gs-search 0.1.5 → 0.1.7

package/lib/browser.cjs

@@ -1 +1,68 @@
-"use strict";exports.BrowserStorage=class{#a;constructor(a){this.#a=a}async#t(){return await(await navigator.storage.getDirectory()).getDirectoryHandle(this.#a,{create:!0})}async write(a,t){const e=await(await(await this.#t()).getFileHandle(a,{create:!0})).createWritable();await e.write(t),await e.close()}async append(a,t){const e=await this.#t();let i;try{i=await e.getFileHandle(a,{create:!0})}catch{i=await e.getFileHandle(a,{create:!0})}const r=await i.getFile(),s=await i.createWritable({keepExistingData:!0});await s.seek(r.size),await s.write(t),await s.close()}async read(a){const t=await this.#t();try{return await(await(await t.getFileHandle(a)).getFile()).arrayBuffer()}catch{return null}}async readRange(a,t,e){const i=await this.#t();try{return await(await(await i.getFileHandle(a)).getFile()).slice(t,e).arrayBuffer()}catch{return null}}async remove(a){const t=await this.#t();try{await t.removeEntry(a)}catch{}}async listFiles(){const a=await this.#t(),t=[];for await(const e of a.keys())t.push(e);return t}async clearAll(){const a=await this.#t();for await(const t of a.keys())await a.removeEntry(t,{recursive:!0})}async getFileSize(a){const t=await this.#t();try{return(await(await t.getFileHandle(a)).getFile()).size}catch{return 0}}};
+"use strict";
+class BrowserStorage {
+  #baseDir;
+  constructor(baseDir) {
+    this.#baseDir = baseDir;
+  }
+  // 获取当前实例的专属目录句柄
+  async #getDirHandle() {
+    return await (await navigator.storage.getDirectory()).getDirectoryHandle(this.#baseDir, { create: !0 });
+  }
+  async write(filename, data) {
+    const writable = await (await (await this.#getDirHandle()).getFileHandle(filename, { create: !0 })).createWritable();
+    await writable.write(data), await writable.close();
+  }
+  async append(filename, data) {
+    const dir = await this.#getDirHandle();
+    let fileHandle;
+    try {
+      fileHandle = await dir.getFileHandle(filename, { create: !0 });
+    } catch {
+      fileHandle = await dir.getFileHandle(filename, { create: !0 });
+    }
+    const file = await fileHandle.getFile(), writable = await fileHandle.createWritable({ keepExistingData: !0 });
+    await writable.seek(file.size), await writable.write(data), await writable.close();
+  }
+  async read(filename) {
+    const dir = await this.#getDirHandle();
+    try {
+      return await (await (await dir.getFileHandle(filename)).getFile()).arrayBuffer();
+    } catch {
+      return null;
+    }
+  }
+  async readRange(filename, start, end) {
+    const dir = await this.#getDirHandle();
+    try {
+      return await (await (await dir.getFileHandle(filename)).getFile()).slice(start, end).arrayBuffer();
+    } catch {
+      return null;
+    }
+  }
+  async remove(filename) {
+    const dir = await this.#getDirHandle();
+    try {
+      await dir.removeEntry(filename);
+    } catch {
+    }
+  }
+  async listFiles() {
+    const dir = await this.#getDirHandle(), keys = [];
+    for await (const key of dir.keys()) keys.push(key);
+    return keys;
+  }
+  async clearAll() {
+    const dir = await this.#getDirHandle();
+    for await (const key of dir.keys())
+      await dir.removeEntry(key, { recursive: !0 });
+  }
+  async getFileSize(filename) {
+    const dir = await this.#getDirHandle();
+    try {
+      return (await (await dir.getFileHandle(filename)).getFile()).size;
+    } catch {
+      return 0;
+    }
+  }
+}
+exports.BrowserStorage = BrowserStorage;

package/lib/browser.js

@@ -1 +1,69 @@
-class a{#a;constructor(a){this.#a=a}async#t(){return await(await navigator.storage.getDirectory()).getDirectoryHandle(this.#a,{create:!0})}async write(a,t){const e=await(await(await this.#t()).getFileHandle(a,{create:!0})).createWritable();await e.write(t),await e.close()}async append(a,t){const e=await this.#t();let i;try{i=await e.getFileHandle(a,{create:!0})}catch{i=await e.getFileHandle(a,{create:!0})}const r=await i.getFile(),n=await i.createWritable({keepExistingData:!0});await n.seek(r.size),await n.write(t),await n.close()}async read(a){const t=await this.#t();try{return await(await(await t.getFileHandle(a)).getFile()).arrayBuffer()}catch{return null}}async readRange(a,t,e){const i=await this.#t();try{return await(await(await i.getFileHandle(a)).getFile()).slice(t,e).arrayBuffer()}catch{return null}}async remove(a){const t=await this.#t();try{await t.removeEntry(a)}catch{}}async listFiles(){const a=await this.#t(),t=[];for await(const e of a.keys())t.push(e);return t}async clearAll(){const a=await this.#t();for await(const t of a.keys())await a.removeEntry(t,{recursive:!0})}async getFileSize(a){const t=await this.#t();try{return(await(await t.getFileHandle(a)).getFile()).size}catch{return 0}}}export{a as BrowserStorage};
+class BrowserStorage {
+  #baseDir;
+  constructor(baseDir) {
+    this.#baseDir = baseDir;
+  }
+  // 获取当前实例的专属目录句柄
+  async #getDirHandle() {
+    return await (await navigator.storage.getDirectory()).getDirectoryHandle(this.#baseDir, { create: !0 });
+  }
+  async write(filename, data) {
+    const writable = await (await (await this.#getDirHandle()).getFileHandle(filename, { create: !0 })).createWritable();
+    await writable.write(data), await writable.close();
+  }
+  async append(filename, data) {
+    const dir = await this.#getDirHandle();
+    let fileHandle;
+    try {
