Printable View
I've always wondered how one would make a script that operates like a search bar.
Ex: Imagine one is searching for a word called "Avengers", but only spells out "Aven". The script recognizes this and isolates any word containing the fragment "Aven", and pushes it into an array. Is there a method to do this?
There are pre-programmed functions to find matches in any string by almost any modern language. SQL even allows such things with simple wildcards as Wizkidje posted.
However many developers even use regular expressions to archive this in a more professional way for more complex search engines. Sadly this requires to load all items before, but it rly allows extended features.
Sample: Most codes split off the search terms by specific delimiters like as space and comma, and perform a regular expression match with each term on all composed items search strings (containing name and other information).
At the end the code decides for AND conditions (to filter closer by typing more than one search term) or OR conditions (to expand the relating search terms for the items to find).
In your given case it would be enough to compose the regex (regular expression) "Aven" to match all "Aven" in the search string. You can use specific methods to determinate the regular expression has been found or not to decide to show the item or simply not. So, you can use regex to find matches or even find out it contains matches. A better example is to only find all words that start with "Aven", so your regex has to be "\bAven" while \b stands for the beginning of a word (by beginning of the string or by any space).
You can even do funnier things like finding all full numbers (not single digits) in a string that are not starting by a 0 but ending with a "$" or "€" (no matter it has a space between the number and the sign or not):
\b[1-9]\d*[$€]\b
This contains possible signs and digit ranges, more special escaped wildcards, like \d that stands for any digit and following quantifier that decide how many signs are possible (? = one or zero, {3,6} = three to six, + = one or more, * = any, no quantifier means it has to be one).
To get any users term into the regex query you can include it by using variables.
Regex is known as own language that is used in many programming languages.
You might cared. =)
Sincerely, VibeTribe
In practice, the above is the correct answer. To provide a more efficient implementation (again, which in most practical cases is definitely an overkill) for huge dictionaries, the dictionary could be stored in a data structure known as Trie
In practice, you wouldn't use regular expressions or even a trie to implement a searching mechanism. They're too slow, or solve the wrong problem. If you're trying to search, you need full text search and fuzzy matching over a very large database.
Consider your straightforward "search" solution:
1. Split words, construct a regex like: /(?:word1|word2|word3)/i.
2. Iterate over a list of strings you want to "search".
3. For each, test the regexp, if true, print it or add it to an array.
This requires you to search every item in the list in step 2. For large datasets, this is prohibitive. Additionally, it's a very restrictive approach.
Unfortunately, the complexity of a system that performs full text search including fuzzy matching over a large database is non-trivial to describe and includes tens of algorithms. You can read more about the problem here:
https://en.wikipedia.org/wiki/Full_text_search
Additionally, here are two very popular full-text search engines:
https://en.wikipedia.org/wiki/Apache_Solr
https://en.wikipedia.org/wiki/Lucene
In any event, for most "search" needs, you'd be better off using intelligent indexing in a database. Most are able to efficiently index whole strings or even starts-with/ends-with type string matches (using structures like tries).
Love it how easily you disregard trivial answers as plain wrong.. you should consider the audience here.
Where have you been anyway? Been awfully dull around here without you.
Well that answer really is basically wrong. The cases where that approach would be sufficient are handily solved by database indexing, so there's no code needed except to construct a query. And I wouldn't advise using a regular expression condition in a SQL query as it's not possible to optimize an index for general regular expression matching, so it'll devolve into a full table scan and be just as slow as looping over a large array. This typically matters because nobody wants to "search" if there are only 5 items in the database.Quote:
Originally Posted by Negata
I've been busy working and other nonsense. Cash monies.
Database indexing hardly helps in the general case where you're not using a database or you wish to filter the list on the fly without a request to the server. You know your stuff for sure so no disrespect from me in that sense but you tend to go to deep in my opinion - now if this were a serious programming forum your answers would be right on but frankly I see this as a newbie playground where most users can't grasp what you're talking about or apply it.Quote:
Originally Posted by jMerliN