geirm 00/11/02 19:26:34
Modified: src/java/org/apache/velocity/runtime/parser Parser.jjt
Log:
The Great Escape. Changes escape handling to be less invasive - the only time \
acts as
an escape character is when it is in front of a valid reference in the context, or a
valid
VTL directive (so \#set is output as #set and \#schmoo is output as \#schmoo). See
my post
to the list(s) explaining in more detail if needed.
Revision Changes Path
1.16 +139 -69
jakarta-velocity/src/java/org/apache/velocity/runtime/parser/Parser.jjt
Index: Parser.jjt
===================================================================
RCS file:
/home/cvs/jakarta-velocity/src/java/org/apache/velocity/runtime/parser/Parser.jjt,v
retrieving revision 1.15
retrieving revision 1.16
diff -u -r1.15 -r1.16
--- Parser.jjt 2000/11/02 03:53:04 1.15
+++ Parser.jjt 2000/11/03 03:26:32 1.16
@@ -127,7 +127,7 @@
*
* @author <a href="mailto:[EMAIL PROTECTED]">Jason van Zyl</a>
* @author <a href="mailto:[EMAIL PROTECTED]">Geir Magnusson Jr.</a>
- * @version $Id: Parser.jjt,v 1.15 2000/11/02 03:53:04 geirm Exp $
+ * @version $Id: Parser.jjt,v 1.16 2000/11/03 03:26:32 geirm Exp $
*/
public class Parser
{
@@ -167,16 +167,19 @@
{
token_source.clearStateVars();
ReInit(stream);
-
- // return process();
-
- SimpleNode n = null;
+ return process();
- try {
- n = process();
- } catch (Exception e ) { System.out.println( e ); }
+ /*
+ * leave the following. Useful for debugging the parser. thx - geir
+ */
- return n;
+ //SimpleNode n = null;
+ //
+ // try {
+ // n = process();
+ // } catch (Exception e ) { System.out.println( e ); }
+ //
+ // return n;
}
public void setDirectives(Hashtable directives)
@@ -472,7 +475,7 @@
SwitchTo( REFMOD2 );
}
-| <RPAREN: ")" (" ")* ("\n")?>
+| <RPAREN: ")" (" ")* ( "\n" | "\r" | "\r\n" )?>
{
RPARENHandler();
}
@@ -490,14 +493,48 @@
}
}
+/*----------------------------------------------
+ *
+ * escape "\\" handling for the built-in directives
+ *
+ *--------------------------------------------- */
+TOKEN:
+{
+ /*
+ * We have to do this, because we want these to be a Text node, and
+ * whatever follows to be peer to this text in the tree.
+ * And we don't want to try an and deal with this w/in the AST.
+ *
+ * This is really simplistic. I actually would prefer to find them in
+ * grammatical context, but I am neither smart nor rested, a receipe
+ * for disaster, another long night with Mr. Parser, or both.
+ */
+
+ <ESCAPE_SET_DIRECTIVE : "\\#set">
+ { matchedToken.image = "#set"; }
+| <ESCAPE_IF_DIRECTIVE : "\\#if">
+ { matchedToken.image = "#if"; }
+| <ESCAPE_END_DIRECTIVE : "\\#end">
+ { matchedToken.image = "#end"; }
+| <ESCAPE_INCLUDE_DIRECTIVE: "\\#include">
+ { matchedToken.image = "#include"; }
+| <ESCAPE_ELSEIF_DIRECTIVE: "\\#elseif">
+ { matchedToken.image = "#elseif"; }
+| <ESCAPE_ELSE_DIRECTIVE: "\\#else">
+ { matchedToken.image = "#else"; }
+| <ESCAPE_STOP_DIRECTIVE: "\\#stop">
+ { matchedToken.image = "#stop"; }
+
+}
+
<*>
MORE :
{
- //
- // Note : DOLLARBANG is a duplicate of DOLLAR. They must be identical.