+      fileHandle = await dir.getFileHandle(filename, { create: !0 });
+    } catch {
+      fileHandle = await dir.getFileHandle(filename, { create: !0 });
+    }
+    const file = await fileHandle.getFile(), writable = await fileHandle.createWritable({ keepExistingData: !0 });
+    await writable.seek(file.size), await writable.write(data), await writable.close();
+  }
+  async read(filename) {
+    const dir = await this.#getDirHandle();
+    try {
+      return await (await (await dir.getFileHandle(filename)).getFile()).arrayBuffer();
+    } catch {
+      return null;
+    }
+  }
+  async readRange(filename, start, end) {
+    const dir = await this.#getDirHandle();
+    try {
+      return await (await (await dir.getFileHandle(filename)).getFile()).slice(start, end).arrayBuffer();
+    } catch {
+      return null;
+    }
+  }
+  async remove(filename) {
+    const dir = await this.#getDirHandle();
+    try {
+      await dir.removeEntry(filename);
+    } catch {
+    }
+  }
+  async listFiles() {
+    const dir = await this.#getDirHandle(), keys = [];
+    for await (const key of dir.keys()) keys.push(key);
+    return keys;
+  }
+  async clearAll() {
+    const dir = await this.#getDirHandle();
+    for await (const key of dir.keys())
+      await dir.removeEntry(key, { recursive: !0 });
+  }
+  async getFileSize(filename) {
+    const dir = await this.#getDirHandle();
+    try {
+      return (await (await dir.getFileHandle(filename)).getFile()).size;
+    } catch {
+      return 0;
+    }
+  }
+}
+export {
+  BrowserStorage
+};

package/lib/core.cjs

@@ -1 +1,504 @@
-"use strict";const e="search_meta.json",t="deleted_ids.bin",s="added_ids.bin";class n{#e;#t={wordSegments:[],charSegments:[]};#s=new Set;#n=new Set;constructor(e){this.#e=e}async load(){const n=await this.#e.read(e);if(n){const e=(new TextDecoder).decode(n);this.#t=JSON.parse(e)}else this.#t={wordSegments:[],charSegments:[]};const i=await this.#e.read(t);if(i){const e=new DataView(i);let t=0;const s=i.byteLength;for(;t<s&&!(t+4>s);){const n=e.getUint32(t,!0);this.#s.add(n),t+=4,t<s&&30===e.getUint8(t)&&(t+=1)}}const a=await this.#e.read(s);if(a){const e=new DataView(a);let t=0;const s=a.byteLength;for(;t<s&&!(t+4>s);){const n=e.getUint32(t,!0);this.#n.add(n),t+=4,t<s&&30===e.getUint8(t)&&(t+=1)}}}async save(){const n=JSON.stringify(this.#t);if(await this.#e.write(e,(new TextEncoder).encode(n).buffer),0===this.#s.size)await this.#e.remove(t);else{const e=4*this.#s.size+this.#s.size,s=new ArrayBuffer(e),n=new DataView(s);let i=0;for(const e of this.#s)n.setUint32(i,e,!0),i+=4,n.setUint8(i,30),i+=1;await this.#e.write(t,s)}if(0===this.#n.size)await this.#e.remove(s);else{const e=4*this.#n.size+this.#n.size,t=new ArrayBuffer(e),n=new DataView(t);let i=0;for(const e of this.#n)n.setUint32(i,e,!0),i+=4,n.setUint8(i,30),i+=1;await this.#e.write(s,t)}}getSegments(e){return"word"===e?this.#t.wordSegments:this.#t.charSegments}getDeletedIds(){return this.#s}addDeletedId(e){this.#s.add(e)}isDeleted(e){return this.#s.has(e)}addAddedId(e){this.#n.add(e)}removeAddedId(e){this.#n.delete(e)}isAdded(e){return this.#n.has(e)}getAddedIds(){return this.#n}hasDocument(e){return this.#n.has(e)||this.#s.has(e)}getLastSegmentInfo(e){const t=this.getSegments(e);return 0===t.length?null:t[t.length-1]}updateSegment(e,t,s,n,i,a){const o="word"===e?this.#t.wordSegments:this.#t.charSegments;if(a)o.push({filename:t,start:s,end:n,tokenCount:i});else{const e=o[o.length-1];e&&e.filename===t&&(e.end=n,e.tokenCount=i)}}reset(){this.#t={wordSegments:[],charSegments:[]},this.#s.clear(),this.#n.clear()}}class i{static SEPARATOR=30;#e;constructor(e){this.#e=e}async appendBatch(e,t){if(0===t.length)return await this.#e.getFileSize(e);const s=new TextEncoder;let n=0;for(const e of t){n+=8;for(const t of e.tokens){n+=2+Math.min(s.encode(t).byteLength,65535)}n+=1}const a=new Uint8Array(n);let o=0;for(const e of t){const t=[];for(const n of e.tokens){const e=s.encode(n),i=e.byteLength>65535?e.slice(0,65535):e;t.push(i)}const n=new DataView(a.buffer,o);n.setUint32(0,e.id,!0),n.setUint32(4,t.length,!0),o+=8;for(const e of t)new DataView(a.buffer,o).setUint16(0,e.byteLength,!0),o+=2,a.set(e,o),o+=e.byteLength;a[o++]=i.SEPARATOR}return await this.#e.append(e,a.buffer),await this.#e.getFileSize(e)}async readRange(e,t,s){const n=await this.#e.readRange(e,t,s);if(!n||0===n.byteLength)return[];const a=new DataView(n),o=new Uint8Array(n),r=new TextDecoder,h=[];let d=0;const c=n.byteLength;for(;d<c&&!(d+8>c);){const e=a.getUint32(d,!0);d+=4;const t=a.getUint32(d,!0);d+=4;const s=[];for(let e=0;e<t&&!(d+2>c);e++){const e=a.getUint16(d,!0);if(d+=2,d+e>c)break;const t=new Uint8Array(n,d,e);s.push(r.decode(t)),d+=e}d<c&&o[d]===i.SEPARATOR&&(d+=1),h.push({id:e,tokens:s})}return h}async getCurrentSize(e){return await this.#e.getFileSize(e)}}function a(e,t=305419896){const s=e.length,n=s>>2;let i=0;for(;i<n;){let s=255&e.charCodeAt(i)|(255&e.charCodeAt(++i))<<8|(255&e.charCodeAt(++i))<<16|(255&e.charCodeAt(++i))<<24;++i,s=3432918353*(65535&s)+((3432918353*(s>>>16)&65535)<<16)&4294967295,s=s<<15|s>>>17,s=461845907*(65535&s)+((461845907*(s>>>16)&65535)<<16)&4294967295,t=27492+(65535&(t=5*(65535&(t=(t^=s)<<13|t>>>19))+((5*(t>>>16)&65535)<<16)&4294967295))+(((t>>>16)+58964&65535)<<16)}let a=0;const o=3&s;return o>0&&(o>=3&&(a^=(255&e.charCodeAt(i+2))<<16),o>=2&&(a^=(255&e.charCodeAt(i+1))<<8),o>=1&&(a^=255&e.charCodeAt(i)),a=3432918353*(65535&a)+((3432918353*(a>>>16)&65535)<<16)&4294967295,a=a<<15|a>>>17,a=461845907*(65535&a)+((461845907*(a>>>16)&65535)<<16)&4294967295,t^=a),t^=s,t=2246822507*(65535&(t^=t>>>16))+((2246822507*(t>>>16)&65535)<<16)&4294967295,t=3266489909*(65535&(t^=t>>>13))+((3266489909*(t>>>16)&65535)<<16)&4294967295,(t^=t>>>16)>>>0}class o{#i;#e;#a=null;#o=null;static hash(e){return a(e)}constructor(e,t){this.