more cleanup and working towards a parser module calling interface

I cleaned up a lot of config variable access along the way. This
version compiles and runs, but does not yet offer any enhanced
functionality. pmrfc5424 is just a dummy that is not yet being used.

9 files changed, 151 insertions, 21 deletions

diff --git a/runtime/glbl.c b/runtime/glbl.c
index f27b8e73..71c2ed0d 100644
--- a/runtime/glbl.c
+++ b/runtime/glbl.c
@@ -56,6 +56,7 @@ DEFobjCurrIf(prop)
  */
 static uchar *pszWorkDir = NULL;
 static int bOptimizeUniProc = 1;	/* enable uniprocessor optimizations */
+static int bParseHOSTNAMEandTAG = 1;	/* parser modification (based on startup params!) */
 static int bPreserveFQDN = 0;		/* should FQDNs always be preserved? */
 static int iMaxLine = 2048;		/* maximum length of a syslog message */
 static int iDefPFFamily = PF_UNSPEC;     /* protocol family (IPv4, IPv6 or both) */
@@ -94,6 +95,7 @@ static dataType Get##nameFunc(void) \
 	return(nameVar); \
 }
 
+SIMP_PROP(ParseHOSTNAMEandTAG, bParseHOSTNAMEandTAG, int)
 SIMP_PROP(OptimizeUniProc, bOptimizeUniProc, int)
 SIMP_PROP(PreserveFQDN, bPreserveFQDN, int)
 SIMP_PROP(MaxLine, iMaxLine, int)
@@ -183,6 +185,7 @@ finalize_it:
 	RETiRet;
 }
 
+
 /* return our local hostname as a string property
  */
 static prop_t*
@@ -266,6 +269,7 @@ CODESTARTobjQueryInterface(glbl)
 	pIf->Set##name = Set##name;
 	SIMP_PROP(MaxLine);
 	SIMP_PROP(OptimizeUniProc);
+	SIMP_PROP(ParseHOSTNAMEandTAG);
 	SIMP_PROP(PreserveFQDN);
 	SIMP_PROP(DefPFFamily);
 	SIMP_PROP(DropMalPTRMsgs);
diff --git a/runtime/glbl.h b/runtime/glbl.h
index 0d0c8210..7506f16b 100644
--- a/runtime/glbl.h
+++ b/runtime/glbl.h
@@ -64,9 +64,11 @@ BEGINinterface(glbl) /* name must also be changed in ENDinterface macro! */
 	/* added v4, 2009-07-20 */
 	int (*GetGlobalInputTermState)(void);
 	void (*SetGlobalInputTermination)(void);
+	/* added v5, 2009-11-03 */
+	SIMP_PROP(ParseHOSTNAMEandTAG, int)
 #undef	SIMP_PROP
 ENDinterface(glbl)
-#define glblCURR_IF_VERSION 4 /* increment whenever you change the interface structure! */
+#define glblCURR_IF_VERSION 5 /* increment whenever you change the interface structure! */
 /* version 2 had PreserveFQDN added - rgerhards, 2008-12-08 */
 
 /* the remaining prototypes */
diff --git a/runtime/module-template.h b/runtime/module-template.h
index d49da2c9..b136b6a3 100644
--- a/runtime/module-template.h
+++ b/runtime/module-template.h
@@ -46,6 +46,9 @@
 	DEFobjCurrIf(obj)
 #define DEF_LMOD_STATIC_DATA \
 	DEF_MOD_STATIC_DATA
+#define DEF_PMOD_STATIC_DATA \
+	DEFobjCurrIf(obj) \
+	DEF_MOD_STATIC_DATA
 
 
 /* Macro to define the module type. Each module can only have a single type. If
@@ -65,6 +68,7 @@ static rsRetVal modGetType(eModType_t *modType) \
 
