@ssttevee/streamsearch 0.3.0 → 0.4.0

package/lib/{index.js → search.cjs}

@@ -1,272 +1,188 @@
 'use strict';
 
-Object.defineProperty(exports, '__esModule', { value: true });
+var concat = require('uint8arrays/concat');
+var fromString = require('uint8arrays/from-string');
 
-var u8Utils = require('@ssttevee/u8-utils');
-
-/*
-  Based heavily on the Streaming Boyer-Moore-Horspool C++ implementation
-  by Hongli Lai at: https://github.com/FooBarWidget/boyer-moore-horspool
-*/
-function coerce(a) {
-    if (a instanceof Uint8Array) {
-        return (index) => a[index];
-    }
-    return a;
-}
-function jsmemcmp(buf1, pos1, buf2, pos2, len) {
-    const fn1 = coerce(buf1);
-    const fn2 = coerce(buf2);
-    for (var i = 0; i < len; ++i) {
-        if (fn1(pos1 + i) !== fn2(pos2 + i)) {
-            return false;
-        }
-    }
-    return true;
-}
-function createOccurenceTable(s) {
-    // Populate occurrence table with analysis of the needle,
-    // ignoring last letter.
-    const table = new Array(256).fill(s.length);
-    if (s.length > 1) {
-        for (let i = 0; i < s.length - 1; i++) {
-            table[s[i]] = s.length - 1 - i;
-        }
-    }
-    return table;
-}
-const MATCH = Symbol('Match');
-class StreamSearch {
-    constructor(needle) {
-        this._lookbehind = new Uint8Array();
-        if (typeof needle === 'string') {
-            this._needle = needle = u8Utils.stringToArray(needle);
-        }
-        else {
-            this._needle = needle;
-        }
-        this._lastChar = needle[needle.length - 1];
-        this._occ = createOccurenceTable(needle);
-    }
-    feed(chunk) {
-        let pos = 0;
-        let tokens;
-        const allTokens = [];
-        while (pos !== chunk.length) {
-            [pos, ...tokens] = this._feed(chunk, pos);
-            allTokens.push(...tokens);
-        }
-        return allTokens;
-    }
-    end() {
-        const tail = this._lookbehind;
-        this._lookbehind = new Uint8Array();
-        return tail;
-    }
-    _feed(data, buf_pos) {
-        const tokens = [];
-        // Positive: points to a position in `data`
-        //           pos == 3 points to data[3]
-        // Negative: points to a position in the lookbehind buffer
-        //           pos == -2 points to lookbehind[lookbehind_size - 2]
-        let pos = -this._lookbehind.length;
-        if (pos < 0) {
-            // Lookbehind buffer is not empty. Perform Boyer-Moore-Horspool
-            // search with character lookup code that considers both the
-            // lookbehind buffer and the current round's haystack data.
-            //
-            // Loop until (condition 1)
-            //   there is a match.
-            // or until
-            //   we've moved past the position that requires the
-            //   lookbehind buffer. In this case we switch to the
-            //   optimized loop.
-            // or until (condition 3)
-            //   the character to look at lies outside the haystack.
-            while (pos < 0 && pos <= data.length - this._needle.length) {
-                const ch = this._charAt(data, pos + this._needle.length - 1);
-                if (ch === this._lastChar && this._memcmp(data, pos, this._needle.length - 1)) {
-                    if (pos > -this._lookbehind.length) {
-                        tokens.push(this._lookbehind.slice(0, this._lookbehind.length + pos));
-                    }
-                    tokens.push(MATCH);
-                    this._lookbehind = new Uint8Array();
-                    return [pos + this._needle.length, ...tokens];
-                }
-                else {
-                    pos += this._occ[ch];
-                }
-            }
-            // No match.
-            if (pos < 0) {
-                // There's too little data for Boyer-Moore-Horspool to run,
-                // so we'll use a different algorithm to skip as much as
-                // we can.
-                // Forward pos until
-                //   the trailing part of lookbehind + data
-                //   looks like the beginning of the needle
-                // or until
-                //   pos == 0
-                while (pos < 0 && !this._memcmp(data, pos, data.length - pos)) {
-                    pos++;
-                }
-            }
-            if (pos >= 0) {
-                // Discard lookbehind buffer.