#i=e,this.#e=t}async loadIndex(){return!!this.#a||(this.#a=await this.#e.read(this.#i),!!this.#a&&(this.#o=new DataView(this.#a),!0))}async buildAndSave(e){const t=new Map;for(const s of e){const e=new Map;for(const n of s.tokens)e.has(n)||(e.set(n,!0),t.has(n)||t.set(n,{hash:o.hash(n),postings:[]}),t.get(n).postings.push(s.id))}const s=Array.from(t.entries());s.sort(([e,{hash:t}],[s,{hash:n}])=>t!==n?t-n:e.localeCompare(s));let n=0,i=0;for(const[e,{postings:t}]of s)n+=t.length,i+=e.length+1;const a=20*s.length,r=12+a+4*n,h=new ArrayBuffer(r+i),d=new DataView(h);d.setUint32(0,1229866072),d.setUint32(4,s.length),d.setUint32(8,r);let c=12,g=12+a,f=r;for(const[e,{hash:t,postings:n}]of s){d.setUint32(c,t),d.setUint32(c+4,e.length),d.setUint32(c+8,f),d.setUint32(c+12,g),d.setUint32(c+16,n.length),c+=20;for(let e=0;e<n.length;e++)d.setUint32(g,n[e],!0),g+=4;const s=(new TextEncoder).encode(e);for(let e=0;e<s.length;e++)d.setUint8(f++,s[e]);d.setUint8(f++,0)}await this.#e.write(this.#i,h),this.#a=h,this.#o=d}search(e){if(!this.#o||!this.#a)return[];const t=o.hash(e),s=this.#o.getUint32(4);let n=0,i=s-1;const a=12,r=20,h=new TextDecoder;for(;n<=i;){const o=n+i>>>1,d=a+o*r,c=this.#o.getUint32(d);if(c<t)n=o+1;else{if(!(c>t)){if(!(o>0&&this.#o.getUint32(a+(o-1)*r)===t||o<s-1&&this.#o.getUint32(a+(o+1)*r)===t)){const e=this.#o.getUint32(a+o*r+12),t=this.#o.getUint32(a+o*r+16),s=[];for(let n=0;n<t;n++)s.push(this.#o.getUint32(e+4*n,!0));return s}let n=o;for(;n>0;){const e=a+(n-1)*r;if(this.#o.getUint32(e)!==t)break;n--}for(let i=n;i<s;i++){const s=a+i*r;if(this.#o.getUint32(s)!==t)break;const n=this.#o.getUint32(s+4),o=this.#o.getUint32(s+8),d=new Uint8Array(this.#a,o,n);if(h.decode(d)===e){const e=this.#o.getUint32(s+12),t=this.#o.getUint32(s+16),n=[];for(let s=0;s<t;s++)n.push(this.#o.getUint32(e+4*s,!0));return n}}return[]}i=o-1}}return[]}}const r=({text:e})=>{try{if(typeof Intl<"u"&&"function"==typeof Intl.Segmenter&&"function"==typeof Array.from){const t=new Intl.Segmenter([],{granularity:"word"}).segment(e);if("object"==typeof t&&null!==t)return Array.from(t).filter(e=>e?.isWordLike).map(e=>e?.segment?.toLowerCase()||"")}}catch{}return e.toLowerCase().split(/[^a-z0-9\u4e00-\u9fa5]+/g).filter(e=>e.length>0)},h="word_cache.bin",d="char_cache.bin";exports.SearchEngine=class{#e;#t;#r;#h;#d=!1;#c;#g=!1;#f={word:0,char:0};constructor(e){if(this.#c={wordSegmentTokenThreshold:1e5,charSegmentTokenThreshold:5e5,minWordTokenSave:0,minCharTokenSave:0,indexingTokenizer:e.indexingTokenizer||r,...e},(this.#c.minWordTokenSave||0)>=(this.#c.wordSegmentTokenThreshold||1e5))throw new Error("minWordTokenSave must be less than wordSegmentTokenThreshold");if((this.#c.minCharTokenSave||0)>=(this.#c.charSegmentTokenThreshold||5e5))throw new Error("minCharTokenSave must be less than charSegmentTokenThreshold");this.#e=e.storage,this.#t=new n(this.#e),this.#r=new i(this.#e),this.#h=new Map}startBatch(){this.#g=!0,this.#f={word:0,char:0}}async endBatch(){this.#g=!1,this.#f.word>0&&await this.#l("word",this.#f.word),this.#f.char>0&&await this.#l("char",this.#f.char),this.#f={word:0,char:0},await this.#t.save()}async addDocument(e){return this.addDocuments([e])}async addDocumentIfMissing(e){return this.addDocumentsIfMissing([e])}async addDocumentsIfMissing(e){if(this.#d||await this.#w(),0===e.length)return;const t=this.#t.getDeletedIds(),s=[],n=[],i=[];for(const a of e){if(t.has(a.id)||this.#t.isAdded(a.id))continue;const e=this.#m(a),o=[],r=[];for(const t of e)t.length>1?o.push(t):1===t.length&&r.push(t);o.length>0&&s.push({id:a.id,tokens:o}),r.length>0&&n.push({id:a.id,tokens:r}),i.push(a)}if(0===i.length)return;let a=0,o=0;if(s.length>0){await this.#r.appendBatch(h,s);for(const e of s)a+=e.tokens.length}if(n.length>0){await this.#r.appendBatch(d,n);for(const e of n)o+=e.tokens.length}for(const e of i)this.#t.addAddedId(e.id);this.#g?(this.#f.word+=a,this.#f.char+=o):(a>0&&await this.#l("word",a),o>0&&await this.#l("char",o),await this.#t.save())}async addDocuments(e){if(this.#d||await this.#w(),0===e.length)return;const t=this.#t.getDeletedIds(),s=[],n=[];for(const i of e){if(t.has(i.id))throw new Error(`Document ID ${i.id} has been deleted and cannot be re-added.`);if(this.#t.isAdded(i.id))throw new Error(`Document ID ${i.id} already exists.`);const e=this.#m(i),a=[],o=[];for(const t of e)t.length>1?a.push(t):1===t.length&&o.push(t);a.length>0&&s.push({id:i.id,tokens:a}),o.length>0&&n.push({id:i.id,tokens:o})}let i=0,a=0;if(s.length>0){await this.#r.appendBatch(h,s);for(const e of s)i+=e.tokens.length}if(n.length>0){await this.#r.appendBatch(d,n);for(const e of n)a+=e.tokens.length}for(const t of e)this.#t.addAddedId(t.id);this.#g?(this.#f.word+=i,this.#f.char+=a):(i>0&&await this.#l("word",i),a>0&&await this.#l("char",a),await this.#t.save())}async search(e,t){this.#d||await this.#w();const s="string"==typeof e?{text:e}:e,n=this.#u(s),i=n.filter(e=>e.length>1),a=n.filter(e=>1===e.length),r=this.