/* rainerscript.c - routines to support RainerScript config language * * Module begun 2011-07-01 by Rainer Gerhards * * Copyright 2011-2014 Rainer Gerhards and Adiscon GmbH. * * This file is part of the rsyslog runtime library. * * The rsyslog runtime library is free software: you can redistribute it and/or modify * it under the terms of the GNU Lesser General Public License as published by * the Free Software Foundation, either version 3 of the License, or * (at your option) any later version. * * The rsyslog runtime library is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public License * along with the rsyslog runtime library. If not, see . * * A copy of the GPL can be found in the file "COPYING" in this distribution. * A copy of the LGPL can be found in the file "COPYING.LESSER" in this distribution. */ #include "config.h" #include #include #include #include #include #include #include #include #include #include #include #include "rsyslog.h" #include "rainerscript.h" #include "conf.h" #include "parserif.h" #include "parse.h" #include "rsconf.h" #include "grammar.h" #include "queue.h" #include "srUtils.h" #include "regexp.h" #include "obj.h" #include "modules.h" #include "ruleset.h" #include "msg.h" #include "wti.h" #include "unicode-helper.h" DEFobjCurrIf(obj) DEFobjCurrIf(regexp) struct cnfexpr* cnfexprOptimize(struct cnfexpr *expr); static void cnfstmtOptimizePRIFilt(struct cnfstmt *stmt); static void cnfarrayPrint(struct cnfarray *ar, int indent); struct cnffunc * cnffuncNew_prifilt(int fac); /* debug support: convert token to a human-readable string. Note that * this function only supports a single thread due to a static buffer. * This is deemed a solid solution, as it is intended to be used during * startup, only. * NOTE: This function MUST be updated if new tokens are defined in the * grammar. */ const char * tokenToString(const int token) { char *tokstr; static char tokbuf[512]; switch(token) { case NAME: tokstr = "NAME"; break; case FUNC: tokstr = "FUNC"; break; case BEGINOBJ: tokstr ="BEGINOBJ"; break; case ENDOBJ: tokstr ="ENDOBJ"; break; case BEGIN_ACTION: tokstr ="BEGIN_ACTION"; break; case BEGIN_PROPERTY: tokstr ="BEGIN_PROPERTY"; break; case BEGIN_CONSTANT: tokstr ="BEGIN_CONSTANT"; break; case BEGIN_TPL: tokstr ="BEGIN_TPL"; break; case BEGIN_RULESET: tokstr ="BEGIN_RULESET"; break; case STOP: tokstr ="STOP"; break; case SET: tokstr ="SET"; break; case UNSET: tokstr ="UNSET"; break; case CONTINUE: tokstr ="CONTINUE"; break; case CALL: tokstr ="CALL"; break; case LEGACY_ACTION: tokstr ="LEGACY_ACTION"; break; case LEGACY_RULESET: tokstr ="LEGACY_RULESET"; break; case PRIFILT: tokstr ="PRIFILT"; break; case PROPFILT: tokstr ="PROPFILT"; break; case IF: tokstr ="IF"; break; case THEN: tokstr ="THEN"; break; case ELSE: tokstr ="ELSE"; break; case OR: tokstr ="OR"; break; case AND: tokstr ="AND"; break; case NOT: tokstr ="NOT"; break; case VAR: tokstr ="VAR"; break; case STRING: tokstr ="STRING"; break; case NUMBER: tokstr ="NUMBER"; break; case CMP_EQ: tokstr ="CMP_EQ"; break; case CMP_NE: tokstr ="CMP_NE"; break; case CMP_LE: tokstr ="CMP_LE"; break; case CMP_GE: tokstr ="CMP_GE"; break; case CMP_LT: tokstr ="CMP_LT"; break; case CMP_GT: tokstr ="CMP_GT"; break; case CMP_CONTAINS: tokstr ="CMP_CONTAINS"; break; case CMP_CONTAINSI: tokstr ="CMP_CONTAINSI"; break; case CMP_STARTSWITH: tokstr ="CMP_STARTSWITH"; break; case CMP_STARTSWITHI: tokstr ="CMP_STARTSWITHI"; break; case UMINUS: tokstr ="UMINUS"; break; default: snprintf(tokbuf, sizeof(tokbuf), "%c[%d]", token, token); tokstr = tokbuf; break; } return tokstr; } const char* getFIOPName(const unsigned iFIOP) { char *pRet; switch(iFIOP) { case FIOP_CONTAINS: pRet = "contains"; break; case FIOP_ISEQUAL: pRet = "isequal"; break; case FIOP_STARTSWITH: pRet = "startswith"; break; case FIOP_REGEX: pRet = "regex"; break; case FIOP_EREREGEX: pRet = "ereregex"; break; case FIOP_ISEMPTY: pRet = "isempty"; break; default: pRet = "NOP"; break; } return pRet; } /* This function takes the filter part of a property * based filter and decodes it. It processes the line up to the beginning * of the action part. */ static rsRetVal DecodePropFilter(uchar *pline, struct cnfstmt *stmt) { rsParsObj *pPars = NULL; cstr_t *pCSCompOp = NULL; cstr_t *pCSPropName = NULL; int iOffset; /* for compare operations */ DEFiRet; ASSERT(pline != NULL); DBGPRINTF("Decoding property-based filter '%s'\n", pline); /* create parser object starting with line string without leading colon */ if((iRet = rsParsConstructFromSz(&pPars, pline+1)) != RS_RET_OK) { parser_errmsg("error %d constructing parser object", iRet); ABORT_FINALIZE(iRet); } /* read property */ iRet = parsDelimCStr(pPars, &pCSPropName, ',', 1, 1, 1); if(iRet != RS_RET_OK) { parser_errmsg("error %d parsing filter property", iRet); rsParsDestruct(pPars); ABORT_FINALIZE(iRet); } CHKiRet(msgPropDescrFill(&stmt->d.s_propfilt.prop, cstrGetSzStrNoNULL(pCSPropName), cstrLen(pCSPropName))); /* read operation */ iRet = parsDelimCStr(pPars, &pCSCompOp, ',', 1, 1, 1); if(iRet != RS_RET_OK) { parser_errmsg("error %d compare operation property - ignoring selector", iRet); rsParsDestruct(pPars); ABORT_FINALIZE(iRet); } /* we now first check if the condition is to be negated. To do so, we first * must make sure we have at least one char in the param and then check the * first one. * rgerhards, 2005-09-26 */ if(rsCStrLen(pCSCompOp) > 0) { if(*rsCStrGetBufBeg(pCSCompOp) == '!') { stmt->d.s_propfilt.isNegated = 1; iOffset = 1; /* ignore '!' */ } else { stmt->d.s_propfilt.isNegated = 0; iOffset = 0; } } else { stmt->d.s_propfilt.isNegated = 0; iOffset = 0; } if(!rsCStrOffsetSzStrCmp(pCSCompOp, iOffset, (uchar*) "contains", 8)) { stmt->d.s_propfilt.operation = FIOP_CONTAINS; } else if(!rsCStrOffsetSzStrCmp(pCSCompOp, iOffset, (uchar*) "isequal", 7)) { stmt->d.s_propfilt.operation = FIOP_ISEQUAL; } else if(!rsCStrOffsetSzStrCmp(pCSCompOp, iOffset, (uchar*) "isempty", 7)) { stmt->d.s_propfilt.operation = FIOP_ISEMPTY; } else if(!rsCStrOffsetSzStrCmp(pCSCompOp, iOffset, (uchar*) "startswith", 10)) { stmt->d.s_propfilt.operation = FIOP_STARTSWITH; } else if(!rsCStrOffsetSzStrCmp(pCSCompOp, iOffset, (unsigned char*) "regex", 5)) { stmt->d.s_propfilt.operation = FIOP_REGEX; } else if(!rsCStrOffsetSzStrCmp(pCSCompOp, iOffset, (unsigned char*) "ereregex", 8)) { stmt->d.s_propfilt.operation = FIOP_EREREGEX; } else { parser_errmsg("error: invalid compare operation '%s'", (char*) rsCStrGetSzStrNoNULL(pCSCompOp)); ABORT_FINALIZE(RS_RET_ERR); } if(stmt->d.s_propfilt.operation != FIOP_ISEMPTY) { /* read compare value */ iRet = parsQuotedCStr(pPars, &stmt->d.s_propfilt.pCSCompValue); if(iRet != RS_RET_OK) { parser_errmsg("error %d compare value property", iRet); rsParsDestruct(pPars); ABORT_FINALIZE(iRet); } } finalize_it: if(pPars != NULL) rsParsDestruct(pPars); if(pCSCompOp != NULL) rsCStrDestruct(&pCSCompOp); if(pCSPropName != NULL) cstrDestruct(&pCSPropName); RETiRet; } static void prifiltInvert(struct funcData_prifilt *__restrict__ const prifilt) { int i; for(i = 0 ; i < LOG_NFACILITIES+1 ; ++i) { prifilt->pmask[i] = ~prifilt->pmask[i]; } } /* set prifilt so that it matches for some severities, sev is its numerical * value. Mode is one of the compop tokens CMP_EQ, CMP_LT, CMP_LE, CMP_GT, * CMP_GE, CMP_NE. */ static void prifiltSetSeverity(struct funcData_prifilt *prifilt, int sev, int mode) { static int lessthanmasks[] = { 0x00, 0x01, 0x03, 0x07, 0x0f, 0x1f, 0x3f, 0x7f, 0xff }; int i; for(i = 0 ; i < LOG_NFACILITIES+1 ; ++i) { if(mode == CMP_EQ || mode == CMP_NE) prifilt->pmask[i] = 1 << sev; else if(mode == CMP_LT) prifilt->pmask[i] = lessthanmasks[sev]; else if(mode == CMP_LE) prifilt->pmask[i] = lessthanmasks[sev+1]; else if(mode == CMP_GT) prifilt->pmask[i] = ~lessthanmasks[sev+1]; else if(mode == CMP_GE) prifilt->pmask[i] = ~lessthanmasks[sev]; else DBGPRINTF("prifiltSetSeverity: program error, invalid mode %s\n", tokenToString(mode)); } if(mode == CMP_NE) prifiltInvert(prifilt); } /* set prifilt so that it matches for some facilities, fac is its numerical * value. Mode is one of the compop tokens CMP_EQ, CMP_LT, CMP_LE, CMP_GT, * CMP_GE, CMP_NE. For the given facilities, all severities are enabled. * NOTE: fac MUST be in the range 0..24 (not multiplied by 8)! */ static void prifiltSetFacility(struct funcData_prifilt *__restrict__ const prifilt, const int fac, const int mode) { int i; memset(prifilt->pmask, 0, sizeof(prifilt->pmask)); switch(mode) { case CMP_EQ: prifilt->pmask[fac] = TABLE_ALLPRI; break; case CMP_NE: prifilt->pmask[fac] = TABLE_ALLPRI; prifiltInvert(prifilt); break; case CMP_LT: for(i = 0 ; i < fac ; ++i) prifilt->pmask[i] = TABLE_ALLPRI; break; case CMP_LE: for(i = 0 ; i < fac+1 ; ++i) prifilt->pmask[i] = TABLE_ALLPRI; break; case CMP_GE: for(i = fac ; i < LOG_NFACILITIES+1 ; ++i) prifilt->pmask[i] = TABLE_ALLPRI; break; case CMP_GT: for(i = fac+1 ; i < LOG_NFACILITIES+1 ; ++i) prifilt->pmask[i] = TABLE_ALLPRI; break; default:break; } } /* combine a prifilt with AND/OR (the respective token values are * used to keep things simple). */ static void prifiltCombine(struct funcData_prifilt *__restrict__ const prifilt, struct funcData_prifilt *__restrict__ const prifilt2, const int mode) { int i; for(i = 0 ; i < LOG_NFACILITIES+1 ; ++i) { if(mode == AND) prifilt->pmask[i] = prifilt->pmask[i] & prifilt2->pmask[i]; else prifilt->pmask[i] = prifilt->pmask[i] | prifilt2->pmask[i]; } } void readConfFile(FILE * const fp, es_str_t **str) { char ln[10240]; char buf[512]; int lenBuf; int bWriteLineno = 0; int len, i; int start; /* start index of to be submitted text */ int bContLine = 0; int lineno = 0; *str = es_newStr(4096); while(fgets(ln, sizeof(ln), fp) != NULL) { ++lineno; if(bWriteLineno) { bWriteLineno = 0; lenBuf = sprintf(buf, "PreprocFileLineNumber(%d)\n", lineno); es_addBuf(str, buf, lenBuf); } len = strlen(ln); /* if we are continuation line, we need to drop leading WS */ if(bContLine) { for(start = 0 ; start < len && isspace(ln[start]) ; ++start) /* JUST SCAN */; } else { start = 0; } for(i = len - 1 ; i >= start && isspace(ln[i]) ; --i) /* JUST SCAN */; if(i >= 0) { if(ln[i] == '\\') { --i; bContLine = 1; } else { if(bContLine) /* write line number if we had cont line */ bWriteLineno = 1; bContLine = 0; } /* add relevant data to buffer */ es_addBuf(str, ln+start, i+1 - start); } if(!bContLine) es_addChar(str, '\n'); } /* indicate end of buffer to flex */ es_addChar(str, '\0'); es_addChar(str, '\0'); } /* comparison function for qsort() and bsearch() string array compare */ static int qs_arrcmp(const void *s1, const void *s2) { return es_strcmp(*((es_str_t**)s1), *((es_str_t**)s2)); } struct objlst* objlstNew(struct cnfobj *o) { struct objlst *lst; if((lst = malloc(sizeof(struct objlst))) != NULL) { lst->next = NULL; lst->obj = o; } cnfobjPrint(o); return lst; } /* add object to end of object list, always returns pointer to root object */ struct objlst* objlstAdd(struct objlst *root, struct cnfobj *o) { struct objlst *l; struct objlst *newl; newl = objlstNew(o); if(root == 0) { root = newl; } else { /* find last, linear search ok, as only during config phase */ for(l = root ; l->next != NULL ; l = l->next) ; l->next = newl; } return root; } /* add stmt to current script, always return root stmt pointer */ struct cnfstmt* scriptAddStmt(struct cnfstmt *root, struct cnfstmt *s) { struct cnfstmt *l; if(root == NULL) { root = s; } else { /* find last, linear search ok, as only during config phase */ for(l = root ; l->next != NULL ; l = l->next) ; l->next = s; } return root; } void objlstDestruct(struct objlst *lst) { struct objlst *toDel; while(lst != NULL) { toDel = lst; lst = lst->next; cnfobjDestruct(toDel->obj); free(toDel); } } void objlstPrint(struct objlst *lst) { dbgprintf("objlst %p:\n", lst); while(lst != NULL) { cnfobjPrint(lst->obj); lst = lst->next; } } struct nvlst* nvlstNewStr(es_str_t *value) { struct nvlst *lst; if((lst = malloc(sizeof(struct nvlst))) != NULL) { lst->next = NULL; lst->val.datatype = 'S'; lst->val.d.estr = value; lst->bUsed = 0; } return lst; } struct nvlst* nvlstNewArray(struct cnfarray *ar) { struct nvlst *lst; if((lst = malloc(sizeof(struct nvlst))) != NULL) { lst->next = NULL; lst->val.datatype = 'A'; lst->val.d.ar = ar; lst->bUsed = 0; } return lst; } struct nvlst* nvlstSetName(struct nvlst *lst, es_str_t *name) { lst->name = name; return lst; } void nvlstDestruct(struct nvlst *lst) { struct nvlst *toDel; while(lst != NULL) { toDel = lst; lst = lst->next; es_deleteStr(toDel->name); varDelete(&toDel->val); free(toDel); } } void nvlstPrint(struct nvlst *lst) { char *name, *value; dbgprintf("nvlst %p:\n", lst); while(lst != NULL) { name = es_str2cstr(lst->name, NULL); switch(lst->val.datatype) { case 'A': dbgprintf("\tname: '%s':\n", name); cnfarrayPrint(lst->val.d.ar, 5); break; case 'S': value = es_str2cstr(lst->val.d.estr, NULL); dbgprintf("\tname: '%s', value '%s'\n", name, value); free(value); break; default:dbgprintf("nvlstPrint: unknown type '%s'\n", tokenToString(lst->val.datatype)); break; } free(name); lst = lst->next; } } /* find a name starting at node lst. Returns node with this * name or NULL, if none found. */ struct nvlst* nvlstFindName(struct nvlst *lst, es_str_t *name) { while(lst != NULL && es_strcmp(lst->name, name)) lst = lst->next; return lst; } /* find a name starting at node lst. Same as nvlstFindName, but * for classical C strings. This is useful because the config system * uses C string constants. */ static inline struct nvlst* nvlstFindNameCStr(struct nvlst *lst, char *name) { es_size_t lenName = strlen(name); while(lst != NULL && es_strcasebufcmp(lst->name, (uchar*)name, lenName)) lst = lst->next; return lst; } /* check if there are duplicate names inside a nvlst and emit * an error message, if so. */ static inline void nvlstChkDupes(struct nvlst *lst) { char *cstr; while(lst != NULL) { if(nvlstFindName(lst->next, lst->name) != NULL) { cstr = es_str2cstr(lst->name, NULL); parser_errmsg("duplicate parameter '%s' -- " "interpretation is ambigious, one value " "will be randomly selected. Fix this problem.", cstr); free(cstr); } lst = lst->next; } } /* check for unused params and emit error message is found. This must * be called after all config params have been pulled from the object * (otherwise the flags are not correctly set). */ void nvlstChkUnused(struct nvlst *lst) { char *cstr; while(lst != NULL) { if(!lst->bUsed) { cstr = es_str2cstr(lst->name, NULL); parser_errmsg("parameter '%s' not known -- " "typo in config file?", cstr); free(cstr); } lst = lst->next; } } static inline int doGetSize(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { unsigned char *c; es_size_t i; long long n; int r; c = es_getBufAddr(valnode->val.d.estr); n = 0; i = 0; while(i < es_strlen(valnode->val.d.estr) && isdigit(*c)) { n = 10 * n + *c - '0'; ++i; ++c; } if(i < es_strlen(valnode->val.d.estr)) { ++i; switch(*c) { /* traditional binary-based definitions */ case 'k': n *= 1024; break; case 'm': n *= 1024 * 1024; break; case 'g': n *= 1024 * 1024 * 1024; break; case 't': n *= (int64) 1024 * 1024 * 1024 * 1024; break; /* tera */ case 'p': n *= (int64) 1024 * 1024 * 1024 * 1024 * 1024; break; /* peta */ case 'e': n *= (int64) 1024 * 1024 * 1024 * 1024 * 1024 * 1024; break; /* exa */ /* and now the "new" 1000-based definitions */ case 'K': n *= 1000; break; case 'M': n *= 1000000; break; case 'G': n *= 1000000000; break; /* we need to use the multiplication below because otherwise * the compiler gets an error during constant parsing */ case 'T': n *= (int64) 1000 * 1000000000; break; /* tera */ case 'P': n *= (int64) 1000000 * 1000000000; break; /* peta */ case 'E': n *= (int64) 1000000000 * 1000000000; break; /* exa */ default: --i; break; /* indicates error */ } } if(i == es_strlen(valnode->val.d.estr)) { val->val.datatype = 'N'; val->val.d.n = n; r = 1; } else { parser_errmsg("parameter '%s' does not contain a valid size", param->name); r = 0; } return r; } static inline int doGetBinary(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { int r = 1; val->val.datatype = 'N'; if(!es_strbufcmp(valnode->val.d.estr, (unsigned char*) "on", 2)) { val->val.d.n = 1; } else if(!es_strbufcmp(valnode->val.d.estr, (unsigned char*) "off", 3)) { val->val.d.n = 0; } else { parser_errmsg("parameter '%s' must be \"on\" or \"off\" but " "is neither. Results unpredictable.", param->name); val->val.d.n = 0; r = 0; } return r; } static inline int doGetQueueType(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { char *cstr; int r = 1; if(!es_strcasebufcmp(valnode->val.d.estr, (uchar*)"fixedarray", 10)) { val->val.d.n = QUEUETYPE_FIXED_ARRAY; } else if(!es_strcasebufcmp(valnode->val.d.estr, (uchar*)"linkedlist", 10)) { val->val.d.n = QUEUETYPE_LINKEDLIST; } else if(!es_strcasebufcmp(valnode->val.d.estr, (uchar*)"disk", 4)) { val->val.d.n = QUEUETYPE_DISK; } else if(!es_strcasebufcmp(valnode->val.d.estr, (uchar*)"direct", 6)) { val->val.d.n = QUEUETYPE_DIRECT; } else { cstr = es_str2cstr(valnode->val.d.estr, NULL); parser_errmsg("param '%s': unknown queue type: '%s'", param->name, cstr); free(cstr); r = 0; } val->val.datatype = 'N'; return r; } /* A file create-mode must be a four-digit octal number * starting with '0'. */ static inline int doGetFileCreateMode(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { int fmtOK = 0; char *cstr; uchar *c; if(es_strlen(valnode->val.d.estr) == 4) { c = es_getBufAddr(valnode->val.d.estr); if( (c[0] == '0') && (c[1] >= '0' && c[1] <= '7') && (c[2] >= '0' && c[2] <= '7') && (c[3] >= '0' && c[3] <= '7') ) { fmtOK = 1; } } if(fmtOK) { val->val.datatype = 'N'; val->val.d.n = (c[1]-'0') * 64 + (c[2]-'0') * 8 + (c[3]-'0'); } else { cstr = es_str2cstr(valnode->val.d.estr, NULL); parser_errmsg("file modes need to be specified as " "4-digit octal numbers starting with '0' -" "parameter '%s=\"%s\"' is not a file mode", param->name, cstr); free(cstr); } return fmtOK; } static inline int doGetGID(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { char *cstr; int r; struct group *resultBuf; struct group wrkBuf; char stringBuf[2048]; /* 2048 has been proven to be large enough */ cstr = es_str2cstr(valnode->val.d.estr, NULL); getgrnam_r(cstr, &wrkBuf, stringBuf, sizeof(stringBuf), &resultBuf); if(resultBuf == NULL) { parser_errmsg("parameter '%s': ID for group %s could not " "be found", param->name, cstr); r = 0; } else { val->val.datatype = 'N'; val->val.d.n = resultBuf->gr_gid; dbgprintf("param '%s': uid %d obtained for group '%s'\n", param->name, (int) resultBuf->gr_gid, cstr); r = 1; } free(cstr); return r; } static inline int doGetUID(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { char *cstr; int r; struct passwd *resultBuf; struct passwd wrkBuf; char stringBuf[2048]; /* 2048 has been proven to be large enough */ cstr = es_str2cstr(valnode->val.d.estr, NULL); getpwnam_r(cstr, &wrkBuf, stringBuf, sizeof(stringBuf), &resultBuf); if(resultBuf == NULL) { parser_errmsg("parameter '%s': ID for user %s could not " "be found", param->name, cstr); r = 0; } else { val->val.datatype = 'N'; val->val.d.n = resultBuf->pw_uid; dbgprintf("param '%s': uid %d obtained for user '%s'\n", param->name, (int) resultBuf->pw_uid, cstr); r = 1; } free(cstr); return r; } /* note: we support all integer formats that es_str2num support, * so hex and octal representations are also valid. */ static inline int doGetInt(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { long long n; int bSuccess; n = es_str2num(valnode->val.d.estr, &bSuccess); if(!bSuccess) { parser_errmsg("parameter '%s' is not a proper number", param->name); } val->val.datatype = 'N'; val->val.d.n = n; return bSuccess; } static inline int doGetNonNegInt(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { int bSuccess; if((bSuccess = doGetInt(valnode, param, val))) { if(val->val.d.n < 0) { parser_errmsg("parameter '%s' cannot be less than zero (was %lld)", param->name, val->val.d.n); bSuccess = 0; } } return bSuccess; } static inline int doGetPositiveInt(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { int bSuccess; if((bSuccess = doGetInt(valnode, param, val))) { if(val->val.d.n < 1) { parser_errmsg("parameter '%s' cannot be less than one (was %lld)", param->name, val->val.d.n); bSuccess = 0; } } return bSuccess; } static inline int doGetWord(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { es_size_t i; int r = 1; unsigned char *c; val->val.datatype = 'S'; val->val.d.estr = es_newStr(32); c = es_getBufAddr(valnode->val.d.estr); for(i = 0 ; i < es_strlen(valnode->val.d.estr) && !isspace(c[i]) ; ++i) { es_addChar(&val->val.d.estr, c[i]); } if(i != es_strlen(valnode->val.d.estr)) { parser_errmsg("parameter '%s' contains whitespace, which is not " "permitted", param->name); r = 0; } return r; } static inline int doGetArray(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { int r = 1; switch(valnode->val.datatype) { case 'S': /* a constant string is assumed to be a single-element array */ val->val.datatype = 'A'; val->val.d.ar = cnfarrayNew(es_strdup(valnode->val.d.estr)); break; case 'A': val->val.datatype = 'A'; val->val.d.ar = cnfarrayDup(valnode->val.d.ar); break; default:parser_errmsg("parameter '%s' must be an array, but is a " "different datatype", param->name); r = 0; break; } return r; } static inline int doGetChar(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { int r = 1; if(es_strlen(valnode->val.d.estr) != 1) { parser_errmsg("parameter '%s' must contain exactly one character " "but contains %d - cannot be processed", param->name, es_strlen(valnode->val.d.estr)); r = 0; } val->val.datatype = 'S'; val->val.d.estr = es_strdup(valnode->val.d.estr); return r; } /* get a single parameter according to its definition. Helper to * nvlstGetParams. returns 1 if success, 0 otherwise */ static inline int nvlstGetParam(struct nvlst *valnode, struct cnfparamdescr *param, struct cnfparamvals *val) { uchar *cstr; int r; DBGPRINTF("nvlstGetParam: name '%s', type %d, valnode->bUsed %d\n", param->name, (int) param->type, valnode->bUsed); if(valnode->val.datatype != 'S' && param->type != eCmdHdlrArray) { parser_errmsg("parameter '%s' is not a string, which is not " "permitted", param->name); r = 0; goto done; } valnode->bUsed = 1; val->bUsed = 1; switch(param->type) { case eCmdHdlrQueueType: r = doGetQueueType(valnode, param, val); break; case eCmdHdlrUID: r = doGetUID(valnode, param, val); break; case eCmdHdlrGID: r = doGetGID(valnode, param, val); break; case eCmdHdlrBinary: r = doGetBinary(valnode, param, val); break; case eCmdHdlrFileCreateMode: r = doGetFileCreateMode(valnode, param, val); break; case eCmdHdlrInt: r = doGetInt(valnode, param, val); break; case eCmdHdlrNonNegInt: r = doGetNonNegInt(valnode, param, val); break; case eCmdHdlrPositiveInt: r = doGetPositiveInt(valnode, param, val); break; case eCmdHdlrSize: r = doGetSize(valnode, param, val); break; case eCmdHdlrGetChar: r = doGetChar(valnode, param, val); break; case eCmdHdlrFacility: cstr = (uchar*) es_str2cstr(valnode->val.d.estr, NULL); val->val.datatype = 'N'; val->val.d.n = decodeSyslogName(cstr, syslogFacNames); free(cstr); r = 1; break; case eCmdHdlrSeverity: cstr = (uchar*) es_str2cstr(valnode->val.d.estr, NULL); val->val.datatype = 'N'; val->val.d.n = decodeSyslogName(cstr, syslogPriNames); free(cstr); r = 1; break; case eCmdHdlrGetWord: r = doGetWord(valnode, param, val); break; case eCmdHdlrString: val->val.datatype = 'S'; val->val.d.estr = es_strdup(valnode->val.d.estr); r = 1; break; case eCmdHdlrArray: r = doGetArray(valnode, param, val); break; case eCmdHdlrGoneAway: parser_errmsg("parameter '%s' is no longer supported", param->name); r = 1; /* this *is* valid! */ break; default: dbgprintf("error: invalid param type\n"); r = 0; break; } done: return r; } /* obtain conf params from an nvlst and emit error messages if * necessary. If an already-existing param value is passed, that is * used. If NULL is passed instead, a new one is allocated. In that case, * it is the caller's duty to free it when no longer needed. * NULL is returned on error, otherwise a pointer to the vals array. */ struct cnfparamvals* nvlstGetParams(struct nvlst *lst, struct cnfparamblk *params, struct cnfparamvals *vals) { int i; int bValsWasNULL; int bInError = 0; struct nvlst *valnode; struct cnfparamdescr *param; if(params->version != CNFPARAMBLK_VERSION) { dbgprintf("nvlstGetParams: invalid param block version " "%d, expected %d\n", params->version, CNFPARAMBLK_VERSION); return NULL; } if(vals == NULL) { bValsWasNULL = 1; if((vals = calloc(params->nParams, sizeof(struct cnfparamvals))) == NULL) return NULL; } else { bValsWasNULL = 0; } for(i = 0 ; i < params->nParams ; ++i) { param = params->descr + i; if((valnode = nvlstFindNameCStr(lst, param->name)) == NULL) { if(param->flags & CNFPARAM_REQUIRED) { parser_errmsg("parameter '%s' required but not specified - " "fix config", param->name); bInError = 1; } continue; } if(vals[i].bUsed) { parser_errmsg("parameter '%s' specified more than once - " "one instance is ignored. Fix config", param->name); continue; } if(!nvlstGetParam(valnode, param, vals + i)) { bInError = 1; } } if(bInError) { if(bValsWasNULL) cnfparamvalsDestruct(vals, params); vals = NULL; } return vals; } /* check if at least one cnfparamval is actually set * returns 1 if so, 0 otherwise */ int cnfparamvalsIsSet(struct cnfparamblk *params, struct cnfparamvals *vals) { int i; if(vals == NULL) return 0; if(params->version != CNFPARAMBLK_VERSION) { dbgprintf("nvlstGetParams: invalid param block version " "%d, expected %d\n", params->version, CNFPARAMBLK_VERSION); return 0; } for(i = 0 ; i < params->nParams ; ++i) { if(vals[i].bUsed) return 1; } return 0; } void cnfparamsPrint(struct cnfparamblk *params, struct cnfparamvals *vals) { int i; char *cstr; for(i = 0 ; i < params->nParams ; ++i) { dbgprintf("%s: ", params->descr[i].name); if(vals[i].bUsed) { // TODO: other types! switch(vals[i].val.datatype) { case 'S': cstr = es_str2cstr(vals[i].val.d.estr, NULL); dbgprintf(" '%s'", cstr); free(cstr); break; case 'A': cnfarrayPrint(vals[i].val.d.ar, 0); break; case 'N': dbgprintf("%lld", vals[i].val.d.n); break; default: dbgprintf("(unsupported datatype %c)", vals[i].val.datatype); } } else { dbgprintf("(unset)"); } dbgprintf("\n"); } } struct cnfobj* cnfobjNew(enum cnfobjType objType, struct nvlst *lst) { struct cnfobj *o; if((o = malloc(sizeof(struct cnfobj))) != NULL) { nvlstChkDupes(lst); o->objType = objType; o->nvlst = lst; o->subobjs = NULL; o->script = NULL; } return o; } void cnfobjDestruct(struct cnfobj *o) { if(o != NULL) { nvlstDestruct(o->nvlst); objlstDestruct(o->subobjs); free(o); } } void cnfobjPrint(struct cnfobj *o) { dbgprintf("obj: '%s'\n", cnfobjType2str(o->objType)); nvlstPrint(o->nvlst); } struct cnfexpr* cnfexprNew(unsigned nodetype, struct cnfexpr *l, struct cnfexpr *r) { struct cnfexpr *expr; /* optimize some constructs during parsing */ if(nodetype == 'M' && r->nodetype == 'N') { ((struct cnfnumval*)r)->val *= -1; expr = r; goto done; } if((expr = malloc(sizeof(struct cnfexpr))) != NULL) { expr->nodetype = nodetype; expr->l = l; expr->r = r; } done: return expr; } /* ensure that retval is a number; if string is no number, * try to convert it to one. The semantics from es_str2num() * are used (bSuccess tells if the conversion went well or not). */ static long long var2Number(struct var *r, int *bSuccess) { long long n; if(r->datatype == 'S') { n = es_str2num(r->d.estr, bSuccess); } else { if(r->datatype == 'J') { #ifdef HAVE_JSON_OBJECT_NEW_INT64 n = (r->d.json == NULL) ? 0 : json_object_get_int64(r->d.json); #else /* HAVE_JSON_OBJECT_NEW_INT64 */ n = (r->d.json == NULL) ? 0 : json_object_get_int(r->d.json); #endif /* HAVE_JSON_OBJECT_NEW_INT64 */ } else { n = r->d.n; } if(bSuccess != NULL) *bSuccess = 1; } return n; } /* ensure that retval is a string */ static es_str_t * var2String(struct var *__restrict__ const r, int *__restrict__ const bMustFree) { es_str_t *estr; char *cstr; rs_size_t lenstr; if(r->datatype == 'N') { *bMustFree = 1; estr = es_newStrFromNumber(r->d.n); } else if(r->datatype == 'J') { *bMustFree = 1; if(r->d.json == NULL) { cstr = "", lenstr = 0; } else { cstr = (char*)json_object_get_string(r->d.json); lenstr = strlen(cstr); } estr = es_newStrFromCStr(cstr, lenstr); } else { *bMustFree = 0; estr = r->d.estr; } return estr; } static uchar* var2CString(struct var *__restrict__ const r, int *__restrict__ const bMustFree) { uchar *cstr; es_str_t *estr; estr = var2String(r, bMustFree); cstr = (uchar*) es_str2cstr(estr, NULL); if(*bMustFree) es_deleteStr(estr); *bMustFree = 1; return cstr; } /* frees struct var members, but not the struct itself. This is because * it usually is allocated on the stack. Callers why dynamically allocate * struct var need to free the struct themselfes! */ static void varFreeMembers(struct var *r) { /* Note: we do NOT need to free JSON objects, as we use * json_object_object_get() to obtain the values, which does not * increment the reference count. So json_object_put() [free] is * neither required nor permitted (would free the original object!). * So for the time being the string data type is the only one that * we currently need to free. */ if(r->datatype == 'S') es_deleteStr(r->d.estr); } static rsRetVal doExtractFieldByChar(uchar *str, uchar delim, const int matchnbr, uchar **resstr) { int iCurrFld; int iLen; uchar *pBuf; uchar *pFld; uchar *pFldEnd; DEFiRet; /* first, skip to the field in question */ iCurrFld = 1; pFld = str; while(*pFld && iCurrFld < matchnbr) { /* skip fields until the requested field or end of string is found */ while(*pFld && (uchar) *pFld != delim) ++pFld; /* skip to field terminator */ if(*pFld == delim) { ++pFld; /* eat it */ ++iCurrFld; } } dbgprintf("field() field requested %d, field found %d\n", matchnbr, iCurrFld); if(iCurrFld == matchnbr) { /* field found, now extract it */ /* first of all, we need to find the end */ pFldEnd = pFld; while(*pFldEnd && *pFldEnd != delim) ++pFldEnd; --pFldEnd; /* we are already at the delimiter - so we need to * step back a little not to copy it as part of the field. */ /* we got our end pointer, now do the copy */ iLen = pFldEnd - pFld + 1; /* the +1 is for an actual char, NOT \0! */ CHKmalloc(pBuf = MALLOC((iLen + 1) * sizeof(uchar))); /* now copy */ memcpy(pBuf, pFld, iLen); pBuf[iLen] = '\0'; /* terminate it */ *resstr = pBuf; } else { ABORT_FINALIZE(RS_RET_FIELD_NOT_FOUND); } finalize_it: RETiRet; } static rsRetVal doExtractFieldByStr(uchar *str, char *delim, const rs_size_t lenDelim, const int matchnbr, uchar **resstr) { int iCurrFld; int iLen; uchar *pBuf; uchar *pFld; uchar *pFldEnd; DEFiRet; if (str == NULL || delim == NULL) ABORT_FINALIZE(RS_RET_FIELD_NOT_FOUND); /* first, skip to the field in question */ iCurrFld = 1; pFld = str; while(pFld != NULL && iCurrFld < matchnbr) { if((pFld = (uchar*) strstr((char*)pFld, delim)) != NULL) { pFld += lenDelim; ++iCurrFld; } } dbgprintf("field() field requested %d, field found %d\n", matchnbr, iCurrFld); if(iCurrFld == matchnbr) { /* field found, now extract it */ /* first of all, we need to find the end */ pFldEnd = (uchar*) strstr((char*)pFld, delim); if(pFldEnd == NULL) { iLen = strlen((char*) pFld); } else { /* found delmiter! Note that pFldEnd *is* already on * the first delmi char, we don't need that. */ iLen = pFldEnd - pFld; } /* we got our end pointer, now do the copy */ CHKmalloc(pBuf = MALLOC((iLen + 1) * sizeof(uchar))); /* now copy */ memcpy(pBuf, pFld, iLen); pBuf[iLen] = '\0'; /* terminate it */ *resstr = pBuf; } else { ABORT_FINALIZE(RS_RET_FIELD_NOT_FOUND); } finalize_it: RETiRet; } static inline void doFunc_re_extract(struct cnffunc *func, struct var *ret, void* usrptr) { size_t submatchnbr; short matchnbr; regmatch_t pmatch[50]; int bMustFree; es_str_t *estr = NULL; /* init just to keep compiler happy */ char *str; struct var r[CNFFUNC_MAX_ARGS]; int iLenBuf; unsigned iOffs; short iTry = 0; uchar bFound = 0; iOffs = 0; sbool bHadNoMatch = 0; cnfexprEval(func->expr[0], &r[0], usrptr); /* search string is already part of the compiled regex, so we don't * need it here! */ cnfexprEval(func->expr[2], &r[2], usrptr); cnfexprEval(func->expr[3], &r[3], usrptr); str = (char*) var2CString(&r[0], &bMustFree); matchnbr = (short) var2Number(&r[2], NULL); submatchnbr = (size_t) var2Number(&r[3], NULL); if(submatchnbr >= sizeof(pmatch)/sizeof(regmatch_t)) { DBGPRINTF("re_extract() submatch %zd is too large\n", submatchnbr); bHadNoMatch = 1; goto finalize_it; } /* first see if we find a match, iterating through the series of * potential matches over the string. */ while(!bFound) { int iREstat; iREstat = regexp.regexec(func->funcdata, (char*)(str + iOffs), submatchnbr+1, pmatch, 0); dbgprintf("re_extract: regexec return is %d\n", iREstat); if(iREstat == 0) { if(pmatch[0].rm_so == -1) { dbgprintf("oops ... start offset of successful regexec is -1\n"); break; } if(iTry == matchnbr) { bFound = 1; } else { dbgprintf("re_extract: regex found at offset %d, new offset %d, tries %d\n", iOffs, (int) (iOffs + pmatch[0].rm_eo), iTry); iOffs += pmatch[0].rm_eo; ++iTry; } } else { break; } } dbgprintf("re_extract: regex: end search, found %d\n", bFound); if(!bFound) { bHadNoMatch = 1; goto finalize_it; } else { /* Match- but did it match the one we wanted? */ /* we got no match! */ if(pmatch[submatchnbr].rm_so == -1) { bHadNoMatch = 1; goto finalize_it; } /* OK, we have a usable match - we now need to malloc pB */ iLenBuf = pmatch[submatchnbr].rm_eo - pmatch[submatchnbr].rm_so; estr = es_newStrFromBuf(str + iOffs + pmatch[submatchnbr].rm_so, iLenBuf); } finalize_it: if(bMustFree) free(str); varFreeMembers(&r[0]); varFreeMembers(&r[2]); varFreeMembers(&r[3]); if(bHadNoMatch) { cnfexprEval(func->expr[4], &r[4], usrptr); estr = var2String(&r[4], &bMustFree); /* Note that we do NOT free the string that was returned/created * for r[4]. We pass it to the caller, which in turn frees it. * This saves us doing one unnecessary memory alloc & write. */ } ret->datatype = 'S'; ret->d.estr = estr; return; } /* note that we do not need to evaluate any parameters, as the template pointer * is set during initialization(). * TODO: think if we can keep our buffer; but that may not be trival thinking about * multiple threads. */ static void doFunc_exec_template(struct cnffunc *__restrict__ const func, struct var *__restrict__ const ret, msg_t *const pMsg) { rsRetVal localRet; actWrkrIParams_t iparam; wtiInitIParam(&iparam); localRet = tplToString(func->funcdata, pMsg, &iparam, NULL); if(localRet == RS_RET_OK) { ret->d.estr = es_newStrFromCStr((char*)iparam.param, iparam.lenStr); } else { ret->d.estr = es_newStrFromCStr("", 0); } ret->datatype = 'S'; free(iparam.param); return; } /* Perform a function call. This has been moved out of cnfExprEval in order * to keep the code small and easier to maintain. */ static inline void doFuncCall(struct cnffunc *__restrict__ const func, struct var *__restrict__ const ret, void *__restrict__ const usrptr) { char *fname; char *envvar; int bMustFree; es_str_t *estr; char *str; uchar *resStr; int retval; struct var r[CNFFUNC_MAX_ARGS]; int delim; int matchnbr; struct funcData_prifilt *pPrifilt; rsRetVal localRet; dbgprintf("rainerscript: executing function id %d\n", func->fID); switch(func->fID) { case CNFFUNC_STRLEN: if(func->expr[0]->nodetype == 'S') { /* if we already have a string, we do not need to * do one more recursive call. */ ret->d.n = es_strlen(((struct cnfstringval*) func->expr[0])->estr); } else { cnfexprEval(func->expr[0], &r[0], usrptr); estr = var2String(&r[0], &bMustFree); ret->d.n = es_strlen(estr); if(bMustFree) es_deleteStr(estr); if(r[0].datatype == 'S') es_deleteStr(r[0].d.estr); } ret->datatype = 'N'; break; case CNFFUNC_GETENV: /* note: the optimizer shall have replaced calls to getenv() * with a constant argument to a single string (once obtained via * getenv()). So we do NOT need to check if there is just a * string following. */ cnfexprEval(func->expr[0], &r[0], usrptr); estr = var2String(&r[0], &bMustFree); str = (char*) es_str2cstr(estr, NULL); envvar = getenv(str); if(envvar == NULL) { ret->d.estr = es_newStr(0); } else { ret->d.estr = es_newStrFromCStr(envvar, strlen(envvar)); } ret->datatype = 'S'; if(bMustFree) es_deleteStr(estr); varFreeMembers(&r[0]); free(str); break; case CNFFUNC_TOLOWER: cnfexprEval(func->expr[0], &r[0], usrptr); estr = var2String(&r[0], &bMustFree); if(!bMustFree) /* let caller handle that M) */ estr = es_strdup(estr); es_tolower(estr); ret->datatype = 'S'; ret->d.estr = estr; varFreeMembers(&r[0]); break; case CNFFUNC_CSTR: cnfexprEval(func->expr[0], &r[0], usrptr); estr = var2String(&r[0], &bMustFree); if(!bMustFree) /* let caller handle that M) */ estr = es_strdup(estr); ret->datatype = 'S'; ret->d.estr = estr; varFreeMembers(&r[0]); break; case