milestone: lexer now correctly identifies (almost) all constructs

2 files changed, 107 insertions, 4 deletions

diff --git a/grammar/debian.new b/grammar/debian.new
index a6574f62..4d55735c 100644
--- a/grammar/debian.new
+++ b/grammar/debian.new
@@ -103,5 +103,35 @@ global (dnscache=yes arg1="1 2" arg2 = "1 2" arg3 ="1=2\"3")
 # samples added to get full "flavor" of what we need to support...
 :msg, contains, "error" /var/log/somelog
 action(type=omfile target=/var/log/mail/log)
-*.* *  # test
+*.* /* comment */ *  # test
 *.info :ommysql:, tra, la , la # comment (comment to be part of old style line!)
+
+# from SUSE:
+if	( \
+	    /* kernel up to warning except of firewall  */ \
+	    ($syslogfacility-text == 'kern')      and      \
+	    ($syslogseverity <= 4 /* warning */ ) and not  \
+	    ($msg contains 'IN=' and $msg contains 'OUT=') \
+	) or ( \
+	    /* up to errors except of facility authpriv */ \
+	    ($syslogseverity <= 3 /* errors  */ ) and not  \
+	    ($syslogfacility-text == 'authpriv')           \
+	) \
+then	/dev/tty10
+&	|/dev/xconsole
+#
+# slightly modified to not use continuation lines
+if	(   /* kernel up to warning except of firewall  */ 
+	    ($syslogfacility-text == 'kern')      and      
+	    ($syslogseverity <= 4 /* warning */ ) and not  
+	    ($msg contains 'IN=' and $msg contains 'OUT=') 
+	) or ( 
+	    /* up to errors except of facility authpriv */ 
+	    ($syslogseverity <= 3 /* errors  */ ) and not  
+	    ($syslogfacility-text == 'authpriv')           
+	) 
+then	/dev/tty10
+&	|/dev/xconsole
+
+*.* rger # write to user (ugly...)
+ruleset name
diff --git a/grammar/rscript.l b/grammar/rscript.l
index b5d23d7a..10ce1748 100644
--- a/grammar/rscript.l
+++ b/grammar/rscript.l
@@ -1,14 +1,77 @@
+ /* lex file for rsyslog config format v2.
+  * Please note: this file introduces the new config format, but maintains
+  * backward compatibility. In order to do so, the grammar is not 100% clean,
+  * but IMHO still sufficiently easy both to understand for programmers
+  * maitaining the code as well as users writing the config file. Users are,
+  * of course, encouraged to use new constructs only. But it needs to be noted
+  * that some of the legacy constructs (specifically the in-front-of-action
+  * PRI filter) are very hard to beat in ease of use, at least for simpler
+  * cases. So while we hope that cfsysline support can be dropped some time in
+  * the future, we will probably keep these useful constructs.
+  *
+  * Copyright (C) 2011 by Rainer Gerhards and Adiscon GmbH
+  * Released under the GNU GPL v3. For details see LICENSE file.
+  */
+
 %option noyywrap nodefault case-insensitive
  /*%option noyywrap nodefault case-insensitive */
 
 %x INOBJ
 	/* INOBJ is selected if we are inside an object (name/value pairs!) */
+%x COMMENT
+	/* COMMENT is "the usual trick" to handle C-style comments */
+%x EXPR
+	/* EXPR is a bit ugly, but we need it to support pre v6-syntax. The problem
+	 * is that cfsysline statement start with $..., the same like variables in
+	 * an expression. However, cfsysline statements can never appear inside an
+	 * expression. So we create a specific expr mode, which is turned on after 
+	 * we lexed a keyword that needs to be followed by an expression (using 
+	 * knowledge from the upper layer...). In expr mode, we strictly do
+	 * expression-based parsing. Expr mode is stopped when we reach a token
+	 * that can not be part of an expression (currently only "then"). As I
+	 * wrote this ugly, but the price needed to pay in order to remain
+	 * compatible to the previous format.
+	 */
 %{
 #include <libestr.h>
+static int preCommentState;
 %}
 
