.replace(/([^c])\1/g,"$1"),t=e=>e.match(/^(kn|gn|pn|ae|wr)/)?e.substr(1,e.length-1):e,a=e=>e.replace(/mb$/,"m"),l=e=>(e=(e=(e=e.replace(/([^s]|^)(c)(h)/g,"$1x$3").trim()).replace(/cia/g,"xia")).replace(/c(i|e|y)/g,"s$1")).replace(/c/g,"k"),c=e=>(e=e.replace(/d(ge|gy|gi)/g,"j$1")).replace(/d/g,"t"),i=e=>(e=e.replace(/gh(^$|[^aeiou])/g,"h$1")).replace(/g(n|ned)$/g,"$1"),h=e=>(e=(e=(e=e.replace(/gh/g,"f")).replace(/([^g]|^)(g)(i|e|y)/g,"$1j$3")).replace(/gg/g,"g")).replace(/g/g,"k"),n=e=>e.replace(/([aeiou])h([^aeiou]|$)/g,"$1$2"),r=e=>e.replace(/ck/g,"k"),g=e=>e.replace(/ph/g,"f"),o=e=>e.replace(/q/g,"k"),u=e=>e.replace(/s(h|io|ia)/g,"x$1"),p=e=>e=e.replace(/t(ia[^n]|io)/g,"x$1"),s=e=>e.replace(/tch/g,"ch"),m=e=>e.replace(/v/g,"f"),f=e=>e.replace(/^wh/,"w"),$=e=>e.replace(/w([^aeiou]|$)/g,"$1"),y=e=>(e=e.replace(/^x/,"s")).replace(/x/g,"ks"),d=e=>e.replace(/y([^aeiou]|$)/g,"$1"),k=e=>e.replace(/z/,"s"),x=e=>e,b=/^(eu)/i,E=/^[^aeiou][e]([^d]|$)/,w=/^([^aeiouy])[aeiouy]\1[aeiouy]/,j=/^[^aeiouy]([aeiouy])[^aeiouy]\1/,v=/^([tg][aeiouy]){2}/,L=/^[^aeiouy]+$/,q=/[aeiouy]$/,z=/^[^aeiouy][h]?[aeiouy]/,A=[/^[^aeiou]?ion/,/^[^aeiou]?ised/,/^[^aeiou]?iled/,/[aeiou][n][gt]$/,/\wa[gt]e$/],B=/ +/g,C=/[aeiouy]$/,D=/^[^aeiouy][h]?[aeiouy]/,F=/^e[sm]/,G=/^e/,H=/(eo|eu|ia|oa|ua|ui)$/i,I=/[aiouy]/,J=/ee$/,K=function(e){let t=[],a=e.split(""),l="",c="",i="";for(let e=0;e<a.length;e++){l=a.slice(0,e).join(""),i=a[e],c=a.slice(e+1,a.length).join("");let h=l+a[e];if(l.match(C)&&!i.match(C))return c.match(F)&&(h+="e",c=c.replace(G,"")),t.push(h),t.concat(K(c));if(h.match(H))return t.push(l),t.push(i),t.concat(K(c));if(h.match(C)&&c.match(D))return t.push(h),t.concat(K(c))}if(e.match(I)||e.match(J))t.push(e);else if(e){let a=t.length-1;a<0&&(a=0),t[a]=(t[a]||"")+e}return t};let M=function(e){let t=[];return e?((e=e.replace(/[.,?]/g,"")).split(B).map(e=>{t=t.concat(K(e))}),t=function(e){let t=(e=(e=e.map((function(e){return e.trim()}))).filter((function(e){return""!==e}))).length;if(t>1){let a=e[t-2]+e[t-1];for(let l=0;l<A.length;l++)a.match(A[l])&&(e[t-2]=e[t-2]+e[t-1],e.pop())}if(e.length>1){let t=(1===e[0].length||e[0].match(z))&&e[0].match(q),a=e[1].match(E);if(t&&a){let t=e[0]+e[1];t.match(w)||t.match(j)||t.match(v)||(e[0]=e[0]+e[1],e.splice(1,1))}}if(e.length>1){let t=e[e.length-2].match(z)&&e[e.length-2].match(q),a=e[e.length-1].match(E)&&A.every(t=>!e[e.length-1].match(t));if(t&&a){let t=e[e.length-2]+e[e.length-1];t.match(w)||t.match(j)||t.match(v)||(e[e.length-2]=e[e.length-2]+e[e.length-1],e.splice(e.length-1,1))}}if(e.length>1){let t=e[0]+e[1];t.match(b)&&(e[0]=t,e.splice(1,1))}return e.length>1&&e[e.length-1].match(L)&&(e[e.length-2]=e[e.length-2]+e[e.length-1],e.splice(e.length-1,1)),e}(t),0===t.length&&(t=[e]),t=t.filter(e=>e),t):t};var N={api:function(e){e.prototype.syllables=function(){this.compute("syllables");let e=[];return this.docs.forEach(t=>{t.forEach(t=>{e=e.concat(t.syllables)})}),e}},compute:{soundsLike:function(b){b.docs.forEach(b=>{b.forEach(b=>{var E;b.soundsLike=(E=b.normal||b.text,E=e(E),E=t(E),E=a(E),E=r(E),E=l(E),E=c(E),E=i(E),E=h(E),E=n(E),E=g(E),E=o(E),E=u(E),E=y(E),E=p(E),E=s(E),E=m(E),E=f(E),E=$(E),E=d(E),E=k(E),(E=x(E)).trim())})})},syllables:function(e){e.docs.forEach(e=>{e.forEach(e=>{e.syllables=M(e.normal||e.text)})})}}};export{N as default};
+var e={dedup:e=>e.replace(/([^c])\1/g,"$1"),dropInitialLetters:e=>e.match(/^(kn|gn|pn|ae|wr)/)?e.substr(1,e.length-1):e,dropBafterMAtEnd:e=>e.replace(/mb$/,"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};