CNFFUNC_CNUM: if(func->expr[0]->nodetype == 'N') { ret->d.n = ((struct cnfnumval*)func->expr[0])->val; } else if(func->expr[0]->nodetype == 'S') { ret->d.n = es_str2num(((struct cnfstringval*) func->expr[0])->estr, NULL); } else { cnfexprEval(func->expr[0], &r[0], usrptr); ret->d.n = var2Number(&r[0], NULL); varFreeMembers(&r[0]); } ret->datatype = 'N'; break; case CNFFUNC_RE_MATCH: cnfexprEval(func->expr[0], &r[0], usrptr); str = (char*) var2CString(&r[0], &bMustFree); retval = regexp.regexec(func->funcdata, str, 0, NULL, 0); if(retval == 0) ret->d.n = 1; else { ret->d.n = 0; if(retval != REG_NOMATCH) { DBGPRINTF("re_match: regexec returned error %d\n", retval); } } ret->datatype = 'N'; if(bMustFree) free(str); varFreeMembers(&r[0]); break; case CNFFUNC_RE_EXTRACT: doFunc_re_extract(func, ret, usrptr); break; case CNFFUNC_EXEC_TEMPLATE: doFunc_exec_template(func, ret, (msg_t*) usrptr); break; case CNFFUNC_FIELD: cnfexprEval(func->expr[0], &r[0], usrptr); cnfexprEval(func->expr[1], &r[1], usrptr); cnfexprEval(func->expr[2], &r[2], usrptr); str = (char*) var2CString(&r[0], &bMustFree); matchnbr = var2Number(&r[2], NULL); if(r[1].datatype == 'S') { char *delimstr; delimstr = (char*) es_str2cstr(r[1].d.estr, NULL); localRet = doExtractFieldByStr((uchar*)str, delimstr, es_strlen(r[1].d.estr), matchnbr, &resStr); free(delimstr); } else { delim = var2Number(&r[1], NULL); localRet = doExtractFieldByChar((uchar*)str, (char) delim, matchnbr, &resStr); } if(localRet == RS_RET_OK) { ret->d.estr = es_newStrFromCStr((char*)resStr, strlen((char*)resStr)); free(resStr); } else if(localRet == RS_RET_FIELD_NOT_FOUND) { ret->d.estr = es_newStrFromCStr("***FIELD NOT FOUND***", sizeof("***FIELD NOT FOUND***")-1); } else { ret->d.estr = es_newStrFromCStr("***ERROR in field() FUNCTION***", sizeof("***ERROR in field() FUNCTION***")-1); } ret->datatype = 'S'; if(bMustFree) free(str); varFreeMembers(&r[0]); varFreeMembers(&r[1]); varFreeMembers(&r[2]); break; case CNFFUNC_PRIFILT: pPrifilt = (struct funcData_prifilt*) func->funcdata; if( (pPrifilt->pmask[((msg_t*)usrptr)->iFacility] == TABLE_NOPRI) || ((pPrifilt->pmask[((msg_t*)usrptr)->iFacility] & (1<<((msg_t*)usrptr)->iSeverity)) == 0) ) ret->d.n = 0; else ret->d.n = 1; ret->datatype = 'N'; break; case CNFFUNC_LOOKUP: dbgprintf("DDDD: executing lookup\n"); ret->datatype = 'S'; if(func->funcdata == NULL) { ret->d.estr = es_newStrFromCStr("TABLE-NOT-FOUND", sizeof("TABLE-NOT-FOUND")-1); break; } cnfexprEval(func->expr[1], &r[1], usrptr); str = (char*) var2CString(&r[1], &bMustFree); ret->d.estr = lookupKey_estr(func->funcdata, (uchar*)str); if(bMustFree) free(str); if(r[1].datatype == 'S') es_deleteStr(r[1].d.estr); break; default: if(Debug) { fname = es_str2cstr(func->fname, NULL); dbgprintf("rainerscript: invalid function id %u (name '%s')\n", (unsigned) func->fID, fname); free(fname); } ret->datatype = 'N'; ret->d.n = 0; } } static inline void evalVar(struct cnfvar *__restrict__ const var, void *__restrict__ const usrptr, struct var *__restrict__ const ret) { rs_size_t propLen; uchar *pszProp = NULL; unsigned short bMustBeFreed = 0; rsRetVal localRet; struct json_object *json; if(var->prop.id == PROP_CEE || var->prop.id == PROP_LOCAL_VAR || var->prop.id == PROP_GLOBAL_VAR ) { localRet = msgGetJSONPropJSON((msg_t*)usrptr, &var->prop, &json); ret->datatype = 'J'; ret->d.json = (localRet == RS_RET_OK) ? json : NULL; DBGPRINTF("rainerscript: var %d:%s: '%s'\n", var->prop.id, var->prop.name, (ret->d.json == NULL) ? "" : json_object_get_string(ret->d.json)); } else { ret->datatype = 'S'; pszProp = (uchar*) MsgGetProp((msg_t*)usrptr, NULL, &var->prop, &propLen, &bMustBeFreed, NULL); ret->d.estr = es_newStrFromCStr((char*)pszProp, propLen); DBGPRINTF("rainerscript: var %d: '%s'\n", var->prop.id, pszProp); if(bMustBeFreed) free(pszProp); } } /* perform a string comparision operation against a while array. Semantic is * that one one comparison is true, the whole construct is true. * TODO: we can obviously optimize this process. One idea is to * compile a regex, which should work faster than serial comparison. * Note: compiling a regex does NOT work at all. I experimented with that * and it was generally 5 to 10 times SLOWER than what we do here... */ static int evalStrArrayCmp(es_str_t *const estr_l, struct cnfarray *__restrict__ const ar, const int cmpop) { int i; int r = 0; es_str_t **res; if(cmpop == CMP_EQ) { res = bsearch(&estr_l, ar->arr, ar->nmemb, sizeof(es_str_t*), qs_arrcmp); r = res != NULL; } else if(cmpop == CMP_NE) { res = bsearch(&estr_l, ar->arr, ar->nmemb, sizeof(es_str_t*), qs_arrcmp); r = res == NULL; } else { for(i = 0 ; (r == 0) && (i < ar->nmemb) ; ++i) { switch(cmpop) { case CMP_STARTSWITH: r = es_strncmp(estr_l, ar->arr[i], es_strlen(ar->arr[i])) == 0; break; case CMP_STARTSWITHI: r = es_strncasecmp(estr_l, ar->arr[i], es_strlen(ar->arr[i])) == 0; break; case CMP_CONTAINS: r = es_strContains(estr_l, ar->arr[i]) != -1; break; case CMP_CONTAINSI: r = es_strCaseContains(estr_l, ar->arr[i]) != -1; break; } } } return r; } #define FREE_BOTH_RET \ varFreeMembers(&r); \ varFreeMembers(&l) #define COMP_NUM_BINOP(x) \ cnfexprEval(expr->l, &l, usrptr); \ cnfexprEval(expr->r, &r, usrptr); \ ret->datatype = 'N'; \ ret->d.n = var2Number(&l, &convok_l) x var2Number(&r, &convok_r); \ FREE_BOTH_RET #define COMP_NUM_BINOP_DIV(x) \ cnfexprEval(expr->l, &l, usrptr); \ cnfexprEval(expr->r, &r, usrptr); \ ret->datatype = 'N'; \ if((ret->d.n = var2Number(&r, &convok_r)) == 0) { \ /* division by zero */ \ } else { \ ret->d.n = var2Number(&l, &convok_l) x ret->d.n; \ } \ FREE_BOTH_RET /* NOTE: array as right-hand argument MUST be handled by user */ #define PREP_TWO_STRINGS \ cnfexprEval(expr->l, &l, usrptr); \ estr_l = var2String(&l, &bMustFree2); \ if(expr->r->nodetype == 'S') { \ estr_r = ((struct cnfstringval*)expr->r)->estr;\ bMustFree = 0; \ } else if(expr->r->nodetype != 'A') { \ cnfexprEval(expr->r, &r, usrptr); \ estr_r = var2String(&r, &bMustFree); \ } else { \ /* Note: this is not really necessary, but if we do not */ \ /* do it, we get a very irritating compiler warning... */ \ estr_r = NULL; \ } #define FREE_TWO_STRINGS \ if(bMustFree) es_deleteStr(estr_r); \ if(expr->r->nodetype != 'S' && expr->r->nodetype != 'A') varFreeMembers(&r); \ if(bMustFree2) es_deleteStr(estr_l); \ varFreeMembers(&l) /* evaluate an expression. * Note that we try to avoid malloc whenever possible (because of * the large overhead it has, especially on highly threaded programs). * As such, the each caller level must provide buffer space for the * result on its stack during recursion. This permits the callee to store * the return value without malloc. As the value is a somewhat larger * struct, we could otherwise not return it without malloc. * Note that we implement boolean shortcut operations. For our needs, there * simply is no case where full evaluation would make any sense at all. */ void cnfexprEval(const struct cnfexpr *__restrict__ const expr, struct var *__restrict__ const ret, void *__restrict__ const usrptr) { struct var r, l; /* memory for subexpression results */ es_str_t *__restrict__ estr_r, *__restrict__ estr_l; int convok_r, convok_l; int bMustFree, bMustFree2; long long n_r, n_l; DBGPRINTF("eval expr %p, type '%s'\n", expr, tokenToString(expr->nodetype)); switch(expr->nodetype) { /* note: comparison operations are extremely similar. The code can be copyied, only * places flagged with "CMP" need to be changed. */ case CMP_EQ: /* this is optimized in regard to right param as a PoC for all compOps * So this is a NOT yet the copy template! */ cnfexprEval(expr->l, &l, usrptr); ret->datatype = 'N'; if(l.datatype == 'S') { if(expr->r->nodetype == 'S') { ret->d.n = !es_strcmp(l.d.estr, ((struct cnfstringval*)expr->r)->estr); /*CMP*/ } else if(expr->r->nodetype == 'A') { ret->d.n = evalStrArrayCmp(l.d.estr, (struct cnfarray*) expr->r, CMP_EQ); } else { cnfexprEval(expr->r, &r, usrptr); if(r.datatype == 'S') { ret->d.n = !es_strcmp(l.d.estr, r.d.estr); /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l == r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree); ret->d.n = !es_strcmp(l.d.estr, estr_r); /*CMP*/ if(bMustFree) es_deleteStr(estr_r); } } varFreeMembers(&r); } } else if(l.datatype == 'J') { estr_l = var2String(&l, &bMustFree); if(expr->r->nodetype == 'S') { ret->d.n = !es_strcmp(estr_l, ((struct cnfstringval*)expr->r)->estr); /*CMP*/ } else if(expr->r->nodetype == 'A') { ret->d.n = evalStrArrayCmp(estr_l, (struct cnfarray*) expr->r, CMP_EQ); } else { cnfexprEval(expr->r, &r, usrptr); if(r.datatype == 'S') { ret->d.n = !es_strcmp(estr_l, r.d.estr); /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l == r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree2); ret->d.n = !es_strcmp(estr_l, estr_r); /*CMP*/ if(bMustFree2) es_deleteStr(estr_r); } } varFreeMembers(&r); } if(bMustFree) es_deleteStr(estr_l); } else { cnfexprEval(expr->r, &r, usrptr); if(r.datatype == 'S') { n_r = var2Number(&r, &convok_r); if(convok_r) { ret->d.n = (l.d.n == n_r); /*CMP*/ } else { estr_l = var2String(&l, &bMustFree); ret->d.n = !es_strcmp(r.d.estr, estr_l); /*CMP*/ if(bMustFree) es_deleteStr(estr_l); } } else { ret->d.n = (l.d.n == r.d.n); /*CMP*/ } varFreeMembers(&r); } varFreeMembers(&l); break; case CMP_NE: cnfexprEval(expr->l, &l, usrptr); cnfexprEval(expr->r, &r, usrptr); ret->datatype = 'N'; if(l.datatype == 'S') { if(expr->r->nodetype == 'S') { ret->d.n = es_strcmp(l.d.estr, ((struct cnfstringval*)expr->r)->estr); /*CMP*/ } else if(expr->r->nodetype == 'A') { ret->d.n = evalStrArrayCmp(l.d.estr, (struct cnfarray*) expr->r, CMP_NE); } else { if(r.datatype == 'S') { ret->d.n = es_strcmp(l.d.estr, r.d.estr); /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l != r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree); ret->d.n = es_strcmp(l.d.estr, estr_r); /*CMP*/ if(bMustFree) es_deleteStr(estr_r); } } } } else if(l.datatype == 'J') { estr_l = var2String(&l, &bMustFree); if(r.datatype == 'S') { ret->d.n = es_strcmp(estr_l, r.d.estr); /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l != r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree2); ret->d.n = es_strcmp(estr_l, estr_r); /*CMP*/ if(bMustFree2) es_deleteStr(estr_r); } } if(bMustFree) es_deleteStr(estr_l); } else { if(r.datatype == 'S') { n_r = var2Number(&r, &convok_r); if(convok_r) { ret->d.n = (l.d.n != n_r); /*CMP*/ } else { estr_l = var2String(&l, &bMustFree); ret->d.n = es_strcmp(r.d.estr, estr_l); /*CMP*/ if(bMustFree) es_deleteStr(estr_l); } } else { ret->d.n = (l.d.n != r.d.n); /*CMP*/ } } FREE_BOTH_RET; break; case CMP_LE: cnfexprEval(expr->l, &l, usrptr); cnfexprEval(expr->r, &r, usrptr); ret->datatype = 'N'; if(l.datatype == 'S') { if(r.datatype == 'S') { ret->d.n = es_strcmp(l.d.estr, r.d.estr) <= 0; /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l <= r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree); ret->d.n = es_strcmp(l.d.estr, estr_r) <= 0; /*CMP*/ if(bMustFree) es_deleteStr(estr_r); } } } else if(l.datatype == 'J') { estr_l = var2String(&l, &bMustFree); if(r.datatype == 'S') { ret->d.n = es_strcmp(estr_l, r.d.estr) <= 0; /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l <= r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree2); ret->d.n = es_strcmp(estr_l, estr_r) <= 0; /*CMP*/ if(bMustFree2) es_deleteStr(estr_r); } } if(bMustFree) es_deleteStr(estr_l); } else { if(r.datatype == 'S') { n_r = var2Number(&r, &convok_r); if(convok_r) { ret->d.n = (l.d.n <= n_r); /*CMP*/ } else { estr_l = var2String(&l, &bMustFree); ret->d.n = es_strcmp(r.d.estr, estr_l) <= 0; /*CMP*/ if(bMustFree) es_deleteStr(estr_l); } } else { ret->d.n = (l.d.n <= r.d.n); /*CMP*/ } } FREE_BOTH_RET; break; case CMP_GE: cnfexprEval(expr->l, &l, usrptr); cnfexprEval(expr->r, &r, usrptr); ret->datatype = 'N'; if(l.datatype == 'S') { if(r.datatype == 'S') { ret->d.n = es_strcmp(l.d.estr, r.d.estr) >= 0; /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l >= r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree); ret->d.n = es_strcmp(l.d.estr, estr_r) >= 0; /*CMP*/ if(bMustFree) es_deleteStr(estr_r); } } } else if(l.datatype == 'J') { estr_l = var2String(&l, &bMustFree); if(r.datatype == 'S') { ret->d.n = es_strcmp(estr_l, r.d.estr) >= 