Thanks for the help Regis - I applied the patch and regenerated the
java classes using the latest javacc but still hit a problem on z/OS.
I figured out that we also need to add "/u0085" to the CHAR
definition in url.g and filter.g [1]. This fixes the test failures on
z/OS for me, and there are no new failures on Windows x86 either.
If you agree that the patch below is right then I will apply it along
with the newly generated java classes.
Regards,
Oliver
[1]
Index: filter.g
===================================================================
--- filter.g (revision 793921)
+++ filter.g (working copy)
@@ -214,7 +214,7 @@
|
<SEMI: ";">
|
-<CHAR : ~["*", "\\", "(", ")", "\n"] >
+<CHAR : ~["*", "\\", "(", ")", "\n", "\u0085"] >
}
String option():
@@ -317,7 +317,7 @@
}
}
- parse() "\n" test() | LOOKAHEAD(2) "\n" | "\n" <EOF> |
<EOF>
+ parse() ("\n" | "\u0085") test() | LOOKAHEAD(2) ("\n" |
"\u0085") | ("\n" | "\u0085") <EOF> | <EOF>
}
// FIXME: get string representation of AttributeValue, then use
Rdn.unescapeValue(String) to get value
String value():
Index: url.g
===================================================================
--- url.g (revision 793921)
+++ url.g (working copy)
@@ -181,7 +181,7 @@
|
<ZERO : "0">
|
-<CHAR : ~["/", "?", "\n"] >
+<CHAR : ~["/", "?", "\n", "\u0085"] >
}
@@ -394,5 +394,5 @@
void test():
{}
{
- parseURL() "\n" test() | LOOKAHEAD(2) "\n" | "\n" <EOF>
| <EOF>
+ parseURL() ("\n" | "\u0085") test() | LOOKAHEAD(2) ("\n"
| "\u0085") | ("\n" | "\u0085") <EOF> | <EOF>
}
Regis wrote:
Oliver Deakin wrote:
Yes, that's right actually - Ive been looking at the grammar files
but can't figure out the right place to make the change. Does
anyone have a good enough knowledge of the javacc tool to figure
out the correct change?
Regards,
Oliver
Nathan Beyer wrote:
Is this class generated code? Should the change be made to a BNF
file instead?
Sent from my iPhone
On Jul 13, 2009, at 8:54 AM, odea...@apache.org wrote:
Author: odeakin
Date: Mon Jul 13 13:54:57 2009
New Revision: 793587
URL: http://svn.apache.org/viewvc?rev=793587&view=rev
Log:
Handle the z/OS NEL character correctly when parsing tokens.
Modified:
harmony/enhanced/classlib/trunk/modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/FilterParserTokenManager.java
harmony/enhanced/classlib/trunk/modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/LdapUrlParserTokenManager.java
Modified:
harmony/enhanced/classlib/trunk/modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/FilterParserTokenManager.java
URL:
http://svn.apache.org/viewvc/harmony/enhanced/classlib/trunk/modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/FilterParserTokenManager.java?rev=793587&r1=793586&r2=793587&view=diff
==============================================================================
---
harmony/enhanced/classlib/trunk/modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/FilterParserTokenManager.java
(original)
+++
harmony/enhanced/classlib/trunk/modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/FilterParserTokenManager.java
Mon Jul 13 13:54:57 2009
@@ -63,6 +63,7 @@
switch(curChar)
{
case 10:
+ case 133: // NEL character for z/OS new lines
return jjStopAtPos(0, 24);
case 33:
return jjStopAtPos(0, 10);
Modified:
harmony/enhanced/classlib/trunk/modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/LdapUrlParserTokenManager.java
URL:
http://svn.apache.org/viewvc/harmony/enhanced/classlib/trunk/modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/LdapUrlParserTokenManager.java?rev=793587&r1=793586&r2=793587&view=diff
==============================================================================
---
harmony/enhanced/classlib/trunk/modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/LdapUrlParserTokenManager.java
(original)
+++
harmony/enhanced/classlib/trunk/modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/LdapUrlParserTokenManager.java
Mon Jul 13 13:54:57 2009
@@ -61,6 +61,7 @@
switch(curChar)
{
case 10:
+ case 133: // NEL character for z/OS new lines
return jjStopAtPos(0, 17);
case 33:
return jjStopAtPos(0, 10);
[1] is a quick fix for this, I don't have z/OS machine, so can't
test if it works.
Actually this part of code is only by method test(), a convenient
way to verify whether parsers work correctly. It doesn't break
parsing functions if removing test() method and "new lines" should
be not a problem, but some test cases will be broken. I'll try to
remove test() method and rewrite test cases depend on it.
[1]
Index:
modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/filter.g
=====================================================================
---
modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/filter.g
+++
modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/filter.g
@@ -317,7 +317,7 @@ void test():
}
}
- parse() "\n" test() | LOOKAHEAD(2) "\n" | "\n" <EOF> |
<EOF>
+ parse() ("\n" | "\u0085") test() | LOOKAHEAD(2) ("\n" |
"\u0085") | ("\n" | "\u0085") <EOF> | <EOF>
}
// FIXME: get string representation of AttributeValue, then use
Rdn.unescapeValue(String) to get value
String value():
Index:
modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/url.g
=====================================================================
---
modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/url.g
+++
modules/jndi/src/main/java/org/apache/harmony/jndi/provider/ldap/parser/url.g
@@ -394,5 +394,5 @@ String oid():
void test():
{}
{
- parseURL() "\n" test() | LOOKAHEAD(2) "\n" | "\n" <EOF>
| <EOF>
+ parseURL() ("\n" | "\u0085") test() | LOOKAHEAD(2)
("\n" | "\u0085") | ("\n" | "\u0085") <EOF> | <EOF>
}