- //
+ /*
+ * Note : DOLLARBANG is a duplicate of DOLLAR. They must be identical.
+ */
- <DOLLAR: "$">
+ <DOLLAR: ("\\")? "$">
{
if (! inComment)
{
@@ -548,7 +585,7 @@
SwitchTo( IN_MULTI_LINE_COMMENT );
}
-| <HASH : "#">
+| <HASH : ("\\")? "#">
{
if (! inComment)
{
@@ -565,7 +602,7 @@
TOKEN :
{
- <ESCAPE_SEQUENCE: "\\" ~[] >
+ <ESCAPE: "\\">
| <TEXT: (~["$", "#", "\\"])+ >
}
@@ -581,7 +618,6 @@
{
inComment = false;
stateStackPop();
- // was :DEFAULT
}
}
@@ -593,7 +629,6 @@
{
inComment = false;
stateStackPop();
- // was :DEFAULT
}
}
@@ -604,7 +639,6 @@
{
inComment = false;
stateStackPop();
- // was :DEFAULT
}
}
@@ -642,10 +676,6 @@
}
incMode = false;
- // from jason :
- // if (lparen == 0)
- // SwitchTo(DEFAULT);
-
/*
* - if we are in REFERENCE || REFMODIFIER then " is an ender
* - if we are in DIRECTIVE and haven't seen ( yet, then also drop out.
@@ -706,7 +736,7 @@
<DIRECTIVE>
TOKEN :
{
- <END: "end" ("\n")?>
+ <END: "end" ( "\n" | "\r" | "\r\n" )?>
{
inDirective = false;
stateStackPop();
@@ -760,7 +790,7 @@
*
* This is more than a single state, because of the structure of
* the VTL references. We use three states because the set of tokens
- * for each state is different.
+ * for each state can be different.
*
* $foo.bar( "arg" )
* ^ ^ ^
@@ -817,7 +847,7 @@
* These can terminate a reference, but are needed in subsequent states
*/
- if (image.toString().equals("=") || image.toString().equals("\n"))
+ if (image.toString().equals("=") || image.toString().equals("\n") ||
image.toString().equals("\r") )
{
input_stream.backup(1);
}
@@ -938,18 +968,31 @@
}
{
/*
- * Get the directive identifier and check to see if this is a directive.
- * Note that because '#' is a <MORE> token, the identifier is preceeded by
the '#'
+ * There are two things we do here :
+ * 1) If the first char is a \, then this has been escaped, so don't
+ * continue. We will take care of the rest in ASTDirective.java
+ * 2) Get the directive identifier and check to see if this is a directive.
+ * Note that because '#' is a <MORE> token, the identifier is preceeded by
the '#'
*/
+
t = <WORD>
{
- d = (Directive) directives.get(t.image.substring(1));
-
- if (d == null)
+ if( t.image.startsWith("\\"))
{
token_source.stateStackPop();
token_source.inDirective = false;
- return jjtThis;
+ return jjtThis;
+ }
+ else
+ {
+ d = (Directive) directives.get(t.image.substring(1));
+
+ if (d == null)
+ {
+ token_source.stateStackPop();
+ token_source.inDirective = false;
+ return jjtThis;
+ }
}
}
@@ -1040,11 +1083,14 @@
| <LPAREN>
| <NUMBER_LITERAL>
| <STRING_LITERAL>
-| t = <ESCAPE_SEQUENCE>
- {
- t.image = t.image.substring(1);
- }
-
+| <ESCAPE>
+| <ESCAPE_SET_DIRECTIVE>
+| <ESCAPE_IF_DIRECTIVE>
+| <ESCAPE_END_DIRECTIVE>
+| <ESCAPE_INCLUDE_DIRECTIVE>
+| <ESCAPE_ELSEIF_DIRECTIVE>
+| <ESCAPE_ELSE_DIRECTIVE>
+| <ESCAPE_STOP_DIRECTIVE>
}
/* -----------------------------------------------------------------------
@@ -1194,36 +1240,60 @@
/* ======================================================================
- *
- * Background
- * ----------
- *
- * VTL == Velocity Template Language
- *
- * The problem with parsing VTL is that an input stream consists generally of
little
- * bits of VTL mixed in with 'other stuff' (hereafter referred to as 'schmoo').
