I've been asked to audit a PHP application. No framework, no router, no model. Pure PHP. Few shared functions. HTML, CSS, and JS all mixed together. I've discovered numerous places where SQL injection would be easily possible. There are other problems with the application (XSS vulnerabilities, rampant inline CSS, code copy-pasted everywhere) but this is the biggest. Sometimes they escape inputs, not using a prepared query or even mysql_real_escape_string(), mind you, but using addslashes(). Often, though, their queries look exactly like this (pasted from their code but with columns and variable names changed):
$user = mysql_query("select * from profile where profile_id='".$_REQUEST["profile_id"]."'");
The developers in question claimed that they were unable to hack their application. I tried, and found mod_security to be enabled, resulting in HTTP 406 for some obvious SQL injection attacks. I believe there to be sophisticated workarounds for mod_security, but I don't have time to chase them down.
They claim that this is a "conceptual" matter and not a "practical" one since the application can't easily be hacked. Their internal auditor agreed that there were problems, but emphasized the conceptual nature of the issues.
They also use this conceptual/practical argument to defend against inline CSS and JS, absence of code organization, XSS vulnerabilities, and massive amounts of repetition.
My client (rightly so, perhaps) just wants this to go away so they can launch their product. The site works. You can log in, do what you need to do, and things are visibly functional, if slow. SQL Injection would indeed be hard to do, given mod_security. Further, their talk of "conceptual vs. practical" is rhetorically brilliant, considering that my client doesn't understand web application security. I worry that they've succeeded in making me sound like an angry puritan.
In many ways, this is a problem of politics, not technology, but I am at a loss. As a developer, I want to tell them to toss the whole project and start over with a new team, but I face a strong defense from the team that built it and a client who really needs to ship their product.
Is my position here too harsh? Even if they fix the SQL Injection and XSS problems can I ever endorse the release of an unmaintainable tangle of spaghetti code?
EDIT: As to my role on this project, I'm not a security expert, I'm a web developer generalist. the client initially asked me to load test the app and make recommendations for scalability. I haven't gotten that far yet, because when I first read the output of the home page, I saw inline CSS, table based layouts, and all sorts of crazy JS, so I requested a copy of the code, where I found the unholy mess described above. They don't even abstract the HTML header and footer. Every PHP file has a slightly different pasted version.
And as to the app itself, it does store personal information, and it processes (but does not store) credit card data.