min-mphash 0.5.1 → 0.5.2

package/README.en.md

@@ -4,7 +4,7 @@
 
 [中文 README](./README.zh.md)
 
-`MinMPHash` can map a set of n strings to the integer range `[0, n-1]` without any collisions.
+`MinMPHash` can map a set of n strings to the integer range `[0, n-1]` without any collisions. If only the hash function is used, the package size is less than 3KB (Gzip).
 
 `MinMPLookup` is a minimal perfect lookup table tool implemented based on `MinMPHash`.
 
@@ -121,6 +121,19 @@ You might think that hash functions like MD5 or SHA-1 can also generate identifi
 npm install min-mphash
 ```
 
+This package only provides ESM, and using build tools can effectively perform tree-shaking to reduce the size.
+
+```
+// hash build & hush function
+min-mphash/index.js         34.7 kB
+min-mphash/index.min.js     14.6 kB   4.9 kB（Gzip）
+
+
+// hush function only
+min-mphash/runtime.js       18.3 kB
+min-mphash/runtime.min.js   7.9 kB    2.8 kB（Gzip）
+```
+
 ### MinMPHash Usage
 
 This is the core function, used to map a set of strings to integers in `[0, n-1]`.

package/README.md

@@ -6,7 +6,7 @@
 
 [中文 README](./README.zh.md)
 
-TypeScript/JavaScript 平台上的最小完美哈希与查找工具实现。
+TypeScript/JavaScript 平台上的最小完美哈希与查找工具实现。如果只使用哈希函数包体积小于 3KB(Gzip)。
 
 `MinMPHash` 可以把一组数量为 n 的字符串映射到 `[0, n-1]` 的整数范围内，且不会有冲突。
 
@@ -18,7 +18,7 @@ TypeScript/JavaScript 平台上的最小完美哈希与查找工具实现。
 
 ---
 
-`MinMPHash` can map a set of n strings to the integer range `[0, n-1]` without any collisions.
+`MinMPHash` can map a set of n strings to the integer range `[0, n-1]` without any collisions. If only the hash function is used, the package size is less than 3KB (Gzip).
 
 `MinMPLookup` is a minimal perfect lookup table tool implemented based on `MinMPHash`.
 
@@ -135,6 +135,19 @@ You might think that hash functions like MD5 or SHA-1 can also generate identifi
 npm install min-mphash
 ```
 
+This package only provides ESM, and using build tools can effectively perform tree-shaking to reduce the size.
+
+```
+// hash build & hush function
+min-mphash/index.js         34.7 kB
+min-mphash/index.min.js     14.6 kB   4.9 kB（Gzip）
+
+
+// hush function only
+min-mphash/runtime.js       18.3 kB
+min-mphash/runtime.min.js   7.9 kB    2.8 kB（Gzip）
+```
+
 ### MinMPHash Usage
 
 This is the core function, used to map a set of strings to integers in `[0, n-1]`.

package/README.zh.md

@@ -4,7 +4,7 @@
 
 [English README](./README.en.md)
 
-TypeScript/JavaScript 平台上的最小完美哈希与查找工具实现。
+TypeScript/JavaScript 平台上的最小完美哈希与查找工具实现。如果只使用哈希函数包体积小于 3KB(Gzip)。
 
 `MinMPHash` 可以把一组数量为 n 的字符串映射到 `[0, n-1]` 的整数范围内，且不会有冲突。
 
@@ -123,6 +123,19 @@ let found = Object.entries(FontMapWithHash).find(([family, hashes]) =>
 npm install min-mphash
 ```
 
+此包仅提供 ESM，使用构建工具可以很的进行 tree-shaking 以减小体积。
+
+```
+// hash build & hush function
+min-mphash/index.js         34.7 kB
+min-mphash/index.min.js     14.6 kB   4.9 kB（Gzip）
+
+
+// hush function only
+min-mphash/runtime.js       18.3 kB
+min-mphash/runtime.min.js   7.9 kB    2.8 kB（Gzip）
+```
+
 ### MinMPHash 最小完美哈希的使用
 
 这是最核心的功能，用于将一组字符串映射到 `[0, n-1]` 的整数。

package/dist/MinMPLookup.d.ts

@@ -93,12 +93,12 @@ export declare function serializeMinMPLookupDict(dict: IMinMPLookupDict): Uint8A
  *```
  */
 export declare class MinMPLookup {
-    private dict;
     private mph;
+    private dict;
     private _invertedIndex;
     static fromCompressed(data: Uint8Array): Promise<MinMPLookup>;
     static fromBinary(data: Uint8Array): MinMPLookup;
-    constructor(dict: IMinMPLookupDict);
+    constructor(dict: IMinMPLookupDict | Uint8Array);
     private buildInvertedIndex;
     /**
      * 查找 value 对应的第一个 key

package/dist/index.js

@@ -693,7 +693,7 @@ function serializeMinMPLookupDict(dict) {
         parts.push(keyBytes);
     }
     if (dict.valueToKeyIndexes && void 0 !== dict.bitsPerKey) {
-        parts.push(writeU32(0xFFFFFFFF));
+        parts.push(writeU32(0xffffffff));
         parts.push(writeU32(dict.bitsPerKey));
         parts.push(writeU32(dict.valueToKeyIndexes.length));
         parts.push(dict.valueToKeyIndexes);
@@ -775,7 +775,7 @@ function deserializeLookupDict(data) {
         keys.push(decoder.decode(kBytes));
     }
     const sectionLen = readU32();
-    if (0xFFFFFFFF === sectionLen) {
+    if (0xffffffff === sectionLen) {
         const bitsPerKey = readU32();
         const dataLen = readU32();
         const valueToKeyIndexes = data.subarray(offset, offset + dataLen);
@@ -925,11 +925,11 @@ class MinMPLookup {
         return this.dict.keys;
     }
     constructor(dict){
-        MinMPLookup_define_property(this, "dict", void 0);
         MinMPLookup_define_property(this, "mph", void 0);
-        MinMPLookup_define_property(this, "_invertedIndex", void 0);
+        MinMPLookup_define_property(this, "dict", void 0);
+        MinMPLookup_define_property(this, "_invertedIndex", null);
+        if (dict instanceof Uint8Array) dict = deserializeLookupDict(dict);
         this.dict = dict;
-        this._invertedIndex = null;
         this.mph = new MinMPHash(dict.mmpHashDictBin);
         if (dict.keyToHashes) this.buildInvertedIndex();
     }

package/dist/index.min.js

@@ -1 +1 @@
-function e(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}function t(e,t,i){let n=0;for(;e>=128;){let r=127&e|128;Array.isArray(t)?t.push(r):void 0!==i&&(t[i+n]=r),e>>>=7,n++}return Array.isArray(t)?t.push(e):void 0!==i&&(t[i+n]=e),n+1}function i(e,t){let i=0,n=0,r=0;for(;;){let s=e[t+r];if(i|=(127&s)<<n,r++,(128&s)==0)break;n+=7}return{value:i,bytes:r}}let n={encodeInt(e,t){e<24?t.push(0|e):e<=255?t.push(24,e):e<=65535?t.push(25,e>>8,255&e):t.push(26,e>>>24&255,e>>>16&255,e>>>8&255,255&e)},encodeBytes(e,t){let i=e.byteLength;i<24?t.push(64|i):i<=255?t.push(88,i):i<=65535?t.push(89,i>>8,255&i):t.push(90,i>>>24&255,i>>>16&255,i>>>8&255,255&i);for(let n=0;n<i;n++)t.push(e[n])},encodeNull(e){e.push(246)},encodeArrayHead(e,t){e<24&&t.push(128|e)},decode(e,t){let i=e.getUint8(t.current++),r=224&i,s=31&i,l=0;if(s<24)l=s;else if(24===s)l=e.getUint8(t.current),t.current+=1;else if(25===s)l=e.getUint16(t.current,!1),t.current+=2;else if(26===s)l=e.getUint32(t.current,!1),t.current+=4;else throw Error("Unsupported CBOR size");if(0===r)return l;if(64===r){let i=l,n=new Uint8Array(e.buffer.slice(e.byteOffset+t.current,e.byteOffset+t.current+i));return t.current+=i,n}if(128===r){let i=l,r=[];for(let s=0;s<i;s++)r.push(n.decode(e,t));return r}if(246===i)return null;throw Error(`Unknown CBOR type: ${i.toString(16)}`)}},r={none:0,4:1,8:2,16:3,32:4,2:5},s=["none","4","8","16","32","2"];function l(e){let t=[];if(n.encodeArrayHead(9,t),n.encodeInt(e.n,t),n.encodeInt(e.m,t),n.encodeInt(e.seed0,t),n.encodeBytes(e.bucketSizes,t),n.encodeBytes(e.seedStream,t),n.encodeInt(r[e.validationMode],t),e.fingerprints&&"none"!