 #define MODULE_TYPE_INPUT MODULE_TYPE(eMOD_IN)
 #define MODULE_TYPE_OUTPUT MODULE_TYPE(eMOD_OUT)
+#define MODULE_TYPE_PARSER MODULE_TYPE(eMOD_PARSER)
 #define MODULE_TYPE_LIB \
 	DEF_LMOD_STATIC_DATA \
 	MODULE_TYPE(eMOD_LIB)
@@ -400,6 +404,16 @@ static rsRetVal queryEtryPt(uchar *name, rsRetVal (**pEtryPoint)())\
 #define CODEqueryEtryPt_STD_LIB_QUERIES \
 	CODEqueryEtryPt_STD_MOD_QUERIES
 
+/* the following definition is the standard block for queryEtryPt for PARSER
+ * modules. This can be used if no specific handling (e.g. to cover version
+ * differences) is needed.
+ */
+#define CODEqueryEtryPt_STD_PMOD_QUERIES \
+	CODEqueryEtryPt_STD_MOD_QUERIES \
+	else if(!strcmp((char*) name, "parse")) {\
+		*pEtryPoint = parse;\
+	}
+
 /* modInit()
  * This has an extra parameter, which is the specific name of the modInit
  * function. That is needed for built-in modules, which must have unique
@@ -590,5 +604,21 @@ static rsRetVal doHUP(instanceData __attribute__((unused)) *pData)\
 }
 
 
+
+/* parse() - main entry point of parser modules
+ */
+#define BEGINparse \
+static rsRetVal parse(msg_t *pMsg)\
+{\
+	DEFiRet;
+
+#define CODESTARTparse \
+	assert(pMsg != NULL);
+
+#define ENDparse \
+	RETiRet;\
+}
+
+
 /* vim:set ai:
  */
diff --git a/runtime/modules.c b/runtime/modules.c
index bdb15e7f..5321c5e8 100644
--- a/runtime/modules.c
+++ b/runtime/modules.c
@@ -475,6 +475,9 @@ doModInit(rsRetVal (*modInit)(int, int*, rsRetVal(**)(), rsRetVal(*)(), modInfo_
 			break;
 		case eMOD_LIB:
 			break;
+		case eMOD_PARSER:
+			CHKiRet((*pNew->modQueryEtryPt)((uchar*)"parse", &pNew->mod.pm.parse));
+			break;
 	}
 
 	pNew->pszName = (uchar*) strdup((char*)name); /* we do not care if strdup() fails, we can accept that */
@@ -521,6 +524,9 @@ static void modPrintList(void)
 		case eMOD_LIB:
 			dbgprintf("library");
 			break;
+		case eMOD_PARSER:
+			dbgprintf("parser");
+			break;
 		}
 		dbgprintf(" module.\n");
 		dbgprintf("Entry points:\n");
diff --git a/runtime/modules.h b/runtime/modules.h
index 71e3199c..e92760b8 100644
--- a/runtime/modules.h
+++ b/runtime/modules.h
@@ -51,9 +51,10 @@
 #define CURR_MOD_IF_VERSION 5
 
 typedef enum eModType_ {
-	eMOD_IN,	/* input module */
-	eMOD_OUT,	/* output module */
-	eMOD_LIB	/* library module - this module provides one or many interfaces */
+	eMOD_IN = 0,	/* input module */
+	eMOD_OUT = 1,	/* output module */
+	eMOD_LIB = 2,	/* library module */
+	eMOD_PARSER = 3	/* parser module */
 } eModType_t;
 
 
@@ -117,7 +118,10 @@ typedef struct modInfo_s {
 			rsRetVal (*parseSelectorAct)(uchar**, void**,omodStringRequest_t**);
 		} om;
 		struct { /* data for library modules */
-		} fm;
+		} lm;
+		struct { /* data for parser modules */
+			rsRetVal (*parse)(msg_t*);
+		} pm;
 	} mod;
 	void *pModHdlr; /* handler to the dynamic library holding the module */
 #	ifdef DEBUG
diff --git a/runtime/parser.c b/runtime/parser.c
index 89e59f87..e2ad69e4 100644
--- a/runtime/parser.c
+++ b/runtime/parser.c
@@ -39,8 +39,10 @@
 #include "obj.h"
 #include "datetime.h"
 #include "errmsg.h"
+#include "parser.h"
 #include "unicode-helper.h"
 #include "dirty.h"
+#include "cfsysline.h"
 
