How do you do wildcard searches with Mongoid in a Ruby on Rails environment?

Question

The Mongoid documentation only gives one example of doing a wildcard search:

Person.where(first_name: /^d/i)

This finds all people with the first name that starts with "d".

What do the /^ and /i represent?

How do I find all people with their first name having an "na" in the middle of the string? E.g., this query would find "jonathan" since "na" is a substring of the entire string.

Is there website or guide with this information?

Solution 1

You need this to find people with "na" in the name.

Person.where(first_name: /na/i)

As for your example:

Person.where(first_name: /^d/i)

^ means "beginning of the line". This regex will match all strings where first letter is "d". /i means "do case-insensitive matches". So it'll match both "d" and "D".

Note: only prefix regexes (with ^ in front) are able to use indexes.

Is there website or guide with this information?

Here's my favourite.

Solution 2

This is not a "wildcard" search, this is called a regular expression.

/^d/i

• The two slashes are only the regex delimiters, you search for what is in between those two slashes.
• The following i is a modifier or option. It changes the matching behaviour of your regex, the i stands for case insensitive, means it matches "d" and "D".
• The first character ^ is an anchor, it anchors the search pattern to the start of the string, means match "d" only at the start of the string

A good tutorial about regular expressions is the tutorial on regular-expressions.info

If you want to search for a string anywhere in the string, just remove the anchor that binds the pattern to the start, /na/ will find "na" anywhere in the string.

Comments

[Jagdeep Singh]

Is it possible to do query like Person.where(first_name: /na/i) if the field first_name is of type Mongoid::EncryptedString. I am using mongoid 3.1.6 and getting TypeError: no implicit conversion of Regexp into String.

[TimP]

@JagdeepSingh, not directly unfortunately. You can iterate through the whole data-set though, something like Person.all.select { |p| p.first_name ~= /na/i }, which is obviously less efficient.