Fixed search indexing of multi-word tags

This commit is contained in:
squidfunk 2022-06-08 20:06:51 +02:00
parent ff9e41f329
commit cfe83ccbba
4 changed files with 9 additions and 6 deletions

View File

@ -43,6 +43,6 @@
* Copyright(c) 2015 Andreas Lubbe
* Copyright(c) 2015 Tiancheng "Timothy" Gu
* MIT Licensed
*/var Le=/["'&<>]/;ne.exports=we;function we(t){var e=""+t,r=Le.exec(e);if(!r)return e;var n,i="",s=0,o=0;for(s=r.index;s<e.length;s++){switch(e.charCodeAt(s)){case 34:n="&quot;";break;case 38:n="&amp;";break;case 39:n="&#39;";break;case 60:n="&lt;";break;case 62:n="&gt;";break;default:continue}o!==s&&(i+=e.substring(o,s)),o=s+1,i+=n}return o!==s?i+e.substring(o,s):i}});var le=H(re());Object.entries||(Object.entries=function(t){let e=[];for(let r of Object.keys(t))e.push([r,t[r]]);return e});Object.values||(Object.values=function(t){let e=[];for(let r of Object.keys(t))e.push(t[r]);return e});typeof Element!="undefined"&&(Element.prototype.scrollTo||(Element.prototype.scrollTo=function(t,e){typeof t=="object"?(this.scrollLeft=t.left,this.scrollTop=t.top):(this.scrollLeft=t,this.scrollTop=e)}),Element.prototype.replaceWith||(Element.prototype.replaceWith=function(...t){let e=this.parentNode;if(e){t.length===0&&e.removeChild(this);for(let r=t.length-1;r>=0;r--){let n=t[r];typeof n!="object"?n=document.createTextNode(n):n.parentNode&&n.parentNode.removeChild(n),r?e.insertBefore(this.previousSibling,n):e.replaceChild(n,this)}}}));var ie=H(q());function se(t){let e=new Map,r=new Set;for(let n of t){let[i,s]=n.location.split("#"),o=n.location,a=n.title,u=n.tags,c=(0,ie.default)(n.text).replace(/\s+(?=[,.:;!?])/g,"").replace(/\s+/g," ");if(s){let h=e.get(i);r.has(h)?e.set(o,{location:o,title:a,text:c,parent:h}):(h.title=n.title,h.text=c,r.add(h))}else e.set(o,M({location:o,title:a,text:c},u&&{tags:u}))}return e}var oe=H(q());function ae(t,e){let r=new RegExp(t.separator,"img"),n=(i,s,o)=>`${s}<mark data-md-highlight>${o}</mark>`;return i=>{i=i.replace(/[\s*+\-:~^]+/g," ").trim();let s=new RegExp(`(^|${t.separator})(${i.replace(/[|\\{}()[\]^$+*?.-]/g,"\\$&").replace(r,"|")})`,"img");return o=>(e?(0,oe.default)(o):o).replace(s,n).replace(/<\/mark>(\s+)<mark[^>]*>/img,"$1")}}function ue(t){let e=new lunr.Query(["title","text"]);return new lunr.QueryParser(t,e).parse(),e.clauses}function ce(t,e){var i;let r=new Set(t),n={};for(let s=0;s<e.length;s++)for(let o of r)e[s].startsWith(o.term)&&(n[o.term]=!0,r.delete(o));for(let s of r)(i=lunr.stopWordFilter)!=null&&i.call(lunr,s.term)&&(n[s.term]=!1);return n}function Ee(t,e){let[r,n]=[new Set(t),new Set(e)];return[...new Set([...r].filter(i=>!n.has(i)))]}var U=class{constructor({config:e,docs:r,options:n}){this.options=n,this.documents=se(r),this.highlight=ae(e,!1),lunr.tokenizer.separator=new RegExp(e.separator),this.index=lunr(function(){e.lang.length===1&&e.lang[0]!=="en"?this.use(lunr[e.lang[0]]):e.lang.length>1&&this.use(lunr.multiLanguage(...e.lang));let i=Ee(["trimmer","stopWordFilter","stemmer"],n.pipeline);for(let s of e.lang.map(o=>o==="en"?lunr:lunr[o]))for(let o of i)this.pipeline.remove(s[o]),this.searchPipeline.remove(s[o]);this.ref("location"),this.field("title",{boost:1e3}),this.field("text"),this.field("tags",{boost:1e6});for(let s of r)this.add(s,{boost:s.boost})})}search(e){if(e)try{let r=this.highlight(e),n=ue(e).filter(o=>o.presence!==lunr.Query.presence.PROHIBITED),i=this.index.search(`${e}*`).reduce((o,{ref:a,score:u,matchData:c})=>{let h=this.documents.get(a);if(typeof h!