#t.getDeletedIds(),h=new Map,d=new Map,c=e=>{const t=this.#t.getSegments(e);for(const e of t){const t=e.filename;!this.#h.has(t)&&!d.has(t)&&d.set(t,new o(t,this.#e))}};c("word"),c("char"),await Promise.all(Array.from(d.entries()).map(([e,t])=>t.loadIndex().then(s=>{s&&this.#h.set(e,t)})));const g=async(e,t)=>{if(0===t.length)return;const s=this.#t.getSegments(e);for(const e of s){const s=e.filename,n=this.#h.get(s);if(n)for(const e of t){const t=n.search(e),s=1+.1*e.length;for(const n of t)if(!r.has(n))if(h.has(n)){const t=h.get(n);t.score+=s,t.tokens.add(e)}else h.set(n,{score:0,tokens:new Set([e])})}}};await g("word",i),await g("char",a);const f=[];return h.forEach((e,t)=>{f.push({id:t,score:e.score,tokens:Array.from(e.tokens)})}),f.sort((e,t)=>t.score-e.score),"number"==typeof t&&t>0?f.slice(0,t):f}async removeDocument(e){this.#d||await this.#w(),this.#t.addDeletedId(e),this.#t.removeAddedId(e),await this.#t.save()}async clearAll(){await this.#e.clearAll(),this.#h.clear(),this.#t.reset(),this.#d=!1,this.#g=!1,this.#f={word:0,char:0}}async getStatus(){return this.#d||await this.#w(),{wordSegments:this.#t.getSegments("word").length,charSegments:this.#t.getSegments("char").length,deleted:this.#t.getDeletedIds().size,wordCacheSize:await this.#r.getCurrentSize(h),charCacheSize:await this.#r.getCurrentSize(d),inBatch:this.#g}}async hasDocument(e){return this.#d||await this.#w(),this.#t.hasDocument(e)}async#w(){if(this.#d)return;await this.#t.load();const e=[...this.#t.getSegments("word"),...this.#t.getSegments("char")];for(const t of e)this.#h.has(t.filename)||this.#h.set(t.filename,new o(t.filename,this.#e)),await this.#h.get(t.filename).loadIndex();this.#d=!0}#m(e){return this.#c.indexingTokenizer(e)}#u(e){return this.#c.searchTokenizer?this.#c.searchTokenizer(e):this.#m(e)}async#l(e,t){const s="word"===e?h:d,n=await this.#r.getCurrentSize(s),i="word"===e?this.#c.wordSegmentTokenThreshold||1e5:this.#c.charSegmentTokenThreshold||5e5,a="word"===e?this.#c.minWordTokenSave||0:this.#c.minCharTokenSave||0,r=this.#t.getLastSegmentInfo(e);let c,g,f,l;const w=()=>{const t=this.#t.getSegments(e).length+1;return`${e}_seg_${t}.bin`};if(r){const e=r.tokenCount;e>=i||e+t>=i?(c=w(),f=!0,g=r.end,l=t):(c=r.filename,f=!1,g=r.start,l=e+t)}else c=w(),f=!0,g=0,l=t;if(l<a)return void this.#t.updateSegment(e,c,g,n,l,f);const m=await this.#r.readRange(s,g,n);let u=this.#h.get(c);u||(u=new o(c,this.#e),this.#h.set(c,u)),await u.buildAndSave(m),this.#t.updateSegment(e,c,g,n,l,f)}},exports.hash=a,exports.murmur3_32=a;
+"use strict";
+const META_FILE = "search_meta.json", DELETED_IDS_FILE = "deleted_ids.bin", ADDED_IDS_FILE = "added_ids.bin", SEPARATOR = 30;
+class MetaManager {
+  #storage;
+  #meta = {
+    wordSegments: [],
+    charSegments: []
+  };
+  #deletedIds = /* @__PURE__ */ new Set();
+  #addedIds = /* @__PURE__ */ new Set();
+  constructor(storage) {
+    this.#storage = storage;
+  }
+  async load() {
+    const buffer = await this.#storage.read(META_FILE);
+    if (buffer) {
+      const json = new TextDecoder().decode(buffer);
+      this.#meta = JSON.parse(json);
+    } else
+      this.#meta = { wordSegments: [], charSegments: [] };
+    const deletedBuffer = await this.#storage.read(DELETED_IDS_FILE);
+    if (deletedBuffer) {
+      const view = new DataView(deletedBuffer);
+      let offset = 0;
+      const max = deletedBuffer.byteLength;
+      for (; offset < max && !(offset + 4 > max); ) {
+        const id = view.getUint32(offset, !0);
+        this.#deletedIds.add(id), offset += 4, offset < max && view.getUint8(offset) === 30 && (offset += 1);
+      }
+    }
+    const addedBuffer = await this.#storage.read(ADDED_IDS_FILE);
+    if (addedBuffer) {
+      const view = new DataView(addedBuffer);
+      let offset = 0;
+      const max = addedBuffer.byteLength;
+      for (; offset < max && !(offset + 4 > max); ) {
+        const id = view.getUint32(offset, !0);
+        this.#addedIds.add(id), offset += 4, offset < max && view.getUint8(offset) === 30 && (offset += 1);
+      }
+    }
+  }
+  async save() {
+    const json = JSON.stringify(this.#meta);
+    if (await this.#storage.write(META_FILE, new TextEncoder().encode(json).buffer), this.#deletedIds.size === 0)
+      await this.#storage.remove(DELETED_IDS_FILE);
+    else {
+      const totalSize = this.#deletedIds.size * 4 + this.#deletedIds.size, buffer = new ArrayBuffer(totalSize), view = new DataView(buffer);
+      let offset = 0;
+      for (const id of this.#deletedIds)
+        view.setUint32(offset, id, !0), offset += 4, view.setUint8(offset, 30), offset += 1;
+      await this.#storage.write(DELETED_IDS_FILE, buffer);
+    }
+    if (this.#addedIds.size === 0)
+      await this.#storage.remove(ADDED_IDS_FILE);
+    else {
+      const totalSize = this.#addedIds.size * 4 + this.#addedIds.size, buffer = new ArrayBuffer(totalSize), view = new DataView(buffer);
+      let offset = 0;
+      for (const id of this.#addedIds)
+        view.setUint32(offset, id, !0), offset += 4, view.setUint8(offset, 30), offset += 1;
+      await this.#storage.write(ADDED_IDS_FILE, buffer);
+    }
+  }
+  getSegments(type) {
+    return type === "word" ? this.#meta.wordSegments : this.#meta.charSegments;
+  }
+  getDeletedIds() {
+    return this.#deletedIds;
+  }
+  addDeletedId(id) {
+    this.#deletedIds.add(id);
+  }
+  isDeleted(id) {
+    return this.#deletedIds.has(id);
