diff options
Diffstat (limited to 'grammar')
-rw-r--r-- | grammar/grammar.y | 100 | ||||
-rw-r--r-- | grammar/lexer.l | 77 | ||||
-rw-r--r-- | grammar/parserif.h | 2 | ||||
-rw-r--r-- | grammar/rainerscript.c | 2099 | ||||
-rw-r--r-- | grammar/rainerscript.h | 138 |
5 files changed, 2090 insertions, 326 deletions
diff --git a/grammar/grammar.y b/grammar/grammar.y index cdb19c3d..c5bad689 100644 --- a/grammar/grammar.y +++ b/grammar/grammar.y @@ -6,10 +6,9 @@ * of course, encouraged to use new constructs only. But it needs to be noted * that some of the legacy constructs (specifically the in-front-of-action * PRI filter) are very hard to beat in ease of use, at least for simpler - * cases. So while we hope that cfsysline support can be dropped some time in - * the future, we will probably keep these useful constructs. + * cases. * - * Copyright 2011 Rainer Gerhards and Adiscon GmbH. + * Copyright 2011-2012 Rainer Gerhards and Adiscon GmbH. * * This file is part of the rsyslog runtime library. * @@ -37,7 +36,7 @@ #define YYDEBUG 1 extern int yylineno; -/* keep compile rule cleam of errors */ +/* keep compile rule clean of errors */ extern int yylex(void); extern int yyerror(char*); %} @@ -48,26 +47,29 @@ extern int yyerror(char*); es_str_t *estr; enum cnfobjType objType; struct cnfobj *obj; + struct cnfstmt *stmt; struct nvlst *nvlst; struct objlst *objlst; - struct cnfactlst *actlst; struct cnfexpr *expr; - struct cnfrule *rule; + struct cnfarray *arr; struct cnffunc *func; struct cnffparamlst *fparams; } %token <estr> NAME -%token <estr> VALUE %token <estr> FUNC %token <objType> BEGINOBJ %token ENDOBJ -%token <s> CFSYSLINE %token BEGIN_ACTION %token BEGIN_PROPERTY %token BEGIN_CONSTANT %token BEGIN_TPL +%token BEGIN_RULESET %token STOP +%token SET +%token UNSET +%token CONTINUE +%token <cnfstmt> CALL %token <s> LEGACY_ACTION %token <s> LEGACY_RULESET %token <s> PRIFILT @@ -76,6 +78,7 @@ extern int yyerror(char*); %token <s> BSD_HOST_SELECTOR %token IF %token THEN +%token ELSE %token OR %token AND %token NOT @@ -93,33 +96,23 @@ extern int yyerror(char*); %token CMP_STARTSWITH %token CMP_STARTSWITHI -%type <nvlst> nv nvlst +%type <nvlst> nv nvlst value %type <obj> obj property constant %type <objlst> propconst -%type <actlst> actlst -%type <actlst> act -%type <s> cfsysline -%type <actlst> block %type <expr> expr -%type <rule> rule -%type <rule> scriptfilt +%type <stmt> stmt s_act actlst block script %type <fparams> fparams +%type <arr> array arrayelt %left AND OR %left CMP_EQ CMP_NE CMP_LE CMP_GE CMP_LT CMP_GT CMP_CONTAINS CMP_CONTAINSI CMP_STARTSWITH CMP_STARTSWITHI -%left '+' '-' +%left '+' '-' '&' %left '*' '/' '%' %nonassoc UMINUS NOT -%expect 3 -/* these shift/reduce conflicts are created by the CFSYSLINE construct, which we - * unfortunately can not avoid. The problem is that CFSYSLINE can occur both in - * global context as well as within an action. It's not permitted somewhere