Reply to thread

Message: [QUOTE="Ken Bloom, post: 4593168"] sqlstatement - Generate complex SQL statements programmatically =============================================================== The main goal of this library is to be able to construct an SQL statement from "slices" that concern different aspects of the final query (perhaps in different places in your code) and then combine them all together into one statement easily. Another important goal of this library is to give some consistent Ruby syntax to three statements (INSERT, SELECT, and UPDATE) that seem to have different enough syntax that one has two write different code to generate each kind of statement. I use my SQL database (specifically MySQL) largely as a bulk data processing engine, by doing INSERTâ€¦SELECT or CREATE TABLEâ€¦SELECT statements. This library is intended to make that kind of coding easier. I expect that Object Relational mappers (such as ActiveRecord) are more useful for most people, who are performing queries and inserting/updating/ querying for individual records. I have nevertheless added INSERTâ€¦VALUES statements, and will add other statements soon, for consistency. This library is inspired by CLSQL for Common LISP, or SchemeQL for Scheme, although it is very different from these two libraries. Scheme and LISPâ€˜s use of s-expressions make it very easy to construct an entire sublanguage for the WHERE clause, simply by list parsing. The Criteria library for Ruby has attempted this, but in a more limited manner than SchemeQL or CLSQL. My library aims to cover much of the functionality in these libraries. This library doesnâ€˜t try to abstract out the limitations of your DBMS, and I think that the SQL it uses should be fairly portable, in large measure because it hasnâ€˜t attempted to deal with serious CREATE TABLE statements, where a lot of syntax concerning types, keys and sequences is much more variable. This library can be downloaded from rubyforge.org/projects/sqlstatement Changes in this release ======================= * Vastly redesign the DSL. The methods for adding components to an SQL statement have all been renamed, and internal state is no longer directly exposed to the outside. The interface is now much more intutive and consistent. Please read the documentation to find out how this version of the library works. * Left joins have been added * Thereâ€˜s no more SelectParts class â€” the corresponding semantics have been built directly into the Select class. * Select.new do |s| â€¦ end syntax has been added * Select statements now remember the order of fields and tables. This makes the next two changes feasible. o Support for Mysqlâ€˜s STRAIGHT_JOIN modifier has been added o Unit tests have been added Example ======= If we wanted to generate SQL code similar to SELECT `value1`,`value2` FROM `jointable`,`dictionary1`,`dictionary2` WHERE `jointable`.`id1`=`dictionary1`.`id` AND `jointable`.`id2`=`dictionary2`.`id` we could generate it as follows: def foo(x) Select.new do |s| s.field :"value#{x}" s.table :"dictionary#{x}" s.condition string_func("`jointable`.`id#{x}`=`dictionary#{x} `.`id`") end end stmt=Select.new stmt.table :jointable (1..2).each do |x| stmt << foo(x) end dbh.execute(stmt) or like this: stmt=Select.new do |s| s.table :jointable [1,2].each do |num| s.field "value#{num}".dbid s.table "dictionary#{num}".dbid s.condition string_func("`jointable`.`id#{num}`=`dictionary#{num} `.`id`") end end dbh.execute stmt [/QUOTE]

Verification