finalized parser module calling interface

looks like we are almost done and need only to add the ruleset parser-specific
config options.

4 files changed, 76 insertions, 28 deletions

diff --git a/runtime/modules.c b/runtime/modules.c
index 41645a27..fd3468d8 100644
--- a/runtime/modules.c
+++ b/runtime/modules.c
@@ -493,10 +493,14 @@ doModInit(rsRetVal (*modInit)(int, int*, rsRetVal(**)(), rsRetVal(*)(), modInfo_
 			CHKiRet(parser.Construct(&pParser));
 
 			/* check some features */
-			localRet = pNew->isCompatibleWithFeature(sFEATUREAtomaticSanitazion);
+			localRet = pNew->isCompatibleWithFeature(sFEATUREAutomaticSanitazion);
 			if(localRet == RS_RET_OK){
 				CHKiRet(parser.SetDoSanitazion(pParser, TRUE));
 			}
+			localRet = pNew->isCompatibleWithFeature(sFEATUREAutomaticPRIParsing);
+			if(localRet == RS_RET_OK){
+				CHKiRet(parser.SetDoPRIParsing(pParser, TRUE));
+			}
 
 			CHKiRet(parser.SetName(pParser, pParserName));
 			CHKiRet(parser.SetModPtr(pParser, pNew));
diff --git a/runtime/parser.c b/runtime/parser.c
index 2c764739..e87068fc 100644
--- a/runtime/parser.c
+++ b/runtime/parser.c
@@ -376,36 +376,19 @@ finalize_it:
 	RETiRet;
 }
 
-
-/* Parse a received message. The object's rawmsg property is taken and
- * parsed according to the relevant standards. This can later be
- * extended to support configured parsers.
- * rgerhards, 2008-10-09
+/* A standard parser to parse out the PRI. This is made available in
+ * this module as it is expected that allmost all parsers will need
+ * that functionality and so they do not need to implement it themsleves.
  */
-static rsRetVal
-ParseMsg(msg_t *pMsg)
+static inline rsRetVal
+ParsePRI(msg_t *pMsg)
 {
-	uchar *msg;
 	int pri;
+	uchar *msg;
 	int lenMsg;
 	int iPriText;
-	rsRetVal localRet;
-	parserList_t *pParserList;
-	static int iErrMsgRateLimiter = 0;
 	DEFiRet;
 
-	if(pMsg->iLenRawMsg == 0)
-		ABORT_FINALIZE(RS_RET_EMPTY_MSG);
-
-#	ifdef USE_NETZIP
-	CHKiRet(uncompressMessage(pMsg));
-#	endif
-
-	CHKiRet(SanitizeMsg(pMsg));
-
-	/* we needed to sanitize first, because we otherwise do not have a C-string we can print... */
-	DBGPRINTF("msg parser: flags %x, from '%s', msg '%s'\n", pMsg->msgFlags, getRcvFrom(pMsg), pMsg->pszRawMsg);
-
 	/* pull PRI */
 	lenMsg = pMsg->iLenRawMsg;
 	msg = pMsg->pszRawMsg;
@@ -428,9 +411,43 @@ ParseMsg(msg_t *pMsg)
 	pMsg->iFacility = LOG_FAC(pri);
 	pMsg->iSeverity = LOG_PRI(pri);
 	MsgSetAfterPRIOffs(pMsg, msg - pMsg->pszRawMsg);
+	RETiRet;
+}
+
+
+/* Parse a received message. The object's rawmsg property is taken and
+ * parsed according to the relevant standards. This can later be
+ * extended to support configured parsers.
+ * rgerhards, 2008-10-09
+ */
+static rsRetVal
+ParseMsg(msg_t *pMsg)
+{
+	rsRetVal localRet;
+	parserList_t *pParserList;
+	parser_t *pParser;
+	bool bIsSanitized;
+	bool bPRIisParsed;
+	static int iErrMsgRateLimiter = 0;
+	DEFiRet;
+
+	if(pMsg->iLenRawMsg == 0)
+		ABORT_FINALIZE(RS_RET_EMPTY_MSG);
+
+#	ifdef USE_NETZIP
+	CHKiRet(uncompressMessage(pMsg));
+#	endif
+
+	/* we take the risk to print a non-sanitized string, because this is the best we can get
+	 * (and that functionality is too important for debugging to drop it...).
+	 */
+	DBGPRINTF("msg parser: flags %x, from '%s', msg '%.50s'\n", pMsg->msgFlags,
+		  getRcvFrom(pMsg), pMsg->pszRawMsg);
 
 	/* we now need to go through our list of parsers and see which one is capable of
-	 * parsing the message.
+	 * parsing the message. Note that the first parser that requires message sanitization
+	 * will cause it to happen. After that, access to the unsanitized message is no
+	 * loger possible.
 	 */
 	pParserList = ruleset.GetParserList(pMsg);
 	if(pParserList == NULL)
@@ -438,7 +455,18 @@ ParseMsg(msg_t *pMsg)
 	DBGPRINTF("Using parser list %p%s.\n", pParserList,
 		  (pParserList == pDfltParsLst) ? " (the default list)" : "");
 