-                tokens.push(this._lookbehind);
-                this._lookbehind = new Uint8Array();
-            }
-            else {
-                // Cut off part of the lookbehind buffer that has
-                // been processed and append the entire haystack
-                // into it.
-                const bytesToCutOff = this._lookbehind.length + pos;
-                if (bytesToCutOff > 0) {
-                    // The cut off data is guaranteed not to contain the needle.
-                    tokens.push(this._lookbehind.slice(0, bytesToCutOff));
-                    this._lookbehind = this._lookbehind.slice(bytesToCutOff);
-                }
-                this._lookbehind = Uint8Array.from(new Array(this._lookbehind.length + data.length), (_, i) => this._charAt(data, i - this._lookbehind.length));
-                return [data.length, ...tokens];
-            }
-        }
-        pos += buf_pos;
-        // Lookbehind buffer is now empty. Perform Boyer-Moore-Horspool
-        // search with optimized character lookup code that only considers
-        // the current round's haystack data.
-        while (pos <= data.length - this._needle.length) {
-            const ch = data[pos + this._needle.length - 1];
-            if (ch === this._lastChar
-                && data[pos] === this._needle[0]
-                && jsmemcmp(this._needle, 0, data, pos, this._needle.length - 1)) {
-                if (pos > buf_pos) {
-                    tokens.push(data.slice(buf_pos, pos));
-                }
-                tokens.push(MATCH);
-                return [pos + this._needle.length, ...tokens];
-            }
-            else {
-                pos += this._occ[ch];
-            }
-        }
-        // There was no match. If there's trailing haystack data that we cannot
-        // match yet using the Boyer-Moore-Horspool algorithm (because the trailing
-        // data is less than the needle size) then match using a modified
-        // algorithm that starts matching from the beginning instead of the end.
-        // Whatever trailing data is left after running this algorithm is added to
-        // the lookbehind buffer.
-        if (pos < data.length) {
-            while (pos < data.length && (data[pos] !== this._needle[0]
-                || !jsmemcmp(data, pos, this._needle, 0, data.length - pos))) {
-                ++pos;
-            }
-            if (pos < data.length) {
-                this._lookbehind = data.slice(pos);
-            }
-        }
-        // Everything until pos is guaranteed not to contain needle data.
-        if (pos > 0) {
-            tokens.push(data.slice(buf_pos, pos < data.length ? pos : data.length));
-        }
-        return [data.length, ...tokens];
-    }
-    _charAt(data, pos) {
-        if (pos < 0) {
-            return this._lookbehind[this._lookbehind.length + pos];
-        }
-        return data[pos];
-    }
-    ;
-    _memcmp(data, pos, len) {
-        return jsmemcmp(this._charAt.bind(this, data), pos, this._needle, 0, len);
-    }
-    ;
+/*
+  Based heavily on the Streaming Boyer-Moore-Horspool C++ implementation
+  by Hongli Lai at: https://github.com/FooBarWidget/boyer-moore-horspool
+*/
+function jsmemcmp(buf1, pos1, buf2, pos2, len) {
+    for (let i = 0; i < len; i++) {
+        if (buf1[pos1 + i] !== buf2[pos2 + i]) {
+            return false;
+        }
+    }
+    return true;
 }
-
-class ReadableStreamSearch {
-    constructor(needle, _readableStream) {
-        this._readableStream = _readableStream;
-        this._search = new StreamSearch(needle);
-    }
-    async *[Symbol.asyncIterator]() {
-        const reader = this._readableStream.getReader();
-        try {
-            while (true) {
-                const result = await reader.read();
-                if (result.done) {
-                    break;
-                }
-                yield* this._search.feed(result.value);
-            }
-            const tail = this._search.end();
-            if (tail.length) {
-                yield tail;
-            }
-        }
-        finally {
-            reader.releaseLock();
-        }
-    }
+function createOccurenceTable(s) {
+    // Populate occurrence table with analysis of the needle,
+    // ignoring last letter.