0; /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l >= r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree2); ret->d.n = es_strcmp(estr_l, estr_r) >= 0; /*CMP*/ if(bMustFree2) es_deleteStr(estr_r); } } if(bMustFree) es_deleteStr(estr_l); } else { if(r.datatype == 'S') { n_r = var2Number(&r, &convok_r); if(convok_r) { ret->d.n = (l.d.n >= n_r); /*CMP*/ } else { estr_l = var2String(&l, &bMustFree); ret->d.n = es_strcmp(r.d.estr, estr_l) >= 0; /*CMP*/ if(bMustFree) es_deleteStr(estr_l); } } else { ret->d.n = (l.d.n >= r.d.n); /*CMP*/ } } FREE_BOTH_RET; break; case CMP_LT: cnfexprEval(expr->l, &l, usrptr); cnfexprEval(expr->r, &r, usrptr); ret->datatype = 'N'; if(l.datatype == 'S') { if(r.datatype == 'S') { ret->d.n = es_strcmp(l.d.estr, r.d.estr) < 0; /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l < r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree); ret->d.n = es_strcmp(l.d.estr, estr_r) < 0; /*CMP*/ if(bMustFree) es_deleteStr(estr_r); } } } else if(l.datatype == 'J') { estr_l = var2String(&l, &bMustFree); if(r.datatype == 'S') { ret->d.n = es_strcmp(estr_l, r.d.estr) < 0; /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l < r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree2); ret->d.n = es_strcmp(estr_l, estr_r) < 0; /*CMP*/ if(bMustFree2) es_deleteStr(estr_r); } } if(bMustFree) es_deleteStr(estr_l); } else { if(r.datatype == 'S') { n_r = var2Number(&r, &convok_r); if(convok_r) { ret->d.n = (l.d.n < n_r); /*CMP*/ } else { estr_l = var2String(&l, &bMustFree); ret->d.n = es_strcmp(r.d.estr, estr_l) < 0; /*CMP*/ if(bMustFree) es_deleteStr(estr_l); } } else { ret->d.n = (l.d.n < r.d.n); /*CMP*/ } } FREE_BOTH_RET; break; case CMP_GT: cnfexprEval(expr->l, &l, usrptr); cnfexprEval(expr->r, &r, usrptr); ret->datatype = 'N'; if(l.datatype == 'S') { if(r.datatype == 'S') { ret->d.n = es_strcmp(l.d.estr, r.d.estr) > 0; /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l > r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree); ret->d.n = es_strcmp(l.d.estr, estr_r) > 0; /*CMP*/ if(bMustFree) es_deleteStr(estr_r); } } } else if(l.datatype == 'J') { estr_l = var2String(&l, &bMustFree); if(r.datatype == 'S') { ret->d.n = es_strcmp(estr_l, r.d.estr) > 0; /*CMP*/ } else { n_l = var2Number(&l, &convok_l); if(convok_l) { ret->d.n = (n_l > r.d.n); /*CMP*/ } else { estr_r = var2String(&r, &bMustFree2); ret->d.n = es_strcmp(estr_l, estr_r) > 0; /*CMP*/ if(bMustFree2) es_deleteStr(estr_r); } } if(bMustFree) es_deleteStr(estr_l); } else { if(r.datatype == 'S') { n_r = var2Number(&r, &convok_r); if(convok_r) { ret->d.n = (l.d.n > n_r); /*CMP*/ } else { estr_l = var2String(&l, &bMustFree); ret->d.n = es_strcmp(r.d.estr, estr_l) > 0; /*CMP*/ if(bMustFree) es_deleteStr(estr_l); } } else { ret->d.n = (l.d.n > r.d.n); /*CMP*/ } } FREE_BOTH_RET; break; case CMP_STARTSWITH: PREP_TWO_STRINGS; ret->datatype = 'N'; if(expr->r->nodetype == 'A') { ret->d.n = evalStrArrayCmp(estr_l, (struct cnfarray*) expr->r, CMP_STARTSWITH); bMustFree = 0; } else { ret->d.n = es_strncmp(estr_l, estr_r, estr_r->lenStr) == 0; } FREE_TWO_STRINGS; break; case CMP_STARTSWITHI: PREP_TWO_STRINGS; ret->datatype = 'N'; if(expr->r->nodetype == 'A') { ret->d.n = evalStrArrayCmp(estr_l, (struct cnfarray*) expr->r, CMP_STARTSWITHI); bMustFree = 0; } else { ret->d.n = es_strncasecmp(estr_l, estr_r, estr_r->lenStr) == 0; } FREE_TWO_STRINGS; break; case CMP_CONTAINS: PREP_TWO_STRINGS; ret->datatype = 'N'; if(expr->r->nodetype == 'A') { ret->d.n = evalStrArrayCmp(estr_l, (struct cnfarray*) expr->r, CMP_CONTAINS); bMustFree = 0; } else { ret->d.n = es_strContains(estr_l, estr_r) != -1; } FREE_TWO_STRINGS; break; case CMP_CONTAINSI: PREP_TWO_STRINGS; ret->datatype = 'N'; if(expr->r->nodetype == 'A') { ret->d.n = evalStrArrayCmp(estr_l, (struct cnfarray*) expr->r, CMP_CONTAINSI); bMustFree = 0; } else { ret->d.n = es_strCaseContains(estr_l, estr_r) != -1; } FREE_TWO_STRINGS; break; case OR: cnfexprEval(expr->l, &l, usrptr); ret->datatype = 'N'; if(var2Number(&l, &convok_l)) { ret->d.n = 1ll; } else { cnfexprEval(expr->r, &r, usrptr); if(var2Number(&r, &convok_r)) ret->d.n = 1ll; else ret->d.n = 0ll; varFreeMembers(&r); } varFreeMembers(&l); break; case AND: cnfexprEval(expr->l, &l, usrptr); ret->datatype = 'N'; if(var2Number(&l, &convok_l)) { cnfexprEval(expr->r, &r, usrptr); if(var2Number(&r, &convok_r)) ret->d.n = 1ll; else ret->d.n = 0ll; varFreeMembers(&r); } else { ret->d.n = 0ll; } varFreeMembers(&l); break; case NOT: cnfexprEval(expr->r, &r, usrptr); ret->datatype = 'N'; ret->d.n = !var2Number(&r, &convok_r); varFreeMembers(&r); break; case 'N': ret->datatype = 'N'; ret->d.n = ((struct cnfnumval*)expr)->val; break; case 'S': ret->datatype = 'S'; ret->d.estr = es_strdup(((struct cnfstringval*)expr)->estr); break; case 'A': /* if an array is used with "normal" operations, it just evaluates * to its first element. */ ret->datatype = 'S'; ret->d.estr = es_strdup(((struct cnfarray*)expr)->arr[0]); break; case 'V': evalVar((struct cnfvar*)expr, usrptr, ret); break; case '&': /* TODO: think about optimization, should be possible ;) */ PREP_TWO_STRINGS; if(expr->r->nodetype == 'A') { estr_r = ((struct cnfarray*)expr->r)->arr[0]; bMustFree = 0; } ret->datatype = 'S'; ret->d.estr = es_strdup(estr_l); es_addStr(&ret->d.estr, estr_r); FREE_TWO_STRINGS; break; case '+': COMP_NUM_BINOP(+); break; case '-': COMP_NUM_BINOP(-); break; case '*': COMP_NUM_BINOP(*); break; case '/': COMP_NUM_BINOP_DIV(/); break; case '%': COMP_NUM_BINOP_DIV(%); break; case 'M': cnfexprEval(expr->r, &r, usrptr); ret->datatype = 'N'; ret->d.n = -var2Number(&r, &convok_r); varFreeMembers(&r); break; case 'F': doFuncCall((struct cnffunc*) expr, ret, usrptr); break; default: ret->datatype = 'N'; ret->d.n = 0ll; dbgprintf("eval error: unknown nodetype %u['%c']\n", (unsigned) expr->nodetype, (char) expr->nodetype); break; } DBGPRINTF("eval expr %p, return datatype '%c'\n", expr, ret->datatype); } //--------------------------------------------------------- void cnfarrayContentDestruct(struct cnfarray *ar) { unsigned short i; for(i = 0 ; i < ar->nmemb ; ++i) { es_deleteStr(ar->arr[i]); } free(ar->arr); } static inline void cnffuncDestruct(struct cnffunc *func) { unsigned short i; for(i = 0 ; i < func->nParams ; ++i) { cnfexprDestruct(func->expr[i]); } /* some functions require special destruction */ switch(func->fID) { case CNFFUNC_RE_MATCH: case CNFFUNC_RE_EXTRACT: if(func->funcdata != NULL) regexp.regfree(func->funcdata); break; default:break; } if(func->fID != CNFFUNC_EXEC_TEMPLATE) free(func->funcdata); free(func->fname); } /* Destruct an expression and all sub-expressions contained in it. */ void cnfexprDestruct(struct cnfexpr *__restrict__ const expr) { if(expr == NULL) { /* this is valid and can happen during optimizer run! */ DBGPRINTF("cnfexprDestruct got NULL ptr - valid, so doing nothing\n"); return; } DBGPRINTF("cnfexprDestruct expr %p, type '%s'\n", expr, tokenToString(expr->nodetype)); switch(expr->nodetype) { case CMP_NE: case CMP_EQ: case CMP_LE: case CMP_GE: case CMP_LT: case CMP_GT: case CMP_STARTSWITH: case CMP_STARTSWITHI: case CMP_CONTAINS: case CMP_CONTAINSI: case OR: case AND: case '&': case '+': case '-': case '*': case '/': case '%': /* binary */ cnfexprDestruct(expr->l); cnfexprDestruct(expr->r); break; case NOT: case 'M': /* unary */ cnfexprDestruct(expr->r); break; case 'N': break; case 'S': es_deleteStr(((struct cnfstringval*)expr)->estr); break; case 'V': free(((struct cnfvar*)expr)->name); msgPropDescrDestruct(&(((struct cnfvar*)expr)->prop)); break; case 'F': cnffuncDestruct((struct cnffunc*)expr); break; case 'A': cnfarrayContentDestruct((struct cnfarray*)expr); break; default:break; } free(expr); } //---- END /* Evaluate an expression as a bool. This is added because expressions are * mostly used inside filters, and so this function is quite common and * important. */ int cnfexprEvalBool(struct cnfexpr *__restrict__ const expr, void *__restrict__ const usrptr) { int convok; struct var ret; cnfexprEval(expr, &ret, usrptr); return var2Number(&ret, &convok); } inline static void doIndent(int indent) { int i; for(i = 0 ; i < indent ; ++i) dbgprintf(" "); } static void pmaskPrint(uchar *pmask, int indent) { int i; doIndent(indent); dbgprintf("pmask: "); for (i = 0; i <= LOG_NFACILITIES; i++) if (pmask[i] == TABLE_NOPRI) dbgprintf(" X "); else dbgprintf("%2X ", pmask[i]); dbgprintf("\n"); } static void cnfarrayPrint(struct cnfarray *ar, int indent) { int i; doIndent(indent); dbgprintf("ARRAY:\n"); for(i = 0 ; i < ar->nmemb ; ++i) { doIndent(indent+1); cstrPrint("string '", ar->arr[i]); dbgprintf("'\n"); } } void cnfexprPrint(struct cnfexpr *expr, int indent) { struct cnffunc *func; int i; switch(expr->nodetype) { case CMP_EQ: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf("==\n"); cnfexprPrint(expr->r, indent+1); break; case CMP_NE: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf("!=\n"); cnfexprPrint(expr->r, indent+1); break; case CMP_LE: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf("<=\n"); cnfexprPrint(expr->r, indent+1); break; case CMP_GE: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf(">=\n"); cnfexprPrint(expr->r, indent+1); break; case CMP_LT: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf("<\n"); cnfexprPrint(expr->r, indent+1); break; case CMP_GT: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf(">\n"); cnfexprPrint(expr->r, indent+1); break; case CMP_CONTAINS: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf("CONTAINS\n"); cnfexprPrint(expr->r, indent+1); break; case CMP_CONTAINSI: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf("CONTAINS_I\n"); cnfexprPrint(expr->r, indent+1); break; case CMP_STARTSWITH: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf("STARTSWITH\n"); cnfexprPrint(expr->r, indent+1); break; case CMP_STARTSWITHI: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf("STARTSWITH_I\n"); cnfexprPrint(expr->r, indent+1); break; case OR: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf("OR\n"); cnfexprPrint(expr->r, indent+1); break; case AND: cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf("AND\n"); cnfexprPrint(expr->r, indent+1); break; case NOT: doIndent(indent); dbgprintf("NOT\n"); cnfexprPrint(expr->r, indent+1); break; case 'S': doIndent(indent); cstrPrint("string '", ((struct cnfstringval*)expr)->estr); dbgprintf("'\n"); break; case 'A': cnfarrayPrint((struct cnfarray*)expr, indent); break; case 'N': doIndent(indent); dbgprintf("%lld\n", ((struct cnfnumval*)expr)->val); break; case 'V': doIndent(indent); dbgprintf("var '%s'\n", ((struct cnfvar*)expr)->name); break; case 'F': doIndent(indent); func = (struct cnffunc*) expr; cstrPrint("function '", func->fname); dbgprintf("' (id:%d, params:%hu)\n", func->fID, func->nParams); if(func->fID == CNFFUNC_PRIFILT) { struct funcData_prifilt *pD; pD = (struct funcData_prifilt*) func->funcdata; pmaskPrint(pD->pmask, indent+1); } for(i = 0 ; i < func->nParams ; ++i) { cnfexprPrint(func->expr[i], indent+1); } break; case '&': case '+': case '-': case '*': case '/': case '%': case 'M': if(expr->l != NULL) cnfexprPrint(expr->l, indent+1); doIndent(indent); dbgprintf("%c\n", (char) expr->nodetype); cnfexprPrint(expr->r, indent+1); break; default: dbgprintf("error: unknown nodetype %u['%c']\n", (unsigned) expr->nodetype, (char) expr->nodetype); break; } } /* print only the given stmt * if "subtree" equals 1, the full statement subtree is printed, else * really only the statement. */ void cnfstmtPrintOnly(struct cnfstmt *stmt, int indent, sbool subtree) { char *cstr; switch(stmt->nodetype) { case S_NOP: doIndent(indent); dbgprintf("NOP\n"); break; case S_STOP: doIndent(indent); dbgprintf("STOP\n"); break; case S_CALL: cstr = es_str2cstr(stmt->d.s_call.name, NULL); doIndent(indent); dbgprintf("CALL [%s, queue:%d]\n", cstr, stmt->d.s_call.ruleset == NULL ? 