package/demo/index.html

@@ -0,0 +1,28 @@
+<!DOCTYPE html>
+<html>
+
+<head>
+
+  <script src="https://unpkg.com/compromise"></script>
+  <!-- <script src="https://unpkg.com/compromise-speech"></script> -->
+  <script src="../builds/compromise-speech.min.js"></script>
+
+</head>
+
+<body>
+  <p>compromise-speech demo:</p>
+  <ul>
+    <kbd>chocolate microscopes </kbd>
+    <pre id="res">loading</pre>
+  </ul>
+
+  <script defer>
+    nlp.plugin(compromiseSpeech) // window.compromiseSpeech
+    let txt = document.querySelector('kbd').innerText
+    let doc = nlp(txt)
+    let json = doc.syllables()
+    document.querySelector('#res').innerHTML = JSON.stringify(json, null, 2)
+  </script>
+</body>
+
+</html>

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "compromise-speech",
   "description": "plugin for nlp-compromise",
-  "version": "0.0.1",
+  "version": "0.0.2",
   "author": "Spencer Kelly <spencermountain@gmail.com> (http://spencermounta.in)",
   "main": "./src/plugin.js",
   "unpkg": "./builds/compromise-speech.min.js",

package/rollup.config.js

@@ -1,22 +1,23 @@
 import { terser } from 'rollup-plugin-terser'
 import sizeCheck from 'rollup-plugin-filesize-check'
+import { nodeResolve } from '@rollup/plugin-node-resolve';
 