+    const table = new Array(256).fill(s.length);
+    if (s.length > 1) {
+        for (let i = 0; i < s.length - 1; i++) {
+            table[s[i]] = s.length - 1 - i;
+        }
+    }
+    return table;
 }
-
-function splitChunks(chunks, needle) {
-    const search = new StreamSearch(needle);
-    const outchunks = [[]];
-    for (const chunk of chunks) {
-        for (const token of search.feed(chunk)) {
-            if (token === MATCH) {
-                outchunks.push([]);
-            }
-            else {
-                outchunks[outchunks.length - 1].push(token);
-            }
-        }
-    }
-    const end = search.end();
-    outchunks[outchunks.length - 1].push(end);
-    return outchunks.map((chunks) => u8Utils.mergeArrays(...chunks));
-}
-function split(buf, needle) {
-    return splitChunks([buf], needle);
-}
-async function* chunksIterator(iter) {
-    let chunks = [];
-    for await (const value of iter) {
-        if (value === MATCH) {
-            yield chunks;
-            chunks = [];
-        }
-        else {
-            chunks.push(value);
-        }
-    }
-    yield chunks;
-}
-async function* stringIterator(iter) {
-    for await (const chunk of chunksIterator(iter)) {
-        yield chunk.map(u8Utils.arrayToString).join('');
-    }
-}
-async function allStrings(iter) {
-    const segments = [];
-    for await (const value of stringIterator(iter)) {
-        segments.push(value);
-    }
-    return segments;
-}
-async function* arrayIterator(iter) {
-    for await (const chunk of chunksIterator(iter)) {
-        yield u8Utils.mergeArrays(...chunk);
-    }
+const MATCH = Symbol("Match");
+class StreamSearch {
+    _needle;
+    _lastChar;
+    _occ;
+    _lookbehind = new Uint8Array();
+    constructor(needle) {
+        if (typeof needle === "string") {
+            this._needle = needle = fromString.fromString(needle);
+        }
+        else {
+            this._needle = needle;
+        }
+        this._lastChar = needle[needle.length - 1];
+        this._occ = createOccurenceTable(needle);
+    }
+    feed(chunk) {
+        let pos = 0;
+        let tokens;
+        const allTokens = [];
+        while (pos !== chunk.length) {
+            [pos, ...tokens] = this._feed(chunk, pos);
+            allTokens.push(...tokens);
+        }
+        return allTokens;
+    }
+    end() {
+        const tail = this._lookbehind;
+        this._lookbehind = new Uint8Array();
+        return tail;
+    }
+    _feed(data, buf_pos) {
+        const tokens = [];
+        // Positive: points to a position in `data`
+        //           pos == 3 points to data[3]
+        // Negative: points to a position in the lookbehind buffer
+        //           pos == -2 points to lookbehind[lookbehind_size - 2]
+        let pos = -this._lookbehind.length;
+        if (pos < 0) {
+            // Lookbehind buffer is not empty. Perform Boyer-Moore-Horspool
+            // search with character lookup code that considers both the
+            // lookbehind buffer and the current round's haystack data.
+            //
+            // Loop until (condition 1)
+            //   there is a match.
+            // or until
+            //   we've moved past the position that requires the
+            //   lookbehind buffer. In this case we switch to the
+            //   optimized loop.
+            // or until (condition 3)
+            //   the character to look at lies outside the haystack.
+            while (pos < 0 && pos <= data.length - this._needle.length) {
+                const cpos = pos + this._needle.length - 1;
+                const ch = cpos < 0
+                    ? this._lookbehind[this._lookbehind.length + cpos]
+                    : data[cpos];
+                if (ch === this._lastChar &&
+                    this._memcmp(data, pos, this._needle.length - 1)) {
+                    if (pos > -this._lookbehind.length) {
+                        tokens.push(this._lookbehind.subarray(0, this._lookbehind.length + pos));
+                    }
+                    tokens.push(MATCH);
+                    this._lookbehind = new Uint8Array();
+                    return [pos + this._needle.length, ...tokens];
+                }
+                else {
+                    pos += this._occ[ch];
+                }
+            }
+            // No match.
+            if (pos < 0) {
+                // There's too little data for Boyer-Moore-Horspool to run,
+                // so we'll use a different algorithm to skip as much as
+                // we can.
