Text to SQL Search is a shard written in Crystal for translating textual searches into SQL WHERE clauses. All the necessary options to configure the specifics of the translation process exist.
The search specifications are generally intended to come from three sources: free-form search inputs on websites, submitted form fields, and command lines. However, there is nothing in the code that would prevent any other uses as long as the input is text.
This shard is for you when you use an SQL data store and want to generate WHERE clauses in a simple and user-intuitive way. It is much better than asking your users to write SQL directly or you manually writing conversions from form inputs into SQL and ORM wrappers' syntax.
The shard works with any database for which the generated SQL WHERE clause is valid. The resulting SQL is usable in Model.all("WHERE ...") invocations in ORM models such as Granite ORM, in any other ORMs that support direct SQL input, and even in manual SQL search invocations.
When search identifiers are recognized to be column names, generated SQL WHEREs default to searching in those columns. When search terms are not identifiers, they are searched in custom-definable sets of fields with custom-definable operators.
Let's start with a couple examples to get you going. The examples do not show specific configuration settings or char-by-char accurate SQL that is generated; they provide a general intro to what is supported and what you get out of the box.
Let's say we want to provide a text search for use in automobile sales. These examples work and are part of the spec suite:
INPUT: sedan 4 doors > 2000 ccm price < 20k with no downpayment
WHERE: "type"='sedan' AND "doors"='4' AND "ccm">'2000' AND "price"<'20000' AND not("downpayment">'0')
INPUT: > 3000 ccm or with stereo
WHERE: ccm>3000 or stereo>0
INPUT: ((4 door and color = blue) or !downpayment) and price < than 5000
WHERE: ((doors=4 AND color=blue) OR not(downpayment>0)) AND price<5000
INPUT: color = "metallic red" or year: 2015
WHERE: color='metallic red' or year > 2015
Ensure you have the necessary dependencies:
git: Use your platform specific package manager to installgitcrystal: Follow the instructions to getcrystalon this page: https://crystal-lang.org/docs/installation/index.html
Then:
$ git clone https://github.com/docelic/text_to_sql_search.cr.git
$ cd text_to_sql_search.cr
$ make
$ make spec- Fork it (https://github.com/docelic/text_to_sql_search.cr/fork)
- Create your feature branch (git checkout -b my-new-feature)
- Commit your changes (git commit -am 'Add some feature')
- Push to the branch (git push origin my-new-feature)
- Create a new Pull Request
This project is licensed under the MIT License - see the LICENSE file for details