==e.validationMode){let i;if(e.fingerprints instanceof Uint8Array)i=e.fingerprints;else if(e.fingerprints instanceof Uint16Array||e.fingerprints instanceof Uint32Array)i=new Uint8Array(e.fingerprints.buffer,e.fingerprints.byteOffset,e.fingerprints.byteLength);else{let t,n=e.validationMode;t="2"===n||"4"===n||"8"===n?new Uint8Array(e.fingerprints):"16"===n?new Uint16Array(e.fingerprints):new Uint32Array(e.fingerprints),i=new Uint8Array(t.buffer,t.byteOffset,t.byteLength)}n.encodeBytes(i,t)}else n.encodeNull(t);return e.seedZeroBitmap?n.encodeBytes(e.seedZeroBitmap,t):n.encodeNull(t),n.encodeInt(e.hashSeed||0,t),new Uint8Array(t)}async function o(e){let t=new Blob([e]).stream().pipeThrough(new CompressionStream("gzip"));return new Uint8Array(await new Response(t).arrayBuffer())}async function f(e){let t=new Blob([e]).stream().pipeThrough(new DecompressionStream("gzip"));return new Uint8Array(await new Response(t).arrayBuffer())}class h{write(e,t){for(let i=0;i<t;i++){let t=e>>i&1;this.currentByte|=t<<this.bitCount,this.bitCount++,8===this.bitCount&&(this.buffer.push(this.currentByte),this.currentByte=0,this.bitCount=0)}}flush(){this.bitCount>0&&(this.buffer.push(this.currentByte),this.currentByte=0,this.bitCount=0)}getData(){return this.flush(),new Uint8Array(this.buffer)}constructor(){e(this,"buffer",[]),e(this,"currentByte",0),e(this,"bitCount",0)}}class a{read(e){let t=0;for(let i=0;i<e;i++){if(this.byteOffset>=this.buffer.length)return 0;t|=(this.buffer[this.byteOffset]>>this.bitOffset&1)<<i,this.bitOffset++,8===this.bitOffset&&(this.byteOffset++,this.bitOffset=0)}return t}constructor(t){e(this,"buffer",void 0),e(this,"byteOffset",0),e(this,"bitOffset",0),this.buffer=t}}function u(e,t,i){let n=0,r=t;for(let t=0;t<i;t++){let i=r>>>3,s=7&r;if(i>=e.length)return 0;n|=(e[i]>>s&1)<<t,r++}return n}function d(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}function c(e,t){let i=e.length;if(0===i){let e={n:0,m:0,seed0:0,seedStream:new Uint8Array(0),bucketSizes:new Uint8Array(0),validationMode:"none"};return t?.outputBinary?l(e):e}let n=t?.level??5,r="none";t?.onlySet===!0?r="8":"string"==typeof t?.onlySet&&(r=t.onlySet);let s=i>5e5?Math.max(1,.9*n):n,f=Math.max(1,Math.ceil(i/s)),h=new Uint32Array(i),a=new Uint32Array(i),u=0;for(;;){let t=new Map,n=new Map,r=!1;for(let s=0;s<i;s++){let i=b(e[s],u),l=b(e[s],~u),o=!1;if(n.has(i)){let e=n.get(i);e.has(l)?o=!0:e.add(l)}else if(t.has(i)){let e=t.get(i);if(e===l)o=!0;else{let r=new Set;r.add(e),r.add(l),n.set(i,r),t.delete(i)}}else t.set(i,l);if(o){r=!0;break}h[s]=i,a[s]=l}if(!r)break;if(++u>100)throw Error(`Could not find a collision-free hash seed after ${u} attempts.`)}let d=new Int32Array(f).fill(-1),c=new Int32Array(i).fill(-1),g=0,m=1/0,M=new Int32Array(f),v=new Int32Array(i),A=new Int32Array(f);for(let e=0;e<2e3;e++){let t=Math.floor(0xffffffff*Math.random());M.fill(-1),A.fill(0);let n=0;for(let e=0;e<i;e++){let i=Math.floor(((p(h[e],t)^a[e])>>>0)/0x100000000*f);v[e]=M[i],M[i]=e,A[i]++,A[i]>n&&(n=A[i])}if(n<13){g=t,d.set(M),c.set(v),m=n;break}if(n<m&&(m=n,g=t,d.set(M),c.set(v)),m<16&&e>50)break}if(m>=16)throw Error(`MPHF Build Failed: Could not find a bucket distribution with max size < 16 (best: ${m}). Try reducing the optimization level (current: ${t?.level??5}).`);let U=new Uint8Array(Math.ceil(f/2));for(let e=0;e<f;e++){let t=0,i=d[e];for(;-1!==i;)t++,i=c[i];let n=e>>>1;(1&e)==0?U[n]|=t:U[n]|=t<<4}let x=new w,k=new Uint8Array(Math.ceil(f/8));for(let e=0;e<f;e++){let t=0,i=d[e];for(;-1!==i;)t++,i=c[i];if(t<=1){k[e>>>3]|=1<<(7&e);continue}let n=0,r=!1,s=t>14?5e7:5e6;for(;!r;){let i=0,l=!1,o=d[e];for(;-1!==o;){let e=((p(h[o],n)^a[o])>>>0)%t;if((i&1<<e)!=0){l=!0;break}i|=1<<e,o=c[o]}if(l){if(++n>s)throw Error(`MPHF Failed: Bucket ${e} (size ${t}) is too hard.`)}else 0===n?k[e>>>3]|=1<<(7&e):x.write(n),r=!0}}let B={n:i,m:f,seed0:g,hashSeed:u,seedStream:x.toUint8Array(),bucketSizes:U,seedZeroBitmap:k,validationMode:r};if("none"!==r){let t;t="2"===r?new Uint8Array(Math.ceil(i/4)):"4"===r?new Uint8Array(Math.ceil(i/2)):"8"===r?new Uint8Array(i):"16"===r?new Uint16Array(i):new Uint32Array(i);let n=new y({...B,validationMode:"none"});for(let s=0;s<i;s++){let l=e[s],o=n.hash(l);if(o>=0&&o<i){let e=b(l,0x1234abcd);if("2"===r){let i=3&e,n=o>>>2,r=(3&o)<<1;t[n]|=i<<r}else if("4"===r){let i=15&e,n=o>>>1;(1&o)==0?t[n]|=i:t[n]|=i<<4}else"8"===r?t[o]=255&e:"16"===r?t[o]=65535&e:t[o]=e>>>0}}B.fingerprints=t}if(t?.outputBinary){let e=l(B);return t.enableCompression?o(e):e}return B}class y{static async fromCompressed(e){return new y(await f(e))}hash(e){if(0===this.n)return -1;let t=b(e,this.hashSeed),i=b(e,~this.hashSeed),n=Math.floor(((p(t,this.seed0)^i)>>>0)/0x100000000*this.m),r=this.offsets[n],s=this.offsets[n+1]-r;if(0===s)return -1;let l=0;if(l=1===s?r:r+((p(t,this.seeds[n])^i)>>>0)%s,"none"!==this.validationMode&&this.fingerprints){let t=b(e,y.FP_SEED);if("2"===this.validationMode){let e=l>>>2,i=(3&l)<<1;if((this.fingerprints[e]>>>i&3)!=(3&t))return -1}else if("4"===this.validationMode){let e=l>>>1,i=this.fingerprints[e];if(((1&l)==0?15&i:i>>>4&15)!=(15&t))return -1}else if("8"===this.validationMode){if(this.fingerprints[l]!==(255&t))return -1}else if("16"===this.validationMode){if(this.fingerprints[l]!==(65535&t))return -1}else if(this.fingerprints[l]!==t>>>0)return -1}return l}constructor(e){if(d(this,"n",void 0),d(this,"m",void 0),d(this,"seed0",void 0),d(this,"hashSeed",void 0),d(this,"offsets",void 0),d(this,"seeds",void 0),d(this,"validationMode",void 0),d(this,"fingerprints",null),e instanceof Uint8Array&&(e=function(e){let t,i=new DataView(e.buffer,e.byteOffset,e.byteLength),r=n.decode(i,{current:0});if(!Array.isArray(r)||r.length<7)throw Error("Invalid CBOR format");let[l,o,f,h,a,u,d,c,y]=r,p=s[u]||"none";return d&&"none"!