+                // Forward pos until
+                //   the trailing part of lookbehind + data
+                //   looks like the beginning of the needle
+                // or until
+                //   pos == 0
+                while (pos < 0 && !this._memcmp(data, pos, data.length - pos)) {
+                    pos++;
+                }
+            }
+            if (pos >= 0) {
+                // Discard lookbehind buffer.
+                tokens.push(this._lookbehind);
+                this._lookbehind = new Uint8Array();
+            }
+            else {
+                // Cut off part of the lookbehind buffer that has
+                // been processed and append the entire haystack
+                // into it.
+                const bytesToCutOff = this._lookbehind.length + pos;
+                if (bytesToCutOff > 0) {
+                    // The cut off data is guaranteed not to contain the needle.
+                    tokens.push(this._lookbehind.subarray(0, bytesToCutOff));
+                    this._lookbehind = this._lookbehind.subarray(bytesToCutOff);
+                }
+                this._lookbehind = concat.concat([this._lookbehind, data]);
+                return [data.length, ...tokens];
+            }
+        }
+        pos += buf_pos;
+        // Lookbehind buffer is now empty. Perform Boyer-Moore-Horspool
+        // search with optimized character lookup code that only considers
+        // the current round's haystack data.
+        while (pos <= data.length - this._needle.length) {
+            const ch = data[pos + this._needle.length - 1];
+            if (ch === this._lastChar &&
+                data[pos] === this._needle[0] &&
+                jsmemcmp(this._needle, 0, data, pos, this._needle.length - 1)) {
+                if (pos > buf_pos) {
+                    tokens.push(data.subarray(buf_pos, pos));
+                }
+                tokens.push(MATCH);
+                return [pos + this._needle.length, ...tokens];
+            }
+            else {
+                pos += this._occ[ch];
+            }
+        }
+        // There was no match. If there's trailing haystack data that we cannot
+        // match yet using the Boyer-Moore-Horspool algorithm (because the trailing
+        // data is less than the needle size) then match using a modified
+        // algorithm that starts matching from the beginning instead of the end.
+        // Whatever trailing data is left after running this algorithm is added to
+        // the lookbehind buffer.
+        if (pos < data.length) {
+            while (pos < data.length &&
+                (data[pos] !== this._needle[0] ||
+                    !jsmemcmp(data, pos, this._needle, 0, data.length - pos))) {
+                ++pos;
+            }
+            if (pos < data.length) {
+                this._lookbehind = data.subarray(pos);
+            }
+        }
+        // Everything until pos is guaranteed not to contain needle data.
+        if (pos > 0) {
+            tokens.push(data.subarray(buf_pos, pos < data.length ? pos : data.length));
+        }
+        return [data.length, ...tokens];
+    }
+    _memcmp(data, pos, len) {
+        if (pos < 0) {
+            if (!jsmemcmp(this._lookbehind, this._lookbehind.length + pos, this._needle, 0, Math.min(-pos, len))) {
+                return false;