="undefined"){let{location:y,title:g,text:b,tags:m,parent:Q}=h,p=ce(n,Object.keys(c.metadata)),d=+!Q+ +Object.values(p).every(w=>w);o.push(Z(M({location:y,title:r(g),text:r(b)},m&&{tags:m.map(r)}),{score:u*(1+d),terms:p}))}return o},[]).sort((o,a)=>a.score-o.score).reduce((o,a)=>{let u=this.documents.get(a.location);if(typeof u!="undefined"){let c="parent"in u?u.parent.location:u.location;o.set(c,[...o.get(c)||[],a])}return o},new Map),s;if(this.options.suggestions){let o=this.index.query(a=>{for(let u of n)a.term(u.term,{fields:["title"],presence:lunr.Query.presence.REQUIRED,wildcard:lunr.Query.wildcard.TRAILING})});s=o.length?Object.keys(o[0].matchData.metadata):[]}return M({items:[...i.values()]},typeof s!="undefined"&&{suggestions:s})}catch(r){console.warn(`Invalid query: ${e} \u2013 see https://bit.ly/2s3ChXG`)}return{items:[]}}};var Y;function ke(t){return z(this,null,function*(){let e="../lunr";if(typeof parent!="undefined"&&"IFrameWorker"in parent){let n=document.querySelector("script[src]"),[i]=n.src.split("/worker");e=e.replace("..",i)}let r=[];for(let n of t.lang){switch(n){case"ja":r.push(`${e}/tinyseg.js`);break;case"hi":case"th":r.push(`${e}/wordcut.js`);break}n!=="en"&&r.push(`${e}/min/lunr.${n}.min.js`)}t.lang.length>1&&r.push(`${e}/min/lunr.multi.min.js`),r.length&&(yield importScripts(`${e}/min/lunr.stemmer.support.min.js`,...r))})}function Te(t){return z(this,null,function*(){switch(t.type){case 0:return yield ke(t.data.config),Y=new U(t.data),{type:1};case 2:return{type:3,data:Y?Y.search(t.data):{items:[]}};default:throw new TypeError("Invalid message type")}})}self.lunr=le.default;addEventListener("message",t=>z(void 0,null,function*(){postMessage(yield Te(t.data))}));})();
//# sourceMappingURL=search.b028fd86.min.js.map
*/var Le=/["'&<>]/;ne.exports=we;function we(t){var e=""+t,r=Le.exec(e);if(!r)return e;var n,i="",s=0,o=0;for(s=r.index;s<e.length;s++){switch(e.charCodeAt(s)){case 34:n="&quot;";break;case 38:n="&amp;";break;case 39:n="&#39;";break;case 60:n="&lt;";break;case 62:n="&gt;";break;default:continue}o!==s&&(i+=e.substring(o,s)),o=s+1,i+=n}return o!==s?i+e.substring(o,s):i}});var le=H(re());Object.entries||(Object.entries=function(t){let e=[];for(let r of Object.keys(t))e.push([r,t[r]]);return e});Object.values||(Object.values=function(t){let e=[];for(let r of Object.keys(t))e.push(t[r]);return e});typeof Element!="undefined"&&(Element.prototype.scrollTo||(Element.prototype.scrollTo=function(t,e){typeof t=="object"?(this.scrollLeft=t.left,this.scrollTop=t.top):(this.scrollLeft=t,this.scrollTop=e)}),Element.prototype.replaceWith||(Element.prototype.replaceWith=function(...t){let e=this.parentNode;if(e){t.length===0&&e.removeChild(this);for(let r=t.length-1;r>=0;r--){let n=t[r];typeof n!="object"?n=document.createTextNode(n):n.parentNode&&n.parentNode.removeChild(n),r?e.insertBefore(this.previousSibling,n):e.replaceChild(n,this)}}}));var ie=H(q());function se(t){let e=new Map,r=new Set;for(let n of t){let[i,s]=n.location.split("#"),o=n.location,a=n.title,u=n.tags,c=(0,ie.default)(n.text).replace(/\s+(?=[,.:;!?])/g,"").replace(/\s+/g," ");if(s){let h=e.get(i);r.has(h)?e.set(o,{location:o,title:a,text:c,parent:h}):(h.title=n.title,h.text=c,r.add(h))}else e.set(o,M({location:o,title:a,text:c},u&&{tags:u}))}return e}var oe=H(q());function ae(t,e){let r=new RegExp(t.separator,"img"),n=(i,s,o)=>`${s}<mark data-md-highlight>${o}</mark>`;return i=>{i=i.replace(/[\s*+\-:~^]+/g," ").trim();let s=new RegExp(`(^|${t.separator})(${i.replace(/[|\\{}()[\]^$+*?.-]/g,"\\$&").replace(r,"|")})`,"img");return o=>(e?(0,oe.default)(o):o).replace(s,n).replace(/<\/mark>(\s+)<mark[^>]*>/img,"$1")}}function ue(t){let e=new lunr.Query(["title","text"]);return new lunr.QueryParser(t,e).parse(),e.clauses}function ce(t,e){var i;let r=new Set(t),n={};for(let s=0;s<e.length;s++)for(let o of r)e[s].startsWith(o.term)&&(n[o.term]=!0,r.delete(o));for(let s of r)(i=lunr.stopWordFilter)!