==p&&("2"===p||"4"===p||"8"===p?t=d:"16"===p?t=new Uint16Array(d.buffer,d.byteOffset,d.byteLength/2):"32"===p&&(t=new Uint32Array(d.buffer,d.byteOffset,d.byteLength/4))),{n:l,m:o,seed0:f,hashSeed:y||0,bucketSizes:h,seedStream:a,validationMode:p,fingerprints:t,seedZeroBitmap:c||void 0}}(e)),this.n=e.n,this.m=e.m,this.seed0=e.seed0,this.hashSeed=e.hashSeed||0,this.validationMode=e.validationMode||"none",0===this.n){this.offsets=new Uint32Array(0),this.seeds=new Int32Array(0);return}this.offsets=new Uint32Array(this.m+1);let t=0;for(let i=0;i<this.m;i++){this.offsets[i]=t;let n=e.bucketSizes[i>>>1];t+=1&i?n>>>4:15&n}this.offsets[this.m]=t,this.seeds=new Int32Array(this.m);let i=0,r=e.seedStream,l=e.seedZeroBitmap;for(let e=0;e<this.m;e++){let t=!1;if(l&&(l[e>>>3]&1<<(7&e))!=0&&(t=!0),t)this.seeds[e]=0;else{let t=0,n=0;for(;;){let e=r[i++];if(t|=(127&e)<<n,(128&e)==0)break;n+=7}this.seeds[e]=t}}if("none"!==this.validationMode&&e.fingerprints){let t=e.fingerprints;"2"===this.validationMode||"4"===this.validationMode||"8"===this.validationMode?this.fingerprints=t instanceof Uint8Array?t:new Uint8Array(t):"16"===this.validationMode?this.fingerprints=t instanceof Uint16Array?t:new Uint16Array(t):this.fingerprints=t instanceof Uint32Array?t:new Uint32Array(t)}}}function p(e,t){return e^=t,e=Math.imul(e,0x85ebca6b),e^=e>>>13,e=Math.imul(e,0xc2b2ae35),(e^=e>>>16)>>>0}function b(e,t){let i=t;for(let t=0;t<e.length;t++){let n=e.charCodeAt(t);i^=n=Math.imul(n=(n=Math.imul(n,0xcc9e2d51))<<15|n>>>17,0x1b873593),i=Math.imul(i=i<<13|i>>>19,5)+0xe6546b64}return i^=e.length,i^=i>>>16,i=Math.imul(i,0x85ebca6b),i^=i>>>13,i=Math.imul(i,0xc2b2ae35),(i^=i>>>16)>>>0}d(y,"FP_SEED",0x1234abcd);class w{write(e){for(;e>=128;)this.buffer.push(127&e|128),e>>>=7;this.buffer.push(e)}toUint8Array(){return new Uint8Array(this.buffer)}constructor(){d(this,"buffer",[])}}function g(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}function m(e,i){let n,r=Object.keys(e),s=new Set;for(let t of r)for(let i of e[t])s.add(i);let l=Array.from(s),f=c(l,{level:i?.level,outputBinary:!0,onlySet:i?.onlySet??"8"}),a=new y(f),u=new Map;for(let t=0;t<r.length;t++)for(let i of e[r[t]])u.has(i)||u.set(i,[]),u.get(i).push(t);let d=0;for(let[e,t]of u)t.length>1&&d++;if(d<.1*l.length){let e=Math.ceil(Math.log2(r.length+1)),t=new h,i=new Map,s=new Int32Array(a.n).fill(-1);for(let[e,t]of u){let n=a.hash(e);n>=0&&(1===t.length?s[n]=t[0]:(s[n]=r.length,i.set(n,t)))}for(let i=0;i<a.n;i++){let n=s[i];t.write(n>=0?n:0,e)}n={mmpHashDictBin:f,keys:r,valueToKeyIndexes:t.getData(),bitsPerKey:e,collisionMap:i.size>0?i:void 0}}else{let t=[];for(let i of r){let n=e[i],r=[];for(let e of n){let t=a.hash(e);t>=0&&r.push(t)}r.sort((e,t)=>e-t),t.push(new Uint32Array(r))}n={mmpHashDictBin:f,keys:r,keyToHashes:t}}if(i?.outputBinary){let e=function(e){let i=[],n=new TextEncoder,r=e=>{let t=new Uint8Array(4);return new DataView(t.buffer).setUint32(0,e,!1),t};for(let t of(i.push(r(e.mmpHashDictBin.length)),i.push(e.mmpHashDictBin),i.push(r(e.keys.length)),e.keys)){let e=n.encode(t);i.push(r(e.length)),i.push(e)}if(e.valueToKeyIndexes&&void 0!==e.bitsPerKey)if(i.push(r(0xffffffff)),i.push(r(e.bitsPerKey)),i.push(r(e.valueToKeyIndexes.length)),i.push(e.valueToKeyIndexes),e.collisionMap&&e.collisionMap.size>0){let n=[];t(e.collisionMap.size,n);let s=Array.from(e.collisionMap.keys()).sort((e,t)=>e-t),l=0;for(let i of s){t(i-l,n),l=i;let r=e.collisionMap.get(i);t(r.length,n),r.sort((e,t)=>e-t);let s=0;for(let e of r)t(e-s,n),s=e}let o=new Uint8Array(n);i.push(r(o.length)),i.push(o)}else i.push(r(0));else if(e.keyToHashes){let n=[];for(let i of e.keyToHashes){if(t(i.length,n),0===i.length)continue;let e=0,r=0,s=[];for(let t=0;t<i.length;t++){let n=i[t],l=n-r;s.push(l),l>e&&(e=l),r=n}let l=0;e>0&&(l=Math.ceil(Math.log2(e+1))),n.push(l);let o=new h;for(let e of s)o.write(e,l);let f=o.getData();for(let e=0;e<f.length;e++)n.push(f[e])}let s=new Uint8Array(n);i.push(r(s.length)),i.push(s)}let s=new Uint8Array(i.reduce((e,t)=>e+t.length,0)),l=0;for(let e of i)s.set(e,l),l+=e.length;return s}(n);return i?.enableCompression?o(e):e}return i?.enableCompression?Promise.resolve(n):n}function M(e){let t=new DataView(e.buffer,e.byteOffset,e.byteLength),n=0,r=new TextDecoder,s=()=>{let e=t.getUint32(n,!1);return n+=4,e},l=s(),o=e.subarray(n,n+l);n+=l;let f=s(),h=[];for(let t=0;t<f;t++){let t=s(),i=e.subarray(n,n+t);n+=t,h.push(r.decode(i))}let u=s();if(0xffffffff===u){let t,r=s(),l=s(),f=e.subarray(n,n+l);n+=l;let a=s();if(a>0){let r=e.subarray(n,n+a);n+=a,t=new Map;let s=0,{value:l,bytes:o}=i(r,0);s+=o;let f=0;for(let e=0;e<l;e++){let{value:e,bytes:n}=i(r,s);s+=n;let l=f+e;f=l;let{value:o,bytes:h}=i(r,s);s+=h;let a=[],u=0;for(let e=0;e<o;e++){let{value:e,bytes:t}=i(r,s);s+=t;let n=u+e;u=n,a.push(n)}t.set(l,a)}}return{mmpHashDictBin:o,keys:h,valueToKeyIndexes:f,bitsPerKey:r,collisionMap:t}}{let t=e.subarray(n,n+u);n+=u;let r=[],s=0;for(let e=0;e<f;e++){let{value:e,bytes:n}=i(t,s);if(s+=n,0===e){r.push(new Uint32Array(0));continue}let l=t[s];s+=1;let o=Math.ceil(l*e/8),f=t.subarray(s,s+o);s+=o;let h=new a(f),u=new Uint32Array(e),d=0;for(let t=0;t<e;t++)d+=h.read(l),u[t]=d;r.push(u)}return{mmpHashDictBin:o,keys:h,keyToHashes:r}}}class v{static async fromCompressed(e){return new v(M(await f(e)))}static fromBinary(e){return new v(M(e))}buildInvertedIndex(){if(!this.dict.keyToHashes)return;let e=this.mph.n;this._invertedIndex=Array.from({length:e},()=>[]);for(let t=0;t<this.dict.keys.length;t++){let i=this.dict.keyToHashes[t];for(let n=0;n<i.length;n++){let r=i[n];r<e&&this._invertedIndex[r].push(t)}}}query(e){if(this.dict.valueToKeyIndexes&&this.dict.bitsPerKey){let t=this.mph.hash(e);if(t<0||t>=this.mph.n)return null;let i=u(this.dict.valueToKeyIndexes,t*this.dict.bitsPerKey,this.dict.bitsPerKey);if(i===this.dict.keys.length){if(this.dict.collisionMap&&this.dict.collisionMap.has(t)){let e=this.dict.collisionMap.get(t);return e.length>0?this.dict.keys[e[0]]:null}return null}return i>=this.dict.keys.length?null:this.dict.keys[i]}let t=this.queryAll(e);return t&&t.length>0?t[0]:null}queryAll(e){if(this.dict.valueToKeyIndexes&&this.dict.bitsPerKey){let t=this.mph.hash(e);if(t<0||t>=this.mph.n)return null;let i=u(this.dict.valueToKeyIndexes,t*this.dict.bitsPerKey,this.dict.bitsPerKey);return i===this.dict.keys.length?this.dict.collisionMap&&this.dict.collisionMap.has(t)?this.dict.collisionMap.get(t).map(e=>this.dict.keys[e]):null:i>=this.dict.keys.length?null:[this.dict.keys[i]]}let t=this.mph.hash(e);if(t<0||!this._invertedIndex||t>=this._invertedIndex.length)return null;let i=this._invertedIndex[t];if(0===i.length)return null;let n=[];for(let e of i)n.push(this.dict.keys[e]);return n.length>0?n:null}keys(){return this.dict.keys}constructor(e){g(this,"dict",void 0),g(this,"mph",void 0),g(this,"_invertedIndex",void 0),this.dict=e,this._invertedIndex=null,this.mph=new y(e.mmpHashDictBin),e.keyToHashes&&this.buildInvertedIndex()}}export{y as MinMPHash,v as MinMPLookup,c as createMinMPHashDict,m as createMinMPLookupDict};