+	bIsSanitized = FALSE;
+	bPRIisParsed = FALSE;
 	while(pParserList != NULL) {
+		pParser = pParserList->pParser;
+		if(pParser->bDoSanitazion && bIsSanitized == FALSE) {
+			CHKiRet(SanitizeMsg(pMsg));
+			if(pParser->bDoPRIParsing && bPRIisParsed == FALSE) {
+				CHKiRet(ParsePRI(pMsg));
+				bPRIisParsed = TRUE;
+			}
+			bIsSanitized = TRUE;
+		}
 		localRet = pParserList->pParser->pModule->mod.pm.parse(pMsg);
 		if(localRet != RS_RET_COULD_NOT_PARSE)
 			break;
@@ -453,14 +481,14 @@ ParseMsg(msg_t *pMsg)
 	if(localRet != RS_RET_OK) {
 		if(++iErrMsgRateLimiter > 1000) {
 			errmsg.LogError(0, localRet, "Error: one message could not be processed by "
-			 	"any parser, message is being discarded (start of raw msg: '%50s')", 
+			 	"any parser, message is being discarded (start of raw msg: '%.50s')", 
 				pMsg->pszRawMsg);
 		}
 		DBGPRINTF("No parser could process the message (state %d), we need to discard it.\n", localRet);
 		ABORT_FINALIZE(localRet);
 	}
 
-	/* finalize message object */
+	/* "finalize" message object */
 	pMsg->msgFlags &= ~NEEDS_PARSING; /* this message is now parsed */
 
 finalize_it:
@@ -516,6 +544,18 @@ SetDoSanitazion(parser_t *pThis, int bDoIt)
 }
 
 
+/* Specify if we should do standard PRI parsing before we pass the data
+ * down to the parser module.
+ */
+static rsRetVal
+SetDoPRIParsing(parser_t *pThis, int bDoIt)
+{
+	ISOBJ_TYPE_assert(pThis, parser);
+	pThis->bDoPRIParsing = bDoIt;
+	return RS_RET_OK;
+}
+
+
 /* queryInterface function-- rgerhards, 2009-11-03
  */
 BEGINobjQueryInterface(parser)
@@ -535,6 +575,7 @@ CODESTARTobjQueryInterface(parser)
 	pIf->SetName = SetName;
 	pIf->SetModPtr = SetModPtr;
 	pIf->SetDoSanitazion = SetDoSanitazion;
+	pIf->SetDoPRIParsing = SetDoPRIParsing;
 	pIf->ParseMsg = ParseMsg;
 	pIf->SanitizeMsg = SanitizeMsg;
 	pIf->InitParserList = InitParserList;
diff --git a/runtime/parser.h b/runtime/parser.h
index d600fb19..a92b9920 100644
--- a/runtime/parser.h
+++ b/runtime/parser.h
@@ -39,6 +39,7 @@ struct parser_s {
 	uchar *pName;		/* name of this parser */
 	modInfo_t *pModule;	/* pointer to parser's module */
 	bool bDoSanitazion;	/* do standard message sanitazion before calling parser? */
+	bool bDoPRIParsing;	/* do standard PRI parsing before calling parser? */
 };
 
 /* interfaces */
@@ -50,6 +51,7 @@ BEGINinterface(parser) /* name must also be changed in ENDinterface macro! */
 	rsRetVal (*SetName)(parser_t *pThis, uchar *name);
 	rsRetVal (*SetModPtr)(parser_t *pThis, modInfo_t *pMod);
 	rsRetVal (*SetDoSanitazion)(parser_t *pThis, int);
+	rsRetVal (*SetDoPRIParsing)(parser_t *pThis, int);
 	rsRetVal (*FindParser)(parser_t **ppThis, uchar*name);
 	rsRetVal (*InitParserList)(parserList_t **pListRoot);
 	rsRetVal (*AddParserToList)(parserList_t **pListRoot, parser_t *pParser);
diff --git a/runtime/syslogd-types.h b/runtime/syslogd-types.h
index 49eeaa6a..fcebd985 100644
--- a/runtime/syslogd-types.h
+++ b/runtime/syslogd-types.h
@@ -58,7 +58,8 @@
 typedef enum _syslogFeature {
 	sFEATURERepeatedMsgReduction = 1,	/* for output modules */
 	sFEATURENonCancelInputTermination = 2,	/* for input modules */
-	sFEATUREAtomaticSanitazion = 3		/* for parser modules */
+	sFEATUREAutomaticSanitazion = 3,	/* for parser modules */
+	sFEATUREAutomaticPRIParsing = 4		/* for parser modules */
 } syslogFeature;
 
 /* we define our own facility and severities */