+            }
+            if (len < -pos) {
+                return true;
+            }
+            len += pos;
+        }
+        return jsmemcmp(data, Math.max(0, pos), this._needle, -Math.min(0, pos), len);
+    }
 }
 
 exports.MATCH = MATCH;
-exports.ReadableStreamSearch = ReadableStreamSearch;
 exports.StreamSearch = StreamSearch;
-exports.allStrings = allStrings;
-exports.arrayIterator = arrayIterator;
-exports.chunksIterator = chunksIterator;
-exports.split = split;
-exports.splitChunks = splitChunks;
-exports.stringIterator = stringIterator;
-//# sourceMappingURL=index.js.map
+//# sourceMappingURL=search.cjs.map

package/lib/search.cjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"search.cjs","sources":["../src/search.ts"],"sourcesContent":[null],"names":["fromString","concat"],"mappings":";;;;;AAAA;;;AAGE;AAKF,SAAS,QAAQ,CAChB,IAAgB,EAChB,IAAY,EACZ,IAAgB,EAChB,IAAY,EACZ,GAAW,EAAA;AAEX,IAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE;AAC7B,QAAA,IAAI,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC,KAAK,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC,EAAE;AACtC,YAAA,OAAO,KAAK;QACb;IACD;AAEA,IAAA,OAAO,IAAI;AACZ;AAEA,SAAS,oBAAoB,CAAC,CAAa,EAAA;;;AAG1C,IAAA,MAAM,KAAK,GAAG,IAAI,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,CAAC;AAC3C,IAAA,IAAI,CAAC,CAAC,MAAM,GAAG,CAAC,EAAE;AACjB,QAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE;AACtC,YAAA,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,MAAM,GAAG,CAAC,GAAG,CAAC;QAC/B;IACD;AAEA,IAAA,OAAO,KAAK;AACb;MAEa,KAAK,GAAG,MAAM,CAAC,OAAO;MAItB,YAAY,CAAA;AAChB,IAAA,OAAO;AACP,IAAA,SAAS;AACT,IAAA,IAAI;AAEJ,IAAA,WAAW,GAAe,IAAI,UAAU,EAAE;AAElD,IAAA,WAAA,CAAmB,MAA2B,EAAA;AAC7C,QAAA,IAAI,OAAO,MAAM,KAAK,QAAQ,EAAE;YAC/B,IAAI,CAAC,OAAO,GAAG,MAAM,GAAGA,qBAAU,CAAC,MAAM,CAAC;QAC3C;aAAO;AACN,YAAA,IAAI,CAAC,OAAO,GAAG,MAAM;QACtB;QAEA,IAAI,CAAC,SAAS,GAAG,MAAM,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC;AAC1C,QAAA,IAAI,CAAC,IAAI,GAAG,oBAAoB,CAAC,MAAM,CAAC;IACzC;AAEO,IAAA,IAAI,CAAC,KAAiB,EAAA;QAC5B,IAAI,GAAG,GAAG,CAAC;AACX,QAAA,IAAI,MAAe;QACnB,MAAM,SAAS,GAAY,EAAE;AAC7B,QAAA,OAAO,GAAG,KAAK,KAAK,CAAC,MAAM,EAAE;AAC5B,YAAA,CAAC,GAAG,EAAE,GAAG,MAAM,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC;AACzC,YAAA,SAAS,CAAC,IAAI,CAAC,GAAG,MAAM,CAAC;QAC1B;AAEA,QAAA,OAAO,SAAS;IACjB;IAEO,GAAG,GAAA;AACT,QAAA,MAAM,IAAI,GAAG,IAAI,CAAC,WAAW;AAC7B,QAAA,IAAI,CAAC,WAAW,GAAG,IAAI,UAAU,EAAE;AACnC,QAAA,OAAO,IAAI;IACZ;IAEQ,KAAK,CAAC,IAAgB,EAAE,OAAe,EAAA;QAC9C,MAAM,MAAM,GAAY,EAAE;;;;;QAM1B,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC,WAAW,CAAC,MAAM;AAElC,QAAA,IAAI,GAAG,GAAG,CAAC,EAAE;;;;;;;;;;;;;AAaZ,YAAA,OAAO,GAAG,GAAG,CAAC,IAAI,GAAG,IAAI,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE;gBAC3D,MAAM,IAAI,GAAG,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC;AAC1C,gBAAA,MAAM,EAAE,GACP,IAAI,GAAG;AACN,sBAAE,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,WAAW,CAAC,MAAM,GAAG,IAAI;AACjD,sBAAE,IAAI,CAAC,IAAI,CAAC;AAEd,gBAAA,IACC,EAAE,KAAK,IAAI,CAAC,SAAS;AACrB,oBAAA,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,GAAG,EAAE,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,EAC/C;oBACD,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC,WAAW,CAAC,MAAM,EAAE;wBACnC,MAAM,CAAC,IAAI,CACV,IAAI,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAC,EAAE,IAAI,CAAC,WAAW,CAAC,MAAM,GAAG,GAAG,