Server evaluation script

[lorax]

I'm at the point in my job where tools which automate certain tasks are becoming increasingly necessary. Case-in-point: evaluating a client's existing webspace for certain criteria like platform and version, whether it supports mod-rewrite, PHP modules installed, etc...

I've been doing this sort of analysis manually but the workload has increased such that it is nolonger feasible. I want to create or obtain a script that does the same analysis for me and formats it into a nice report. Does anyone know of/have a ready made script for such reporting or am I going to have to create this?

[DrDoc]

phpinfo() will tell you most of this :)

[lorax]

>> phpinfo

Not really DrDoc. It will supply some of what I'm looking for but it won't evaluate the majority of what I need evaluated (which I neglected to mention) like whether files have DOCTYPE declarations, HTML or XHTML, using tables or CSS-2P, etc... Not to mention it returns a bunch of stuff that the client will have no need to know.

BUT! It is a start and I think it does create an array with all of that information in it which (if so) I could read and take what I need from.

[ergophobe]

DOCTYPE declarations, HTML or XHTML, using tables or CSS-2P

Wow, that's a tall order and obviously phpinfo isn't much help there.

Nothing constructive to add, but I'm curious about a number of things.

How do you propose to decide whether or not it's HTML or XHTML? In other words, are you basing this on

- mime type

- doctype - so you trust that if it says it's XHTML, it actually is, regardless of mime type or underlying code

- code itself - in other words, writing a validator.

Similarly, how do you distinguish between
1) a site that has CSS for all layout, but still makes extensive use of tables for tabular data?
2) a site that uses some CSS-P, but also depends heavily on tables for layout?

I don't think you could do that automatically without some highly sophisticated AI.

Tom

[lorax]

It shouldn't need to be that complicated.

If no DOCTYPE has been established then that'll be flagged and HTML (unknown version and flavor) will be assumed.

I envision using cURL to open and read the file as delivered by the webserver. I should be able to gather the file type as well. If they're doing mod-rewrite and masking scripted pages as html then obviously I won't be able to pick that up. But if it's a .php or .asp file then I can safely assume the script server they're using.

As for the CSS, it's simply a matter of looking for the various ways of including/linking an external css file and looking for inline css.

Distinguishing CSS-2P versus tables is easy. Look for tables and look for the CSS elements that are used for positioning.

I'm really looking to get a flavor of the page's structural makeup and whether or not the previous web developer had a clue as to what they were doing.

As for the PHP information - I'll have to look into it but I believe I phpinfo returns an array that is used to build that information page. I can call that, suppress the output, and sift through the array for what I want. I can also use the script to query their server for DNS info and the other details I want to know.

Add on the number of images, page weight, percentage of code to visible text, whether or not there are <h1> (h2,h3,etc.) elements, what the title, meta kw, and meta desc are, plus a host of other tidbits and you can see what I'm after.