+function e(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}function t(e,t,i){let n=0;for(;e>=128;){let r=127&e|128;Array.isArray(t)?t.push(r):void 0!==i&&(t[i+n]=r),e>>>=7,n++}return Array.isArray(t)?t.push(e):void 0!==i&&(t[i+n]=e),n+1}function i(e,t){let i=0,n=0,r=0;for(;;){let s=e[t+r];if(i|=(127&s)<<n,r++,(128&s)==0)break;n+=7}return{value:i,bytes:r}}let n={encodeInt(e,t){e<24?t.push(0|e):e<=255?t.push(24,e):e<=65535?t.push(25,e>>8,255&e):t.push(26,e>>>24&255,e>>>16&255,e>>>8&255,255&e)},encodeBytes(e,t){let i=e.byteLength;i<24?t.push(64|i):i<=255?t.push(88,i):i<=65535?t.push(89,i>>8,255&i):t.push(90,i>>>24&255,i>>>16&255,i>>>8&255,255&i);for(let n=0;n<i;n++)t.push(e[n])},encodeNull(e){e.push(246)},encodeArrayHead(e,t){e<24&&t.push(128|e)},decode(e,t){let i=e.getUint8(t.current++),r=224&i,s=31&i,l=0;if(s<24)l=s;else if(24===s)l=e.getUint8(t.current),t.current+=1;else if(25===s)l=e.getUint16(t.current,!1),t.current+=2;else if(26===s)l=e.getUint32(t.current,!1),t.current+=4;else throw Error("Unsupported CBOR size");if(0===r)return l;if(64===r){let i=l,n=new Uint8Array(e.buffer.slice(e.byteOffset+t.current,e.byteOffset+t.current+i));return t.current+=i,n}if(128===r){let i=l,r=[];for(let s=0;s<i;s++)r.push(n.decode(e,t));return r}if(246===i)return null;throw Error(`Unknown CBOR type: ${i.toString(16)}`)}},r={none:0,4:1,8:2,16:3,32:4,2:5},s=["none","4","8","16","32","2"];function l(e){let t=[];if(n.encodeArrayHead(9,t),n.encodeInt(e.n,t),n.encodeInt(e.m,t),n.encodeInt(e.seed0,t),n.encodeBytes(e.bucketSizes,t),n.encodeBytes(e.seedStream,t),n.encodeInt(r[e.validationMode],t),e.fingerprints&&"none"!==e.validationMode){let i;if(e.fingerprints instanceof Uint8Array)i=e.fingerprints;else if(e.fingerprints instanceof Uint16Array||e.fingerprints instanceof Uint32Array)i=new Uint8Array(e.fingerprints.buffer,e.fingerprints.byteOffset,e.fingerprints.byteLength);else{let t,n=e.validationMode;t="2"===n||"4"===n||"8"===n?new Uint8Array(e.fingerprints):"16"===n?new Uint16Array(e.fingerprints):new Uint32Array(e.fingerprints),i=new Uint8Array(t.buffer,t.byteOffset,t.byteLength)}n.encodeBytes(i,t)}else n.encodeNull(t);return e.seedZeroBitmap?n.encodeBytes(e.seedZeroBitmap,t):n.encodeNull(t),n.encodeInt(e.hashSeed||0,t),new Uint8Array(t)}async function o(e){let t=new Blob([e]).stream().pipeThrough(new CompressionStream("gzip"));return new Uint8Array(await new Response(t).arrayBuffer())}async function f(e){let t=new Blob([e]).stream().pipeThrough(new DecompressionStream("gzip"));return new Uint8Array(await new Response(t).arrayBuffer())}class h{write(e,t){for(let i=0;i<t;i++){let t=e>>i&1;this.currentByte|=t<<this.bitCount,this.bitCount++,8===this.bitCount&&(this.buffer.push(this.currentByte),this.currentByte=0,this.bitCount=0)}}flush(){this.bitCount>0&&(this.buffer.push(this.currentByte),this.currentByte=0,this.bitCount=0)}getData(){return this.flush(),new Uint8Array(this.buffer)}constructor(){e(this,"buffer",[]),e(this,"currentByte",0),e(this,"bitCount",0)}}class a{read(e){let t=0;for(let i=0;i<e;i++){if(this.byteOffset>=this.buffer.length)return 0;t|=(this.buffer[this.byteOffset]>>this.bitOffset&1)<<i,this.bitOffset++,8===this.bitOffset&&(this.byteOffset++,this.bitOffset=0)}return t}constructor(t){e(this,"buffer",void 0),e(this,"byteOffset",0),e(this,"bitOffset",0),this.buffer=t}}function u(e,t,i){let n=0,r=t;for(let t=0;t<i;t++){let i=r>>>3,s=7&r;if(i>=e.length)return 0;n|=(e[i]>>s&1)<<t,r++}return n}function d(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}function c(e,t){let i=e.length;if(0===i){let e={n:0,m:0,seed0:0,seedStream:new Uint8Array(0),bucketSizes:new Uint8Array(0),validationMode:"none"};return t?.outputBinary?l(e):e}let n=t?.level??5,r="none";t?.onlySet===!0?r="8":"string"==typeof t?.onlySet&&(r=t.onlySet);let s=i>5e5?Math.max(1,.9*n):n,f=Math.max(1,Math.ceil(i/s)),h=new Uint32Array(i),a=new Uint32Array(i),u=0;for(;;){let t=new Map,n=new Map,r=!1;for(let s=0;s<i;s++){let i=b(e[s],u),l=b(e[s],~u),o=!1;if(n.has(i)){let e=n.get(i);e.has(l)?o=!0:e.add(l)}else if(t.has(i)){let e=t.get(i);if(e===l)o=!0;else{let r=new Set;r.add(e),r.add(l),n.set(i,r),t.delete(i)}}else t.set(i,l);if(o){r=!0;break}h[s]=i,a[s]=l}if(!r)break;if(++u>100)throw Error(`Could not find a collision-free hash seed after ${u} attempts.`)}let d=new Int32Array(f).fill(-1),c=new Int32Array(i).fill(-1),g=0,m=1/0,A=new Int32Array(f),M=new Int32Array(i),v=new Int32Array(f);for(let e=0;e<2e3;e++){let t=Math.floor(0xffffffff*Math.random());A.fill(-1),v.fill(0);let n=0;for(let e=0;e<i;e++){let i=Math.floor(((p(h[e],t)^a[e])>>>0)/0x100000000*f);M[e]=A[i],A[i]=e,v[i]++,v[i]>n&&(n=v[i])}if(n<13){g=t,d.set(A),c.set(M),m=n;break}if(n<m&&(m=n,g=t,d.set(A),c.set(M)),m<16&&e>50)break}if(m>=16)throw Error(`MPHF Build Failed: Could not find a bucket distribution with max size < 16 (best: ${m}). Try reducing the optimization level (current: ${t?.level??5}).`);let U=new Uint8Array(Math.ceil(f/2));for(let e=0;e<f;e++){let t=0,i=d[e];for(;-1!==i;)t++,i=c[i];let n=e>>>1;(1&e)==0?U[n]|=t:U[n]|=t<<4}let k=new w,x=new Uint8Array(Math.ceil(f/8));for(let e=0;e<f;e++){let t=0,i=d[e];for(;-1!==i;)t++,i=c[i];if(t<=1){x[e>>>3]|=1<<(7&e);continue}let n=0,r=!1,s=t>14?5e7:5e6;for(;!r;){let i=0,l=!1,o=d[e];for(;-1!==o;){let e=((p(h[o],n)^a[o])>>>0)%t;if((i&1<<e)!=0){l=!0;break}i|=1<<e,o=c[o]}if(l){if(++n>s)throw Error(`MPHF Failed: Bucket ${e} (size ${t}) is too hard.`)}else 0===n?x[e>>>3]|=1<<(7&e):k.write(n),r=!0}}let B={n:i,m:f,seed0:g,hashSeed:u,seedStream:k.toUint8Array(),bucketSizes:U,seedZeroBitmap:x,validationMode:r};if("none"!