transparent SSL proxy myths and facts
If you want to filter on domain names, you have two possible ways: you could take the name from the CONNECT method issued by the client if it knows that it has to use a proxy for HTTPS conenctions and filter on that one (Squids supports that BTW). Or, if you really really need to do it transparently, you would need to look into the (encrypted) request headers.
If you want to see encrypted request headers, you need to have a key. If you want to have a key, you need a certificate which is a) trusted by the client to be the "correct" certificate and b) certifies every possible host (wildcard-everything).
So what you would need to do is
- set up a certificate for your proxy. It depends on your software how to do that - you might use stunnel to terminate the SSL connection at the proxy's side, have some filtering HTTP proxy behind it and re-establish SSL for all outgoing traffic using iptables DNAT targets and stunnel again. There might be "boxed" solutions for MitM SSL proxying as well.
- install the aforementioned certificate on all of your clients that are going use the proxy
Mostly, if you need transparent proxying it is because you do not want or cannot reconfigure the clients to use the proxy. If this is the case with your request as well, you probably would not have the option to install certificates on clients and mark them as "trusted". So even though there is a technical way for transparent SSL proxying, I suspect not much will be won for you.
I know this is an old question, but if the OP only wants to blacklist/whitelist certain domain names, they don't have to use a proxy at all, they could just use a DNS based blacklist.
- setup your onsite DNS servers to return 127.0.0.1 for any domain you want to blacklist
- at your internet gateway block all IPs except your DNS servers from accessing TCP/UDP ports 53, so only your DNS servers can make DNS requests from internet based servers.
Any other domain will be allowed. All web traffic SSL or otherwise will pass through the gateway unchanged.
Edit: Seeing as the OP is prepared to force users through his lists, he can block other methods of accessing DNS. This way, if the user tries to use one of those other DNS methods that are blocked, the websites just will not work. aka 'My way or the highway'
For DNS-over-HTTPS that @wheeler mentioned, you can just block regular DNS looksups to sites such as https://dns.google.com and https://cloudflare-dns.com/dns-query and https://doh.cleanbrowsing.org/doh/family-filter/. Though this will quickly become untenable as more and more services come online.
You may also need a way of blocking MIME types such as application/dns-udpwireformat for other methods of DNS being developed.
Delegate can work as Man-In-The-Middle proxy for HTTPS.
For transparent mode, Inline IPS (Suricata, Snort) system can help you to block ssl sites.
https://redmine.openinfosecfoundation.org/projects/suricata/wiki/Setting_up_IPSinline_for_Linux
Block SSL Site Rules for Specific User:
drop ip any 443 -> 192.168.3.x any (content:".facebook.com"; msg:"Simplewall block facebook.com ";sid:7;rev:1;)
drop ip any 443 -> 192.168.3.204 any (content:".youtube"; msg:"Simplewall block youtube.com" ;sid:4;rev:1;)
Block File Download Rules Based on Extensions:
drop ip any any -> 192.168.3.63 any (msg:"File exe block"; fileext:"exe"; filestore ;sid:1;rev:1;)
drop ip any ssl -> 192.168.3.63 any (msg:"File mp3 block"; fileext:"mp3"; filestore ;sid:11;rev:1;)
drop ip any ssl -> 192.168.3.63 any (msg:"File pdf block"; fileext:"pdf"; filestore ;sid:21;rev:1;)
Try this out with SimpleWall
Its very very simple to add block rule with Simplewall web interface.
You can also add rules under Simplewall => Content Filter use same rules for IPS along with http content filter.