else, - * but this is suficient for conflicts. The "dangling else" built-in resolution - * works well to solve this issue, so we accept it (it's a wonder that our - * old style grammar doesn't work at all, so we better do not complain...). - * Use "bison -v rscript.y" if more conflicts arise and check rscript.out for - * were exactly these conflicts exits. +%expect 1 /* dangling else */ +/* If more erors show up, Use "bison -v grammar.y" if more conflicts arise and + * check grammar.output for were exactly these conflicts exits. */ %% /* note: we use left recursion below, because that saves stack space AND @@ -128,41 +121,55 @@ extern int yyerror(char*); */ conf: /* empty (to end recursion) */ | conf obj { cnfDoObj($2); } - | conf rule { cnfDoRule($2); } - | conf cfsysline { cnfDoCfsysline($2); } + | conf stmt { cnfDoScript($2); } | conf LEGACY_RULESET { cnfDoCfsysline($2); } | conf BSD_TAG_SELECTOR { cnfDoBSDTag($2); } | conf BSD_HOST_SELECTOR { cnfDoBSDHost($2); } obj: BEGINOBJ nvlst ENDOBJ { $$ = cnfobjNew($1, $2); } - | BEGIN_ACTION nvlst ENDOBJ { $$ = cnfobjNew(CNFOBJ_ACTION, $2); } | BEGIN_TPL nvlst ENDOBJ { $$ = cnfobjNew(CNFOBJ_TPL, $2); } | BEGIN_TPL nvlst ENDOBJ '{' propconst '}' { $$ = cnfobjNew(CNFOBJ_TPL, $2); $$->subobjs = $5; } + | BEGIN_RULESET nvlst ENDOBJ '{' script '}' + { $$ = cnfobjNew(CNFOBJ_RULESET, $2); + $$->script = $5; + } propconst: { $$ = NULL; } | propconst property { $$ = objlstAdd($1, $2); } | propconst constant { $$ = objlstAdd($1, $2); } property: BEGIN_PROPERTY nvlst ENDOBJ { $$ = cnfobjNew(CNFOBJ_PROPERTY, $2); } constant: BEGIN_CONSTANT nvlst ENDOBJ { $$ = cnfobjNew(CNFOBJ_CONSTANT, $2); } -cfsysline: CFSYSLINE { $$ = $1; } nvlst: { $$ = NULL; } | nvlst nv { $2->next = $1; $$ = $2; } -nv: NAME '=' VALUE { $$ = nvlstNew($1, $3); } -rule: PRIFILT actlst { $$ = cnfruleNew(CNFFILT_PRI, $2); $$->filt.s = $1; } - | PROPFILT actlst { $$ = cnfruleNew(CNFFILT_PROP, $2); $$->filt.s = $1; } - | scriptfilt { $$ = $1; } - -scriptfilt: IF expr THEN actlst { $$ = cnfruleNew(CNFFILT_SCRIPT, $4); - $$->filt.expr = $2; } -block: actlst { $$ = $1; } - | block actlst { $2->next = $1; $$ = $2; } -actlst: act { $$=$1; } - | actlst '&' act { $3->next = $1; $$ = $3; } - | actlst cfsysline { $$ = cnfactlstAddSysline($1, $2); } - | '{' block '}' { $$ = $2; } -act: BEGIN_ACTION nvlst ENDOBJ { $$ = cnfactlstNew(CNFACT_V2, $2, NULL); } - | LEGACY_ACTION { $$ = cnfactlstNew(CNFACT_LEGACY, NULL, $1); } +nv: NAME '=' value { $$ = nvlstSetName($3, $1); } +value: STRING { $$ = nvlstNewStr($1); } + | array { $$ = nvlstNewArray($1); } +script: stmt { $$ = $1; } + | script stmt { $$ = scriptAddStmt($1, $2); } +stmt: actlst { $$ = $1; } + | IF expr THEN block { $$ = cnfstmtNew(S_IF); + $$->d.s_if.expr = $2; + $$->d.s_if.t_then = $4; + $$->d.s_if.t_else = NULL; } + | IF expr THEN block ELSE block { $$ = cnfstmtNew(S_IF); + $$->d.s_if.expr = $2; + $$->d.s_if.t_then = $4; + $$->d.s_if.t_else = $6; } + | SET VAR '=' expr ';' { $$ = cnfstmtNewSet($2, $4); } + | UNSET VAR ';' { $$ = cnfstmtNewUnset($2); } + | PRIFILT block { $$ = cnfstmtNewPRIFILT($1, $2); } + | PROPFILT block { $$ = cnfstmtNewPROPFILT($1, $2); } +block: stmt { $$ = $1; } + | '{' script '}' { $$ = $2; } +actlst: s_act { $$ = $1; } + | actlst '&' s_act { $$ = scriptAddStmt($1, $3); } +/* s_act are actions and action-like statements */ +s_act: BEGIN_ACTION nvlst ENDOBJ { $$ = cnfstmtNewAct($2); } + | LEGACY_ACTION { $$ = cnfstmtNewLegaAct($1); } + | STOP { $$ = cnfstmtNew(S_STOP); } + | CALL NAME { $$ = cnfstmtNewCall($2); } + | CONTINUE { $$ = cnfstmtNewContinue(); } expr: expr AND expr { $$ = cnfexprNew(AND, $1, $3); } | expr OR expr { $$ = cnfexprNew(OR, $1, $3); } | NOT expr { $$ = cnfexprNew(NOT, NULL, $2); } @@ -176,6 +183,7 @@ expr: expr AND expr { $$ = cnfexprNew(AND, $1, $3); } | expr CMP_CONTAINSI expr { $$ = cnfexprNew(CMP_CONTAINSI, $1, $3); } | expr CMP_STARTSWITH expr { $$ = cnfexprNew(CMP_STARTSWITH, $1, $3); } | expr CMP_STARTSWITHI expr { $$ = cnfexprNew(CMP_STARTSWITHI, $1, $3); } + | expr '&' expr { $$ = cnfexprNew('&', $1, $3); } | expr '+' expr { $$ = cnfexprNew('+', $1, $3); } | expr '-' expr { $$ = cnfexprNew('-', $1, $3); } | expr '*' expr { $$ = cnfexprNew('*', $1, $3); } @@ -188,8 +196,12 @@ expr: expr AND expr { $$ = cnfexprNew(AND, $1, $3); } | NUMBER { $$ = (struct cnfexpr*) cnfnumvalNew($1); } | STRING { $$ = (struct cnfexpr*) cnfstringvalNew($1); } | VAR { $$ = (struct cnfexpr*) cnfvarNew($1); } + | array { $$ = (struct cnfexpr*) $1; } fparams: expr { $$ = cnffparamlstNew($1, NULL); } | expr ',' fparams { $$ = cnffparamlstNew($1, $3); } +array: '[' arrayelt ']' { $$ = $2; } +arrayelt: STRING { $$ = cnfarrayNew($1); } + | arrayelt ',' STRING { $$ = cnfarrayAdd($1, $3); } %% /* diff --git a/grammar/lexer.l b/grammar/lexer.l index 86fd97b4..ed5d8a80 100644 --- a/grammar/lexer.l +++ b/grammar/lexer.l @@ -45,6 +45,8 @@ /* INCL is in $IncludeConfig processing (skip to include file) */ %x LINENO /* LINENO: support for setting the linenumber */ +%x INCALL + /* INCALL: support for the call statement */ %x EXPR /* EXPR is a bit ugly, but we need it to support pre v6-syntax. The problem * is that cfsysline statement start with $..., the same like variables in @@ -86,8 +88,9 @@ extern int yydebug; /* somehow, I need these prototype even though the headers are * included. I guess that's some autotools magic I don't understand... */ -//char *strdup(char*); +#if !defined(__FreeBSD__) int fileno(FILE *stream); +#endif %} @@ -96,15 +99,20 @@ int fileno(FILE *stream); /* keywords */ "if" { BEGIN EXPR; return IF; } <EXPR>"then" { BEGIN INITIAL; return THEN; } +<EXPR>";" { BEGIN INITIAL; return ';'; } <EXPR>"or" { return OR; } <EXPR>"and" { return AND; } <EXPR>"not" { return NOT; } +<EXPR>"=" | <EXPR>"," | <EXPR>"*" | <EXPR>"/" | <EXPR>"%" | <EXPR>"+" | +<EXPR>"&" | <EXPR>"-" | +<EXPR>"[" | +<EXPR>"]" | <EXPR>"(" | <EXPR>")" { return yytext[0]; } <EXPR>"==" { return CMP_EQ; } @@ -121,7 +129,7 @@ int fileno(FILE *stream); <EXPR>0[0-7]+ | /* octal number */ <EXPR>0x[0-7a-f] | /* hex number, following rule is dec; strtoll handles all! */ <EXPR>([1-9][0-9]*|0) { yylval.n = strtoll(yytext, NULL, 0); return NUMBER; } -<EXPR>\$[$!]