+  }
+  addAddedId(id) {
+    this.#addedIds.add(id);
+  }
+  removeAddedId(id) {
+    this.#addedIds.delete(id);
+  }
+  isAdded(id) {
+    return this.#addedIds.has(id);
+  }
+  getAddedIds() {
+    return this.#addedIds;
+  }
+  /**
+   * 检查文档ID是否曾经添加过（包括已删除的）
+   * @param id 文档ID
+   * @returns 文档是否曾经添加过的布尔值
+   */
+  hasDocument(id) {
+    return this.#addedIds.has(id) || this.#deletedIds.has(id);
+  }
+  getLastSegmentInfo(type) {
+    const segments = this.getSegments(type);
+    return segments.length === 0 ? null : segments[segments.length - 1];
+  }
+  updateSegment(type, filename, start, end, tokenCount, isNew) {
+    const segments = type === "word" ? this.#meta.wordSegments : this.#meta.charSegments;
+    if (isNew)
+      segments.push({ filename, start, end, tokenCount });
+    else {
+      const last = segments[segments.length - 1];
+      last && last.filename === filename && (last.end = end, last.tokenCount = tokenCount);
+    }
+  }
+  reset() {
+    this.#meta = { wordSegments: [], charSegments: [] }, this.#deletedIds.clear(), this.#addedIds.clear();
+  }
+}
+class IntermediateCache {
+  static SEPARATOR = 30;
+  #storage;
+  constructor(storage) {
+    this.#storage = storage;
+  }
+  async appendBatch(filename, docs) {
+    if (docs.length === 0)
+      return await this.#storage.getFileSize(filename);
+    const encoder = new TextEncoder();
+    let totalLen = 0;
+    for (const doc of docs) {
+      totalLen += 8;
+      for (const token of doc.tokens) {
+        const tokenLen = Math.min(encoder.encode(token).byteLength, 65535);
+        totalLen += 2 + tokenLen;
+      }
+      totalLen += 1;
+    }
+    const combined = new Uint8Array(totalLen);
+    let pos = 0;
+    for (const doc of docs) {
+      const tokenBuffers = [];
+      for (const token of doc.tokens) {
+        const buf = encoder.encode(token), finalBuf = buf.byteLength > 65535 ? buf.slice(0, 65535) : buf;
+        tokenBuffers.push(finalBuf);
+      }
+      const view = new DataView(combined.buffer, pos);
+      view.setUint32(0, doc.id, !0), view.setUint32(4, tokenBuffers.length, !0), pos += 8;
+      for (const buf of tokenBuffers)
+        new DataView(combined.buffer, pos).setUint16(0, buf.byteLength, !0), pos += 2, combined.set(buf, pos), pos += buf.byteLength;
+      combined[pos++] = IntermediateCache.SEPARATOR;
+    }
+    return await this.#storage.append(filename, combined.buffer), await this.#storage.getFileSize(filename);
+  }
+  async readRange(filename, start, end) {
+    const buffer = await this.#storage.readRange(filename, start, end);
+    if (!buffer || buffer.byteLength === 0) return [];
+    const view = new DataView(buffer), uint8 = new Uint8Array(buffer), decoder = new TextDecoder(), docs = [];
+    let offset = 0;
+    const max = buffer.byteLength;
+    for (; offset < max && !(offset + 8 > max); ) {
+      const id = view.getUint32(offset, !0);
+      offset += 4;
+      const count = view.getUint32(offset, !0);
+      offset += 4;
+      const tokens = [];
+      for (let i = 0; i < count && !(offset + 2 > max); i++) {
+        const len = view.getUint16(offset, !0);
+        if (offset += 2, offset + len > max) break;
+        const textBuf = new Uint8Array(buffer, offset, len);
+        tokens.push(decoder.decode(textBuf)), offset += len;
+      }
+      offset < max && uint8[offset] === IntermediateCache.SEPARATOR && (offset += 1), docs.push({ id, tokens });
+    }
+    return docs;
+  }
+  async getCurrentSize(filename) {
+    return await this.#storage.getFileSize(filename);
+  }
+}
+function murmur3_32(str, h = 305419896) {
+  const len = str.length, nBlocks = len >> 2;
+  let i = 0;
+  for (; i < nBlocks; ) {
+    let k12 = str.charCodeAt(i) & 255 | (str.charCodeAt(++i) & 255) << 8 | (str.charCodeAt(++i) & 255) << 16 | (str.charCodeAt(++i) & 255) << 24;
+    ++i, k12 = (k12 & 65535) * 3432918353 + (((k12 >>> 16) * 3432918353 & 65535) << 16) & 4294967295, k12 = k12 << 15 | k12 >>> 17, k12 = (k12 & 65535) * 461845907 + (((k12 >>> 16) * 461845907 & 65535) << 16) & 4294967295, h ^= k12, h = h << 13 | h >>> 19, h = (h & 65535) * 5 + (((h >>> 16) * 5 & 65535) << 16) & 4294967295, h = (h & 65535) + 27492 + (((h >>> 16) + 58964 & 65535) << 16);
+  }
+  let k1 = 0;
+  const remainder = len & 3;
+  return remainder > 0 && (remainder >= 3 && (k1 ^= (str.charCodeAt(i + 2) & 255) << 16), remainder >= 2 && (k1 ^= (str.charCodeAt(i + 1) & 255) << 8), remainder >= 1 && (k1 ^= str.charCodeAt(i) & 255), k1 = (k1 & 65535) * 3432918353 + (((k1 >>> 16) * 3432918353 & 65535) << 16) & 4294967295, k1 = k1 << 15 | k1 >>> 17, k1 = (k1 & 65535) * 461845907 + (((k1 >>> 16) * 461845907 & 65535) << 16) & 4294967295, h ^= k1), h ^= len, h ^= h >>> 16, h = (h & 65535) * 2246822507 + (((h >>> 16) * 2246822507 & 65535) << 16) & 4294967295, h ^= h >>> 13, h = (h & 65535) * 3266489909 + (((h >>> 16) * 3266489909 & 65535) << 16) & 4294967295, h ^= h >>> 16, h >>> 0;
+}
+class IndexSegment {
+  #filename;
+  #storage;
+  #buffer = null;
+  #view = null;
+  /**
+   * 使用MurmurHash3计算字符串哈希值
+   * @param str 要哈希的字符串
+   * @returns 32位无符号哈希值
+   */
+  static hash(str) {
+    return murmur3_32(str);
+  }
+  constructor(filename, storage) {
+    this.#filename = filename, this.#storage = storage;
+  }
+  async loadIndex() {
+    return this.#buffer ? !0 : (this.#buffer = await this.#storage.read(this.#filename), this.#buffer ? (this.#view = new DataView(this.#buffer), !0) : !1);