CAAC,CAC3D;oBACF;AAEA,oBAAA,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC;AAElB,oBAAA,IAAI,CAAC,WAAW,GAAG,IAAI,UAAU,EAAE;AAEnC,oBAAA,OAAO,CAAC,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,MAAM,CAAC;gBAC9C;qBAAO;AACN,oBAAA,GAAG,IAAI,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;gBACrB;YACD;;AAIA,YAAA,IAAI,GAAG,GAAG,CAAC,EAAE;;;;;;;;;gBASZ,OAAO,GAAG,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,GAAG,EAAE,IAAI,CAAC,MAAM,GAAG,GAAG,CAAC,EAAE;AAC9D,oBAAA,GAAG,EAAE;gBACN;YACD;AAEA,YAAA,IAAI,GAAG,IAAI,CAAC,EAAE;;AAEb,gBAAA,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,WAAW,CAAC;AAC7B,gBAAA,IAAI,CAAC,WAAW,GAAG,IAAI,UAAU,EAAE;YACpC;iBAAO;;;;gBAIN,MAAM,aAAa,GAAG,IAAI,CAAC,WAAW,CAAC,MAAM,GAAG,GAAG;AAEnD,gBAAA,IAAI,aAAa,GAAG,CAAC,EAAE;;AAEtB,oBAAA,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAC,EAAE,aAAa,CAAC,CAAC;oBACxD,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,WAAW,CAAC,QAAQ,CAAC,aAAa,CAAC;gBAC5D;AAEA,gBAAA,IAAI,CAAC,WAAW,GAAGC,aAAM,CAAC,CAAC,IAAI,CAAC,WAAW,EAAE,IAAI,CAAC,CAAC;gBAEnD,OAAO,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,MAAM,CAAC;YAChC;QACD;QAEA,GAAG,IAAI,OAAO;;;;AAKd,QAAA,OAAO,GAAG,IAAI,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE;AAChD,YAAA,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC;AAE9C,YAAA,IACC,EAAE,KAAK,IAAI,CAAC,SAAS;gBACrB,IAAI,CAAC,GAAG,CAAC,KAAK,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;gBAC7B,QAAQ,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,EAC5D;AACD,gBAAA,IAAI,GAAG,GAAG,OAAO,EAAE;AAClB,oBAAA,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,OAAO,EAAE,GAAG,CAAC,CAAC;gBACzC;AAEA,gBAAA,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC;AAElB,gBAAA,OAAO,CAAC,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,MAAM,CAAC;YAC9C;iBAAO;AACN,gBAAA,GAAG,IAAI,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACrB;QACD;;;;;;;AAQA,QAAA,IAAI,GAAG,GAAG,IAAI,CAAC,MAAM,EAAE;AACtB,YAAA,OACC,GAAG,GAAG,IAAI,CAAC,MAAM;iBAChB,IAAI,CAAC,GAAG,CAAC,KAAK,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;oBAC7B,CAAC,QAAQ,CAAC,IAAI,EAAE,GAAG,EAAE,IAAI,CAAC,OAAO,EAAE,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,GAAG,CAAC,CAAC,EACzD;AACD,gBAAA,EAAE,GAAG;YACN;AAEA,YAAA,IAAI,GAAG,GAAG,IAAI,CAAC,MAAM,EAAE;gBACtB,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC;YACtC;QACD;;AAGA,QAAA,IAAI,GAAG,GAAG,CAAC,EAAE;YACZ,MAAM,CAAC,IAAI,CACV,IAAI,CAAC,QAAQ,CAAC,OAAO,EAAE,GAAG,GAAG,IAAI,CAAC,MAAM,GAAG,GAAG,GAAG,IAAI,CAAC,MAAM,CAAC,CAC7D;QACF;QAEA,OAAO,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,MAAM,CAAC;IAChC;AAEQ,IAAA,OAAO,CAAC,IAAgB,EAAE,GAAW,EAAE,GAAW,EAAA;AACzD,QAAA,IAAI,GAAG,GAAG,CAAC,EAAE;AACZ,YAAA,IACC,CAAC,QAAQ,CACR,IAAI,CAAC,WAAW,EAChB,IAAI,CAAC,WAAW,CAAC,MAAM,GAAG,GAAG,EAC7B,IAAI,CAAC,OAAO,EACZ,CAAC,EACD,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,CAAC,CACnB,EACA;AACD,gBAAA,OAAO,KAAK;YACb;AAEA,YAAA,IAAI,GAAG,GAAG,CAAC,GAAG,EAAE;AACf,gBAAA,OAAO,IAAI;YACZ;YAEA,GAAG,IAAI,GAAG;QACX;AAEA,QAAA,OAAO,QAAQ,CACd,IAAI,EACJ,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,GAAG,CAAC,EAChB,IAAI,CAAC,OAAO,EACZ,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,GAAG,CAAC,EACjB,GAAG,CACH;IACF;AACA;;;;;"}

package/lib/search.d.ts

@@ -1,14 +1,13 @@
-export declare const MATCH: unique symbol;
-export declare type Token = Uint8Array | typeof MATCH;