 const opts = { keep_classnames: true, module: true }
 
 export default [
   {
     input: 'src/plugin.js',
-    output: [{ file: 'builds/compromise-stream.cjs', format: 'umd', name: 'compromiseStream' }],
-    plugins: [sizeCheck({ expect: 11, warn: 15 })],
+    output: [{ file: 'builds/compromise-speech.cjs', format: 'umd', name: 'compromiseSpeech' }],
+    plugins: [nodeResolve(), sizeCheck({ expect: 11, warn: 15 })],
   },
   {
     input: 'src/plugin.js',
-    output: [{ file: 'builds/compromise-stream.min.js', format: 'umd', name: 'compromiseStream' }],
-    plugins: [terser(opts), sizeCheck({ expect: 4, warn: 15 })],
+    output: [{ file: 'builds/compromise-speech.min.js', format: 'umd', name: 'compromiseSpeech' }],
+    plugins: [nodeResolve(), terser(opts), sizeCheck({ expect: 4, warn: 15 })],
   },
   {
     input: 'src/plugin.js',
-    output: [{ file: 'builds/compromise-stream.mjs', format: 'esm' }],
-    plugins: [terser(opts), sizeCheck({ expect: 4, warn: 15 })],
+    output: [{ file: 'builds/compromise-speech.mjs', format: 'esm' }],
+    plugins: [nodeResolve(), terser(opts), sizeCheck({ expect: 4, warn: 15 })],
   }
 ]

package/tests/_lib.js

@@ -1,13 +1,15 @@
-/* eslint-disable no-console */
 import build from '../../../builds/one/compromise-one.mjs'
 import src from '../../../src/one.js'
-let nlp = src
-import speech from '../src/plugin.js'
+import plgBuild from '../builds/compromise-speech.mjs'
+import plg from '../src/plugin.js'
+let nlp;
 
-// import dateBuild from '../builds/compromise-speech.js'
 if (process.env.TESTENV === 'prod') {
-  console.warn('== production build test 🚀 ==')
+  console.warn('== production build test 🚀 ==')  // eslint-disable-line
   nlp = build
+  nlp.plugin(plgBuild)
+} else {
+  nlp = src
+  nlp.plugin(plg)
 }
-nlp.plugin(speech)
 export default nlp