0 : 1); free(cstr); break; case S_ACT: doIndent(indent); dbgprintf("ACTION %d [%s:%s]\n", stmt->d.act->iActionNbr, modGetName(stmt->d.act->pMod), stmt->printable); break; case S_IF: doIndent(indent); dbgprintf("IF\n"); cnfexprPrint(stmt->d.s_if.expr, indent+1); if(subtree) { doIndent(indent); dbgprintf("THEN\n"); cnfstmtPrint(stmt->d.s_if.t_then, indent+1); if(stmt->d.s_if.t_else != NULL) { doIndent(indent); dbgprintf("ELSE\n"); cnfstmtPrint(stmt->d.s_if.t_else, indent+1); } doIndent(indent); dbgprintf("END IF\n"); } break; case S_SET: doIndent(indent); dbgprintf("SET %s =\n", stmt->d.s_set.varname); cnfexprPrint(stmt->d.s_set.expr, indent+1); doIndent(indent); dbgprintf("END SET\n"); break; case S_UNSET: doIndent(indent); dbgprintf("UNSET %s\n", stmt->d.s_unset.varname); break; case S_PRIFILT: doIndent(indent); dbgprintf("PRIFILT '%s'\n", stmt->printable); pmaskPrint(stmt->d.s_prifilt.pmask, indent); if(subtree) { cnfstmtPrint(stmt->d.s_prifilt.t_then, indent+1); if(stmt->d.s_prifilt.t_else != NULL) { doIndent(indent); dbgprintf("ELSE\n"); cnfstmtPrint(stmt->d.s_prifilt.t_else, indent+1); } doIndent(indent); dbgprintf("END PRIFILT\n"); } break; case S_PROPFILT: doIndent(indent); dbgprintf("PROPFILT\n"); doIndent(indent); dbgprintf("\tProperty.: '%s'\n", propIDToName(stmt->d.s_propfilt.prop.id)); if(stmt->d.s_propfilt.prop.id == PROP_CEE || stmt->d.s_propfilt.prop.id == PROP_LOCAL_VAR || stmt->d.s_propfilt.prop.id == PROP_GLOBAL_VAR) { doIndent(indent); dbgprintf("\tCEE-Prop.: '%s'\n", stmt->d.s_propfilt.prop.name); } doIndent(indent); dbgprintf("\tOperation: "); if(stmt->d.s_propfilt.isNegated) dbgprintf("NOT "); dbgprintf("'%s'\n", getFIOPName(stmt->d.s_propfilt.operation)); if(stmt->d.s_propfilt.pCSCompValue != NULL) { doIndent(indent); dbgprintf("\tValue....: '%s'\n", rsCStrGetSzStrNoNULL(stmt->d.s_propfilt.pCSCompValue)); } if(subtree) { doIndent(indent); dbgprintf("THEN\n"); cnfstmtPrint(stmt->d.s_propfilt.t_then, indent+1); doIndent(indent); dbgprintf("END PROPFILT\n"); } break; default: dbgprintf("error: unknown stmt type %u\n", (unsigned) stmt->nodetype); break; } } void cnfstmtPrint(struct cnfstmt *root, int indent) { struct cnfstmt *stmt; //dbgprintf("stmt %p, indent %d, type '%c'\n", expr, indent, expr->nodetype); for(stmt = root ; stmt != NULL ; stmt = stmt->next) { cnfstmtPrintOnly(stmt, indent, 1); } } struct cnfnumval* cnfnumvalNew(const long long val) { struct cnfnumval *numval; if((numval = malloc(sizeof(struct cnfnumval))) != NULL) { numval->nodetype = 'N'; numval->val = val; } return numval; } struct cnfstringval* cnfstringvalNew(es_str_t *const estr) { struct cnfstringval *strval; if((strval = malloc(sizeof(struct cnfstringval))) != NULL) { strval->nodetype = 'S'; strval->estr = estr; } return strval; } /* creates array AND adds first element to it */ struct cnfarray* cnfarrayNew(es_str_t *val) { struct cnfarray *ar; if((ar = malloc(sizeof(struct cnfarray))) != NULL) { ar->nodetype = 'A'; ar->nmemb = 1; if((ar->arr = malloc(sizeof(es_str_t*))) == NULL) { free(ar); ar = NULL; goto done; } ar->arr[0] = val; } done: return ar; } struct cnfarray* cnfarrayAdd(struct cnfarray *__restrict__ const ar, es_str_t *__restrict__ val) { es_str_t **newptr; if((newptr = realloc(ar->arr, (ar->nmemb+1)*sizeof(es_str_t*))) == NULL) { DBGPRINTF("cnfarrayAdd: realloc failed, item ignored, ar->arr=%p\n", ar->arr); goto done; } else { ar->arr = newptr; ar->arr[ar->nmemb] = val; ar->nmemb++; } done: return ar; } /* duplicate an array (deep copy) */ struct cnfarray* cnfarrayDup(struct cnfarray *old) { int i; struct cnfarray *ar; ar = cnfarrayNew(es_strdup(old->arr[0])); for(i = 1 ; i < old->nmemb ; ++i) { cnfarrayAdd(ar, es_strdup(old->arr[i])); } return ar; } struct cnfvar* cnfvarNew(char *name) { struct cnfvar *var; if((var = malloc(sizeof(struct cnfvar))) != NULL) { var->nodetype = 'V'; var->name = name; msgPropDescrFill(&var->prop, (uchar*)var->name, strlen(var->name)); } return var; } struct cnfstmt * cnfstmtNew(unsigned s_type) { struct cnfstmt* cnfstmt; if((cnfstmt = malloc(sizeof(struct cnfstmt))) != NULL) { cnfstmt->nodetype = s_type; cnfstmt->printable = NULL; cnfstmt->next = NULL; } return cnfstmt; } void cnfstmtDestructLst(struct cnfstmt *root); /* delete a single stmt */ static void cnfstmtDestruct(struct cnfstmt *stmt) { switch(stmt->nodetype) { case S_NOP: case S_STOP: break; case S_CALL: es_deleteStr(stmt->d.s_call.name); break; case S_ACT: actionDestruct(stmt->d.act); break; case S_IF: cnfexprDestruct(stmt->d.s_if.expr); if(stmt->d.s_if.t_then != NULL) { cnfstmtDestructLst(stmt->d.s_if.t_then); } if(stmt->d.s_if.t_else != NULL) { cnfstmtDestructLst(stmt->d.s_if.t_else); } break; case S_SET: free(stmt->d.s_set.varname); cnfexprDestruct(stmt->d.s_set.expr); break; case S_UNSET: free(stmt->d.s_set.varname); break; case S_PRIFILT: cnfstmtDestructLst(stmt->d.s_prifilt.t_then); cnfstmtDestructLst(stmt->d.s_prifilt.t_else); break; case S_PROPFILT: msgPropDescrDestruct(&stmt->d.s_propfilt.prop); if(stmt->d.s_propfilt.regex_cache != NULL) rsCStrRegexDestruct(&stmt->d.s_propfilt.regex_cache); if(stmt->d.s_propfilt.pCSCompValue != NULL) cstrDestruct(&stmt->d.s_propfilt.pCSCompValue); cnfstmtDestructLst(stmt->d.s_propfilt.t_then); break; default: dbgprintf("error: unknown stmt type during destruct %u\n", (unsigned) stmt->nodetype); break; } free(stmt->printable); free(stmt); } /* delete a stmt and all others following it */ void cnfstmtDestructLst(struct cnfstmt *root) { struct cnfstmt *stmt, *todel; for(stmt = root ; stmt != NULL ; ) { todel = stmt; stmt = stmt->next; cnfstmtDestruct(todel); } } struct cnfstmt * cnfstmtNewSet(char *var, struct cnfexpr *expr) { struct cnfstmt* cnfstmt; if((cnfstmt = cnfstmtNew(S_SET)) != NULL) { cnfstmt->d.s_set.varname = (uchar*) var; cnfstmt->d.s_set.expr = expr; } return cnfstmt; } struct cnfstmt * cnfstmtNewCall(es_str_t *name) { struct cnfstmt* cnfstmt; if((cnfstmt = cnfstmtNew(S_CALL)) != NULL) { cnfstmt->d.s_call.name = name; } return cnfstmt; } struct cnfstmt * cnfstmtNewUnset(char *var) { struct cnfstmt* cnfstmt; if((cnfstmt = cnfstmtNew(S_UNSET)) != NULL) { cnfstmt->d.s_unset.varname = (uchar*) var; } return cnfstmt; } struct cnfstmt * cnfstmtNewContinue(void) { return cnfstmtNew(S_NOP); } struct cnfstmt * cnfstmtNewPRIFILT(char *prifilt, struct cnfstmt *t_then) { struct cnfstmt* cnfstmt; if((cnfstmt = cnfstmtNew(S_PRIFILT)) != NULL) { cnfstmt->printable = (uchar*)prifilt; cnfstmt->d.s_prifilt.t_then = t_then; cnfstmt->d.s_prifilt.t_else = NULL; DecodePRIFilter((uchar*)prifilt, cnfstmt->d.s_prifilt.pmask); } return cnfstmt; } struct cnfstmt * cnfstmtNewPROPFILT(char *propfilt, struct cnfstmt *t_then) { struct cnfstmt* cnfstmt; if((cnfstmt = cnfstmtNew(S_PROPFILT)) != NULL) { cnfstmt->printable = (uchar*)propfilt; cnfstmt->d.s_propfilt.t_then = t_then; cnfstmt->d.s_propfilt.regex_cache = NULL; cnfstmt->d.s_propfilt.pCSCompValue = NULL; if(DecodePropFilter((uchar*)propfilt, cnfstmt) != RS_RET_OK) { cnfstmt->nodetype = S_NOP; /* disable action! */ cnfstmtDestructLst(t_then); /* we do no longer need this */ } } return cnfstmt; } struct cnfstmt * cnfstmtNewAct(struct nvlst *lst) { struct cnfstmt* cnfstmt; char namebuf[256]; rsRetVal localRet; if((cnfstmt = cnfstmtNew(S_ACT)) == NULL) goto done; localRet = actionNewInst(lst, &cnfstmt->d.act); if(localRet == RS_RET_OK_WARN) { parser_errmsg("warnings occured in file '%s' around line %d", cnfcurrfn, yylineno); } else if(localRet != RS_RET_OK) { parser_errmsg("errors occured in file '%s' around line %d", cnfcurrfn, yylineno); cnfstmt->nodetype = S_NOP; /* disable action! */ goto done; } snprintf(namebuf, sizeof(namebuf)-1, "action(type=\"%s\" ...)", modGetName(cnfstmt->d.act->pMod)); namebuf[255] = '\0'; /* be on safe side */ cnfstmt->printable = (uchar*)strdup(namebuf); nvlstChkUnused(lst); nvlstDestruct(lst); done: return cnfstmt; } struct cnfstmt * cnfstmtNewLegaAct(char *actline) { struct cnfstmt* cnfstmt; rsRetVal localRet; if((cnfstmt = cnfstmtNew(S_ACT)) == NULL) goto done; cnfstmt->printable = (uchar*)strdup((char*)actline); localRet = cflineDoAction(loadConf, (uchar**)&actline, &cnfstmt->d.act); if(localRet != RS_RET_OK) { parser_errmsg("%s occured in file '%s' around line %d", (localRet == RS_RET_OK_WARN) ? "warnings" : "errors", cnfcurrfn, yylineno); if(localRet != RS_RET_OK_WARN) { cnfstmt->nodetype = S_NOP; /* disable action! */ goto done; } } done: return cnfstmt; } /* returns 1 if the two expressions are constants, 0 otherwise * if both are constants, the expression subtrees are destructed * (this is an aid for constant folding optimizing) */ static int getConstNumber(struct cnfexpr *expr, long long *l, long long *r) { int ret = 0; cnfexprOptimize(expr->l); cnfexprOptimize(expr->r); if(expr->l->nodetype == 'N') { if(expr->r->nodetype == 'N') { ret = 1; *l = ((struct cnfnumval*)expr->l)->val; *r = ((struct cnfnumval*)expr->r)->val; cnfexprDestruct(expr->l); cnfexprDestruct(expr->r); } else if(expr->r->nodetype == 'S') { ret = 1; *l = ((struct cnfnumval*)expr->l)->val; *r = es_str2num(((struct cnfstringval*)expr->r)->estr, NULL); cnfexprDestruct(expr->l); cnfexprDestruct(expr->r); } } else if(expr->l->nodetype == 'S') { if(expr->r->nodetype == 'N') { ret = 1; *l = es_str2num(((struct cnfstringval*)expr->l)->estr, NULL); *r = ((struct cnfnumval*)expr->r)->val; cnfexprDestruct(expr->l); cnfexprDestruct(expr->r); } else if(expr->r->nodetype == 'S') { ret = 1; *l = es_str2num(((struct cnfstringval*)expr->l)->estr, NULL); *r = es_str2num(((struct cnfstringval*)expr->r)->estr, NULL); cnfexprDestruct(expr->l); cnfexprDestruct(expr->r); } } return ret; } /* constant folding for string concatenation */ static inline void constFoldConcat(struct cnfexpr *expr) { es_str_t *estr; cnfexprOptimize(expr->l); cnfexprOptimize(expr->r); if(expr->l->nodetype == 'S') { if(expr->r->nodetype == 'S') { estr = ((struct cnfstringval*)expr->l)->estr; ((struct cnfstringval*)expr->l)->estr = NULL; es_addStr(&estr, ((struct cnfstringval*)expr->r)->estr); cnfexprDestruct(expr->l); cnfexprDestruct(expr->r); expr->nodetype = 'S'; ((struct cnfstringval*)expr)->estr = estr; } else if(expr->r->nodetype == 'N') { es_str_t *numstr; estr = ((struct cnfstringval*)expr->l)->estr; ((struct cnfstringval*)expr->l)->estr = NULL; numstr = es_newStrFromNumber(((struct cnfnumval*)expr->r)->val); es_addStr(&estr, numstr); es_deleteStr(numstr); cnfexprDestruct(expr->l); cnfexprDestruct(expr->r); expr->nodetype = 'S'; ((struct cnfstringval*)expr)->estr = estr; } } else if(expr->l->nodetype == 'N') { if(expr->r->nodetype == 'S') { estr = es_newStrFromNumber(((struct cnfnumval*)expr->l)->val); es_addStr(&estr, ((struct cnfstringval*)expr->r)->estr); cnfexprDestruct(expr->l); cnfexprDestruct(expr->r); expr->nodetype = 'S'; ((struct cnfstringval*)expr)->estr = estr; } else if(expr->r->nodetype == 'S') { es_str_t *numstr; estr = es_newStrFromNumber(((struct cnfnumval*)expr->l)->val); numstr = es_newStrFromNumber(((struct cnfnumval*)expr->r)->val); es_addStr(&estr, numstr); es_deleteStr(numstr); cnfexprDestruct(expr->l); cnfexprDestruct(expr->r); expr->nodetype = 'S'; ((struct cnfstringval*)expr)->estr = estr; } } } /* optimize comparisons with syslog severity/facility. This is a special * handler as the numerical values also support GT, LT, etc ops. */ static inline struct cnfexpr* cnfexprOptimize_CMP_severity_facility(struct cnfexpr *expr) { struct cnffunc *func; if(expr->l->nodetype != 'V') FINALIZE; if(!strcmp("syslogseverity", ((struct cnfvar*)expr->l)->name)) { if(expr->r->nodetype == 'N') { int sev = (int) ((struct cnfnumval*)expr->r)->val; if(sev >= 0 && sev <= 7) { DBGPRINTF("optimizer: change comparison OP to FUNC prifilt()\n"); func = cnffuncNew_prifilt(0); /* fac is irrelevant, set below... */ prifiltSetSeverity(func->funcdata, sev, expr->nodetype); cnfexprDestruct(expr); expr = (struct cnfexpr*) func; } else { parser_errmsg("invalid syslogseverity %d, expression will always " "evaluate to FALSE", sev); } } } else if(!strcmp("syslogfacility", ((struct cnfvar*)expr->l)->name)) { if(expr->r->nodetype == 'N') { int fac = (int) ((struct cnfnumval*)expr->r)->val; if(fac >= 0 && fac <= 24) { DBGPRINTF("optimizer: change comparison OP to FUNC prifilt()\n"); func = cnffuncNew_prifilt(0); /* fac is irrelevant, set below... */ prifiltSetFacility(func->funcdata, fac, expr->nodetype); cnfexprDestruct(expr); expr = (struct cnfexpr*) func; } else { parser_errmsg("invalid syslogfacility %d, expression will always " "evaluate to FALSE", fac); } } } finalize_it: return expr; } /* optimize a comparison with a variable as left-hand operand * NOTE: Currently support CMP_EQ, CMP_NE only and