+  }
+  async buildAndSave(docs) {
+    const tokenMap = /* @__PURE__ */ new Map();
+    for (const doc of docs) {
+      const uniqueTokens = /* @__PURE__ */ new Map();
+      for (const token of doc.tokens)
+        uniqueTokens.has(token) || (uniqueTokens.set(token, !0), tokenMap.has(token) || tokenMap.set(token, {
+          hash: IndexSegment.hash(token),
+          postings: []
+        }), tokenMap.get(token).postings.push(doc.id));
+    }
+    const entries = Array.from(tokenMap.entries());
+    entries.sort(([a, ah], [b, bh]) => ah.hash !== bh.hash ? ah.hash - bh.hash : a.localeCompare(b));
+    const encoder = new TextEncoder();
+    let totalPostings = 0, totalTokensSize = 0;
+    for (const [token, { postings }] of entries) {
+      totalPostings += postings.length;
+      const bytes = encoder.encode(token);
+      totalTokensSize += bytes.length + 1;
+    }
+    const headerSize = 12, dictSize = entries.length * 20, postingsSize = totalPostings * 4, tokensOffset = headerSize + dictSize + postingsSize, totalSize = tokensOffset + totalTokensSize, buffer = new ArrayBuffer(totalSize), view = new DataView(buffer);
+    view.setUint32(0, 1229866072), view.setUint32(4, entries.length, !0), view.setUint32(8, tokensOffset, !0);
+    let currentDictOffset = headerSize, currentPostingsOffset = headerSize + dictSize, currentTokenOffset = tokensOffset;
+    for (const [token, { hash: hash2, postings }] of entries) {
+      view.setUint32(currentDictOffset, hash2, !0);
+      const tokenBytes = encoder.encode(token);
+      view.setUint32(currentDictOffset + 4, tokenBytes.length, !0), view.setUint32(currentDictOffset + 8, currentTokenOffset, !0), view.setUint32(currentDictOffset + 12, currentPostingsOffset, !0), view.setUint32(currentDictOffset + 16, postings.length, !0), currentDictOffset += 20;
+      for (let i = 0; i < postings.length; i++)
+        view.setUint32(currentPostingsOffset, postings[i], !0), currentPostingsOffset += 4;
+      for (let i = 0; i < tokenBytes.length; i++)
+        view.setUint8(currentTokenOffset++, tokenBytes[i]);
+      view.setUint8(currentTokenOffset++, 0);
+    }
+    await this.#storage.write(this.#filename, buffer), this.#buffer = buffer, this.#view = view;
+  }
+  search(term) {
+    if (!this.#view || !this.#buffer) return [];
+    const h = IndexSegment.hash(term), count = this.#view.getUint32(4, !0);
+    let left = 0, right = count - 1;
+    const headerSize = 12, entrySize = 20, decoder = new TextDecoder();
+    for (; left <= right; ) {
+      const mid = left + right >>> 1, entryPos = headerSize + mid * entrySize, entryHash = this.#view.getUint32(entryPos, !0);
+      if (entryHash < h)
+        left = mid + 1;
+      else if (entryHash > h)
+        right = mid - 1;
+      else {
+        if (!(mid > 0 && this.#view.getUint32(headerSize + (mid - 1) * entrySize, !0) === h || mid < count - 1 && this.#view.getUint32(headerSize + (mid + 1) * entrySize, !0) === h)) {
+          const postingsOffset = this.#view.getUint32(headerSize + mid * entrySize + 12, !0), postingsLen = this.#view.getUint32(headerSize + mid * entrySize + 16, !0), result = [];
+          for (let j = 0; j < postingsLen; j++)
+            result.push(this.#view.getUint32(postingsOffset + j * 4, !0));
+          return result;
+        }
+        let firstMatch = mid;
+        for (; firstMatch > 0; ) {
+          const prevPos = headerSize + (firstMatch - 1) * entrySize;
+          if (this.#view.getUint32(prevPos, !0) === h)
+            firstMatch--;
+          else
+            break;
+        }
+        for (let i = firstMatch; i < count; i++) {
+          const checkPos = headerSize + i * entrySize;
+          if (this.#view.getUint32(checkPos, !0) !== h) break;
+          const tokenLen = this.#view.getUint32(checkPos + 4, !0), tokenOffset = this.#view.getUint32(checkPos + 8, !0), tokenBuffer = new Uint8Array(this.#buffer, tokenOffset, tokenLen);
+          if (decoder.decode(tokenBuffer) === term) {
+            const postingsOffset = this.#view.getUint32(checkPos + 12, !0), postingsLen = this.#view.getUint32(checkPos + 16, !0), result = [];
+            for (let j = 0; j < postingsLen; j++)
+              result.push(this.#view.getUint32(postingsOffset + j * 4, !0));
+            return result;
+          }
+        }
+        return [];
+      }
+    }
+    return [];
+  }
+}
+const defaultTokenize = ({ text }) => {
+  try {
+    if (typeof Intl < "u" && typeof Intl.Segmenter == "function" && typeof Array.from == "function") {
+      const segments = new Intl.Segmenter([], { granularity: "word" }).segment(text);
+      if (typeof segments == "object" && segments !== null)
+        return Array.from(segments).filter((s) => s?.isWordLike).map((s) => s?.segment?.toLowerCase() || "");
+    }
+  } catch {
+  }
+  return text.toLowerCase().split(/[^a-z0-9\u4e00-\u9fa5]+/g).filter((t) => t.length > 0);
+}, WORD_CACHE_FILE = "word_cache.bin", CHAR_CACHE_FILE = "char_cache.bin";
+class SearchEngine {
+  #storage;
+  #meta;
+  #cache;
+  #segments;
+  #initialized = !1;
+  #config;
+  // 批处理状态
+  #inBatch = !1;
+  #pendingTokenCounts = { word: 0, char: 0 };
+  constructor(config) {
+    if (this.#config = {
+      wordSegmentTokenThreshold: 1e5,
+      charSegmentTokenThreshold: 5e5,
+      minWordTokenSave: 0,