package/builds/compromise-stream.cjs

@@ -1,373 +0,0 @@
-(function (global, factory) {
-  typeof exports === 'object' && typeof module !== 'undefined' ? module.exports = factory() :
-  typeof define === 'function' && define.amd ? define(factory) :
-  (global = typeof globalThis !== 'undefined' ? globalThis : global || self, global.compromiseStream = factory());
-})(this, (function () { 'use strict';
-
-  //individual manipulations of the text
-  const transformations = {
-    dedup: (s) => {
-      return s.replace(/([^c])\1/g, '$1')
-    },
-    dropInitialLetters: (s) => {
-      if (s.match(/^(kn|gn|pn|ae|wr)/)) {
-        return s.substr(1, s.length - 1)
-      }
-      return s
-    },
-    dropBafterMAtEnd: (s) => {
-      return s.replace(/mb$/, 'm')
-    },
-    cchange: (s) => {
-      s = s.replace(/([^s]|^)(c)(h)/g, '$1x$3').trim();
-      s = s.replace(/cia/g, 'xia');
-      s = s.replace(/c([iey])/g, 's$1');
-      return s.replace(/c/g, 'k')
-    },
-    dchange: (s) => {
-      s = s.replace(/d(ge|gy|gi)/g, 'j$1');
-      return s.replace(/d/g, 't')
-    },
-    dropG: (s) => {
-      // Drop 'G' if followed by 'H' and 'H' is not at the end or before a vowel. 
-      s = s.replace(/gh(^$|[^aeiou])/g, 'h$1');//eslint-disable-line
-      // Drop 'G' if followed by 'N' or 'NED' and is at the end.
-      return s.replace(/g(n|ned)$/g, '$1')
-    },
-    changeG: (s) => {
-      s = s.replace(/gh/g, 'f');
-      s = s.replace(/([^g]|^)(g)([iey])/g, '$1j$3');
-      s = s.replace(/gg/g, 'g');
-      return s.replace(/g/g, 'k')
-    },
-    dropH: (s) => {
-      return s.replace(/([aeiou])h([^aeiou]|$)/g, '$1$2')
-    },
-    changeCK: (s) => {
-      return s.replace(/ck/g, 'k')
-    },
-    changePH: (s) => {
-      return s.replace(/ph/g, 'f')
-    },
-    changeQ: (s) => {
-      return s.replace(/q/g, 'k')
-    },
-    changeS: (s) => {
-      return s.replace(/s(h|io|ia)/g, 'x$1')
-    },
-    changeT: (s) => {
-      s = s.replace(/t(ia[^n]|io)/g, 'x$1');
-      // return s.replace(/th/, '0')
-      return s
-    },
-    dropT: (s) => {
-      return s.replace(/tch/g, 'ch')
-    },
-    changeV: (s) => {
-      return s.replace(/v/g, 'f')
-    },
-    changeWH: (s) => {
-      return s.replace(/^wh/, 'w')
-    },
-    dropW: (s) => {
-      return s.replace(/w([^aeiou]|$)/g, '$1')
-    },
-    changeX: (s) => {
-      s = s.replace(/^x/, 's');
-      return s.replace(/x/g, 'ks')
-    },
-    dropY: (s) => {
-      return s.replace(/y([^aeiou]|$)/g, '$1')
-    },
-    changeZ: (s) => {
-      return s.replace(/z/, 's')
-    },
-    dropVowels: (s) => {
-      return s //.charAt(0) + s.substr(1, s.length).replace(/[aeiou]/g, '');
-    },
-  };
-
-  //a js version of the metaphone (#1) algorithm
-
-  const metaphone = function (s) {
-    s = transformations.dedup(s);
-    s = transformations.dropInitialLetters(s);
-    s = transformations.dropBafterMAtEnd(s);
-    s = transformations.changeCK(s);
-    s = transformations.cchange(s);
-    s = transformations.dchange(s);
-    s = transformations.dropG(s);
-    s = transformations.changeG(s);
-    s = transformations.dropH(s);
-    s = transformations.changePH(s);
-    s = transformations.changeQ(s);
-    s = transformations.changeS(s);
-    s = transformations.changeX(s);
-    s = transformations.changeT(s);
-    s = transformations.dropT(s);
-    s = transformations.changeV(s);
-    s = transformations.changeWH(s);
-    s = transformations.dropW(s);
-    s = transformations.dropY(s);
-    s = transformations.changeZ(s);
-    s = transformations.dropVowels(s);
-    return s.trim()
-  };
-
-  const soundsLike = function (view) {
