I don't quite like implementing policies that I don't fully understand.
An admirable sentiment :)
It's not a backslash after the hostname; it's a backslash before the following period. Here's a walkthrough since you ask*:
RewriteCond %{HTTP_HOST} !^(example\.com}?$
RewriteRule (.*) http://example.com/$1 [R=301,L]
Piece by piece:
RewriteCond
Set this aside for now. mod_rewrite works on a system of "two steps forward, one step back": it
first looks at the rule itself. Only if the rule can potentially apply does it also look at the condition(s).
RewriteRule
"Here's what will happen".
In mod_rewrite, and certain other Apache mods, a space has syntactic meaning. There's no other punctuation, no braces or "function" labels.
(.*)
First part of rule: the PATTERN.
If there's a request for .* then capture it. Both of these characters have special meanings in Regular Expressions (any flavor, not just mod_rewrite, not just Apache).
The . means "any one character, including spaces". Spaces don't occur in URLs-- but they do occur in other aspects of the request, where they need special handling.
The * means "zero or more of what came before". If the rule is in an htaccess file located in your root directory-- the normal place for htaccess files if you use them-- then a request for the root is seen as "" (nothing). Regular Expressions start as soon as they can and continue as long as they can, so the form .* will cover the entire thing.
Finally the () parentheses mean "capture this part for later re-use". Parentheses can be used for other things, but they always capture too unless you've explicitly told them not to.
http://example.com/$1
Second part of rule: the TARGET. If the pattern matches, then point them to http://example.com/ plus
$1
whatever they asked for in the first place. In targets only, $ followed by a single numeral means "reuse this captured group from the pattern". There can be up to 9, counted from left to right. If there are nested parentheses, count from the opening parenthesis only.
[R=301,L]
Optional final part of a rule: the FLAGS.
Any time a target contains the full protocol-plus-hostname, it turns into a 302 temporary redirect. (This applies to other areas too, such as ErrorDocument directives, not only to mod_rewrite.) To change it to 301 permanent, you have to say so explicitly.
In addition, you need an [L] flag meaning "We're done with mod_rewrite this time around". Some actions carry an implied [L]; a redirect doesn't, so again you have to say so explicitly.
And now we get to the condition. If there's more than one, they are read from top to bottom.
The form .* obviously matches all requests everywhere. So if you don't include a condition, requests will go around in circles forever, winding up with an error message from your browser. Conditions have a little more syntactic variation than rules, but this is the most common pattern.
RewriteCond %{HTTP_HOST}
"If the requested hostname is..."
!^(example\.com)?$
-- or, in this case, "is not..." That's what the leading ! means. Unlike other stuff we've met so far, the ! locution is specific to Apache. (It occurs elsewhere, but is not a property of Regular Expressions as such.)
:: bzzt! typo! bzzt! typo! ::
Oh, lord. I remember one time I made this identical typo myself, and then various other people cut-and-pasted, and poor phranque had to spend all day cleaning up the thread. ^
^ "opening anchor" means "the specified text has to come at the very beginning of the element I'm looking at", here the hostname. So "example" matches but "www.example" doesn't.
(example\.com)
Remember I said that parentheses have uses besides capturing? Here they mean that the contents of the parentheses are to be treated as a package; anything that comes afterward applies to the whole package.
\.
I said before that a . in Regular Expressions means "any one character". So if you mean an actual, literal period, it has to be "escaped". That's the meaning of the \ backslash. It occurs in other situations, such as
\(
"an actual, literal parenthesis" or
\
"an actual, literal space" but it's most important with periods. Consider the difference between "1.2" and "1\.2".
(example\.com)?
The question mark also has a special meaning: "the immediately preceding character-- or, here, the whole part in parentheses-- is optional". So it means "the hostname starts with 'example.com' or starts with nothing".
$
One more special character.
In patterns only, the $ is a closing anchor. It means "the part I'm looking at has to end this way".
So, all together:
!^(example\.com)?$
"The part I'm looking at-- which we've already specified is the HTTP_HOST-- is
not exactly 'example.com' or exactly nothing." The "exactly nothing" part has to do with ancient browsers and/or certain HTTP/1.0 requests that don't include a hostname at all. By convention, these are allowed in, although if you really hate proxies you could omit the parentheses and question mark, so the only condition is "exactly example.com".
Some conditions have flags of their own, but here there aren't any, so we're all done.
All together:
"
If there is a request for anything at all,
then if the requested hostname is anything other than 'example.com' or nothing,
then send them around to 'example.com', keeping the rest of the request in its original form."
Whew.
*
And then, whenever anyone else asks in the next couple of years, we can point them to this post.
