Directing a form input to a web search

Anon

Hello there, PHP Maister.

My challenge is making users search in a form on my website, and have that search "pasted" into the correct URL of the site they want to search on.

Take a look: http://zkole.com.

I found some info in some older post here, which made it possible for me to go this far (at the moment of writing, only the britannica search is coded).

Well, I figure the following has to be done:

Getting the URL set-up for each search correct.
Then pasting the search into the URL format.

The problem is, one-word strings are fine, but what happens when people enter multi strings? Actually, in the case of Britannica, it worked out automatically. The space between the words (can I call them variables), became a %20 or something. But if I were adhering to Britannica's normal standard, it would have to be a +. Even tho it worked out with Britannica, doesnt mean it will work out with all the search enginges.

So, how do I break (format) the search string into individual variables to then seperate them with & or + or whatever the search engine may use?'

Regards,

Stig Eidissen

brandonschnell

spaces and other characters that aren't kosher for the web are encoded in the URL. for example a space would be "%20" in a URL.
you should be able to take the html code & javascript code for a form on a search engine website and use just about the same code to submit your queries to the remote web site.

Anon

That would not work in the case they are using a CGI or something else server-side script. And besides, I want to code PHP. Not Java.

Thanks for help tho.

Anon

Hi there Stig,

I hope i can help you. Why dont you try urldecode(string) ? This makes %#,etc .. back into the original strings.

hope this helps. 🙂 Maybe you want to look at urlencode also.

Allen Ding

Anon

Or maybe you could use regular expressions to parse %20 into + if you wanted to.

Anon

I am not familiar with using regular expressions. Yes, on certain search, making a space into a + is what I need. Other searches, its &, but I guess thats the same code with just changed letters. Im not sure if that technique can be applied into all searches tho.

Anon

Okay, im not great with regular expressions either, but heres a regex that i think will work 🙂

<?php

$string="My%20name%20is%20allen";
echo $string."<BR><BR>";

$string = ereg_replace("%20","+",$string);
echo $string."<BR><BR>";

This will replace the %20's with +'ses. You can change the + to anything else and the %30 to what you want to match.
Hope this helps.

Allen

Anon

I should mention that using PCRE's would be faster in execution, but you probably wont notice the difference, and im a little lazy to look up the syntax 🙁

brandonschnell

this is a html issue more than a php issue and can work regardless of what the remote site is using on the backend. In fact all of the search engines are running some sort of cgi or server side scripting.

if you look at your php code you'll notice that you're emulating the html search forms on the remote search engine so you need to figure out what the html is doing on the search engines. get the html & javascript being used on the remote sites to get a full list of options to use with the site. or search for something on a web site and use the resulting URL for your script.

what you need is some php to translate from your URL variables to the remote search engine form variables.

the below php code works for Britannica, Yahoo, and Google. its a 5 minute hack, but it should be simple to add new sites.
####################################
<?

if (isset($search_terms))
{
if($search_engine == "britannica")
{
header("Location: http://www.britannica.com/search?query=$search_terms");
}
if($search_engine == "yahoo")
{
header("Location: http://search.yahoo.com/bin/search?p=$search_terms");
}
if($search_engine == "google")
{
header("Location: http://www.google.com/search?q=$search_terms");
}
}

<BODY>

<FORM METHOD ACTION="<? echo $PHP_SELF; ?>">
<input type=text name=search_terms size=50>
<input type=image src="/ok.gif" name=ok>
<input type=radio name=search_engine value=britannica checked>Britannica
<input type=radio name=search_engine value=yahoo>Yahoo
<input type=radio name=search_engine value=google>Google
</form>

</BODY>

brandonschnell

go to www.yahoo.com and search using multiple words. now replace the "%20" text with "+" and press enter. for all of the search engines i tried, the "%20" & "+" text is handled exactly the same and no changes would have to be made.

Anon

Yup, youre right. But i think the situation is this. When the user goes to his site and enters two words 'something something',it becomes something%20something. So if in the script this is put to a search engine, the search engine will search for 'something%20something' , which the search engine will not convert into "something something'.
🙂

Hope this helps.

Allen

brandonschnell

maybe. but thats a problem with his script. With better php coding his script won't need the encoding/unencoding.

download my example at:

http://www.phpbuilder.com/forum/read.php3?num=2&id=130702&thread=130686

and you'll find that (at least on my system) no conversion is necessary.

Anon

I managed to get that far, myself. But don't I need to split the string into each seperate word, and then merge them with + or & ?

brandonschnell

type in the url http://search.yahoo.com/bin/search?p= in your browser. now type in at the end multiple words to search on. press enter and see what happens.

this is exactly whats happening when using the header() function with the location header and adding in the search terms. with the browsers i tested, the browser is url encoding the url before heading to the web site.

look at the html source of the forms on the search sites for more info if you have problems getting a site to work.