-    view.docs.forEach(terms => {
-      terms.forEach(term => {
-        term.soundsLike = metaphone(term.normal || term.text);
-      });
-    });
-  };
-
-  const starts_with_single_vowel_combos = /^(eu)/i;
-  const joining_consonant_vowel = /^[^aeiou]e([^d]|$)/;
-  const cvcv_same_consonant = /^([^aeiouy])[aeiouy]\1[aeiouy]/;
-  const cvcv_same_vowel = /^[^aeiouy]([aeiouy])[^aeiouy]\1/;
-  const cvcv_known_consonants = /^([tg][aeiouy]){2}/;
-  const only_one_or_more_c = /^[^aeiouy]+$/;
-
-  const ends_with_vowel$1 = /[aeiouy]$/;
-  const starts_with_consonant_vowel$1 = /^[^aeiouy]h?[aeiouy]/;
-
-  const ones = [
-    /^[^aeiou]?ion/,
-    /^[^aeiou]?ised/,
-    /^[^aeiou]?iled/,
-
-    // -ing, -ent
-    /[aeiou]n[gt]$/,
-
-    // -ate, -age
-    /\wa[gt]e$/,
-  ];
-
-  //suffix fixes
-  const postprocess = function (arr) {
-    //trim whitespace
-    arr = arr.map(function (w) {
-      return w.trim()
-    });
-    arr = arr.filter(function (w) {
-      return w !== ''
-    });
-    // if (arr.length > 2) {
-    //   return arr;
-    // }
-    let l = arr.length;
-    if (l > 1) {
-      let suffix = arr[l - 2] + arr[l - 1];
-      for (let i = 0; i < ones.length; i++) {
-        if (suffix.match(ones[i])) {
-          arr[l - 2] = arr[l - 2] + arr[l - 1];
-          arr.pop();
-        }
-      }
-    }
-
-    // since the open syllable detection is overzealous,
-    // sometimes need to rejoin incorrect splits
-    if (arr.length > 1) {
-      let first_is_open =
-        (arr[0].length === 1 || arr[0].match(starts_with_consonant_vowel$1)) &&
-        arr[0].match(ends_with_vowel$1);
-      let second_is_joining = arr[1].match(joining_consonant_vowel);
-
-      if (first_is_open && second_is_joining) {
-        let possible_combination = arr[0] + arr[1];
-        let probably_separate_syllables =
-          possible_combination.match(cvcv_same_consonant) ||
-          possible_combination.match(cvcv_same_vowel) ||
-          possible_combination.match(cvcv_known_consonants);
-
-        if (!probably_separate_syllables) {
-          arr[0] = arr[0] + arr[1];
-          arr.splice(1, 1);
-        }
-      }
-    }
-
-    if (arr.length > 1) {
-      let second_to_last_is_open =
-        arr[arr.length - 2].match(starts_with_consonant_vowel$1) &&
-        arr[arr.length - 2].match(ends_with_vowel$1);
-      let last_is_joining =
-        arr[arr.length - 1].match(joining_consonant_vowel) &&
-        ones.every(re => !arr[arr.length - 1].match(re));
-
-      if (second_to_last_is_open && last_is_joining) {
-        let possible_combination = arr[arr.length - 2] + arr[arr.length - 1];
-        let probably_separate_syllables =
-          possible_combination.match(cvcv_same_consonant) ||
-          possible_combination.match(cvcv_same_vowel) ||
-          possible_combination.match(cvcv_known_consonants);
-
-        if (!probably_separate_syllables) {
-          arr[arr.length - 2] = arr[arr.length - 2] + arr[arr.length - 1];
-          arr.splice(arr.length - 1, 1);
-        }
-      }
-    }
-
-    if (arr.length > 1) {
-      let single = arr[0] + arr[1];
-      if (single.match(starts_with_single_vowel_combos)) {
-        arr[0] = single;
-        arr.splice(1, 1);
-      }
-    }
-
-    if (arr.length > 1) {
-      if (arr[arr.length - 1].match(only_one_or_more_c)) {
-        arr[arr.length - 2] = arr[arr.length - 2] + arr[arr.length - 1];
-        arr.splice(arr.length - 1, 1);
-      }
-    }
-
-    return arr
-  };
-
-  //chop a string into pronounced syllables
-
-  const all_spaces = / +/g;
-  const ends_with_vowel = /[aeiouy]$/;