 /* some defines */
 #define DEFUPRI		(LOG_USER|LOG_NOTICE)
@@ -52,23 +54,22 @@ DEFobjCurrIf(errmsg)
 DEFobjCurrIf(datetime)
 
 /* static data */
+static int bParseHOSTNAMEandTAG;	/* cache for the equally-named global param - performance enhancement */
+/* config data */
+static uchar cCCEscapeChar = '#';/* character to be used to start an escape sequence for control chars */
+static int bEscapeCCOnRcv = 1; /* escape control characters on reception: 0 - no, 1 - yes */
+static int bDropTrailingLF = 1; /* drop trailing LF's on reception? */
 
 
-/* this is a dummy class init
+/* we need to provide standard constructors and destructors, even though
+ * we only have static methods. The framework requires that.
  */
-rsRetVal parserClassInit(void)
-{
-	DEFiRet;
+BEGINobjConstruct(parser) /* be sure to specify the object type also in END macro! */
+ENDobjConstruct(parser)
 
-	/* request objects we use */
-	CHKiRet(objGetObjInterface(&obj)); /* this provides the root pointer for all other queries */
-	CHKiRet(objUse(glbl, CORE_COMPONENT));
-	CHKiRet(objUse(errmsg, CORE_COMPONENT));
-	CHKiRet(objUse(datetime, CORE_COMPONENT));
-// TODO: free components! see action.c
-finalize_it:
-	RETiRet;
-}
+BEGINobjDestruct(parser) /* be sure to specify the object type also in END and CODESTART macros! */
+CODESTARTobjDestruct(parser)
+ENDobjDestruct(parser)
 
 /***************************RFC 5425 PARSER ******************************************************/
 
@@ -643,7 +644,8 @@ finalize_it:
  * extended to support configured parsers.
  * rgerhards, 2008-10-09
  */
-rsRetVal parseMsg(msg_t *pMsg)
+static rsRetVal
+ParseMsg(msg_t *pMsg)
 {
 	DEFiRet;
 	uchar *msg;
@@ -709,3 +711,57 @@ rsRetVal parseMsg(msg_t *pMsg)
 finalize_it:
 	RETiRet;
 }
+
+
+/* queryInterface function-- rgerhards, 2009-11-03
+ */
+BEGINobjQueryInterface(parser)
+CODESTARTobjQueryInterface(parser)
+	if(pIf->ifVersion != parserCURR_IF_VERSION) { /* check for current version, increment on each change */
+		ABORT_FINALIZE(RS_RET_INTERFACE_NOT_SUPPORTED);
+	}
+
+	/* ok, we have the right interface, so let's fill it
+	 * Please note that we may also do some backwards-compatibility
+	 * work here (if we can support an older interface version - that,
+	 * of course, also affects the "if" above).
+	 */
+	pIf->ParseMsg = ParseMsg;
+finalize_it:
+ENDobjQueryInterface(parser)
+
+
+
+/* Reset config variables to default values.
+ * rgerhards, 2007-07-17
+ */
+static rsRetVal
+resetConfigVariables(uchar __attribute__((unused)) *pp, void __attribute__((unused)) *pVal)
+{
+	cCCEscapeChar = '#';
+	bEscapeCCOnRcv = 1; /* default is to escape control characters */
+	bDropTrailingLF = 1; /* default is to drop trailing LF's on reception */
+
+	return RS_RET_OK;
+}
+
+
+/* Initialize the parser class. Must be called as the very first method
+ * before anything else is called inside this class.
+ * rgerhards, 2009-11-02
+ */
+//BEGINObjClassInit(parser, 1, OBJ_IS_CORE_MODULE) /* class, version */
+BEGINAbstractObjClassInit(parser, 1, OBJ_IS_CORE_MODULE) /* class, version */
+	/* request objects we use */
+	CHKiRet(objUse(glbl, CORE_COMPONENT));
+	CHKiRet(objUse(errmsg, CORE_COMPONENT));
+	CHKiRet(objUse(datetime, CORE_COMPONENT));
+
+	bParseHOSTNAMEandTAG = glbl.GetParseHOSTNAMEandTAG(); /* cache value, is set only during rsyslogd option processing */
+
+	CHKiRet(regCfSysLineHdlr((uchar *)"controlcharacterescapeprefix", 0, eCmdHdlrGetChar, NULL, &cCCEscapeChar, NULL));
+	CHKiRet(regCfSysLineHdlr((uchar *)"droptrailinglfonreception", 0, eCmdHdlrBinary, NULL, &bDropTrailingLF, NULL));
+	CHKiRet(regCfSysLineHdlr((uchar *)"escapecontrolcharactersonreceive", 0, eCmdHdlrBinary, NULL, &bEscapeCCOnRcv, NULL));
+	CHKiRet(regCfSysLineHdlr((uchar *)"resetconfigvariables", 1, eCmdHdlrCustomHandler, resetConfigVariables, NULL, NULL));
+ENDObjClassInit(parser)
+
diff --git a/runtime/parser.h b/runtime/parser.h
index cec9c083..d5d9243d 100644
--- a/runtime/parser.h
+++ b/runtime/parser.h
@@ -24,7 +24,21 @@
 #ifndef INCLUDED_PARSE_H
 #define INCLUDED_PARSE_H
 