+      minCharTokenSave: 0,
+      indexingTokenizer: config.indexingTokenizer || defaultTokenize,
+      ...config
+    }, (this.#config.minWordTokenSave || 0) >= (this.#config.wordSegmentTokenThreshold || 1e5))
+      throw new Error("minWordTokenSave must be less than wordSegmentTokenThreshold");
+    if ((this.#config.minCharTokenSave || 0) >= (this.#config.charSegmentTokenThreshold || 5e5))
+      throw new Error("minCharTokenSave must be less than charSegmentTokenThreshold");
+    this.#storage = config.storage, this.#meta = new MetaManager(this.#storage), this.#cache = new IntermediateCache(this.#storage), this.#segments = /* @__PURE__ */ new Map();
+  }
+  /**
+   * 开启批处理
+   * 批处理期间 addDocuments 只写入缓存，不触发索引段构建
+   */
+  startBatch() {
+    this.#inBatch = !0, this.#pendingTokenCounts = { word: 0, char: 0 };
+  }
+  /**
+   * 结束批处理
+   * 触发索引构建检查并保存元数据
+   */
+  async endBatch() {
+    this.#inBatch = !1, this.#pendingTokenCounts.word > 0 && await this.#processSegmentLogic("word", this.#pendingTokenCounts.word), this.#pendingTokenCounts.char > 0 && await this.#processSegmentLogic("char", this.#pendingTokenCounts.char), this.#pendingTokenCounts = { word: 0, char: 0 }, await this.#meta.save();
+  }
+  async addDocument(doc) {
+    return this.addDocuments([doc]);
+  }
+  /**
+   * 添加单个文档，如果文档ID已存在则跳过
+   * 用于在批量添加中途出错后的恢复添加行为，也可直接用于单个文档添加
+   */
+  async addDocumentIfMissing(doc) {
+    return this.addDocumentsIfMissing([doc]);
+  }
+  /**
+   * 添加多个文档，跳过已存在的文档ID
+   * 用于在批量添加中途出错后的恢复添加行为，也可直接用于批量添加
+   */
+  async addDocumentsIfMissing(docs) {
+    if (this.#initialized || await this.#init(), docs.length === 0) return;
+    const deletedIds = this.#meta.getDeletedIds(), batchWordDocs = [], batchCharDocs = [], newDocs = [];
+    for (const doc of docs) {
+      if (deletedIds.has(doc.id) || this.#meta.isAdded(doc.id))
+        continue;
+      const rawTokens = this.#getIndexingTokens(doc), wordTokens = [], charTokens = [];
+      for (const t of rawTokens)
+        t.length > 1 ? wordTokens.push(t) : t.length === 1 && charTokens.push(t);
+      wordTokens.length > 0 && batchWordDocs.push({ id: doc.id, tokens: wordTokens }), charTokens.length > 0 && batchCharDocs.push({ id: doc.id, tokens: charTokens }), newDocs.push(doc);
+    }
+    if (newDocs.length === 0) return;
+    let addedWordTokens = 0, addedCharTokens = 0;
+    if (batchWordDocs.length > 0) {
+      await this.#cache.appendBatch(WORD_CACHE_FILE, batchWordDocs);
+      for (const d of batchWordDocs) addedWordTokens += d.tokens.length;
+    }
+    if (batchCharDocs.length > 0) {
+      await this.#cache.appendBatch(CHAR_CACHE_FILE, batchCharDocs);
+      for (const d of batchCharDocs) addedCharTokens += d.tokens.length;
+    }
+    for (const doc of newDocs)
+      this.#meta.addAddedId(doc.id);
+    this.#inBatch ? (this.#pendingTokenCounts.word += addedWordTokens, this.#pendingTokenCounts.char += addedCharTokens) : (addedWordTokens > 0 && await this.#processSegmentLogic("word", addedWordTokens), addedCharTokens > 0 && await this.#processSegmentLogic("char", addedCharTokens), await this.#meta.save());
+  }
+  async addDocuments(docs) {
+    if (this.#initialized || await this.#init(), docs.length === 0) return;
+    const deletedIds = this.#meta.getDeletedIds(), batchWordDocs = [], batchCharDocs = [];
+    for (const doc of docs) {
+      if (deletedIds.has(doc.id))
+        throw new Error(`Document ID ${doc.id} has been deleted and cannot be re-added.`);
+      if (this.#meta.isAdded(doc.id))
+        throw new Error(`Document ID ${doc.id} already exists.`);
+      const rawTokens = this.#getIndexingTokens(doc), wordTokens = [], charTokens = [];
+      for (const t of rawTokens)
+        t.length > 1 ? wordTokens.push(t) : t.length === 1 && charTokens.push(t);
+      wordTokens.length > 0 && batchWordDocs.push({ id: doc.id, tokens: wordTokens }), charTokens.length > 0 && batchCharDocs.push({ id: doc.id, tokens: charTokens });
+    }
+    let addedWordTokens = 0, addedCharTokens = 0;
+    if (batchWordDocs.length > 0) {
+      await this.#cache.appendBatch(WORD_CACHE_FILE, batchWordDocs);
+      for (const d of batchWordDocs) addedWordTokens += d.tokens.length;
+    }
+    if (batchCharDocs.length > 0) {
+      await this.#cache.appendBatch(CHAR_CACHE_FILE, batchCharDocs);
+      for (const d of batchCharDocs) addedCharTokens += d.tokens.length;
+    }
+    for (const doc of docs)
+      this.#meta.addAddedId(doc.id);
+    this.#inBatch ? (this.#pendingTokenCounts.word += addedWordTokens, this.#pendingTokenCounts.char += addedCharTokens) : (addedWordTokens > 0 && await this.#processSegmentLogic("word", addedWordTokens), addedCharTokens > 0 && await this.#processSegmentLogic("char", addedCharTokens), await this.#meta.save());
+  }
+  async search(query, limit) {
+    this.#initialized || await this.#init();
+    const queryDoc = typeof query == "string" ? { text: query } : query, rawTokens = this.#getSearchTokens(queryDoc), wordTerms = rawTokens.filter((t) => t.length > 1), charTerms = rawTokens.filter((t) => t.length === 1), deletedIds = this.#meta.getDeletedIds(), docMatches = /* @__PURE__ */ new Map(), segmentsToLoad = /* @__PURE__ */ new Map(), collectSegments = (type) => {
+      const segmentsMeta = this.#meta.getSegments(type);
+      for (const meta of segmentsMeta) {
+        const filename = meta.filename;
