专栏首页游戏杂谈node.js获取图片文件的真实类型

node.js获取图片文件的真实类型

遇到一个需求:假定有一个图片文件,真实的类型为jpg,而有人偷懒把jpg直接复制一张,存为同名的png文件,这样在as3读取文件时不会遇到问题,但手机c++在读取文件时却遇到问题了 - -!

现在就需要写一个程序,遍历所有文件夹下的文件,查找文件格式“不正常”的文件。我们的资源主要是gif、png、jpg,最开始,我到网上找到一篇文章:根据二进制流及文件头获取文件类型mime-type,然后读取文件二进制的头信息,获取其真实的文件类型,对与通过后缀名获得的文件类型进行比较。

var fd = fs.openSync(new_file_path, 'r');
var buffer = new Buffer(8);

var mineType = mime.lookup(new_file_path);
var fileType = mime.extension(mineType);

fs.readSync(fd, buffer, 0, 8, 0);
var newBuf = buffer.slice(0, 4);
var head_1 = newBuf[0].toString(16);
var head_2 = newBuf[1].toString(16);
var head_3 = newBuf[2].toString(16);
var head_4 = newBuf[3].toString(16);
var head_iden = head_1 + head_2;

var tempFileType = FILE_TYPE_CONFIG[head_iden];
if (!tempFileType) {
    head_iden += head_3;

    tempFileType = FILE_TYPE_CONFIG[head_iden];

    if (!tempFileType) {
        var msg = "Unknow fileType " + new_file_path + '-' + fileType;
        showLog(msg);
        continue;
    }
}

if (tempFileType != fileType) {
    var msg = "Error fileType" + new_file_path + '-' + fileType + '|' + tempFileType + '--正确的图像文件格式';
    showLog(msg);

    g_errorFileTypArr.push(msg);
}

后来搜索node image相关的信息时,找到这篇文章:node.js module ranking>> (images)

然后筛选到一个模块“node-imageinfo”,写了一个例子进行测试(故意把jpg文件直接修改后缀名为png):

它的源码,有兴趣可以研究一下:

function readUInt32(buffer, offset, bigEndian) {
    if (buffer.readUInt32) {
        return buffer.readUInt32(offset, bigEndian);
    }

    var value;
    if (bigEndian) {
        if (buffer.readUInt32BE) {
            return buffer.readUInt32BE(offset);
        }
        value = (buffer[offset] << 24) + (buffer[offset+1] << 16) + (buffer[offset+2] << 8) + buffer[offset+3];
    }
    else {
        if (buffer.readUInt32LE) {
            return buffer.readUInt32LE(offset);
        }
        value = buffer[offset] + (buffer[offset+1] << 8) + (buffer[offset+2] << 16) + (buffer[offset+3] << 24);
    }
    return value;
}

function readUInt16(buffer, offset, bigEndian) {
    if (buffer.readUInt16) {
        return buffer.readUInt16(offset, bigEndian);
    }

    var value;
    if (bigEndian) {
        if (buffer.readUInt16BE) {
            return buffer.readUInt16BE(offset);
        }
        value = (buffer[offset] << 8) + buffer[offset+1];
    }
    else {
        if (buffer.readUInt16LE) {
            return buffer.readUInt16LE(offset);
        }
        value = buffer[offset] + (buffer[offset+1] << 8);
    }
    return value;
}

function readBit(buffer, offset, bitOffset) {
    if (bitOffset > 7) {
        offset += Math.floor(bitOffset / 8);
        bitOffset = bitOffset % 8;
    }

    var b = buffer[offset];
    if (bitOffset < 7) {
        b >>>= (7 - bitOffset);
    }

    var val = b & 0x01;
    return val;
}

function readBits(buffer, offset, bitOffset, bitLen, signed) {
    var val = 0;
    
    var neg = false;
    if (signed) {
        if (readBit(buffer, offset, bitOffset) > 0) {
            neg = true;
        }
        bitLen--;
        bitOffset++;
    }

    var bytes = [];
    for (var i = 0; i < bitLen; i++) {
        var b = readBit(buffer, offset, bitOffset + i);
        if (i>0 && (bitLen - i) % 8 == 0) {
            bytes.push(val);
            val = 0;
        }
        val <<= 1;
        val |= b;
    }
    bytes.push(val);

    val = new Buffer(bytes);
    val.negative = neg?true:false;
    return val;
}

function imageInfoPng(buffer) {
    var imageHeader = [0x49, 0x48, 0x44, 0x52],
        pos = 12;

    if (!checkSig(buffer, pos, imageHeader)) {
        return false;
    }

    pos += 4;
    return {
        type: 'image',
        format: 'PNG',
        mimeType: 'image/png',
        width: readUInt32(buffer, pos, true),
        height: readUInt32(buffer, pos+4, true),
    };
}

function imageInfoJpg(buffer) {
    var pos = 2,
        len = buffer.length,
        sizeSig = [0xff, [0xc0, 0xc2]];

    while (pos < len) {
        if (checkSig(buffer, pos, sizeSig)) {
            pos += 5;
            return {
                type: 'image',
                format: 'JPG',
                mimeType: 'image/jpeg',
                width: readUInt16(buffer, pos+2, true),
                height: readUInt16(buffer, pos, true),
            };
        }

        pos += 2;
        var size = readUInt16(buffer, pos, true);
        pos += size;
    }
}

