Re: Problem simulating LEFT or LIKE with RegEx

Indeed, this works fine in the shell.  The PHP issue was a double-escaping
issue (my PHP application defines the Mongo functions) an it has been
resolved.  Thanks for your help!  Thanks to MongoDB I've increased the
performance of Tera-WURFL by almost 2x and I'll be pushing it hard in the
upcoming release.

Thanks,

Steve Kamerman

Sent from my iPhone 

On Sep 3, 2010, at 7:52 AM, Kristina Chodorow <kris### @10gen.com>
wrote:

 I think this is a shell bug.  You can get around it by doing:
 
 user_agent : new RegExp("^Mozilla/")
 
 Similarly, you shouldn't have to escape the / in PHP.
 
 
 On Fri, Sep 3, 2010 at 1:26 AM, Tera-WURFL
<stevek### @gmail.com> wrote:
 I'm trying find all documents with a string that starts with a given
 string.  For example, I would look for "app" and expect to find
 "apple" but not "aptitude".  In my case, I'm actually searching for
 browser User Agent prefixes, which occasionally end in a "/"
 character.  It appears that it is impossible to do a RegEx search for
 a string ending in "/" in MongoDBs PCRE search in an efficient way.
 I'm using the PHP MongoDB client, but I can reproduce the problem from
 the shell:
 
 I'm trying to find all documents with a "user_agent" that starts with
 "Mozilla/" by using the regex /^Mozilla\// (I'm escaping the "/" with
 a "\")
 
 Try #1: failed to parse - Mongo is looking for the end of the RegEx I
 think...
 > db.TeraWurfl_MSIE.find({user_agent:/^Mozilla\//},{user_agent:1})
 ... ^C
 
 Try #2: If I add a space after the "/" it parses, but doesn't find any
 documents since there isn't supposed to be a space there
 > db.TeraWurfl_MSIE.find({user_agent:/^Mozilla\/ /},{user_agent:1})
 
 Try #3: This is less efficient since it scans for an extra optional
 character, but it does work and it's better than using ".*" since that
 would result in a complete string scan.
 >
db.TeraWurfl_MSIE.find({user_agent:/^Mozilla\/.?/},{user_agent:1})
 { "_id" : ObjectId("4c806876077db9a3161a6900"), "user_agent" :
 "Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 5.0; mowser;
 http://www.mowser.com)" }
 { "_id" : ObjectId("4c806876077db9a3161b6900"), "user_agent" :
 "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; XV6875.1)" }
 { "_id" : ObjectId("4c806876077db9a3161c6900"), "user_agent" :
 "Mozilla/4.0 (compatible; MSIE" }
 { "_id" : ObjectId("4c806876077db9a3161d6900"), "user_agent" :
 "Mozilla/4.0 (compatible; MSIE 4.0;" }
 { "_id" : ObjectId("4c806876077db9a3161e6900"), "user_agent" :
 "Mozilla/4.0 (compatible; MSIE 5.0;" }
 { "_id" : ObjectId("4c806876077db9a3161f6900"), "user_agent" :
 "Mozilla/4.0 (compatible; MSIE 5.5;" }
 { "_id" : ObjectId("4c806876077db9a316206900"), "user_agent" :
 "Mozilla/4.0 (compatible; MSIE 6.0;" }
 { "_id" : ObjectId("4c806876077db9a316216900"), "user_agent" :
 "Mozilla/4.0 (compatible; MSIE 7.0;" }
 { "_id" : ObjectId("4c806876077db9a316226900"), "user_agent" :
 "Mozilla/4.0 (compatible; MSIE 8.0;" }
 
 Is there a better way for me to escape strings before inserting them
 in regexps?
 
 Also, I am using MongoDB 1.6.2 on Ubuntu 10.04
 
 Thanks!
 
 Steve Kamerman
 http://www.tera-wurfl.com
 
 --