-export declare class StreamSearch {
-    private _needle;
-    private _lastChar;
-    private _occ;
-    private _lookbehind;
-    constructor(needle: Uint8Array | string);
-    feed(chunk: Uint8Array): Token[];
-    end(): Uint8Array;
-    private _feed;
-    private _charAt;
-    private _memcmp;
-}
+export declare const MATCH: unique symbol;
+export type Token = Uint8Array | typeof MATCH;
+export declare class StreamSearch {
+    private _needle;
+    private _lastChar;
+    private _occ;
+    private _lookbehind;
+    constructor(needle: Uint8Array | string);
+    feed(chunk: Uint8Array): Token[];
+    end(): Uint8Array;
+    private _feed;
+    private _memcmp;
+}

package/lib/search.mjs

@@ -0,0 +1,185 @@
+import { concat } from 'uint8arrays/concat';
+import { fromString } from 'uint8arrays/from-string';
+
+/*
+  Based heavily on the Streaming Boyer-Moore-Horspool C++ implementation
+  by Hongli Lai at: https://github.com/FooBarWidget/boyer-moore-horspool
+*/
+function jsmemcmp(buf1, pos1, buf2, pos2, len) {
+    for (let i = 0; i < len; i++) {
+        if (buf1[pos1 + i] !== buf2[pos2 + i]) {
+            return false;
+        }
+    }
+    return true;
+}
+function createOccurenceTable(s) {
+    // Populate occurrence table with analysis of the needle,
+    // ignoring last letter.
+    const table = new Array(256).fill(s.length);
+    if (s.length > 1) {
+        for (let i = 0; i < s.length - 1; i++) {
+            table[s[i]] = s.length - 1 - i;
+        }
+    }
+    return table;
+}
+const MATCH = Symbol("Match");
+class StreamSearch {
+    _needle;
+    _lastChar;
+    _occ;
+    _lookbehind = new Uint8Array();
+    constructor(needle) {
+        if (typeof needle === "string") {
+            this._needle = needle = fromString(needle);
+        }
+        else {
+            this._needle = needle;
+        }
+        this._lastChar = needle[needle.length - 1];
+        this._occ = createOccurenceTable(needle);
+    }
+    feed(chunk) {
+        let pos = 0;
+        let tokens;
+        const allTokens = [];
+        while (pos !== chunk.length) {
+            [pos, ...tokens] = this._feed(chunk, pos);
+            allTokens.push(...tokens);
+        }
+        return allTokens;
+    }
+    end() {
+        const tail = this._lookbehind;
+        this._lookbehind = new Uint8Array();
+        return tail;
+    }
+    _feed(data, buf_pos) {
+        const tokens = [];
+        // Positive: points to a position in `data`
+        //           pos == 3 points to data[3]
+        // Negative: points to a position in the lookbehind buffer
+        //           pos == -2 points to lookbehind[lookbehind_size - 2]
+        let pos = -this._lookbehind.length;
+        if (pos < 0) {
+            // Lookbehind buffer is not empty. Perform Boyer-Moore-Horspool
+            // search with character lookup code that considers both the
+            // lookbehind buffer and the current round's haystack data.
+            //
+            // Loop until (condition 1)
+            //   there is a match.
+            // or until
+            //   we've moved past the position that requires the
+            //   lookbehind buffer. In this case we switch to the
+            //   optimized loop.
+            // or until (condition 3)
+            //   the character to look at lies outside the haystack.