{0,1}[a-z][a-z0-9\-_\.]* { yylval.s = strdup(yytext); return VAR; } +<EXPR>\$[$!]{0,1}[a-z][!a-z0-9\-_\.]* { yylval.s = strdup(yytext); return VAR; } <EXPR>\'([^'\\]|\\['"\\$bntr]|\\x[0-9a-f][0-9a-f]|\\[0-7][0-7][0-7])*\' { yytext[yyleng-1] = '\0'; unescapeStr((uchar*)yytext+1, yyleng-2); @@ -135,15 +143,28 @@ int fileno(FILE *stream); <EXPR>[ \t\n] <EXPR>[a-z][a-z0-9_]* { yylval.estr = es_newStrFromCStr(yytext, yyleng); return FUNC; } -<EXPR>. { dbgprintf("invalid char in expr: %s\n", yytext); } +<EXPR>. { parser_errmsg("invalid character '%s' in expression " + "- is there an invalid escape sequence somewhere?", + yytext); } +<INCALL>[ \t\n] +<INCALL>. { parser_errmsg("invalid character '%s' in 'call' statement" + "- is there an invalid escape sequence somewhere?", + yytext); } +<INCALL>[a-zA-Z][a-zA-Z0-9_\.]* { yylval.estr = es_newStrFromCStr(yytext, yyleng); + BEGIN INITIAL; + return NAME; } "&" { return '&'; } "{" { return '{'; } "}" { return '}'; } -"ruleset" { dbgprintf("RULESET\n"); } +"stop" { return STOP; } +"else" { return ELSE; } +"call" { BEGIN INCALL; return CALL; } +"set" { BEGIN EXPR; return SET; } +"unset" { BEGIN EXPR; return UNSET; } +"continue" { return CONTINUE; } /* line number support because the "preprocessor" combines lines and so needs * to tell us the real source line. */ -"stop" { dbgprintf("STOP\n"); return STOP; } "preprocfilelinenumber(" { BEGIN LINENO; } <LINENO>[0-9]+ { yylineno = atoi(yytext) - 1; } <LINENO>")" { BEGIN INITIAL; } @@ -159,6 +180,8 @@ int fileno(FILE *stream); BEGIN INOBJ; return BEGINOBJ; } "template"[ \n\t]*"(" { yylval.objType = CNFOBJ_TPL; BEGIN INOBJ; return BEGIN_TPL; } +"ruleset"[ \n\t]*"(" { yylval.objType = CNFOBJ_RULESET; + BEGIN INOBJ; return BEGIN_RULESET; } "property"[ \n\t]*"(" { yylval.objType = CNFOBJ_PROPERTY; BEGIN INOBJ; return BEGIN_PROPERTY; } "constant"[ \n\t]*"(" { yylval.objType = CNFOBJ_CONSTANT; @@ -168,25 +191,33 @@ int fileno(FILE *stream); "module"[ \n\t]*"(" { yylval.objType = CNFOBJ_MODULE; BEGIN INOBJ; return BEGINOBJ; } "action"[ \n\t]*"(" { BEGIN INOBJ; return BEGIN_ACTION; } -^[ \t]*:\$?[a-z\-]+[ ]*,[ ]*!?[a-z]+[ ]*,[ ]*\".*\" { - yylval.s = strdup(yytext); return PROPFILT; } -^[ \t]*[\*a-z][\*a-z]*[0-7]*[\.,][,!=;\.\*a-z0-7]+ { yylval.s = strdup(yytext); return PRIFILT; } +^[ \t]*:\$?[a-z\-]+[ ]*,[ ]*!?[a-z]+[ ]*,[ ]*\"(\\\"|[^\"])*\" { + yylval.s = strdup(rmLeadingSpace(yytext)); + dbgprintf("lexer: propfilt is '%s'\n", yylval.s); + return PROPFILT; + } +^[ \t]*[\*a-z][\*a-z]*[0-7]*[\.,][,!=;\.\*a-z0-7]+ { yylval.s = strdup(rmLeadingSpace(yytext)); return PRIFILT; } "~" | "*" | \-\/[^*][^\n]* | \/[^*][^\n]* | :[a-z0-9]+:[^\n]* | [\|\.