 %%
 
+ /* keywords */
+"if"		{ printf("IF\n");
+		  BEGIN EXPR;
+		}
+<EXPR>"then"		{ printf("THEN\n");
+			  BEGIN INITIAL;
+			}
+<EXPR>"or"		{ printf("OR\n"); }
+<EXPR>"and"		{ printf("AND\n"); }
+<EXPR>"not"		{ printf("NOT\n"); }
+<EXPR>"("		{ printf("LPAREN\n"); }
+<EXPR>")"		{ printf("RPAREN\n"); }
+<EXPR>"=="		{ printf("==\n"); }
+<EXPR>"<="		{ printf("<=\n"); }
+<EXPR>">="		{ printf(">=\n"); }
+<EXPR>"!=" |
+<EXPR>"<>"		{ printf("!=\n"); }
+<EXPR>"<"		{ printf("<\n"); }
+<EXPR>">"		{ printf(">\n"); }
+<EXPR>"contains"	{ printf("CONTAINS\n"); }
+<EXPR>"contains_i"	{ printf("CONTAINS_I\n"); }
+<EXPR>"startswith"	{ printf("STARTSWITH\n"); }
+<EXPR>"startswith_i"	{ printf("STARTSWITH_I\n"); }
+<EXPR>-?0[0-7]+		{ printf("NUMBER (oct) %s\n", yytext); }
+<EXPR>-?0x[0-7a-f]	{ printf("NUMBER (hex) %s\n", yytext); }
+<EXPR>-?([1-9][0-9]*|0)	{ printf("NUMBER (dec) %s\n", yytext); }
+<EXPR>\$[$!]{0,1}[a-z][a-z0-9\-_\.]*		{ printf("VARNAME: '%s'\n", yytext); }
+<EXPR>\'([^'\\]|\\['])*\' { printf("EXPR string: -%s-\n", yytext); }
+<EXPR>[ \t\n]
+<EXPR>.			{ printf("invalid char in expr: %s\n", yytext); }
+"&"		{ printf("AMPER\n"); }
+"ruleset"	{ printf("RULESET\n"); }
+
 "global"[ \n\t]*"("			{ printf("OBJ GLOBAL begin\n");
 					  BEGIN INOBJ;
 					}
@@ -23,17 +86,18 @@
 					}
 ^[ \t]*:\$?[a-z]+[ ]*,[ ]*!?[a-z]+[ ]*,[ ]*\".*\"	{
 					  printf("PROP-FILT: '%s'\n", yytext);
-					  /*BEGIN OLDACT;*/
 					}
 
 ^[ \t]*[,\*a-z]+\.[,!=;\.\*a-z]+	{ printf("PRI-FILT: '%s'\n", yytext);
-					  /*BEGIN OLDACT;*/
 					}
 
 "*" |
-[\|\.\/\-:][^\n]+			{ printf("old style action: '%s'\n", yytext);
+\/[^*][^\n]* |
+[\|\.\-:][^\n]+				{ printf("old style action: '%s'\n", yytext);
 					}
 
+[a-z0-9_\-\+]+				{ printf("name: '%s'\n", yytext); }
+
 <INOBJ>")"				{ printf("OBJ end\n");
 					  BEGIN INITIAL;
 					}
@@ -46,6 +110,15 @@
  			  printf("INOBJ: value '%s'\n", yytext);
 			  BEGIN INOBJ;
 			}
+"/*"		{ preCommentState = YY_START;
+		  BEGIN COMMENT;
+		}
+<EXPR>"/*"		{ preCommentState = YY_START;
+			  BEGIN COMMENT;
+			}
+<COMMENT>"*/"		{ BEGIN preCommentState; }
+<COMMENT>([^*]|\n)+|.
+
 <INOBJ>#.*\n	/* skip comments in input */
 <INOBJ>[ \n\t]
 <INOBJ>.		{ printf("INOBJ: invalid char '%s'\n", yytext); }