-extern rsRetVal parserClassInit(void);
-extern rsRetVal parseMsg(msg_t*);
+/* the parser object, a dummy because we have only static methods */
+typedef struct parser_s {
+	BEGINobjInstance;	/* Data to implement generic object - MUST be the first data element! */
+} parser_t;
+
+/* interfaces */
+BEGINinterface(parser) /* name must also be changed in ENDinterface macro! */
+	INTERFACEObjDebugPrint(var);
+	rsRetVal (*ParseMsg)(msg_t *pMsg);
+ENDinterface(parser)
+#define parserCURR_IF_VERSION 1 /* increment whenever you change the interface above! */
+
+
+/* prototypes */
+PROTOTYPEObj(parser);
+
 
 #endif /* #ifndef INCLUDED_PARSE_H */
diff --git a/runtime/rsyslog.c b/runtime/rsyslog.c
index 443d0f41..8c44d16a 100644
--- a/runtime/rsyslog.c
+++ b/runtime/rsyslog.c
@@ -80,6 +80,7 @@
 #include "prop.h"
 #include "rule.h"
 #include "ruleset.h"
+#include "parser.h"
 
 /* forward definitions */
 static rsRetVal dfltErrLogger(int, uchar *errMsg);
@@ -183,6 +184,8 @@ rsrtInit(char **ppErrObj, obj_if_t *pObjIF)
 		CHKiRet(qqueueClassInit(NULL));
 		if(ppErrObj != NULL) *ppErrObj = "conf";
 		CHKiRet(confClassInit(NULL));
+		if(ppErrObj != NULL) *ppErrObj = "parser";
+		CHKiRet(parserClassInit(NULL));
 
 		/* dummy "classes" */
 		if(ppErrObj != NULL) *ppErrObj = "str";
diff --git a/runtime/rsyslog.h b/runtime/rsyslog.h
index f059a970..ef323c7d 100644
--- a/runtime/rsyslog.h
+++ b/runtime/rsyslog.h
@@ -40,6 +40,17 @@
 #define CONF_TAG_BUFSIZE		32
 #define CONF_HOSTNAME_BUFSIZE		32
 #define CONF_PROP_BUFSIZE		16	/* should be close to sizeof(ptr) or lighly above it */
+#define	CONF_MIN_SIZE_FOR_COMPRESS	60 	/* config param: minimum message size to try compression. The smaller
+						 * the message, the less likely is any compression gain. We check for
+						 * gain before we submit the message. But to do so we still need to
+						 * do the (costly) compress() call. The following setting sets a size
+						 * for which no call to compress() is done at all. This may result in
+						 * a few more bytes being transmited but better overall performance.
+						 * Note: I have not yet checked the minimum UDP packet size. It might be
+						 * that we do not save anything by compressing very small messages, because
+						 * UDP might need to pad ;)
+						 * rgerhards, 2006-11-30
+						 */
 
 
 /* ############################################################# *