T-rev

A bunch of my URIs contain %27 (percent-encoded single quote) but also work with a single quote instead. Is this ancanonical URI issue? Search engines don't differentiate between %27 and single quote, do they?

If so, how can a rewriterule change all single quotes to %27, or vice versa, with a 301 redirect? From what I've read, mod_rewrite sees the request after it's been unencoded to single quotes, so it might just cause an infinite loop.

FredFredrickson

As far as I know, the search engines ignore punctuation anyhow.

FredFredrickson

But adding to that I'd say, you should probably get rid of the need for such characters in your queries. Most "friendly" link schemes utilize a numeric identifier that appears just before the friendly text, making the text completely useless except for to make it more search engine friendly. I'd say just do a replace("%27", "")

T-rev

search engines can't ignore punctuation in URIs or the links wouldn't work. you're talking about searches; i really meant to ask about web-crawlers.

my real question, stated better: are search engines/web crawlers going to consider http://www.xxx.yyy/it's_working and http://www.xxx.yyy/it%27s_working to be two separate pages? i would guess not, but i was wondering if anybody knew for sure.

if the answer to that is no, then question number 2 wouldn't matter (though it's still interesting). if the answer is yes, then i ought to do a 301 redirect so the SE/crawler knows they're really the same page.

thus question number 2: how can you do a mod_rewrite replace of ' with %27?

generally you can do replace all using the [N] flag. but mod_rewrite doesn't see the actual request, it sees it after apache has unencoded all the %27s to 's. so if you were trying to replace %27 with ', it would never even see the %27 to know to replace it, whereas if you're replacing ' with %27, it might be an infinite loop (assuming apache unencoded everything everytime it restarted due to the [N] flag, which i have doubts about).

you can generally get around the unencoding problem by directly calling %{THE_REQUEST}, but i really have doubts that this approach would work in combination with the [N] flag, because %{THE_REQUEST} would be the same everytime so it would certainly loop infinitely.

of course, i could just do the redirect from the script instead of with apache, but i prefer to let apache do that kind of thing, since it's built for it.

as far as using a numeric identifier followed by friendly text, yeah, but it might (depending on how smart the web-crawler) still have to be the same friendly text for the crawler to know it's the same page.

FredFredrickson

Well if you have hyperlinks with %27 in them, search engines should have no trouble with that. As far as text searches go it would ignore the punctuation, but I think you'll be alright with no changes at all.

T-rev

i know. really i'm wondering are search engines/web crawlers going to consider http://www.xxx.yyy/it's_working and http://www.xxx.yyy/it%27s_working to be two separate pages?

FredFredrickson

proabably not. And if so, search engines generally will only show 1 or 2 pages with good matches from one site during a search anyway.