-  const starts_with_consonant_vowel = /^[^aeiouy]h?[aeiouy]/;
-  const starts_with_e_then_specials = /^e[sm]/;
-  const starts_with_e = /^e/;
-  const ends_with_noisy_vowel_combos = /(eo|eu|ia|oa|ua|ui)$/i;
-  const aiouy = /[aiouy]/;
-  const ends_with_ee = /ee$/;
-  // const whitespace_dash = /\s\-/
-
-  //method is nested because it's called recursively
-  const doWord = function (w) {
-    let all = [];
-    let chars = w.split('');
-    let before = '';
-    let after = '';
-    let current = '';
-    for (let i = 0; i < chars.length; i++) {
-      before = chars.slice(0, i).join('');
-      current = chars[i];
-      after = chars.slice(i + 1, chars.length).join('');
-      let candidate = before + chars[i];
-
-      //it's a consonant that comes after a vowel
-      if (before.match(ends_with_vowel) && !current.match(ends_with_vowel)) {
-        if (after.match(starts_with_e_then_specials)) {
-          candidate += 'e';
-          after = after.replace(starts_with_e, '');
-        }
-        all.push(candidate);
-        return all.concat(doWord(after))
-      }
-
-      //unblended vowels ('noisy' vowel combinations)
-      if (candidate.match(ends_with_noisy_vowel_combos)) {
-        //'io' is noisy, not in 'ion'
-        all.push(before);
-        all.push(current);
-        return all.concat(doWord(after)) //recursion
-      }
-
-      // if candidate is followed by a CV, assume consecutive open syllables
-      if (candidate.match(ends_with_vowel) && after.match(starts_with_consonant_vowel)) {
-        all.push(candidate);
-        return all.concat(doWord(after))
-      }
-    }
-    //if still running, end last syllable
-    if (w.match(aiouy) || w.match(ends_with_ee)) {
-      //allow silent trailing e
-      all.push(w);
-    } else if (w) {
-      let last = all.length - 1;
-      if (last < 0) {
-        last = 0;
-      }
-      all[last] = (all[last] || '') + w; //append it to the last one
-    }
-    return all
-  };
-
-  let syllables$1 = function (str) {
-    let all = [];
-    if (!str) {
-      return all
-    }
-    str = str.replace(/[.,?]/g, '');
-    str.split(all_spaces).map(s => {
-      all = all.concat(doWord(s));
-    });
-
-    // str.split(whitespace_dash).forEach(doWord)
-    all = postprocess(all);
-
-    //for words like 'tree' and 'free'
-    if (all.length === 0) {
-      all = [str];
-    }
-    //filter blanks
-    all = all.filter(s => s);
-
-    return all
-  };
-
-  // const defaultObj = { normal: true, text: true, terms: false }
-
-  const syllables = function (view) {
-    view.docs.forEach(terms => {
-      terms.forEach(term => {
-        term.syllables = syllables$1(term.normal || term.text);
-      });
-    });
-  };
-
-  var compute = {
-    soundsLike,
-    syllables
-  };
-
-  const api = function (View) {
-    /** */
-    View.prototype.syllables = function () {
-      this.compute('syllables');
-      let all = [];
-      this.docs.forEach(terms => {
-        let some = [];
-        terms.forEach(term => {
-          some = some.concat(term.syllables);
-        });
-        if (some.length > 0) {
-          all.push(some);
-        }
-      });
-      return all
-    };
-    /** */
-    View.prototype.soundsLike = function () {
-      this.compute('soundsLike');
-      let all = [];
-      this.docs.forEach(terms => {
-        let some = [];
-        terms.forEach(term => {
-          some = some.concat(term.soundsLike);
-        });
-        if (some.length > 0) {
-          all.push(some);
-        }
-      });
-      return all
-    };
-  };
-
-  var plugin = {
-    api,
-    compute
-  };
-
-  return plugin;
-
-}));