==r){let t;t="2"===r?new Uint8Array(Math.ceil(i/4)):"4"===r?new Uint8Array(Math.ceil(i/2)):"8"===r?new Uint8Array(i):"16"===r?new Uint16Array(i):new Uint32Array(i);let n=new y({...B,validationMode:"none"});for(let s=0;s<i;s++){let l=e[s],o=n.hash(l);if(o>=0&&o<i){let e=b(l,0x1234abcd);if("2"===r){let i=3&e,n=o>>>2,r=(3&o)<<1;t[n]|=i<<r}else if("4"===r){let i=15&e,n=o>>>1;(1&o)==0?t[n]|=i:t[n]|=i<<4}else"8"===r?t[o]=255&e:"16"===r?t[o]=65535&e:t[o]=e>>>0}}B.fingerprints=t}if(t?.outputBinary){let e=l(B);return t.enableCompression?o(e):e}return B}class y{static async fromCompressed(e){return new y(await f(e))}hash(e){if(0===this.n)return -1;let t=b(e,this.hashSeed),i=b(e,~this.hashSeed),n=Math.floor(((p(t,this.seed0)^i)>>>0)/0x100000000*this.m),r=this.offsets[n],s=this.offsets[n+1]-r;if(0===s)return -1;let l=0;if(l=1===s?r:r+((p(t,this.seeds[n])^i)>>>0)%s,"none"!==this.validationMode&&this.fingerprints){let t=b(e,y.FP_SEED);if("2"===this.validationMode){let e=l>>>2,i=(3&l)<<1;if((this.fingerprints[e]>>>i&3)!=(3&t))return -1}else if("4"===this.validationMode){let e=l>>>1,i=this.fingerprints[e];if(((1&l)==0?15&i:i>>>4&15)!=(15&t))return -1}else if("8"===this.validationMode){if(this.fingerprints[l]!==(255&t))return -1}else if("16"===this.validationMode){if(this.fingerprints[l]!==(65535&t))return -1}else if(this.fingerprints[l]!==t>>>0)return -1}return l}constructor(e){if(d(this,"n",void 0),d(this,"m",void 0),d(this,"seed0",void 0),d(this,"hashSeed",void 0),d(this,"offsets",void 0),d(this,"seeds",void 0),d(this,"validationMode",void 0),d(this,"fingerprints",null),e instanceof Uint8Array&&(e=function(e){let t,i=new DataView(e.buffer,e.byteOffset,e.byteLength),r=n.decode(i,{current:0});if(!Array.isArray(r)||r.length<7)throw Error("Invalid CBOR format");let[l,o,f,h,a,u,d,c,y]=r,p=s[u]||"none";return d&&"none"!==p&&("2"===p||"4"===p||"8"===p?t=d:"16"===p?t=new Uint16Array(d.buffer,d.byteOffset,d.byteLength/2):"32"===p&&(t=new Uint32Array(d.buffer,d.byteOffset,d.byteLength/4))),{n:l,m:o,seed0:f,hashSeed:y||0,bucketSizes:h,seedStream:a,validationMode:p,fingerprints:t,seedZeroBitmap:c||void 0}}(e)),this.n=e.n,this.m=e.m,this.seed0=e.seed0,this.hashSeed=e.hashSeed||0,this.validationMode=e.validationMode||"none",0===this.n){this.offsets=new Uint32Array(0),this.seeds=new Int32Array(0);return}this.offsets=new Uint32Array(this.m+1);let t=0;for(let i=0;i<this.m;i++){this.offsets[i]=t;let n=e.bucketSizes[i>>>1];t+=1&i?n>>>4:15&n}this.offsets[this.m]=t,this.seeds=new Int32Array(this.m);let i=0,r=e.seedStream,l=e.seedZeroBitmap;for(let e=0;e<this.m;e++){let t=!1;if(l&&(l[e>>>3]&1<<(7&e))!=0&&(t=!0),t)this.seeds[e]=0;else{let t=0,n=0;for(;;){let e=r[i++];if(t|=(127&e)<<n,(128&e)==0)break;n+=7}this.seeds[e]=t}}if("none"!==this.validationMode&&e.fingerprints){let t=e.fingerprints;"2"===this.validationMode||"4"===this.validationMode||"8"===this.validationMode?this.fingerprints=t instanceof Uint8Array?t:new Uint8Array(t):"16"===this.validationMode?this.fingerprints=t instanceof Uint16Array?t:new Uint16Array(t):this.fingerprints=t instanceof Uint32Array?t:new Uint32Array(t)}}}function p(e,t){return e^=t,e=Math.imul(e,0x85ebca6b),e^=e>>>13,e=Math.imul(e,0xc2b2ae35),(e^=e>>>16)>>>0}function b(e,t){let i=t;for(let t=0;t<e.length;t++){let n=e.charCodeAt(t);i^=n=Math.imul(n=(n=Math.imul(n,0xcc9e2d51))<<15|n>>>17,0x1b873593),i=Math.imul(i=i<<13|i>>>19,5)+0xe6546b64}return i^=e.length,i^=i>>>16,i=Math.imul(i,0x85ebca6b),i^=i>>>13,i=Math.imul(i,0xc2b2ae35),(i^=i>>>16)>>>0}d(y,"FP_SEED",0x1234abcd);class w{write(e){for(;e>=128;)this.buffer.push(127&e|128),e>>>=7;this.buffer.push(e)}toUint8Array(){return new Uint8Array(this.buffer)}constructor(){d(this,"buffer",[])}}function g(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}function m(e,i){let n,r=Object.keys(e),s=new Set;for(let t of r)for(let i of e[t])s.add(i);let l=Array.from(s),f=c(l,{level:i?.level,outputBinary:!0,onlySet:i?.onlySet??"8"}),a=new y(f),u=new Map;for(let t=0;t<r.length;t++)for(let i of e[r[t]])u.has(i)||u.set(i,[]),u.get(i).push(t);let d=0;for(let[e,t]of u)t.length>1&&d++;if(d<.1*l.length){let e=Math.ceil(Math.log2(r.length+1)),t=new h,i=new Map,s=new Int32Array(a.n).fill(-1);for(let[e,t]of u){let n=a.hash(e);n>=0&&(1===t.length?s[n]=t[0]:(s[n]=r.length,i.set(n,t)))}for(let i=0;i<a.n;i++){let n=s[i];t.write(n>=0?n:0,e)}n={mmpHashDictBin:f,keys:r,valueToKeyIndexes:t.getData(),bitsPerKey:e,collisionMap:i.size>0?i:void 0}}else{let t=[];for(let i of r){let n=e[i],r=[];for(let e of n){let t=a.hash(e);t>=0&&r.push(t)}r.sort((e,t)=>e-t),t.push(new Uint32Array(r))}n={mmpHashDictBin:f,keys:r,keyToHashes:t}}if(i?.outputBinary){let e=function(e){let i=[],n=new TextEncoder,r=e=>{let t=new Uint8Array(4);return new DataView(t.buffer).setUint32(0,e,!1),t};for(let t of(i.push(r(e.mmpHashDictBin.length)),i.push(e.mmpHashDictBin),i.push(r(e.keys.length)),e.keys)){let e=n.encode(t);i.push(r(e.length)),i.push(e)}if(e.valueToKeyIndexes&&void 0!==e.bitsPerKey)if(i.push(r(0xffffffff)),i.push(r(e.bitsPerKey)),i.push(r(e.valueToKeyIndexes.length)),i.push(e.valueToKeyIndexes),e.collisionMap&&e.collisionMap.size>0){let n=[];t(e.collisionMap.size,n);let s=Array.from(e.collisionMap.keys()).sort((e,t)=>e-t),l=0;for(let i of s){t(i-l,n),l=i;let r=e.collisionMap.get(i);t(r.length,n),r.sort((e,t)=>e-t);let s=0;for(let e of r)t(e-s,n),s=e}let o=new Uint8Array(n);i.push(r(o.length)),i.push(o)}else i.push(r(0));else if(e.keyToHashes){let n=[];for(let i of e.keyToHashes){if(t(i.length,n),0===i.length)continue;let e=0,r=0,s=[];for(let t=0;t<i.length;t++){let n=i[t],l=n-r;s.push(l),l>e&&(e=l),r=n}let l=0;e>0&&(l=Math.ceil(Math.log2(e+1))),n.push(l);let o=new h;for(let e of s)o.write(e,l);let f=o.getData();for(let e=0;e<f.length;e++)n.push(f[e])}let s=new Uint8Array(n);i.push(r(s.length)),i.push(s)}let s=new Uint8Array(i.reduce((e,t)=>e+t.length,0)),l=0;for(let e of i)s.set(e,l),l+=e.length;return s}(n);return i?.enableCompression?o(e):e}return i?.enableCompression?Promise.resolve(n):n}function A(e){let t=new DataView(e.buffer,e.byteOffset,e.byteLength),n=0,r=new TextDecoder,s=()=>{let e=t.getUint32(n,!1);return n+=4,e},l=s(),o=e.subarray(n,n+l);n+=l;let f=s(),h=[];for(let t=0;t<f;t++){let t=s(),i=e.subarray(n,n+t);n+=t,h.push(r.decode(i))}let u=s();if(0xffffffff===u){let t,r=s(),l=s(),f=e.subarray(n,n+l);n+=l;let a=s();if(a>0){let r=e.subarray(n,n+a);n+=a,t=new Map;let s=0,{value:l,bytes:o}=i(r,0);s+=o;let f=0;for(let e=0;e<l;e++){let{value:e,bytes:n}=i(r,s);s+=n;let l=f+e;f=l;let{value:o,bytes:h}=i(r,s);s+=h;let a=[],u=0;for(let e=0;e<o;e++){let{value:e,bytes:t}=i(r,s);s+=t;let