[python]上传文件验证

上传文件验证

上传文件验证分为：1、文件头验证　　2、文件类型验证　　3、文件后缀验证

• 获取文件上传的二进制数据

# 获取上传文件
file = request.files.get('file')
if file:
    # 大文件
    if hasattr(file.stream._file, 'peek'):
        print(file.stream._file.peek())
    # 小文件
    elif hasattr(file.stream._file, 'getvalue'):
        print(file.stream._file.getvalue())
    else:
        pass

• 文件头验证

# 类型字典
def typeList():
    return {
        "FFD8FF": "jpg",
        "89504E47": "png",
        "255044462D312E": "pdf",
        "504B0304": "zip",
        "47494638": "gif",
        "49492A00": "tif",
        "424D": "bmp",
        "41433130": "dwg",
        "38425053": "psd",
        "7B5C727466": "rtf",
        "3C3F786D6C": "xml",
        "68746D6C3E": "html",
        "44656C69766572792D646174653A": "eml",
        "CFAD12FEC5FD746F": "dbx",
        "2142444E": "pst",
        "D0CF11E0": "xls.or.doc",
        "5374616E64617264204A": "mdb",
        "FF575043": "wpd",
        "252150532D41646F6265": "eps.or.ps",
        "AC9EBD8F": "qdf",
        "E3828596": "pwl",
        "52617221": "rar",
        "57415645": "wav",
        "41564920": "avi",
        "2E7261FD": "ram",
        "2E524D46": "rm",
        "000001BA": "mpg",
        "000001B3": "mpg",
        "6D6F6F76": "mov",
        "3026B2758E66CF11": "asf",
        "4D546864": "mid",
        "4D5A": "pe"
    }
 
# 字节码转16进制字符串
def bytes2hex(bytes):
    num = len(bytes)
    hexstr = u""
    for i in range(num):
        t = u"%x" % bytes[i]
        if len(t) % 2:
            hexstr += u"0"
        hexstr += t
    return hexstr.upper()
 
# 获取文件类型
def filetype(binfile): # binfile 上传文件的二进制数据
    tl = typeList()
    ftype = 'unknown'
    for hcode in tl.keys():
        numOfBytes = len(hcode) / 2  # 需要读多少字节
        numOfBytes = int(numOfBytes)
 
        k, hbytes = 0, []
        for i in binfile:
            if k < numOfBytes:
                hbytes.append(i)
                k += 1
            else:
                break
 
        hbytes = tuple(hbytes)
        f_hcode = bytes2hex(hbytes)
        if f_hcode == hcode:
            ftype = tl[hcode]
            break
    # 返回类型字典的value
    return ftype

• 文件类型验证

dict = {
    "jpg": "jpeg",
    "png": "png",
    "zip": "zip",
    "pdf": "pdf",
    "md": "octet-stream"
}
 
# 检测文件content_type
for type in types:
    if dict[type] in file.content_type:
        break
    else:
        pass

　　

• 文件后缀验证

# 检测文件后缀
type = file.filename.split('.')[-1]
if type not in types:
    pass