package/builds/compromise-stream.min.js

@@ -1 +0,0 @@
-var e,t;e=this,t=function(){const e=e=>e.replace(/([^c])\1/g,"$1"),t=e=>e.match(/^(kn|gn|pn|ae|wr)/)?e.substr(1,e.length-1):e,a=e=>e.replace(/mb$/,"m"),l=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"),c=e=>(e=e.replace(/d(ge|gy|gi)/g,"j$1")).replace(/d/g,"t"),i=e=>(e=e.replace(/gh(^$|[^aeiou])/g,"h$1")).replace(/g(n|ned)$/g,"$1"),n=e=>(e=(e=(e=e.replace(/gh/g,"f")).replace(/([^g]|^)(g)([iey])/g,"$1j$3")).replace(/gg/g,"g")).replace(/g/g,"k"),o=e=>e.replace(/([aeiou])h([^aeiou]|$)/g,"$1$2"),h=e=>e.replace(/ck/g,"k"),r=e=>e.replace(/ph/g,"f"),g=e=>e.replace(/q/g,"k"),u=e=>e.replace(/s(h|io|ia)/g,"x$1"),p=e=>e=e.replace(/t(ia[^n]|io)/g,"x$1"),s=e=>e.replace(/tch/g,"ch"),f=e=>e.replace(/v/g,"f"),m=e=>e.replace(/^wh/,"w"),y=e=>e.replace(/w([^aeiou]|$)/g,"$1"),$=e=>(e=e.replace(/^x/,"s")).replace(/x/g,"ks"),d=e=>e.replace(/y([^aeiou]|$)/g,"$1"),k=e=>e.replace(/z/,"s"),b=e=>e,x=/^(eu)/i,E=/^[^aeiou]e([^d]|$)/,j=/^([^aeiouy])[aeiouy]\1[aeiouy]/,w=/^[^aeiouy]([aeiouy])[^aeiouy]\1/,L=/^([tg][aeiouy]){2}/,v=/^[^aeiouy]+$/,T=/[aeiouy]$/,q=/^[^aeiouy]h?[aeiouy]/,z=[/^[^aeiou]?ion/,/^[^aeiou]?ised/,/^[^aeiou]?iled/,/[aeiou]n[gt]$/,/\wa[gt]e$/],S=/ +/g,A=/[aeiouy]$/,B=/^[^aeiouy]h?[aeiouy]/,C=/^e[sm]/,D=/^e/,F=/(eo|eu|ia|oa|ua|ui)$/i,G=/[aiouy]/,H=/ee$/,I=function(e){let t=[],a=e.split(""),l="",c="",i="";for(let e=0;e<a.length;e++){l=a.slice(0,e).join(""),i=a[e],c=a.slice(e+1,a.length).join("");let n=l+a[e];if(l.match(A)&&!i.match(A))return c.match(C)&&(n+="e",c=c.replace(D,"")),t.push(n),t.concat(I(c));if(n.match(F))return t.push(l),t.push(i),t.concat(I(c));if(n.match(A)&&c.match(B))return t.push(n),t.concat(I(c))}if(e.match(G)||e.match(H))t.push(e);else if(e){let a=t.length-1;a<0&&(a=0),t[a]=(t[a]||"")+e}return t};let J=function(e){let t=[];return e?((e=e.replace(/[.,?]/g,"")).split(S).map(e=>{t=t.concat(I(e))}),t=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 l=0;l<z.length;l++)a.match(z[l])&&(e[t-2]=e[t-2]+e[t-1],e.pop())}if(e.length>1){let t=(1===e[0].length||e[0].match(q))&&e[0].match(T),a=e[1].match(E);if(t&&a){let t=e[0]+e[1];t.match(j)||t.match(w)||t.match(L)||(e[0]=e[0]+e[1],e.splice(1,1))}}if(e.length>1){let t=e[e.length-2].match(q)&&e[e.length-2].match(T),a=e[e.length-1].match(E)&&z.every(t=>!e[e.length-1].match(t));if(t&&a){let t=e[e.length-2]+e[e.length-1];t.match(j)||t.match(w)||t.match(L)||(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(x)&&(e[0]=t,e.splice(1,1))}return e.length>1&&e[e.length-1].match(v)&&(e[e.length-2]=e[e.length-2]+e[e.length-1],e.splice(e.length-1,1)),e}(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(x){x.docs.forEach(x=>{x.forEach(x=>{var E;x.soundsLike=(E=x.normal||x.text,E=e(E),E=t(E),E=a(E),E=h(E),E=l(E),E=c(E),E=i(E),E=n(E),E=o(E),E=r(E),E=g(E),E=u(E),E=$(E),E=p(E),E=s(E),E=f(E),E=m(E),E=y(E),E=d(E),E=k(E),(E=b(E)).trim())})})},syllables:function(e){e.docs.forEach(e=>{e.forEach(e=>{e.syllables=J(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).compromiseStream=t();