n=u+e;u=n,a.push(n)}t.set(l,a)}}return{mmpHashDictBin:o,keys:h,valueToKeyIndexes:f,bitsPerKey:r,collisionMap:t}}{let t=e.subarray(n,n+u);n+=u;let r=[],s=0;for(let e=0;e<f;e++){let{value:e,bytes:n}=i(t,s);if(s+=n,0===e){r.push(new Uint32Array(0));continue}let l=t[s];s+=1;let o=Math.ceil(l*e/8),f=t.subarray(s,s+o);s+=o;let h=new a(f),u=new Uint32Array(e),d=0;for(let t=0;t<e;t++)d+=h.read(l),u[t]=d;r.push(u)}return{mmpHashDictBin:o,keys:h,keyToHashes:r}}}class M{static async fromCompressed(e){return new M(A(await f(e)))}static fromBinary(e){return new M(A(e))}buildInvertedIndex(){if(!this.dict.keyToHashes)return;let e=this.mph.n;this._invertedIndex=Array.from({length:e},()=>[]);for(let t=0;t<this.dict.keys.length;t++){let i=this.dict.keyToHashes[t];for(let n=0;n<i.length;n++){let r=i[n];r<e&&this._invertedIndex[r].push(t)}}}query(e){if(this.dict.valueToKeyIndexes&&this.dict.bitsPerKey){let t=this.mph.hash(e);if(t<0||t>=this.mph.n)return null;let i=u(this.dict.valueToKeyIndexes,t*this.dict.bitsPerKey,this.dict.bitsPerKey);if(i===this.dict.keys.length){if(this.dict.collisionMap&&this.dict.collisionMap.has(t)){let e=this.dict.collisionMap.get(t);return e.length>0?this.dict.keys[e[0]]:null}return null}return i>=this.dict.keys.length?null:this.dict.keys[i]}let t=this.queryAll(e);return t&&t.length>0?t[0]:null}queryAll(e){if(this.dict.valueToKeyIndexes&&this.dict.bitsPerKey){let t=this.mph.hash(e);if(t<0||t>=this.mph.n)return null;let i=u(this.dict.valueToKeyIndexes,t*this.dict.bitsPerKey,this.dict.bitsPerKey);return i===this.dict.keys.length?this.dict.collisionMap&&this.dict.collisionMap.has(t)?this.dict.collisionMap.get(t).map(e=>this.dict.keys[e]):null:i>=this.dict.keys.length?null:[this.dict.keys[i]]}let t=this.mph.hash(e);if(t<0||!this._invertedIndex||t>=this._invertedIndex.length)return null;let i=this._invertedIndex[t];if(0===i.length)return null;let n=[];for(let e of i)n.push(this.dict.keys[e]);return n.length>0?n:null}keys(){return this.dict.keys}constructor(e){g(this,"mph",void 0),g(this,"dict",void 0),g(this,"_invertedIndex",null),e instanceof Uint8Array&&(e=A(e)),this.dict=e,this.mph=new y(e.mmpHashDictBin),e.keyToHashes&&this.buildInvertedIndex()}}export{y as MinMPHash,M as MinMPLookup,c as createMinMPHashDict,m as createMinMPLookupDict};

package/dist/runtime.js

@@ -330,7 +330,7 @@ function deserializeLookupDict(data) {
         keys.push(decoder.decode(kBytes));
     }
     const sectionLen = readU32();
-    if (0xFFFFFFFF === sectionLen) {
+    if (0xffffffff === sectionLen) {
         const bitsPerKey = readU32();
         const dataLen = readU32();
         const valueToKeyIndexes = data.subarray(offset, offset + dataLen);
@@ -480,11 +480,11 @@ class MinMPLookup {
         return this.dict.keys;
     }
     constructor(dict){
-        MinMPLookup_define_property(this, "dict", void 0);
         MinMPLookup_define_property(this, "mph", void 0);
-        MinMPLookup_define_property(this, "_invertedIndex", void 0);
+        MinMPLookup_define_property(this, "dict", void 0);
+        MinMPLookup_define_property(this, "_invertedIndex", null);
+        if (dict instanceof Uint8Array) dict = deserializeLookupDict(dict);
         this.dict = dict;
-        this._invertedIndex = null;
         this.mph = new MinMPHash_MinMPHash(dict.mmpHashDictBin);
         if (dict.keyToHashes) this.buildInvertedIndex();
     }

package/dist/runtime.min.js

@@ -1 +1 @@
-function e(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}function t(e,t){let i=0,s=0,r=0;for(;;){let n=e[t+r];if(i|=(127&n)<<s,r++,(128&n)==0)break;s+=7}return{value:i,bytes:r}}let i={encodeInt(e,t){e<24?t.push(0|e):e<=255?t.push(24,e):e<=65535?t.push(25,e>>8,255&e):t.push(26,e>>>24&255,e>>>16&255,e>>>8&255,255&e)},encodeBytes(e,t){let i=e.byteLength;i<24?t.push(64|i):i<=255?t.push(88,i):i<=65535?t.push(89,i>>8,255&i):t.push(90,i>>>24&255,i>>>16&255,i>>>8&255,255&i);for(let s=0;s<i;s++)t.push(e[s])},encodeNull(e){e.push(246)},encodeArrayHead(e,t){e<24&&t.push(128|e)},decode(e,t){let s=e.getUint8(t.current++),r=224&s,n=31&s,h=0;if(n<24)h=n;else if(24===n)h=e.getUint8(t.current),t.current+=1;else if(25===n)h=e.getUint16(t.current,!1),t.current+=2;else if(26===n)h=e.getUint32(t.current,!1),t.current+=4;else throw Error("Unsupported CBOR size");if(0===r)return h;if(64===r){let i=h,s=new Uint8Array(e.buffer.slice(e.byteOffset+t.current,e.byteOffset+t.current+i));return t.current+=i,s}if(128===r){let s=h,r=[];for(let n=0;n<s;n++)r.push(i.decode(e,t));return r}if(246===s)return null;throw Error(`Unknown CBOR type: ${s.toString(16)}`)}},s=["none","4","8","16","32","2"];async function r(e){let t=new Blob([e]).stream().pipeThrough(new DecompressionStream("gzip"));return new Uint8Array(await new Response(t).arrayBuffer())}class n{read(e){let t=0;for(let i=0;i<e;i++){if(this.byteOffset>=this.buffer.length)return 0;t|=(this.buffer[this.byteOffset]>>this.bitOffset&1)<<i,this.bitOffset++,8===this.bitOffset&&(this.byteOffset++,this.bitOffset=0)}return t}constructor(t){e(this,"buffer",void 0),e(this,"byteOffset",0),e(this,"bitOffset",0),this.buffer=t}}function h(e,t,i){let s=0,r=t;for(let t=0;t<i;t++){let i=r>>>3,n=7&r;if(i>=e.length)return 0;s|=(e[i]>>n&1)<<t,r++}return s}function l(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}class f{static async fromCompressed(e){return new f(await r(e))}hash(e){if(0===this.n)return -1;let t=o(e,this.hashSeed),i=o(e,~this.hashSeed),s=Math.floor(((a(t,this.seed0)^i)>>>0)/0x100000000*this.m),r=this.offsets[s],n=this.offsets[s+1]-r;if(0===n)return -1;let h=0;if(h=1===n?r:r+((a(t,this.seeds[s])^i)>>>0)%n,"none"!==this.validationMode&&this.fingerprints){let t=o(e,f.FP_SEED);if("2"===this.validationMode){let e=h>>>2,i=(3&h)<<1;if((this.fingerprints[e]>>>i&3)!=(3&t))return -1}else if("4"===this.validationMode){let e=h>>>1,i=this.fingerprints[e];if(((1&h)==0?15&i:i>>>4&15)!=(15&t))return -1}else if("8"===this.validationMode){if(this.fingerprints[h]!==(255&t))return -1}else if("16"===this.validationMode){if(this.fingerprints[h]!==(65535&t))return -1}else if(this.fingerprints[h]!==t>>>0)return -1}return h}constructor(e){if(l(this,"n",void 0),l(this,"m",void 0),l(this,"seed0",void 0),l(this,"hashSeed",void 0),l(this,"offsets",void 0),l(this,"seeds",void 0),l(this,"validationMode",void 0),l(this,"fingerprints",null),e instanceof Uint8Array&&(e=function(e){let t,r=new DataView(e.buffer,e.byteOffset,e.byteLength),n=i.decode(r,{current:0});if(!Array.isArray(n)||n.length<7)throw Error("Invalid CBOR format");let[h,l,f,a,o,u,d,c,y]=n,b=s[u]||"none";return d&&"none"!