nv-hash-file 1.0.0

package/README.md

@@ -0,0 +1,46 @@
+
+nv-hash-file
+=======================
+- fast and high conflict file content hash
+- for estimate using, normally useless
+
+install
+=======
+- npm install nv-hash-file
+
+usage
+=====
+    
+      const { }   = require("nv-hash-file");
+ 
+
+example
+-------
+
+### 0
+
+
+#### 0_0
+
+#### 0_1
+
+
+### 1 
+
+#### 1_0
+
+#### 1_1
+
+
+METHODS
+========
+
+
+APIS
+=======
+
+
+LICENSE
+=======
+- ISC 
+

package/index.js

@@ -0,0 +1,171 @@
+const _fs   = require("fs");
+const _path = require("path");
+
+const _cmp_file_byte_by_byte = (p0, p1) => {
+    const fd0 = _fs.openSync(p0, 'r');
+    const fd1 = _fs.openSync(p1, 'r');
+    const buf0 = Buffer.alloc(4096);
+    const buf1 = Buffer.alloc(4096);
+
+    let offset = 0;
+    while (true) {
+        const bytesRead0 = _fs.readSync(fd0, buf0, 0, buf0.length, offset);
+        const bytesRead1 = _fs.readSync(fd1, buf1, 0, buf1.length, offset);
+
+        const minLen = Math.min(bytesRead0, bytesRead1);
+
+        for (let i = 0; i < minLen; i++) {
+            if (buf0[i] < buf1[i]) {
+                _fs.closeSync(fd0);
+                _fs.closeSync(fd1);
+                return -1;
+            }
+            if (buf0[i] > buf1[i]) {
+                _fs.closeSync(fd0);
+                _fs.closeSync(fd1);
+                return 1;
+            }
+        }
+
+        // 如果读取的字节数不同，较长的文件更大
+        if (bytesRead0 < bytesRead1) {
+            _fs.closeSync(fd0);
+            _fs.closeSync(fd1);
+            return -1;
+        }
+        if (bytesRead0 > bytesRead1) {
+            _fs.closeSync(fd0);
+            _fs.closeSync(fd1);
+            return 1;
+        }
+
+        // 两者都读完且没有差异
+        if (bytesRead0 === 0) {
+            break;
+        }
+
+        offset += minLen;
+    }
+
+    _fs.closeSync(fd0);
+    _fs.closeSync(fd1);
+    return 0;
+};
+
+const _eq_byte_by_byte = (p0, p1) => _cmp_file_byte_by_byte(p0,p1)===0;
+
+
+const cmp_file_byte_by_byte = (p0,p1)=>{
+    p0 = _path.resolve(p0);
+    p1 = _path.resolve(p1);
+    if(_fs.existsSync(p0)) {
+        if(_fs.existsSync(p1)) {
+            var stat0 = _fs.lstatSync(p0);
+            if(stat0.isFile()) {
+                var stat1 = _fs.lstatSync(p1);
+                if(stat1.isFile()) {
+                    return [true,_cmp_file_byte_by_byte(p0,p1)];                 
+                } else {
+                    return [false, `${p1} NOT file`];
+                }
+            } else {
+                return [false, `${p0} NOT file`];
+            }
+        } else {
+            return [false, `${p1} NOT exist`];
+        }
+    } else {
+        return [false, `${p0} NOT exist`];      
+    }
+}
+const eq_byte_by_byte = (p0, p1) => {
+   var pair = cmp_file_byte_by_byte(p0,p1);
+   if(pair[0]) {
+       return pair[1] === 0;
+   } else {
+       return false;
+   }
+}
+
+
+const _hash = (p) => {
+    const stat = _fs.lstatSync(p);
+    if (!stat.isFile()) { return -1;} 
+    const sz = stat.size;
+    if(sz === 0) {return 0;} 
+    // 打开文件读二进制
+    const fd = _fs.openSync(p, 'r');
+    const buf = Buffer.alloc(1);
+    // 读取第一个字节
+    _fs.readSync(fd, buf, 0, 1, 0);
+    const fst = buf[0];
+    // 读取中间字节
+    let mid;
+    if (sz % 2 === 1) {
+        // 奇数长度：取正中间那一个字节
+        const midPos = Math.floor(sz / 2); // 例如 sz=5 → midPos=2
+        _fs.readSync(fd, buf, 0, 1, midPos);
+        mid = buf[0];
+    } else {
+        // 偶数长度：取中间两个字节，求和后右移 1 位
+        const leftPos = sz / 2 - 1;  // 例如 sz=6 → leftPos=2
+        const rightPos = sz / 2;     // rightPos=3
+        _fs.readSync(fd, buf, 0, 1, leftPos);
+        const leftByte = buf[0];
+        _fs.readSync(fd, buf, 0, 1, rightPos);
+        const rightByte = buf[0];
+        mid = (leftByte + rightByte) >> 1; // 求和后除以 2（取整）
+    }
+
+    // 读取最后一个字节
+    _fs.readSync(fd, buf, 0, 1, sz - 1);
+    const lst = buf[0];
+
+    _fs.closeSync(fd);
+
+    // 组合哈希
+    const h0 = sz % (2 ** 28);                 // 高 28 位：size 低 28 位
+    const h1 = (fst << 16) | (mid << 8) | lst;  // 低 24 位：3 个字节
+    const h = (h0 * (2 ** 24)) + h1;            // 最终哈希
+    return h;
+}
+
+const hash = (p) => {
+    p = _path.resolve(p);
+    if (_fs.existsSync(p)) {
+        return _hash(p);
+    } else {
+        return -2;
+    }
+};
+
+const _eq = (p0,p1)=>{
+   var h0 = _hash(p0);
+   var h1 = _hash(p1);
+   if(h0===h1 && h0 >=0) {
+       return _eq_byte_by_byte(p0,p1);
+   } else {
+       return false;
+   }
+}
+
+const eq = (p0,p1)=>{
+   var h0 = hash(p0);
+   var h1 = hash(p1);
+   if(h0===h1) {
+       return _eq_byte_by_byte(p0,p1);
+   } else {
+       return false;
+   }
+}
+
+
+module.exports = hash;
+module.exports._hash = hash;
+module.exports._cmp_file_byte_by_byte = _cmp_file_byte_by_byte;
+module.exports.cmp_file_byte_by_byte = cmp_file_byte_by_byte;
+module.exports._eq_byte_by_byte = _eq_byte_by_byte;
+module.exports.eq_byte_by_byte  = eq_byte_by_byte;
+module.exports._eq = _eq;
+module.exports.eq  = eq;
+

package/package.json

@@ -0,0 +1,15 @@
+{
+  "name": "nv-hash-file",
+  "version": "1.0.0",
+  "main": "index.js",
+  "scripts": {
+    "test": "echo \"Error: no test specified\" && exit 1"
+  },
+  "author": "",
+  "license": "ISC",
+  "description": "",
+  "devDependencies": {
+    "nv-array-eqgrp": "^1.0.0",
+    "nv-array-group-by": "^1.0.2"
+  }
+}