+            while (pos < 0 && pos <= data.length - this._needle.length) {
+                const cpos = pos + this._needle.length - 1;
+                const ch = cpos < 0
+                    ? this._lookbehind[this._lookbehind.length + cpos]
+                    : data[cpos];
+                if (ch === this._lastChar &&
+                    this._memcmp(data, pos, this._needle.length - 1)) {
+                    if (pos > -this._lookbehind.length) {
+                        tokens.push(this._lookbehind.subarray(0, this._lookbehind.length + pos));
+                    }
+                    tokens.push(MATCH);
+                    this._lookbehind = new Uint8Array();
+                    return [pos + this._needle.length, ...tokens];
+                }
+                else {
+                    pos += this._occ[ch];
+                }
+            }
+            // No match.
+            if (pos < 0) {
+                // There's too little data for Boyer-Moore-Horspool to run,
+                // so we'll use a different algorithm to skip as much as
+                // we can.
+                // Forward pos until
+                //   the trailing part of lookbehind + data
+                //   looks like the beginning of the needle
+                // or until
+                //   pos == 0
+                while (pos < 0 && !this._memcmp(data, pos, data.length - pos)) {
+                    pos++;
+                }
+            }
+            if (pos >= 0) {
+                // Discard lookbehind buffer.
+                tokens.push(this._lookbehind);
+                this._lookbehind = new Uint8Array();
+            }
+            else {
+                // Cut off part of the lookbehind buffer that has
+                // been processed and append the entire haystack
+                // into it.
+                const bytesToCutOff = this._lookbehind.length + pos;
+                if (bytesToCutOff > 0) {
+                    // The cut off data is guaranteed not to contain the needle.
+                    tokens.push(this._lookbehind.subarray(0, bytesToCutOff));
+                    this._lookbehind = this._lookbehind.subarray(bytesToCutOff);
+                }
+                this._lookbehind = concat([this._lookbehind, data]);
+                return [data.length, ...tokens];
+            }
+        }
+        pos += buf_pos;
+        // Lookbehind buffer is now empty. Perform Boyer-Moore-Horspool
+        // search with optimized character lookup code that only considers
+        // the current round's haystack data.
+        while (pos <= data.length - this._needle.length) {
+            const ch = data[pos + this._needle.length - 1];
+            if (ch === this._lastChar &&
+                data[pos] === this._needle[0] &&
+                jsmemcmp(this._needle, 0, data, pos, this._needle.length - 1)) {
+                if (pos > buf_pos) {
+                    tokens.push(data.subarray(buf_pos, pos));
+                }
+                tokens.push(MATCH);
+                return [pos + this._needle.length, ...tokens];
+            }
+            else {
+                pos += this._occ[ch];
+            }
+        }
+        // There was no match. If there's trailing haystack data that we cannot
+        // match yet using the Boyer-Moore-Horspool algorithm (because the trailing
+        // data is less than the needle size) then match using a modified
+        // algorithm that starts matching from the beginning instead of the end.
+        // Whatever trailing data is left after running this algorithm is added to
+        // the lookbehind buffer.
+        if (pos < data.length) {
+            while (pos < data.length &&
+                (data[pos] !== this._needle[0] ||
+                    !jsmemcmp(data, pos, this._needle, 0, data.length - pos))) {
+                ++pos;
+            }
+            if (pos < data.length) {
+                this._lookbehind = data.subarray(pos);
+            }
+        }
+        // Everything until pos is guaranteed not to contain needle data.
+        if (pos > 0) {
+            tokens.push(data.subarray(buf_pos, pos < data.length ? pos : data.length));
+        }
+        return [data.length, ...tokens];
+    }
+    _memcmp(data, pos, len) {
+        if (pos < 0) {
+            if (!jsmemcmp(this._lookbehind, this._lookbehind.length + pos, this._needle, 0, Math.min(-pos, len))) {
+                return false;
+            }
+            if (len < -pos) {
+                return true;
+            }
+            len += pos;
+        }
+        return jsmemcmp(data, Math.max(0, pos), this._needle, -Math.min(0, pos), len);
+    }
+}
+
+export { MATCH, StreamSearch };
+//# sourceMappingURL=search.mjs.map