=null&&i.call(lunr,s.term)&&(n[s.term]=!1);return n}function Ee(t,e){let[r,n]=[new Set(t),new Set(e)];return[...new Set([...r].filter(i=>!n.has(i)))]}var U=class{constructor({config:e,docs:r,options:n}){this.options=n,this.documents=se(r),this.highlight=ae(e,!1),lunr.tokenizer.separator=new RegExp(e.separator),this.index=lunr(function(){e.lang.length===1&&e.lang[0]!=="en"?this.use(lunr[e.lang[0]]):e.lang.length>1&&this.use(lunr.multiLanguage(...e.lang));let i=Ee(["trimmer","stopWordFilter","stemmer"],n.pipeline);for(let s of e.lang.map(o=>o==="en"?lunr:lunr[o]))for(let o of i)this.pipeline.remove(s[o]),this.searchPipeline.remove(s[o]);this.ref("location"),this.field("title",{boost:1e3}),this.field("text"),this.field("tags",{boost:1e6,extractor:s=>{let{tags:o=[]}=s;return o.flatMap(a=>lunr.tokenizer(a))}});for(let s of r)this.add(s,{boost:s.boost})})}search(e){if(e)try{let r=this.highlight(e),n=ue(e).filter(o=>o.presence!==lunr.Query.presence.PROHIBITED),i=this.index.search(`${e}*`).reduce((o,{ref:a,score:u,matchData:c})=>{let h=this.documents.get(a);if(typeof h!="undefined"){let{location:y,title:g,text:b,tags:m,parent:Q}=h,p=ce(n,Object.keys(c.metadata)),d=+!Q+ +Object.values(p).every(w=>w);o.push(Z(M({location:y,title:r(g),text:r(b)},m&&{tags:m.map(r)}),{score:u*(1+d),terms:p}))}return o},[]).sort((o,a)=>a.score-o.score).reduce((o,a)=>{let u=this.documents.get(a.location);if(typeof u!="undefined"){let c="parent"in u?u.parent.location:u.location;o.set(c,[...o.get(c)||[],a])}return o},new Map),s;if(this.options.suggestions){let o=this.index.query(a=>{for(let u of n)a.term(u.term,{fields:["title"],presence:lunr.Query.presence.REQUIRED,wildcard:lunr.Query.wildcard.TRAILING})});s=o.length?Object.keys(o[0].matchData.metadata):[]}return M({items:[...i.values()]},typeof s!="undefined"&&{suggestions:s})}catch(r){console.warn(`Invalid query: ${e} \u2013 see https://bit.ly/2s3ChXG`)}return{items:[]}}};var Y;function ke(t){return z(this,null,function*(){let e="../lunr";if(typeof parent!="undefined"&&"IFrameWorker"in parent){let n=document.querySelector("script[src]"),[i]=n.src.split("/worker");e=e.replace("..",i)}let r=[];for(let n of t.lang){switch(n){case"ja":r.push(`${e}/tinyseg.js`);break;case"hi":case"th":r.push(`${e}/wordcut.js`);break}n!=="en"&&r.push(`${e}/min/lunr.${n}.min.js`)}t.lang.length>1&&r.push(`${e}/min/lunr.multi.min.js`),r.length&&(yield importScripts(`${e}/min/lunr.stemmer.support.min.js`,...r))})}function Te(t){return z(this,null,function*(){switch(t.type){case 0:return yield ke(t.data.config),Y=new U(t.data),{type:1};case 2:return{type:3,data:Y?Y.search(t.data):{items:[]}};default:throw new TypeError("Invalid message type")}})}self.lunr=le.default;addEventListener("message",t=>z(void 0,null,function*(){postMessage(yield Te(t.data))}));})();
//# sourceMappingURL=search.85cb4492.min.js.map

View File

@ -187,7 +187,7 @@
"base": base_url,
"features": features,
"translations": {},
"search": "assets/javascripts/workers/search.b028fd86.min.js" | url
"search": "assets/javascripts/workers/search.85cb4492.min.js" | url
} -%}
{%- if config.extra.version -%}
{%- set _ = app.update({ "version": config.extra.version }) -%}

View File

@ -204,7 +204,10 @@ export class Search {
/* Set up fields */
this.field("title", { boost: 1e3 })
this.field("text")
this.field("tags", { boost: 1e6 })
this.field("tags", { boost: 1e6, extractor: doc => {
const { tags = [] } = doc as SearchDocument
return tags.flatMap(tag => lunr.tokenizer(tag))
} })
/* Index documents */
for (const doc of docs)