by Tdarcos » Mon May 23, 2016 2:11 am
Discussing the new feature I just developed so that I was able to make my messages on Caltrops, when they contain my image reference, I thought it was a brilliant hack.
Originally I wanted to return an SVG image. Those are easy to do and the image is a text file, so it's not hard. While this worked when typed in as a URL, it failed in an IMG tag.
So I had it do a regular image file, and it creates an on-the-fly JPEG as a one-time image. The only thing I was worried about was browser caching, in which if there was one reference it might not go back and retrieve the image again, but apparently either it's not checking, there's no "last saved" value, or it is able to notice the image is not the same as the last one.
At first, I just had it "write" an image with a random number in a box, it was not really intended as an authentication. Then I looked at how Caltrops would send a browser request for the image, and, as it turns out, it was okay for the time being until people started impersonating me again.
Not that I care that much, but I thought it would be a fun exercise to see if I could find a way to allow me to distinguish uses of it between mine and impostors. And it was basically using the same method the telephone company uses to set off an alarm when someone tries to rob a pay phone, vs. being able to tell when a collector comes to empty the coin box legitimately.
Jonsey immediately recognized what I'm doing. Other people looked at the reference and said you could do cache poisoning and dns attacks because I'm not using HTTPS to retrieve the image (I'm too cheap to spend another $20 a year to get an SSL certificate for a non-revenue producing site). But all I'm doing is reading the HTTP_REFERRER to get the &pid= of the posting, then doing a database lookup to see if it's in there. Nothing earth shaking, just a simple proof of concept to see if it could be done.
Oh by the way, Pinback, if you really did forge one of my messages, it just showed that you had enough common sense to write something reasonable, literate and normal, as opposed to the typical three-year-old writing level of too many of Caltrops' inmates.
I just thought of it as a way to do a fun hack and to thumb my nose at the people over there, but nothing serious.
And in total programming time it was probably 2 sessions of 4-5 hours over 8-10 months, most of the time being in the second session to line up the text fields because each color change requires a different write to image statement.
Also, it now shows something different on a right-click and view image in Firefox.
But it definately was fun to do.
"I'm Tansin A. Darcos and I ... what the fuck do I need a Freshness seal for? I log in here!"
Discussing the new feature I just developed so that I was able to make my messages on Caltrops, when they contain my image reference, I thought it was a brilliant hack.
Originally I wanted to return an SVG image. Those are easy to do and the image is a text file, so it's not hard. While this worked when typed in as a URL, it failed in an IMG tag.
So I had it do a regular image file, and it creates an on-the-fly JPEG as a one-time image. The only thing I was worried about was browser caching, in which if there was one reference it might not go back and retrieve the image again, but apparently either it's not checking, there's no "last saved" value, or it is able to notice the image is not the same as the last one.
At first, I just had it "write" an image with a random number in a box, it was not really intended as an authentication. Then I looked at how Caltrops would send a browser request for the image, and, as it turns out, it was okay for the time being until people started impersonating me again.
Not that I care that much, but I thought it would be a fun exercise to see if I could find a way to allow me to distinguish uses of it between mine and impostors. And it was basically using the same method the telephone company uses to set off an alarm when someone tries to rob a pay phone, vs. being able to tell when a collector comes to empty the coin box legitimately.
Jonsey immediately recognized what I'm doing. Other people looked at the reference and said you could do cache poisoning and dns attacks because I'm not using HTTPS to retrieve the image (I'm too cheap to spend another $20 a year to get an SSL certificate for a non-revenue producing site). But all I'm doing is reading the HTTP_REFERRER to get the &pid= of the posting, then doing a database lookup to see if it's in there. Nothing earth shaking, just a simple proof of concept to see if it could be done.
Oh by the way, Pinback, if you really did forge one of my messages, it just showed that you had enough common sense to write something reasonable, literate and normal, as opposed to the typical three-year-old writing level of too many of Caltrops' inmates.
I just thought of it as a way to do a fun hack and to thumb my nose at the people over there, but nothing serious.
And in total programming time it was probably 2 sessions of 4-5 hours over 8-10 months, most of the time being in the second session to line up the text fields because each color change requires a different write to image statement.
Also, it now shows something different on a right-click and view image in Firefox.
But it definately was fun to do.
"I'm Tansin A. Darcos and I ... what the fuck do I need a Freshness seal for? I log in here!"