==b&&("2"===b||"4"===b||"8"===b?t=d:"16"===b?t=new Uint16Array(d.buffer,d.byteOffset,d.byteLength/2):"32"===b&&(t=new Uint32Array(d.buffer,d.byteOffset,d.byteLength/4))),{n:h,m:l,seed0:f,hashSeed:y||0,bucketSizes:a,seedStream:o,validationMode:b,fingerprints:t,seedZeroBitmap:c||void 0}}(e)),this.n=e.n,this.m=e.m,this.seed0=e.seed0,this.hashSeed=e.hashSeed||0,this.validationMode=e.validationMode||"none",0===this.n){this.offsets=new Uint32Array(0),this.seeds=new Int32Array(0);return}this.offsets=new Uint32Array(this.m+1);let t=0;for(let i=0;i<this.m;i++){this.offsets[i]=t;let s=e.bucketSizes[i>>>1];t+=1&i?s>>>4:15&s}this.offsets[this.m]=t,this.seeds=new Int32Array(this.m);let r=0,n=e.seedStream,h=e.seedZeroBitmap;for(let e=0;e<this.m;e++){let t=!1;if(h&&(h[e>>>3]&1<<(7&e))!=0&&(t=!0),t)this.seeds[e]=0;else{let t=0,i=0;for(;;){let e=n[r++];if(t|=(127&e)<<i,(128&e)==0)break;i+=7}this.seeds[e]=t}}if("none"!==this.validationMode&&e.fingerprints){let t=e.fingerprints;"2"===this.validationMode||"4"===this.validationMode||"8"===this.validationMode?this.fingerprints=t instanceof Uint8Array?t:new Uint8Array(t):"16"===this.validationMode?this.fingerprints=t instanceof Uint16Array?t:new Uint16Array(t):this.fingerprints=t instanceof Uint32Array?t:new Uint32Array(t)}}}function a(e,t){return e^=t,e=Math.imul(e,0x85ebca6b),e^=e>>>13,e=Math.imul(e,0xc2b2ae35),(e^=e>>>16)>>>0}function o(e,t){let i=t;for(let t=0;t<e.length;t++){let s=e.charCodeAt(t);i^=s=Math.imul(s=(s=Math.imul(s,0xcc9e2d51))<<15|s>>>17,0x1b873593),i=Math.imul(i=i<<13|i>>>19,5)+0xe6546b64}return i^=e.length,i^=i>>>16,i=Math.imul(i,0x85ebca6b),i^=i>>>13,i=Math.imul(i,0xc2b2ae35),(i^=i>>>16)>>>0}function u(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}function d(e){let i=new DataView(e.buffer,e.byteOffset,e.byteLength),s=0,r=new TextDecoder,h=()=>{let e=i.getUint32(s,!1);return s+=4,e},l=h(),f=e.subarray(s,s+l);s+=l;let a=h(),o=[];for(let t=0;t<a;t++){let t=h(),i=e.subarray(s,s+t);s+=t,o.push(r.decode(i))}let u=h();if(0xffffffff===u){let i,r=h(),n=h(),l=e.subarray(s,s+n);s+=n;let a=h();if(a>0){let r=e.subarray(s,s+a);s+=a,i=new Map;let n=0,{value:h,bytes:l}=t(r,0);n+=l;let f=0;for(let e=0;e<h;e++){let{value:e,bytes:s}=t(r,n);n+=s;let h=f+e;f=h;let{value:l,bytes:a}=t(r,n);n+=a;let o=[],u=0;for(let e=0;e<l;e++){let{value:e,bytes:i}=t(r,n);n+=i;let s=u+e;u=s,o.push(s)}i.set(h,o)}}return{mmpHashDictBin:f,keys:o,valueToKeyIndexes:l,bitsPerKey:r,collisionMap:i}}{let i=e.subarray(s,s+u);s+=u;let r=[],h=0;for(let e=0;e<a;e++){let{value:e,bytes:s}=t(i,h);if(h+=s,0===e){r.push(new Uint32Array(0));continue}let l=i[h];h+=1;let f=Math.ceil(l*e/8),a=i.subarray(h,h+f);h+=f;let o=new n(a),u=new Uint32Array(e),d=0;for(let t=0;t<e;t++)d+=o.read(l),u[t]=d;r.push(u)}return{mmpHashDictBin:f,keys:o,keyToHashes:r}}}l(f,"FP_SEED",0x1234abcd);class c{static async fromCompressed(e){return new c(d(await r(e)))}static fromBinary(e){return new c(d(e))}buildInvertedIndex(){if(!this.dict.keyToHashes)return;let e=this.mph.n;this._invertedIndex=Array.from({length:e},()=>[]);for(let t=0;t<this.dict.keys.length;t++){let i=this.dict.keyToHashes[t];for(let s=0;s<i.length;s++){let r=i[s];r<e&&this._invertedIndex[r].push(t)}}}query(e){if(this.dict.valueToKeyIndexes&&this.dict.bitsPerKey){let t=this.mph.hash(e);if(t<0||t>=this.mph.n)return null;let i=h(this.dict.valueToKeyIndexes,t*this.dict.bitsPerKey,this.dict.bitsPerKey);if(i===this.dict.keys.length){if(this.dict.collisionMap&&this.dict.collisionMap.has(t)){let e=this.dict.collisionMap.get(t);return e.length>0?this.dict.keys[e[0]]:null}return null}return i>=this.dict.keys.length?null:this.dict.keys[i]}let t=this.queryAll(e);return t&&t.length>0?t[0]:null}queryAll(e){if(this.dict.valueToKeyIndexes&&this.dict.bitsPerKey){let t=this.mph.hash(e);if(t<0||t>=this.mph.n)return null;let i=h(this.dict.valueToKeyIndexes,t*this.dict.bitsPerKey,this.dict.bitsPerKey);return i===this.dict.keys.length?this.dict.collisionMap&&this.dict.collisionMap.has(t)?this.dict.collisionMap.get(t).map(e=>this.dict.keys[e]):null:i>=this.dict.keys.length?null:[this.dict.keys[i]]}let t=this.mph.hash(e);if(t<0||!this._invertedIndex||t>=this._invertedIndex.length)return null;let i=this._invertedIndex[t];if(0===i.length)return null;let s=[];for(let e of i)s.push(this.dict.keys[e]);return s.length>0?s:null}keys(){return this.dict.keys}constructor(e){u(this,"dict",void 0),u(this,"mph",void 0),u(this,"_invertedIndex",void 0),this.dict=e,this._invertedIndex=null,this.mph=new f(e.mmpHashDictBin),e.keyToHashes&&this.buildInvertedIndex()}}export{f as MinMPHash,c as MinMPLookup};
+function e(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}function t(e,t){let i=0,s=0,r=0;for(;;){let n=e[t+r];if(i|=(127&n)<<s,r++,(128&n)==0)break;s+=7}return{value:i,bytes:r}}let i={encodeInt(e,t){e<24?t.push(0|e):e<=255?t.push(24,e):e<=65535?t.push(25,e>>8,255&e):t.push(26,e>>>24&255,e>>>16&255,e>>>8&255,255&e)},encodeBytes(e,t){let i=e.byteLength;i<24?t.push(64|i):i<=255?t.push(88,i):i<=65535?t.push(89,i>>8,255&i):t.push(90,i>>>24&255,i>>>16&255,i>>>8&255,255&i);for(let s=0;s<i;s++)t.push(e[s])},encodeNull(e){e.push(246)},encodeArrayHead(e,t){e<24&&t.push(128|e)},decode(e,t){let s=e.getUint8(t.current++),r=224&s,n=31&s,h=0;if(n<24)h=n;else if(24===n)h=e.getUint8(t.current),t.current+=1;else if(25===n)h=e.getUint16(t.current,!1),t.current+=2;else if(26===n)h=e.getUint32(t.current,!1),t.current+=4;else throw Error("Unsupported CBOR size");if(0===r)return h;if(64===r){let i=h,s=new Uint8Array(e.buffer.slice(e.byteOffset+t.current,e.byteOffset+t.current+i));return t.current+=i,s}if(128===r){let s=h,r=[];for(let n=0;n<s;n++)r.push(i.decode(e,t));return r}if(246===s)return null;throw Error(`Unknown CBOR type: ${s.toString(16)}`)}},s=["none","4","8","16","32","2"];async function r(e){let t=new Blob([e]).stream().pipeThrough(new DecompressionStream("gzip"));return new Uint8Array(await new Response(t).arrayBuffer())}class n{read(e){let t=0;for(let i=0;i<e;i++){if(this.byteOffset>=this.buffer.length)return 0;t|=(this.buffer[this.byteOffset]>>this.bitOffset&1)<<i,this.bitOffset++,8===this.bitOffset&&(this.byteOffset++,this.bitOffset=0)}return t}constructor(t){e(this,"buffer",void 0),e(this,"byteOffset",0),e(this,"bitOffset",0),this.buffer=t}}function h(e,t,i){let s=0,r=t;for(let t=0;t<i;t++){let i=r>>>3,n=7&r;if(i>=e.length)return 0;s|=(e[i]>>n&1)<<t,r++}return s}function l(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}class f{static async fromCompressed(e){return new f(await r(e))}hash(e){if(0===this.n)return -1;let t=o(e,this.hashSeed),i=o(e,~this.hashSeed),s=Math.floor(((a(t,this.seed0)^i)>>>0)/0x100000000*this.m),r=this.offsets[s],n=this.offsets[s+1]-r;if(0===n)return -1;let h=0;if(h=1===n?r:r+((a(t,this.seeds[s])^i)>>>0)%n,"none"!