Unlike
- * other languages, like C or Java, where the parser can punt whenever it
encounters
- * input that doesn't conform to the grammar, the VTL parser can't do that.
- *
- * There are a few things that we do here :
- * - define a set of parser states (DEFAULT, DIRECTIVE, REFERENCE, etc)
- * - define for each parser state a set of tokens for each state
- * - define the VTL grammar, expressed (mostly) in the productions such as
Text(), SetStatement(), etc.
- *
- * It is clear that this expression of the VTL grammer (the contents of this .jjt
file) is maturing and
- * evolving as we learn more about how to parse VTL ( and as I learn about
parsing...), so in the event
- * this documentation is in disagreement w/ the source, the source takes
precedence.
- *
- * Parser States
- * -------------
- * DEFAULT : This is the base or starting state, and strangely enough, the
default state.
- *
- * DIRECTIVE : This state is triggered by the '#' character, and is used when
processing both
- * 'defined' directives such as #if() #else #elseif() #end #set #foreach() as
well as the
- * so-called 'pluggable' directives (PDs), which include #foreach(), #macro()
etc. The
- * PDs are able to be implemented entirely outside of the parser.
- * - with the exception of #set, #else and #else, the 'shape' of a directive
is
- * #foo( optional args )
- * -
- */
+
+ Background
+ ----------
+
+ VTL == Velocity Template Language
+
+ The problem with parsing VTL is that an input stream consists generally of little
+ bits of VTL mixed in with 'other stuff' (hereafter referred to as 'schmoo').
Unlike
+ other languages, like C or Java, where the parser can punt whenever it
encounters
+ input that doesn't conform to the grammar, the VTL parser can't do that.
+
+ There are a few things that we do here :
+ - define a set of parser states (DEFAULT, DIRECTIVE, REFERENCE, etc)
+ - define for each parser state a set of tokens for each state
+ - define the VTL grammar, expressed (mostly) in the productions such as
Text(), SetStatement(), etc.
+
+ It is clear that this expression of the VTL grammer (the contents of this .jjt
file) is maturing and
+ evolving as we learn more about how to parse VTL ( and as I learn about
parsing...), so in the event
+ this documentation is in disagreement w/ the source, the source takes precedence.
+
+ Parser States
+ -------------
+ DEFAULT : This is the base or starting state, and strangely enough, the default
state.
+
+ DIRECTIVE : This state is triggered by the '#' character, and is used when
processing both
+ 'defined' directives such as #if() #else #elseif() #end #set #foreach() as
well as the
+ so-called 'pluggable' directives (PDs), which include #foreach(), #macro()
etc. The
+ PDs are able to be implemented entirely outside of the parser.
+ - with the exception of #set, #else and #else, the 'shape' of a directive is
+ #foo( optional args )
+ -
+
+ (cont)
+
+ Escape Sequences
+ ----------------
+ The escape processing in VTL is very simple. The '\' character acts only as an
escape when :
+
+ 1) It preceeds a reference that is in the context.
+
+ 2) It preceeds a defined directive (#set, #if, #end, etc) or a valid
pluggable directive,
+ such as #foreach
+
+ In all other cases the '\' is just another piece of text. The purpose of this
is to allow the non-VTL
+ parts of a template (the 'schmoo') to not have to be altered for processing by
Velocity.
+
+ So if in the context $foo and $bar were defined and $woogie was not
+
+ \$foo \$bar \$woogie
+
+ would output
+
+ $foo $bar \$woogie
+
+
+*/