code NEEDS * TO BE CHANGED fgr other comparisons! */ static inline struct cnfexpr* cnfexprOptimize_CMP_var(struct cnfexpr *expr) { struct cnffunc *func; if(!strcmp("syslogfacility-text", ((struct cnfvar*)expr->l)->name)) { if(expr->r->nodetype == 'S') { char *cstr = es_str2cstr(((struct cnfstringval*)expr->r)->estr, NULL); int fac = decodeSyslogName((uchar*)cstr, syslogFacNames); if(fac == -1) { parser_errmsg("invalid facility '%s', expression will always " "evaluate to FALSE", cstr); } else { /* we can actually optimize! */ DBGPRINTF("optimizer: change comparison OP to FUNC prifilt()\n"); func = cnffuncNew_prifilt(fac); if(expr->nodetype == CMP_NE) prifiltInvert(func->funcdata); cnfexprDestruct(expr); expr = (struct cnfexpr*) func; } free(cstr); } } else if(!strcmp("syslogseverity-text", ((struct cnfvar*)expr->l)->name)) { if(expr->r->nodetype == 'S') { char *cstr = es_str2cstr(((struct cnfstringval*)expr->r)->estr, NULL); int sev = decodeSyslogName((uchar*)cstr, syslogPriNames); if(sev == -1) { parser_errmsg("invalid syslogseverity '%s', expression will always " "evaluate to FALSE", cstr); } else { /* we can acutally optimize! */ DBGPRINTF("optimizer: change comparison OP to FUNC prifilt()\n"); func = cnffuncNew_prifilt(0); prifiltSetSeverity(func->funcdata, sev, expr->nodetype); cnfexprDestruct(expr); expr = (struct cnfexpr*) func; } free(cstr); } } else { expr = cnfexprOptimize_CMP_severity_facility(expr); } return expr; } static inline struct cnfexpr* cnfexprOptimize_NOT(struct cnfexpr *expr) { struct cnffunc *func; if(expr->r->nodetype == 'F') { func = (struct cnffunc *)expr->r; if(func->fID == CNFFUNC_PRIFILT) { DBGPRINTF("optimize NOT prifilt() to inverted prifilt()\n"); expr->r = NULL; cnfexprDestruct(expr); prifiltInvert(func->funcdata); expr = (struct cnfexpr*) func; } } return expr; } static inline struct cnfexpr* cnfexprOptimize_AND_OR(struct cnfexpr *expr) { struct cnffunc *funcl, *funcr; if(expr->l->nodetype == 'F') { if(expr->r->nodetype == 'F') { funcl = (struct cnffunc *)expr->l; funcr = (struct cnffunc *)expr->r; if(funcl->fID == CNFFUNC_PRIFILT && funcr->fID == CNFFUNC_PRIFILT) { DBGPRINTF("optimize combine AND/OR prifilt()\n"); expr->l = NULL; prifiltCombine(funcl->funcdata, funcr->funcdata, expr->nodetype); cnfexprDestruct(expr); expr = (struct cnfexpr*) funcl; } } } return expr; } /* optimize array for EQ/NEQ comparisons. We sort the array in * this case so that we can apply binary search later on. */ static inline void cnfexprOptimize_CMPEQ_arr(struct cnfarray *arr) { DBGPRINTF("optimizer: sorting array of %d members for CMP_EQ/NEQ comparison\n", arr->nmemb); qsort(arr->arr, arr->nmemb, sizeof(es_str_t*), qs_arrcmp); } /* (recursively) optimize an expression */ struct cnfexpr* cnfexprOptimize(struct cnfexpr *expr) { long long ln, rn; struct cnfexpr *exprswap; dbgprintf("optimize expr %p, type '%s'\n", expr, tokenToString(expr->nodetype)); switch(expr->nodetype) { case '&': constFoldConcat(expr); break; case '+': if(getConstNumber(expr, &ln, &rn)) { expr->nodetype = 'N'; ((struct cnfnumval*)expr)->val = ln + rn; } break; case '-': if(getConstNumber(expr, &ln, &rn)) { expr->nodetype = 'N'; ((struct cnfnumval*)expr)->val = ln - rn; } break; case '*': if(getConstNumber(expr, &ln, &rn)) { expr->nodetype = 'N'; ((struct cnfnumval*)expr)->val = ln * rn; } break; case '/': if(getConstNumber(expr, &ln, &rn)) { expr->nodetype = 'N'; if(rn == 0) { /* division by zero */ ((struct cnfnumval*)expr)->val = 0; } else { ((struct cnfnumval*)expr)->val = ln / rn; } } break; case '%': if(getConstNumber(expr, &ln, &rn)) { expr->nodetype = 'N'; if(rn == 0) { /* division by zero */ ((struct cnfnumval*)expr)->val = 0; } else { ((struct cnfnumval*)expr)->val = ln % rn; } } break; case CMP_NE: case CMP_EQ: expr->l = cnfexprOptimize(expr->l); expr->r = cnfexprOptimize(expr->r); if(expr->l->nodetype == 'A') { if(expr->r->nodetype == 'A') { parser_errmsg("warning: '==' or '<>' " "comparison of two constant string " "arrays makes no sense"); } else { /* swap for simpler execution step */ exprswap = expr->l; expr->l = expr->r; expr->r = exprswap; } } if(expr->r->nodetype == 'A') { cnfexprOptimize_CMPEQ_arr((struct cnfarray *)expr->r); } /* This should be evaluated last because it may change expr * to a function. */ if(expr->l->nodetype == 'V') { expr = cnfexprOptimize_CMP_var(expr); } break; case CMP_LE: case CMP_GE: case CMP_LT: case CMP_GT: expr->l = cnfexprOptimize(expr->l); expr->r = cnfexprOptimize(expr->r); expr = cnfexprOptimize_CMP_severity_facility(expr); break; case CMP_CONTAINS: case CMP_CONTAINSI: case CMP_STARTSWITH: case CMP_STARTSWITHI: expr->l = cnfexprOptimize(expr->l); expr->r = cnfexprOptimize(expr->r); break; case AND: case OR: expr->l = cnfexprOptimize(expr->l); expr->r = cnfexprOptimize(expr->r); expr = cnfexprOptimize_AND_OR(expr); break; case NOT: expr->r = cnfexprOptimize(expr->r); expr = cnfexprOptimize_NOT(expr); break; default:/* nodetypes we cannot optimize */ break; } return expr; } /* removes NOPs from a statement list and returns the * first non-NOP entry. */ static inline struct cnfstmt * removeNOPs(struct cnfstmt *root) { struct cnfstmt *stmt, *toDel, *prevstmt = NULL; struct cnfstmt *newRoot = NULL; if(root == NULL) goto done; stmt = root; while(stmt != NULL) { if(stmt->nodetype == S_NOP) { if(prevstmt != NULL) /* end chain, is rebuild if more non-NOPs follow */ prevstmt->next = NULL; toDel = stmt; stmt = stmt->next; cnfstmtDestruct(toDel); } else { if(newRoot == NULL) newRoot = stmt; if(prevstmt != NULL) prevstmt->next = stmt; prevstmt = stmt; stmt = stmt->next; } } done: return newRoot; } static inline void cnfstmtOptimizeIf(struct cnfstmt *stmt) { struct cnfstmt *t_then, *t_else; struct cnfexpr *expr; struct cnffunc *func; struct funcData_prifilt *prifilt; expr = stmt->d.s_if.expr = cnfexprOptimize(stmt->d.s_if.expr); stmt->d.s_if.t_then = removeNOPs(stmt->d.s_if.t_then); stmt->d.s_if.t_else = removeNOPs(stmt->d.s_if.t_else); cnfstmtOptimize(stmt->d.s_if.t_then); cnfstmtOptimize(stmt->d.s_if.t_else); if(stmt->d.s_if.expr->nodetype == 'F') { func = (struct cnffunc*)expr; if(func->fID == CNFFUNC_PRIFILT) { DBGPRINTF("optimizer: change IF to PRIFILT\n"); t_then = stmt->d.s_if.t_then; t_else = stmt->d.s_if.t_else; stmt->nodetype = S_PRIFILT; prifilt = (struct funcData_prifilt*) func->funcdata; memcpy(stmt->d.s_prifilt.pmask, prifilt->pmask, sizeof(prifilt->pmask)); stmt->d.s_prifilt.t_then = t_then; stmt->d.s_prifilt.t_else = t_else; if(func->nParams == 0) stmt->printable = (uchar*)strdup("[Optimizer Result]"); else stmt->printable = (uchar*) es_str2cstr(((struct cnfstringval*)func->expr[0])->estr, NULL); cnfexprDestruct(expr); cnfstmtOptimizePRIFilt(stmt); } } } static inline void cnfstmtOptimizeAct(struct cnfstmt *stmt) { action_t *pAct; pAct = stmt->d.act; if(!strcmp((char*)modGetName(pAct->pMod), "builtin:omdiscard")) { DBGPRINTF("optimizer: replacing omdiscard by STOP\n"); actionDestruct(stmt->d.act); stmt->nodetype = S_STOP; } } static void cnfstmtOptimizePRIFilt(struct cnfstmt *stmt) { int i; int isAlways = 1; struct cnfstmt *subroot, *last; stmt->d.s_prifilt.t_then = removeNOPs(stmt->d.s_prifilt.t_then); cnfstmtOptimize(stmt->d.s_prifilt.t_then); for(i = 0; i <= LOG_NFACILITIES; i++) if(stmt->d.s_prifilt.pmask[i] != 0xff) { isAlways = 0; break; } if(!isAlways) goto done; DBGPRINTF("optimizer: removing always-true PRIFILT %p\n", stmt); if(stmt->d.s_prifilt.t_else != NULL) { parser_errmsg("error: always-true PRI filter has else part!\n"); cnfstmtDestructLst(stmt->d.s_prifilt.t_else); } free(stmt->printable); stmt->printable = NULL; subroot = stmt->d.s_prifilt.t_then; if(subroot == NULL) { /* very strange, we set it to NOP, best we can do * This case is NOT expected in practice */ stmt->nodetype = S_NOP; goto done; } for(last = subroot ; last->next != NULL ; last = last->next) /* find last node in subtree */; last->next = stmt->next; memcpy(stmt, subroot, sizeof(struct cnfstmt)); free(subroot); done: return; } /* we abuse "optimize" a bit. Actually, we obtain a ruleset pointer, as * all rulesets are only known later in the process (now!). */ static void cnfstmtOptimizeCall(struct cnfstmt *stmt) { ruleset_t *pRuleset; rsRetVal localRet; uchar *rsName; rsName = (uchar*) es_str2cstr(stmt->d.s_call.name, NULL); localRet = rulesetGetRuleset(loadConf, &pRuleset, rsName); if(localRet != RS_RET_OK) { /* in that case, we accept that a NOP will "survive" */ parser_errmsg("ruleset '%s' cannot be found\n", rsName); es_deleteStr(stmt->d.s_call.name); stmt->nodetype = S_NOP; goto done; } DBGPRINTF("CALL obtained ruleset ptr %p for ruleset %s [hasQueue:%d]\n", pRuleset, rsName, rulesetHasQueue(pRuleset)); if(rulesetHasQueue(pRuleset)) { stmt->d.s_call.ruleset = pRuleset; } else { stmt->d.s_call.ruleset = NULL; stmt->d.s_call.stmt = pRuleset->root; } done: free(rsName); return; } /* (recursively) optimize a statement */ void cnfstmtOptimize(struct cnfstmt *root) { struct cnfstmt *stmt; if(root == NULL) goto done; for(stmt = root ; stmt != NULL ; stmt = stmt->next) { switch(stmt->nodetype) { case S_IF: cnfstmtOptimizeIf(stmt); break; case S_PRIFILT: cnfstmtOptimizePRIFilt(stmt); break; case S_PROPFILT: stmt->d.s_propfilt.t_then = removeNOPs(stmt->d.s_propfilt.t_then); cnfstmtOptimize(stmt->d.s_propfilt.t_then); break; case S_SET: stmt->d.s_set.expr = cnfexprOptimize(stmt->d.s_set.expr); break; case S_ACT: cnfstmtOptimizeAct(stmt); break; case S_CALL: cnfstmtOptimizeCall(stmt); break; case S_STOP: if(stmt->next != NULL) parser_errmsg("STOP is followed by unreachable statements!\n"); break; case S_UNSET: /* nothing to do */ break; case S_NOP: DBGPRINTF("optimizer error: we see a NOP, how come?\n"); break; default: dbgprintf("error: unknown stmt type %u during optimizer run\n", (unsigned) stmt->nodetype); break; } } done: return; } struct cnffparamlst * cnffparamlstNew(struct cnfexpr *expr, struct cnffparamlst *next) { struct cnffparamlst* lst; if((lst = malloc(sizeof(struct cnffparamlst))) != NULL) { lst->nodetype = 'P'; lst->expr = expr; lst->next = next; } return lst; } /* Obtain function id from name AND number of params. Issues the * relevant error messages if errors are detected. */ static inline enum cnffuncid funcName2ID(es_str_t *fname, unsigned short nParams) { if(!es_strbufcmp(fname, (unsigned char*)"strlen", sizeof("strlen") - 1)) { if(nParams != 1) { parser_errmsg("number of parameters for strlen() must be one " "but is %d.", nParams); return CNFFUNC_INVALID; } return CNFFUNC_STRLEN; } else if(!es_strbufcmp(fname, (unsigned char*)"getenv", sizeof("getenv") - 1)) { if(nParams != 1) { parser_errmsg("number of parameters for getenv() must be one " "but is %d.", nParams); return CNFFUNC_INVALID; } return CNFFUNC_GETENV; } else if(!es_strbufcmp(fname, (unsigned char*)"tolower", sizeof("tolower") - 1)) { if(nParams != 1) { parser_errmsg("number of parameters for tolower() must be one " "but is %d.", nParams); return CNFFUNC_INVALID; } return CNFFUNC_TOLOWER; } else if(!es_strbufcmp(fname, (unsigned char*)"cstr", sizeof("cstr") - 1)) { if(nParams != 1) { parser_errmsg("number of parameters for cstr() must be one " "but is %d.", nParams); return CNFFUNC_INVALID; } return CNFFUNC_CSTR; } else if(!es_strbufcmp(fname, (unsigned char*)"cnum", sizeof("cnum") - 1)) { if(nParams != 1) { parser_errmsg("number of parameters for cnum() must be one " "but is %d.", nParams); return CNFFUNC_INVALID; } return CNFFUNC_CNUM; } else if(!es_strbufcmp(fname, (unsigned char*)"re_match", sizeof("re_match") - 1)) { if(nParams != 2) { parser_errmsg("number of parameters for re_match() must be two " "but is %d.", nParams); return CNFFUNC_INVALID; } return CNFFUNC_RE_MATCH; } else if(!es_strbufcmp(fname, (unsigned char*)"re_extract", sizeof("re_extract") - 1)) { if(nParams != 5) { parser_errmsg("number of parameters for re_extract() must be five " "but is %d.", nParams); return CNFFUNC_INVALID; } return CNFFUNC_RE_EXTRACT; } else if(!es_strbufcmp(fname, (unsigned char*)"field", sizeof("field") - 1)) { if(nParams != 3) { parser_errmsg("number of parameters for field() must be three " "but is %d.", nParams); return CNFFUNC_INVALID; } return CNFFUNC_FIELD; } else if(!es_strbufcmp(fname, (unsigned char*)"exec_template", sizeof("exec_template") - 1)) { if(nParams != 1) { parser_errmsg("number of parameters for exec-template() must be one " "but is %d.", nParams); return CNFFUNC_INVALID; } return CNFFUNC_EXEC_TEMPLATE; } else if(!es_strbufcmp(fname, (unsigned char*)"prifilt", sizeof("prifilt") - 1)) { if(nParams != 1) { parser_errmsg("number of parameters for prifilt() must be one " "but is %d.", nParams); return CNFFUNC_INVALID; } return CNFFUNC_PRIFILT; } else if(!es_strbufcmp(fname, (unsigned char*)"lookup", sizeof("lookup") - 1)) { if(nParams != 2) { parser_errmsg("number of parameters for lookup() must be two " "but is %d.", nParams); return CNFFUNC_INVALID; } return CNFFUNC_LOOKUP; } else { return CNFFUNC_INVALID; } } static inline rsRetVal initFunc_re_match(struct cnffunc *func) { rsRetVal localRet; char *regex = NULL; regex_t *re; DEFiRet; func->funcdata = NULL; if(func->expr[1]->nodetype != 'S') { parser_errmsg("param 2 of re_match/extract() must be a constant string"); FINALIZE; } CHKmalloc(re = malloc(sizeof(regex_t))); func->funcdata = re; regex = es_str2cstr(((struct cnfstringval*) func->expr[1])->estr, NULL); if((localRet = objUse(regexp, LM_REGEXP_FILENAME)) == RS_RET_OK) { if(regexp.regcomp(re, (char*) regex, REG_EXTENDED) != 0) { parser_errmsg("cannot compile regex '%s'", regex); ABORT_FINALIZE(RS_RET_ERR); } } else { /* regexp object could not be loaded */ parser_errmsg("could not load regex support - regex ignored"); ABORT_FINALIZE(RS_RET_ERR); } finalize_it: free(regex); RETiRet; } static rsRetVal initFunc_exec_template(struct cnffunc *func) { char *tplName = NULL; DEFiRet; if(func->expr[0]->nodetype != 'S') { parser_errmsg("exec_template(): param 1 must be a constant string"); FINALIZE; } tplName = es_str2cstr(((struct cnfstringval*) func->expr[0])->estr, NULL); func->funcdata = tplFind(ourConf, tplName, strlen(tplName)); if(func->funcdata == NULL) { parser_errmsg("exec_template(): template '%s' could not be found", tplName); FINALIZE; } finalize_it: free(tplName); RETiRet; } static inline rsRetVal initFunc_prifilt(struct cnffunc *func) { struct funcData_prifilt *pData; uchar *cstr; DEFiRet; func->funcdata = NULL; if(func->expr[0]->nodetype != 'S') { parser_errmsg("param 1 of prifilt() must be a constant string"); FINALIZE; } CHKmalloc(pData = calloc(1, sizeof(struct funcData_prifilt))); func->funcdata = pData; cstr = (uchar*)es_str2cstr(((struct cnfstringval*) func->expr[0])->estr, NULL); CHKiRet(DecodePRIFilter(cstr, pData->pmask)); free(cstr); finalize_it: RETiRet; } static inline rsRetVal initFunc_lookup(struct cnffunc *func) { uchar *cstr = NULL; DEFiRet; func->funcdata = NULL; if(func->expr[0]->nodetype != 'S') { parser_errmsg("table name (param 1) of lookup() must be a constant string"); FINALIZE; } cstr = (uchar*)es_str2cstr(((struct cnfstringval*) func->expr[0])->estr, NULL); if((func->funcdata = lookupFindTable(cstr)) == NULL) { parser_errmsg("lookup table '%s' not found", cstr); FINALIZE; } finalize_it: free(cstr); RETiRet; } struct cnffunc * cnffuncNew(es_str_t *fname, struct cnffparamlst* paramlst) { struct cnffunc* func; struct cnffparamlst *param, *toDel; unsigned short i; unsigned short nParams; /* we first need to find out how many params we have */ nParams = 0; for(param = paramlst ; param != NULL ; param = param->next) ++nParams; if((func = malloc(sizeof(struct cnffunc) + (nParams * sizeof(struct cnfexp*)))) != NULL) { func->nodetype = 'F'; func->fname = fname; func->nParams = nParams; func->funcdata = NULL; func->fID = funcName2ID(fname, nParams); /* shuffle params over to array (access speed!) */ param = paramlst; for(i = 0 ; i < nParams ; ++i) { func->expr[i] = param->expr; toDel = param; param = param->next; free(toDel); } /* some functions require special initialization */ switch(func->fID) { case CNFFUNC_RE_MATCH: case CNFFUNC_RE_EXTRACT: /* need to compile the regexp in param 2, so this MUST be a constant */ initFunc_re_match(func); break; case CNFFUNC_PRIFILT: initFunc_prifilt(func); break; case CNFFUNC_LOOKUP: initFunc_lookup(func); break; case CNFFUNC_EXEC_TEMPLATE: initFunc_exec_template(func); break; default:break; } } return func; } /* A special function to create a prifilt() expression during optimization * phase. */ struct cnffunc * cnffuncNew_prifilt(int fac) { struct cnffunc* func; fac >>= 3; if (fac >= LOG_NFACILITIES + 1 || fac < 0) return NULL; if((func = malloc(sizeof(struct cnffunc))) != NULL) { if ((func->funcdata = calloc(1, sizeof(struct funcData_prifilt))) == NULL) { free(func); return NULL; } func->nodetype = 'F'; func->fname = es_newStrFromCStr("prifilt", sizeof("prifilt")-1); func->nParams = 0; func->fID = CNFFUNC_PRIFILT; ((struct funcData_prifilt *)func->funcdata)->pmask[fac] = TABLE_ALLPRI; } return func; } /* returns 0 if everything is OK and config parsing shall continue, * and 1 if things are so wrong that config parsing shall be aborted. */ int cnfDoInclude(char *name) { char *cfgFile; char *finalName; int i; int result; glob_t cfgFiles; struct stat fileInfo; char nameBuf[MAXFNAME+1]; char cwdBuf[MAXFNAME+1]; finalName = name; if(stat(name, &fileInfo) == 0) { /* stat usually fails if we have a wildcard - so this does NOT indicate error! */ if(S_ISDIR(fileInfo.st_mode)) { /* if we have a directory, we need to add "*" to get its files */ snprintf(nameBuf, sizeof(nameBuf), "%s*", name); finalName = nameBuf; } } /* Use GLOB_MARK to append a trailing slash for directories. */ /* Use GLOB_NOMAGIC to detect wildcards that match nothing. */ #ifdef HAVE_GLOB_NOMAGIC /* Silently ignore wildcards that match nothing */ result = glob(finalName, GLOB_MARK | GLOB_NOMAGIC, NULL, &cfgFiles); if(result == GLOB_NOMATCH) { #else result = glob(finalName, GLOB_MARK, NULL, &cfgFiles); if(result == GLOB_NOMATCH && containsGlobWildcard(finalName)) { #endif /* HAVE_GLOB_NOMAGIC */ return 0; } if(result == GLOB_NOSPACE || result == GLOB_ABORTED) { char errStr[1024]; rs_strerror_r(errno, errStr, sizeof(errStr)); if(getcwd(cwdBuf, sizeof(cwdBuf)) == NULL) strcpy(cwdBuf, "??getcwd() failed??"); parser_errmsg("error accessing config file or directory '%s' [cwd:%s]: %s", finalName, cwdBuf, errStr); return 1; } /* note: bison "stacks" the files, so we need to submit them * in reverse order to the *stack* in order to get the proper * parsing order. Also see * http://bugzilla.adiscon.com/show_bug.cgi?id=411 */ for(i = cfgFiles.gl_pathc - 1; i >= 0 ; i--) { cfgFile = cfgFiles.gl_pathv[i]; if(stat(cfgFile, &fileInfo) != 0) { char errStr[1024]; rs_strerror_r(errno, errStr, sizeof(errStr)); if(getcwd(cwdBuf, sizeof(cwdBuf)) == NULL) strcpy(cwdBuf, "??getcwd() failed??"); parser_errmsg("error accessing config file or directory '%s' " "[cwd: %s]: %s", cfgFile, cwdBuf, errStr); return 1; } if(S_ISREG(fileInfo.st_mode)) { /* config file */ dbgprintf("requested to include config file '%s'\n", cfgFile); cnfSetLexFile(cfgFile); } else if(S_ISDIR(fileInfo.st_mode)) { /* config directory */ dbgprintf("requested to include directory '%s'\n", cfgFile); cnfDoInclude(cfgFile); } else { dbgprintf("warning: unable to process IncludeConfig directive '%s'\n", cfgFile); } } globfree(&cfgFiles); return 0; } void varDelete(struct var *v) { switch(v->datatype) { case 'S': es_deleteStr(v->d.estr); break; case 'A': cnfarrayContentDestruct(v->d.ar); free(v->d.ar); break; default:break; } } void cnfparamvalsDestruct(struct cnfparamvals *paramvals, struct cnfparamblk *blk) { int i; if(paramvals == NULL) return; for(i = 0 ; i < blk->nParams ; ++i) { if(paramvals[i].bUsed) { varDelete(¶mvals[i].val); } } free(paramvals); } /* find the index (or -1!) for a config param by name. This is used to * address the parameter array. Of course, we could use with static * indices, but that would create some extra bug potential. So we * resort to names. As we do this only during the initial config parsing * stage the (considerable!) extra overhead is OK. -- rgerhards, 2011-07-19 */ int cnfparamGetIdx(struct cnfparamblk *params, char *name) { int i; for(i = 0 ; i < params->nParams ; ++i) if(!strcmp(params->descr[i].name, name)) break; if(i == params->nParams) i = -1; /* not found */ return i; } void cstrPrint(char *text, es_str_t *estr) { char *str; str = es_str2cstr(estr, NULL); dbgprintf("%s%s", text, str); free(str); } char * rmLeadingSpace(char *s) { char *p; for(p = s ; *p && isspace(*p) ; ++p) ; return(p); } /* init must be called once before any parsing of the script files start */ rsRetVal initRainerscript(void) { DEFiRet; CHKiRet(objGetObjInterface(&obj)); finalize_it: RETiRet; } /* we need a function to check for octal digits */ static inline int isodigit(uchar c) { return(c >= '0' && c <= '7'); } /** * Get numerical value of a hex digit. This is a helper function. * @param[in] c a character containing 0..9, A..Z, a..z anything else * is an (undetected) error. */ static inline int hexDigitVal(char c) { int r; if(c < 'A') r = c - '0'; else if(c < 'a') r = c - 'A' + 10; else r = c - 'a' + 10; return r; } /* Handle the actual unescaping. * a helper to unescapeStr(), to help make the function easier to read. */ static inline void doUnescape(unsigned char *c, int len, int *iSrc, int iDst) { if(c[*iSrc] == '\\') { if(++(*iSrc) == len) { /* error, incomplete escape, treat as single char */ c[iDst] = '\\'; } /* regular case, unescape */ switch(c[*iSrc]) { case 'a': c[iDst] = '\007'; break; case 'b': c[iDst] = '\b'; break; case 'f': c[iDst] = '\014'; break; case 'n': c[iDst] = '\n'; break; case 'r': c[iDst] = '\r'; break; case 't': c[iDst] = '\t'; break; case '\'': c[iDst] = '\''; break; case '"': c[iDst] = '"'; break; case '?': c[iDst] = '?'; break; case '$': c[iDst] = '$'; break; case '\\': c[iDst] = '\\'; break; case 'x': if( (*iSrc)+2 >= len || !isxdigit(c[(*iSrc)+1]) || !isxdigit(c[(*iSrc)+2])) { /* error, incomplete escape, use as is */ c[iDst] = '\\'; --(*iSrc); } c[iDst] = (hexDigitVal(c[(*iSrc)+1]) << 4) + hexDigitVal(c[(*iSrc)+2]); *iSrc += 2; break; case '0': /* octal escape */ case '1': case '2': case '3': case '4': case '5': case '6': case '7': if( (*iSrc)+2 >= len || !isodigit(c[(*iSrc)+1]) || !isodigit(c[(*iSrc)+2])) { /* error, incomplete escape, use as is */ c[iDst] = '\\'; --(*iSrc); } c[iDst] = ((c[(*iSrc) ] - '0') << 6) + ((c[(*iSrc)+1] - '0') << 3) + ( c[(*iSrc)+2] - '0'); *iSrc += 2; break; default: /* error, incomplete escape, indicate by '?' */ c[iDst] = '?'; break; } } else { /* regular character */ c[iDst] = c[*iSrc]; } } void unescapeStr(uchar *s, int len) { int iSrc, iDst; assert(s != NULL); /* scan for first escape sequence (if we are luky, there is none!) */ iSrc = 0; while(iSrc < len && s[iSrc] != '\\') ++iSrc; /* now we have a sequence or end of string. In any case, we process * all remaining characters (maybe 0!) and unescape. */ if(iSrc != len) { iDst = iSrc; while(iSrc < len) { doUnescape(s, len, &iSrc, iDst); ++iSrc; ++iDst; } s[iDst] = '\0'; } } char * tokenval2str(int tok) { if(tok < 256) return ""; switch(tok) { case NAME: return "NAME"; case FUNC: return "FUNC"; case BEGINOBJ: return "BEGINOBJ"; case ENDOBJ: return "ENDOBJ"; case BEGIN_ACTION: return "BEGIN_ACTION"; case BEGIN_PROPERTY: return "BEGIN_PROPERTY"; case BEGIN_CONSTANT: return "BEGIN_CONSTANT"; case BEGIN_TPL: return "BEGIN_TPL"; case BEGIN_RULESET: return "BEGIN_RULESET"; case STOP: return "STOP"; case SET: return "SET"; case UNSET: return "UNSET"; case CONTINUE: return "CONTINUE"; case CALL: return "CALL"; case LEGACY_ACTION: return "LEGACY_ACTION"; case LEGACY_RULESET: return "LEGACY_RULESET"; case PRIFILT: return "PRIFILT"; case PROPFILT: return "PROPFILT"; case BSD_TAG_SELECTOR: return "BSD_TAG_SELECTOR"; case BSD_HOST_SELECTOR: return "BSD_HOST_SELECTOR"; case IF: return "IF"; case THEN: return "THEN"; case ELSE: return "ELSE"; case OR: return "OR"; case AND: return "AND"; case NOT: return "NOT"; case VAR: return "VAR"; case STRING: return "STRING"; case NUMBER: return "NUMBER"; case CMP_EQ: return "CMP_EQ"; case CMP_NE: return "CMP_NE"; case CMP_LE: return "CMP_LE"; case CMP_GE: return "CMP_GE"; case CMP_LT: return "CMP_LT"; case CMP_GT: return "CMP_GT"; case CMP_CONTAINS: return "CMP_CONTAINS"; case CMP_CONTAINSI: return "CMP_CONTAINSI"; case CMP_STARTSWITH: return "CMP_STARTSWITH"; case CMP_STARTSWITHI: return "CMP_STARTSWITHI"; case UMINUS: return "UMINUS"; default: return "UNKNOWN TOKEN"; } }