由网友(梦醉为红颜丶)分享简介：我想验证一个类型的上传文件（假设该文件具有不正确的扩展名）。I want to validate a type of an uploaded file (suppose the file has incorrect extension).我想，这可以通过检查文件签名的幻数来完成。同样，因为它是在UNIX 的完成。所...

我想验证一个类型的上传文件（假设该文件具有不正确的扩展名）。

I want to validate a type of an uploaded file (suppose the file has incorrect extension).

我想，这可以通过检查文件签名的幻数来完成。同样，因为它是在UNIX 的完成。所以我想我需要阅读的文件的第一个字节，并将它们与幻数映射比较（我需要有映射，我需要保持他们）。

I figure it could be done by checking the magic number of file signature. Similarly, as it is done in unix. So I guess I need to read the first bytes of the file and compare them with the magic number mappings (and I need to have the mappings and I need to maintain them).

所以，我想也许还有一类是在那里只是为了这个目的吗？或者，也许有人已经这样做了，可以在该解决方案分享他们的见解？谢谢

So I thought maybe there is a class out there just for this purpose? Or maybe someone has done this and can share their insights on the solution? Thanks

推荐答案

我怀疑有此问题的银弹解决方案。没有人说，每个文件格式有任何形式的幻数。考虑纯文本文件 - 根本没有合理的指示（可能的字节顺序标记旁一开始，然而，并不一定意味着什么）和纯启发式必须用来猜测它是一个纯文本文件。像明智地认识到一个XML或HTML（即SGML）文件，它是指以查了很多可能性。

I doubt there's a "silver bullet" solution for this problem. Nobody says every file format has any form of a magic number. Consider plain text files — no reasonable indication at all (beside a possible byte-order-mark at the beginning which, however, need not mean anything) and pure heuristics must be used to guess it's a plain text file. Like wise to recognize an XML or HTML (i.e. SGML) file it means to check a lot of possibilities.

您应该采取的办法是：

缩小下来的一组预期的文件格式。 要么做它自己的基础上，整体结构和/或文件格式的特性，你需要支持或者寻找类似的 GNU文件提到重新运行。您还可以在网站，如 MagicDB 一看 - 但是，这个人是特别过时（2005年） 指定意外的文件格式的行为。 Narrow down the set of expected file formats. Either do-it-yourself based on the knowledge of the overall structure and/or characteristics of the file formats you need to support or look for a solution like GNU File mentioned by rerun. You can also have a look on sites like MagicDB — however, this one is particularly outdated (year 2005). Specify behavior for unexpected file formats.

在情况设置窄够了我会考虑一个简单的自定义解决方案是正确的。

In case the set is narrow "enough" I'd consider a simple custom solution as the right one.