I just checked in a change (to be in 2.0pre4) which switches the
named parameter parsing to use ANTLR. It uses JarJar Links to
renamespace ANTLR and bundle it in the jdbi jar, which, sadly, has
made the jar 4x its previous size. I am going to work on further
trimming out the stuff from antlr we don't need.
I am really happy to make this change, though as the old parser was a
regex based nightmare which was basically unchangeable. The new one
is four rules:
header {
package org.skife.jdbi.rewriter.colon;
}
class ColonStatementLexer extends Lexer;
options {
charVocabulary='\u0000'..'\uFFFE';
k=2;
}
LITERAL: ('a'..'z' | 'A'..'Z' | ' ' | '\t' | '0'..'9' | ',' | '*'
| '=' | ';' | '(' | ')' | '[' | ']' | '+' | '-' | '/' |
'>' | '<' )+;
NAMED_PARAM: ':' ('a'..'z' | 'A'..'Z' | '_')+;
POSITIONAAL_PARAM: '?';
QUOTED_TEXT: ('\'' (~'\'')+ '\'');
I am not sure of the LITERAL matching yet, but it passes all the
current unit tests. What are the rules for UTF-8 names and whatnot in
most databases, and anyone have pointers for improving the grammar?
Also, I'd really appreciate anyone who can taking a poke at 2.0. Thanks!
-Brian
---------------------------------------------------------------------
To unsubscribe from this list please visit:
http://xircles.codehaus.org/manage_email