package/builds/compromise-stream.mjs

@@ -1 +0,0 @@
-const e=e=>e.replace(/([^c])\1/g,"$1"),t=e=>e.match(/^(kn|gn|pn|ae|wr)/)?e.substr(1,e.length-1):e,a=e=>e.replace(/mb$/,"m"),c=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"),l=e=>(e=e.replace(/d(ge|gy|gi)/g,"j$1")).replace(/d/g,"t"),i=e=>(e=e.replace(/gh(^$|[^aeiou])/g,"h$1")).replace(/g(n|ned)$/g,"$1"),h=e=>(e=(e=(e=e.replace(/gh/g,"f")).replace(/([^g]|^)(g)([iey])/g,"$1j$3")).replace(/gg/g,"g")).replace(/g/g,"k"),n=e=>e.replace(/([aeiou])h([^aeiou]|$)/g,"$1$2"),o=e=>e.replace(/ck/g,"k"),r=e=>e.replace(/ph/g,"f"),g=e=>e.replace(/q/g,"k"),u=e=>e.replace(/s(h|io|ia)/g,"x$1"),p=e=>e=e.replace(/t(ia[^n]|io)/g,"x$1"),s=e=>e.replace(/tch/g,"ch"),m=e=>e.replace(/v/g,"f"),f=e=>e.replace(/^wh/,"w"),$=e=>e.replace(/w([^aeiou]|$)/g,"$1"),y=e=>(e=e.replace(/^x/,"s")).replace(/x/g,"ks"),d=e=>e.replace(/y([^aeiou]|$)/g,"$1"),k=e=>e.replace(/z/,"s"),x=e=>e,E=/^(eu)/i,b=/^[^aeiou]e([^d]|$)/,w=/^([^aeiouy])[aeiouy]\1[aeiouy]/,L=/^[^aeiouy]([aeiouy])[^aeiouy]\1/,j=/^([tg][aeiouy]){2}/,v=/^[^aeiouy]+$/,q=/[aeiouy]$/,z=/^[^aeiouy]h?[aeiouy]/,A=[/^[^aeiou]?ion/,/^[^aeiou]?ised/,/^[^aeiou]?iled/,/[aeiou]n[gt]$/,/\wa[gt]e$/],B=/ +/g,C=/[aeiouy]$/,D=/^[^aeiouy]h?[aeiouy]/,F=/^e[sm]/,G=/^e/,H=/(eo|eu|ia|oa|ua|ui)$/i,I=/[aiouy]/,J=/ee$/,K=function(e){let t=[],a=e.split(""),c="",l="",i="";for(let e=0;e<a.length;e++){c=a.slice(0,e).join(""),i=a[e],l=a.slice(e+1,a.length).join("");let h=c+a[e];if(c.match(C)&&!i.match(C))return l.match(F)&&(h+="e",l=l.replace(G,"")),t.push(h),t.concat(K(l));if(h.match(H))return t.push(c),t.push(i),t.concat(K(l));if(h.match(C)&&l.match(D))return t.push(h),t.concat(K(l))}if(e.match(I)||e.match(J))t.push(e);else if(e){let a=t.length-1;a<0&&(a=0),t[a]=(t[a]||"")+e}return t};let M=function(e){let t=[];return e?((e=e.replace(/[.,?]/g,"")).split(B).map(e=>{t=t.concat(K(e))}),t=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<A.length;c++)a.match(A[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(z))&&e[0].match(q),a=e[1].match(b);if(t&&a){let t=e[0]+e[1];t.match(w)||t.match(L)||t.match(j)||(e[0]=e[0]+e[1],e.splice(1,1))}}if(e.length>1){let t=e[e.length-2].match(z)&&e[e.length-2].match(q),a=e[e.length-1].match(b)&&A.every(t=>!e[e.length-1].match(t));if(t&&a){let t=e[e.length-2]+e[e.length-1];t.match(w)||t.match(L)||t.match(j)||(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(E)&&(e[0]=t,e.splice(1,1))}return e.length>1&&e[e.length-1].match(v)&&(e[e.length-2]=e[e.length-2]+e[e.length-1],e.splice(e.length-1,1)),e}(t),0===t.length&&(t=[e]),t=t.filter(e=>e),t):t};var N={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=>{var b;E.soundsLike=(b=E.normal||E.text,b=e(b),b=t(b),b=a(b),b=o(b),b=c(b),b=l(b),b=i(b),b=h(b),b=n(b),b=r(b),b=g(b),b=u(b),b=y(b),b=p(b),b=s(b),b=m(b),b=f(b),b=$(b),b=d(b),b=k(b),(b=x(b)).trim())})})},syllables:function(e){e.docs.forEach(e=>{e.forEach(e=>{e.syllables=M(e.normal||e.text)})})}}};export{N as default};