==this.validationMode&&this.fingerprints){let t=o(e,f.FP_SEED);if("2"===this.validationMode){let e=h>>>2,i=(3&h)<<1;if((this.fingerprints[e]>>>i&3)!=(3&t))return -1}else if("4"===this.validationMode){let e=h>>>1,i=this.fingerprints[e];if(((1&h)==0?15&i:i>>>4&15)!=(15&t))return -1}else if("8"===this.validationMode){if(this.fingerprints[h]!==(255&t))return -1}else if("16"===this.validationMode){if(this.fingerprints[h]!==(65535&t))return -1}else if(this.fingerprints[h]!==t>>>0)return -1}return h}constructor(e){if(l(this,"n",void 0),l(this,"m",void 0),l(this,"seed0",void 0),l(this,"hashSeed",void 0),l(this,"offsets",void 0),l(this,"seeds",void 0),l(this,"validationMode",void 0),l(this,"fingerprints",null),e instanceof Uint8Array&&(e=function(e){let t,r=new DataView(e.buffer,e.byteOffset,e.byteLength),n=i.decode(r,{current:0});if(!Array.isArray(n)||n.length<7)throw Error("Invalid CBOR format");let[h,l,f,a,o,u,d,c,y]=n,b=s[u]||"none";return d&&"none"!==b&&("2"===b||"4"===b||"8"===b?t=d:"16"===b?t=new Uint16Array(d.buffer,d.byteOffset,d.byteLength/2):"32"===b&&(t=new Uint32Array(d.buffer,d.byteOffset,d.byteLength/4))),{n:h,m:l,seed0:f,hashSeed:y||0,bucketSizes:a,seedStream:o,validationMode:b,fingerprints:t,seedZeroBitmap:c||void 0}}(e)),this.n=e.n,this.m=e.m,this.seed0=e.seed0,this.hashSeed=e.hashSeed||0,this.validationMode=e.validationMode||"none",0===this.n){this.offsets=new Uint32Array(0),this.seeds=new Int32Array(0);return}this.offsets=new Uint32Array(this.m+1);let t=0;for(let i=0;i<this.m;i++){this.offsets[i]=t;let s=e.bucketSizes[i>>>1];t+=1&i?s>>>4:15&s}this.offsets[this.m]=t,this.seeds=new Int32Array(this.m);let r=0,n=e.seedStream,h=e.seedZeroBitmap;for(let e=0;e<this.m;e++){let t=!1;if(h&&(h[e>>>3]&1<<(7&e))!=0&&(t=!0),t)this.seeds[e]=0;else{let t=0,i=0;for(;;){let e=n[r++];if(t|=(127&e)<<i,(128&e)==0)break;i+=7}this.seeds[e]=t}}if("none"!==this.validationMode&&e.fingerprints){let t=e.fingerprints;"2"===this.validationMode||"4"===this.validationMode||"8"===this.validationMode?this.fingerprints=t instanceof Uint8Array?t:new Uint8Array(t):"16"===this.validationMode?this.fingerprints=t instanceof Uint16Array?t:new Uint16Array(t):this.fingerprints=t instanceof Uint32Array?t:new Uint32Array(t)}}}function a(e,t){return e^=t,e=Math.imul(e,0x85ebca6b),e^=e>>>13,e=Math.imul(e,0xc2b2ae35),(e^=e>>>16)>>>0}function o(e,t){let i=t;for(let t=0;t<e.length;t++){let s=e.charCodeAt(t);i^=s=Math.imul(s=(s=Math.imul(s,0xcc9e2d51))<<15|s>>>17,0x1b873593),i=Math.imul(i=i<<13|i>>>19,5)+0xe6546b64}return i^=e.length,i^=i>>>16,i=Math.imul(i,0x85ebca6b),i^=i>>>13,i=Math.imul(i,0xc2b2ae35),(i^=i>>>16)>>>0}function u(e,t,i){return t in e?Object.defineProperty(e,t,{value:i,enumerable:!0,configurable:!0,writable:!0}):e[t]=i,e}function d(e){let i=new DataView(e.buffer,e.byteOffset,e.byteLength),s=0,r=new TextDecoder,h=()=>{let e=i.getUint32(s,!1);return s+=4,e},l=h(),f=e.subarray(s,s+l);s+=l;let a=h(),o=[];for(let t=0;t<a;t++){let t=h(),i=e.subarray(s,s+t);s+=t,o.push(r.decode(i))}let u=h();if(0xffffffff===u){let i,r=h(),n=h(),l=e.subarray(s,s+n);s+=n;let a=h();if(a>0){let r=e.subarray(s,s+a);s+=a,i=new Map;let n=0,{value:h,bytes:l}=t(r,0);n+=l;let f=0;for(let e=0;e<h;e++){let{value:e,bytes:s}=t(r,n);n+=s;let h=f+e;f=h;let{value:l,bytes:a}=t(r,n);n+=a;let o=[],u=0;for(let e=0;e<l;e++){let{value:e,bytes:i}=t(r,n);n+=i;let s=u+e;u=s,o.push(s)}i.set(h,o)}}return{mmpHashDictBin:f,keys:o,valueToKeyIndexes:l,bitsPerKey:r,collisionMap:i}}{let i=e.subarray(s,s+u);s+=u;let r=[],h=0;for(let e=0;e<a;e++){let{value:e,bytes:s}=t(i,h);if(h+=s,0===e){r.push(new Uint32Array(0));continue}let l=i[h];h+=1;let f=Math.ceil(l*e/8),a=i.subarray(h,h+f);h+=f;let o=new n(a),u=new Uint32Array(e),d=0;for(let t=0;t<e;t++)d+=o.read(l),u[t]=d;r.push(u)}return{mmpHashDictBin:f,keys:o,keyToHashes:r}}}l(f,"FP_SEED",0x1234abcd);class c{static async fromCompressed(e){return new c(d(await r(e)))}static fromBinary(e){return new c(d(e))}buildInvertedIndex(){if(!this.dict.keyToHashes)return;let e=this.mph.n;this._invertedIndex=Array.from({length:e},()=>[]);for(let t=0;t<this.dict.keys.length;t++){let i=this.dict.keyToHashes[t];for(let s=0;s<i.length;s++){let r=i[s];r<e&&this._invertedIndex[r].push(t)}}}query(e){if(this.dict.valueToKeyIndexes&&this.dict.bitsPerKey){let t=this.mph.hash(e);if(t<0||t>=this.mph.n)return null;let i=h(this.dict.valueToKeyIndexes,t*this.dict.bitsPerKey,this.dict.bitsPerKey);if(i===this.dict.keys.length){if(this.dict.collisionMap&&this.dict.collisionMap.has(t)){let e=this.dict.collisionMap.get(t);return e.length>0?this.dict.keys[e[0]]:null}return null}return i>=this.dict.keys.length?null:this.dict.keys[i]}let t=this.queryAll(e);return t&&t.length>0?t[0]:null}queryAll(e){if(this.dict.valueToKeyIndexes&&this.dict.bitsPerKey){let t=this.mph.hash(e);if(t<0||t>=this.mph.n)return null;let i=h(this.dict.valueToKeyIndexes,t*this.dict.bitsPerKey,this.dict.bitsPerKey);return i===this.dict.keys.length?this.dict.collisionMap&&this.dict.collisionMap.has(t)?this.dict.collisionMap.get(t).map(e=>this.dict.keys[e]):null:i>=this.dict.keys.length?null:[this.dict.keys[i]]}let t=this.mph.hash(e);if(t<0||!this._invertedIndex||t>=this._invertedIndex.length)return null;let i=this._invertedIndex[t];if(0===i.length)return null;let s=[];for(let e of i)s.push(this.dict.keys[e]);return s.length>0?s:null}keys(){return this.dict.keys}constructor(e){u(this,"mph",void 0),u(this,"dict",void 0),u(this,"_invertedIndex",null),e instanceof Uint8Array&&(e=d(e)),this.dict=e,this.mph=new f(e.mmpHashDictBin),e.keyToHashes&&this.buildInvertedIndex()}}export{f as MinMPHash,c as MinMPLookup};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "min-mphash",
-  "version": "0.5.1",
+  "version": "0.5.2",
   "type": "module",
   "author": "git@qzrzz.com",
   "license": "MIT",