+        !this.#segments.has(filename) && !segmentsToLoad.has(filename) && segmentsToLoad.set(filename, new IndexSegment(filename, this.#storage));
+      }
+    };
+    collectSegments("word"), collectSegments("char"), await Promise.all(
+      Array.from(segmentsToLoad.entries()).map(([filename, segment]) => segment.loadIndex().then((loaded) => {
+        loaded && this.#segments.set(filename, segment);
+      }))
+    );
+    const processTerms = async (type, terms) => {
+      if (terms.length === 0) return;
+      const segmentsMeta = this.#meta.getSegments(type);
+      for (const meta of segmentsMeta) {
+        const filename = meta.filename, segment = this.#segments.get(filename);
+        if (segment)
+          for (const term of terms) {
+            const hits = segment.search(term), termScore = 1 + term.length * 0.1;
+            for (const id of hits)
+              if (!deletedIds.has(id))
+                if (!docMatches.has(id))
+                  docMatches.set(id, { score: 0, tokens: /* @__PURE__ */ new Set([term]) });
+                else {
+                  const match = docMatches.get(id);
+                  match.score += termScore, match.tokens.add(term);
+                }
+          }
+      }
+    };
+    await processTerms("word", wordTerms), await processTerms("char", charTerms);
+    const results = [];
+    return docMatches.forEach((data, id) => {
+      results.push({
+        id,
+        score: data.score,
+        tokens: Array.from(data.tokens)
+      });
+    }), results.sort((a, b) => b.score - a.score), typeof limit == "number" && limit > 0 ? results.slice(0, limit) : results;
+  }
+  async removeDocument(id) {
+    this.#initialized || await this.#init(), this.#meta.addDeletedId(id), this.#meta.removeAddedId(id), await this.#meta.save();
+  }
+  async clearAll() {
+    await this.#storage.clearAll(), this.#segments.clear(), this.#meta.reset(), this.#initialized = !1, this.#inBatch = !1, this.#pendingTokenCounts = { word: 0, char: 0 };
+  }
+  async getStatus() {
+    return this.#initialized || await this.#init(), {
+      wordSegments: this.#meta.getSegments("word").length,
+      charSegments: this.#meta.getSegments("char").length,
+      deleted: this.#meta.getDeletedIds().size,
+      wordCacheSize: await this.#cache.getCurrentSize(WORD_CACHE_FILE),
+      charCacheSize: await this.#cache.getCurrentSize(CHAR_CACHE_FILE),
+      inBatch: this.#inBatch
+    };
+  }
+  /**
+   * 检查文档ID是否曾经添加过（包括已删除的）
+   * @param id 文档ID
+   * @returns 文档是否曾经添加过的布尔值
+   */
+  async hasDocument(id) {
+    return this.#initialized || await this.#init(), this.#meta.hasDocument(id);
+  }
+  async #init() {
+    if (this.#initialized) return;
+    await this.#meta.load();
+    const allSegments = [
+      ...this.#meta.getSegments("word"),
+      ...this.#meta.getSegments("char")
+    ];
+    for (const seg of allSegments)
+      this.#segments.has(seg.filename) || this.#segments.set(seg.filename, new IndexSegment(seg.filename, this.#storage)), await this.#segments.get(seg.filename).loadIndex();
+    this.#initialized = !0;
+  }
+  #getIndexingTokens(doc) {
+    return this.#config.indexingTokenizer(doc);
+  }
+  #getSearchTokens(doc) {
+    return this.#config.searchTokenizer ? this.#config.searchTokenizer(doc) : this.#getIndexingTokens(doc);
+  }
+  /**
+   * 核心索引段处理逻辑
+   * 负责判断是否需要合并、新建 Segment，并执行构建
+   */
+  async #processSegmentLogic(type, addedTokenCount) {
+    const cacheFilename = type === "word" ? WORD_CACHE_FILE : CHAR_CACHE_FILE, currentCacheSize = await this.#cache.getCurrentSize(cacheFilename), segThreshold = type === "word" ? this.#config.wordSegmentTokenThreshold || 1e5 : this.#config.charSegmentTokenThreshold || 5e5, minSave = type === "word" ? this.#config.minWordTokenSave || 0 : this.#config.minCharTokenSave || 0, lastSegInfo = this.#meta.getLastSegmentInfo(type);
+    let targetSegmentName, startOffset, isNew, newTokenCountTotal;
+    const generateSegmentName = () => {
+      const nextNumber = this.#meta.getSegments(type).length + 1;
+      return `${type}_seg_${nextNumber}.bin`;
+    };
+    if (!lastSegInfo)
+      targetSegmentName = generateSegmentName(), isNew = !0, startOffset = 0, newTokenCountTotal = addedTokenCount;
+    else {
+      const existingTokenCount = lastSegInfo.tokenCount;
+      existingTokenCount >= segThreshold || existingTokenCount + addedTokenCount >= segThreshold ? (targetSegmentName = generateSegmentName(), isNew = !0, startOffset = lastSegInfo.end, newTokenCountTotal = addedTokenCount) : (targetSegmentName = lastSegInfo.filename, isNew = !1, startOffset = lastSegInfo.start, newTokenCountTotal = existingTokenCount + addedTokenCount);
+    }
+    if (newTokenCountTotal < minSave) {
+      this.#meta.updateSegment(type, targetSegmentName, startOffset, currentCacheSize, newTokenCountTotal, isNew);
+      return;
+    }
+    const docsToBuild = await this.#cache.readRange(cacheFilename, startOffset, currentCacheSize);
+    let segment = this.#segments.get(targetSegmentName);
+    segment || (segment = new IndexSegment(targetSegmentName, this.#storage), this.#segments.set(targetSegmentName, segment)), await segment.buildAndSave(docsToBuild), this.#meta.updateSegment(type, targetSegmentName, startOffset, currentCacheSize, newTokenCountTotal, isNew);
+  }
+}
+exports.SearchEngine = SearchEngine, exports.hash = murmur3_32, exports.murmur3_32 = murmur3_32;