function imageInfoGif(buffer) {
    var pos = 6;

    return {
        type: 'image',
        format: 'GIF',
        mimeType: 'image/gif',
        width: readUInt16(buffer, pos, false),
        height: readUInt16(buffer, pos+2, false),
    };
}

function imageInfoSwf(buffer) {
    var pos = 8,
        bitPos = 0,
        val;

    if (buffer[0] === 0x43) {
        try {
            // If you have zlib available ( npm install zlib ) then we can read compressed flash files
            buffer = require('zlib').inflate(buffer.slice(8, 100));
            pos = 0;
        }
        catch (ex) {
            // Can't get width/height of compressed flash files... yet (need zlib)
            return {
                type: 'flash',
                format: 'SWF',
                mimeType: 'application/x-shockwave-flash',
                width: null,
                height: null,
            }
        }
    }

    var numBits = readBits(buffer, pos, bitPos, 5)[0];
    bitPos += 5;
    
    val = readBits(buffer, pos, bitPos, numBits, true);
    var xMin = (numBits > 9 ? readUInt16(val, 0, true) : val[0]) * (val.negative ? -1 : 1);
    bitPos += numBits;

    val = readBits(buffer, pos, bitPos, numBits, true);
    var xMax = (numBits > 9 ? readUInt16(val, 0, true) : val[0]) * (val.negative ? -1 : 1);
    bitPos += numBits;

    val = readBits(buffer, pos, bitPos, numBits, true);
    var yMin = (numBits > 9 ? readUInt16(val, 0, true) : val[0]) * (val.negative ? -1 : 1);
    bitPos += numBits;

    val = readBits(buffer, pos, bitPos, numBits, true);
    var yMax = (numBits > 9 ? readUInt16(val, 0, true) : val[0]) * (val.negative ? -1 : 1);

    return {
        type: 'flash',
        format: 'SWF',
        mimeType: 'application/x-shockwave-flash',
        width: Math.ceil((xMax - xMin) / 20),
        height: Math.ceil((yMax - yMin) / 20),
    };
}

function checkSig(buffer, offset, sig) {
    var len = sig.length;
    for (var i = 0; i < len; i++) {
        var b = buffer[i+offset],
            s = sig[i],
            m = false;

        if ('number' == typeof s) {
            m = s === b;
        }
        else {
            for (var k in s) {
                var o = s[k];
                if (o === b) {
                    m = true;
                }
            }
        }

        if (!m) {
            return false;
        }
    }

    return true;
}

module.exports = function imageInfo(buffer, path) {
    var pngSig = [0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a];
    var jpgSig = [0xff, 0xd8, 0xff];
    var gifSig = [0x47, 0x49, 0x46, 0x38, [0x37, 0x39], 0x61];
    var swfSig = [[0x46, 0x43], 0x57, 0x53];

    if (checkSig(buffer, 0, pngSig)) return imageInfoPng(buffer);
    if (checkSig(buffer, 0, jpgSig)) return imageInfoJpg(buffer);
    if (checkSig(buffer, 0, gifSig)) return imageInfoGif(buffer);
    if (checkSig(buffer, 0, swfSig)) return imageInfoSwf(buffer);

    return false;
};

本文参与腾讯云自媒体分享计划,欢迎正在阅读的你也加入,一起分享。

我来说两句

0 条评论
登录 后参与评论

相关文章

  • Unity中的定时器与延时器

    JavaScript中的定时器与延时器,分别是 setInterval、setTimeout,对应的清理函数是:clearInterval、clearTimeo...

    meteoric
  • as3--简单的文字提示队列

    _______________________________________________________________

    meteoric
  • as3与node.js的socket通信

    SocketManager最多在连接失败时,进行5次重连。建立socket连接代码:

    meteoric
  • 机器学习之卷积神经网络(三)

    假设,有一张大小为32×32×3的输入图片,这是一张RGB模式的图片,你想做手写体数字识别。32×32×3的RGB图片中含有某个数字,比如7,你想识别它是从0-...

    牛油果
  • 谷歌Daydream:我们的征途是星辰大海

    镁客网
  • 【AI TOP 10】苹果Siri智能音箱传英业达代工,前百度研究院博士自动驾驶创业,神经网络领域中的下一件大事

    新智元每日AI TOP 10 新智元 “每日AI ·TOP 10” 是新智元从 2017年5月起推出的一档快读栏目,精选汇集每日国内外最重大产业新闻,第一...

    新智元
  • ETL是BI(商业智能)的基础,调度是ETL的灵魂

    ETL是数据抽取(Extract)、清洗(Cleaning)、转换(Transform)、装载(Load)的过程

    taskctl官方频道
  • 移动VR狼人杀开局,Daydream是选择继续做闭眼玩家,还是决定化身狼人干掉平民?

    VRPinea
  • 如何让自己工资涨的快点?

     有群友提出这样的问题:   他刚进公司没有多久,向项目经理提出了加薪要求,项目经理将这事报告了部门经理,但部门经理一直没有给他结论,也没有找他谈话。他因为这事...

    程序员互动联盟
  • Linux 进程管理之调度和进程切换

    每个CPU都有一个运行队列,每个运行队列中有三个调度队列,task作为调度实体加入到各自的调度队列中。

    刘盼

扫码关注云+社区

领取腾讯云代金券