\-\@\^?~>][^\n]+ | -[a-z0-9_][a-z0-9_\-\+,;]* { yylval.s = strdup(yytext); - dbgprintf("lex: LEGA ACT: '%s'\n", yytext); - return LEGACY_ACTION; } +[a-z0-9_][a-z0-9_\-\+,;]* { yylval.s = yytext; return LEGACY_ACTION; } <INOBJ>")" { BEGIN INITIAL; return ENDOBJ; } <INOBJ>[a-z][a-z0-9_\.]* { yylval.estr = es_newStrFromCStr(yytext, yyleng); return NAME; } +<INOBJ>"," | +<INOBJ>"[" | +<INOBJ>"]" | <INOBJ>"=" { return(yytext[0]); } <INOBJ>\"([^"\\]|\\['"?\\abfnrtv]|\\[0-7]{1,3})*\" { - yylval.estr = es_newStrFromBuf(yytext+1, yyleng-2); - return VALUE; } + yytext[yyleng-1] = '\0'; + unescapeStr((uchar*)yytext+1, yyleng-2); + yylval.estr = es_newStrFromBuf(yytext+1, strlen(yytext)-1); + return STRING; } + /*yylval.estr = es_newStrFromBuf(yytext+1, yyleng-2); + return VALUE; }*/ "/*" { preCommentState = YY_START; BEGIN COMMENT; } <INOBJ>"/*" { preCommentState = YY_START; BEGIN COMMENT; } <EXPR>"/*" { preCommentState = YY_START; BEGIN COMMENT; } @@ -194,7 +225,9 @@ int fileno(FILE *stream); <COMMENT>([^*]|\n)+|. <INOBJ>#.*$ /* skip comments in input */ <INOBJ>[ \n\t] -<INOBJ>. { dbgprintf("INOBJ: invalid char '%s'\n", yytext); } +<INOBJ>. { parser_errmsg("invalid character '%s' in object definition " + "- is there an invalid escape sequence somewhere?", + yytext); } \$[a-z]+.*$ { /* see comment on $IncludeConfig above */ if(!strncasecmp(yytext, "$includeconfig ", 14)) { yyless(14); @@ -203,8 +236,7 @@ int fileno(FILE *stream); yylval.s = strdup(yytext); return LEGACY_RULESET; } else { - yylval.s = strdup(yytext); - return CFSYSLINE; + cnfDoCfsysline(strdup(yytext)); } } ![^ \t\n]+[ \t]*$ { yylval.s = strdup(yytext); return BSD_TAG_SELECTOR; } @@ -213,8 +245,9 @@ int fileno(FILE *stream); ^[ \t]*[+-][a-z0-9.:-]+[ \t]*$ { yylval.s = strdup(yytext); return BSD_HOST_SELECTOR; } \#.*\n /* skip comments in input */ [\n\t ] /* drop whitespace */ -. { dbgprintf("invalid char: %s\n", yytext); - } +. { parser_errmsg("invalid character '%s' " + "- is there an invalid escape sequence somewhere?", + yytext); } <<EOF>> { if(popfile() != 0) yyterminate(); } %% @@ -279,6 +312,7 @@ cnfSetLexFile(char *fname) currbs = bs; cnfcurrfn = bs->fn; yylineno = 1; + dbgprintf("config parser: pushed file %s on top of stack\n", fname); done: if(r != 0) { @@ -306,6 +340,7 @@ popfile(void) * necessary, as otherwise we may provide wrong file name information * at the end of include files as well. -- rgerhards, 2011-07-22 */ + dbgprintf("config parser: reached end of file %s\n", bs->fn); yy_delete_buffer(bs->bs); if(bs->prev != NULL) free(bs->fn); @@ -315,12 +350,16 @@ popfile(void) currbs = bs->prev; free(bs); - if(currbs == NULL) + if(currbs == NULL) { + dbgprintf("config parser: parsing completed\n"); return 1; /* all processed */ + } yy_switch_to_buffer(currbs->bs); yylineno = currbs->lineno; cnfcurrfn = currbs->fn; + dbgprintf("config parser: resume parsing of file %s at line %d\n", + cnfcurrfn, yylineno); return 0; } diff --git a/grammar/parserif.h b/grammar/parserif.h index 597cfe40..aa271ec4 100644 --- a/grammar/parserif.h +++ b/grammar/parserif.h @@ -15,7 +15,7 @@ extern int yylineno; * these functions. */ void cnfDoObj(struct cnfobj *o); -void cnfDoRule(struct cnfrule *rule); +void cnfDoScript(struct cnfstmt *script); void cnfDoCfsysline(char *ln); void cnfDoBSDTag(char *ln); void cnfDoBSDHost(char *ln); diff --git a/grammar/rainerscript.c b/grammar/rainerscript.c index 072bdd5d..0f0b469c 100644 --- a/grammar/rainerscript.c +++ b/grammar/rainerscript.c @@ -2,7 +2,7 @@ * * Module begun 2011-07-01 by Rainer Gerhards * - * Copyright 2011 Rainer Gerhards and Adiscon GmbH. + * Copyright 2011-2013 Rainer Gerhards and Adiscon GmbH. * * This file is part of the rsyslog runtime library. * @@ -22,7 +22,6 @@ * 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 <stdio.h> #include <stdlib.h> @@ -37,16 +36,206 @@ #include <libestr.h> #include "rsyslog.h" #include "rainerscript.h" +#include "conf.h" #include "parserif.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" 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. + */ +char * +tokenToString(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; +} + + +char* +getFIOPName(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; +} + +static void +prifiltInvert(struct funcData_prifilt *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 *prifilt, int fac, 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 *prifilt, struct funcData_prifilt *prifilt2, 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 *fp, es_str_t **str) { @@ -98,6 +287,14 @@ readConfFile(FILE *fp, es_str_t **str) 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) { @@ -107,7 +304,6 @@ objlstNew(struct cnfobj *o) lst->next = NULL; lst->obj = o; } -dbgprintf("AAAA: creating new objlst\n"); cnfobjPrint(o); return lst; @@ -131,6 +327,22 @@ objlstAdd(struct objlst *root, struct cnfobj *o) 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) { @@ -155,13 +367,12 @@ objlstPrint(struct objlst *lst) } struct nvlst* -nvlstNew(es_str_t *name, es_str_t *value) +nvlstNewStr(es_str_t *value) { struct nvlst *lst; if((lst = malloc(sizeof(struct nvlst))) != NULL) { lst->next = NULL; - lst->name = name; lst->val.datatype = 'S'; lst->val.d.estr = value; lst->bUsed = 0; @@ -170,6 +381,28 @@ nvlstNew(es_str_t *name, es_str_t *value) 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) { @@ -179,8 +412,7 @@ nvlstDestruct(struct nvlst *lst) toDel = lst; lst = lst->next; es_deleteStr(toDel->name); - if(toDel->val.datatype == 'S') - es_deleteStr(toDel->val.d.estr); + varDelete(&toDel->val); free(toDel); } } @@ -192,11 +424,21 @@ nvlstPrint(struct nvlst *lst) dbgprintf("nvlst %p:\n", lst); while(lst != NULL) { name = es_str2cstr(lst->name, NULL); - // TODO: support for non-string types - value = es_str2cstr(lst->val.d.estr, NULL); - dbgprintf("\tname: '%s', value '%s'\n", name, value); + 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); - free(value); lst = lst->next; } } @@ -380,17 +622,17 @@ doGetFileCreateMode(struct nvlst *valnode, struct cnfparamdescr *param, 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') ) ) { + 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');; + 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 " @@ -515,6 +757,7 @@ doGetWord(struct nvlst *valnode, struct cnfparamdescr *param, 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); @@ -523,7 +766,7 @@ doGetWord(struct nvlst *valnode, struct cnfparamdescr *param, } if(i != es_strlen(valnode->val.d.estr)) { parser_errmsg("parameter '%s' contains whitespace, which is not " - "permitted - data after first whitespace ignored", + "permitted", param->name); r = 0; } @@ -531,6 +774,30 @@ doGetWord(struct nvlst *valnode, struct cnfparamdescr *param, } 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) { @@ -556,8 +823,15 @@ nvlstGetParam(struct nvlst *valnode, struct cnfparamdescr *param, uchar *cstr; int r; - dbgprintf("XXXX: in nvlstGetParam, name '%s', type %d, valnode->bUsed %d\n", + 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) { @@ -613,6 +887,9 @@ nvlstGetParam(struct nvlst *valnode, struct cnfparamdescr *param, 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); @@ -623,7 +900,7 @@ nvlstGetParam(struct nvlst *valnode, struct cnfparamdescr *param, r = 0; break; } - return r; +done: return r; } @@ -661,8 +938,14 @@ nvlstGetParams(struct nvlst *lst, struct cnfparamblk *params, for(i = 0 ; i < params->nParams ; ++i) { param = params->descr + i; - if((valnode = nvlstFindNameCStr(lst, param->name)) == NULL) + 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); @@ -673,7 +956,6 @@ nvlstGetParams(struct nvlst *lst, struct cnfparamblk *params, } } - if(bInError) { if(bValsWasNULL) cnfparamvalsDestruct(vals, params); @@ -684,6 +966,30 @@ nvlstGetParams(struct nvlst *lst, struct cnfparamblk *params, } +/* 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) { @@ -700,6 +1006,9 @@ cnfparamsPrint(struct cnfparamblk *params, struct cnfparamvals *vals) 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; @@ -724,6 +1033,7 @@ cnfobjNew(enum cnfobjType objType, struct nvlst *lst) o->objType = objType; o->nvlst = lst; o->subobjs = NULL; + o->script = NULL; } return o; @@ -747,116 +1057,6 @@ cnfobjPrint(struct cnfobj *o) } -struct cnfactlst* -cnfactlstNew(enum cnfactType actType, struct nvlst *lst, char *actLine) -{ - struct cnfactlst *actlst; - - if((actlst = malloc(sizeof(struct cnfactlst))) != NULL) { - actlst->next = NULL; - actlst->syslines = NULL; - actlst->actType = actType; - actlst->lineno = yylineno; - actlst->cnfFile = strdup(cnfcurrfn); - if(actType == CNFACT_V2) - actlst->data.lst = lst; - else - actlst->data.legActLine = actLine; - } - return actlst; -} - -struct cnfactlst* -cnfactlstAddSysline(struct cnfactlst* actlst, char *line) -{ - struct cnfcfsyslinelst *cflst; - - if((cflst = malloc(sizeof(struct cnfcfsyslinelst))) != NULL) { - cflst->line = line; - if(actlst->syslines == NULL) { - cflst->next = NULL; - } else { - cflst->next = actlst->syslines; - } - actlst->syslines = cflst; - } - return actlst; -} - - -void -cnfactlstDestruct(struct cnfactlst *actlst) -{ - struct cnfactlst *toDel; - - while(actlst != NULL) { - toDel = actlst; - actlst = actlst->next; - free(toDel->cnfFile); - cnfcfsyslinelstDestruct(toDel->syslines); - if(toDel->actType == CNFACT_V2) - nvlstDestruct(toDel->data.lst); - else - free(toDel->data.legActLine); - free(toDel); - } - -} - -static inline struct cnfcfsyslinelst* -cnfcfsyslinelstReverse(struct cnfcfsyslinelst *lst) -{ - struct cnfcfsyslinelst *curr, *prev; - if(lst == NULL) - return NULL; - prev = NULL; - while(lst != NULL) { - curr = lst; - lst = lst->next; - curr->next = prev; - prev = curr; - } - return prev; -} - -struct cnfactlst* -cnfactlstReverse(struct cnfactlst *actlst) -{ - struct cnfactlst *curr, *prev; - - prev = NULL; - while(actlst != NULL) { - curr = actlst; - actlst = actlst->next; - curr->syslines = cnfcfsyslinelstReverse(curr->syslines); - curr->next = prev; - prev = curr; - } - return prev; -} - -void -cnfactlstPrint(struct cnfactlst *actlst) -{ - struct cnfcfsyslinelst *cflst; - - while(actlst != NULL) { - dbgprintf("aclst %p: ", actlst); - if(actlst->actType == CNFACT_V2) { - dbgprintf("V2 action type: "); - nvlstPrint(actlst->data.lst); - } else { - dbgprintf("legacy action line: '%s'\n", - actlst->data.legActLine); - } - for( cflst = actlst->syslines - ; cflst != NULL ; cflst = cflst->next) { - dbgprintf("action:cfsysline: '%s'\n", cflst->line); - } - actlst = actlst->next; - } -} - struct cnfexpr* cnfexprNew(unsigned nodetype, struct cnfexpr *l, struct cnfexpr *r) { @@ -883,15 +1083,19 @@ done: * try to convert it to one. The semantics from es_str2num() * are used (bSuccess tells if the conversion went well or not). */ -static inline long long +static long long var2Number(struct var *r, int *bSuccess) { long long n; if(r->datatype == 'S') { n = es_str2num(r->d.estr, bSuccess); } else { - n = r->d.n; - if(bSuccess) + if(r->datatype == 'J') { + n = (r->d.json == NULL) ? 0 : json_object_get_int(r->d.json); + } else { + n = r->d.n; + } + if(bSuccess != NULL) *bSuccess = 1; } return n; @@ -902,14 +1106,223 @@ var2Number(struct var *r, int *bSuccess) static inline es_str_t * var2String(struct var *r, int *bMustFree) { + es_str_t *estr; + char *cstr; + rs_size_t lenstr; if(r->datatype == 'N') { *bMustFree = 1; - return es_newStrFromNumber(r->d.n); + 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 *r, int *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; +} + +static rsRetVal +doExtractFieldByChar(uchar *str, uchar delim, 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(char))); + /* now copy */ + memcpy(pBuf, pFld, iLen); + pBuf[iLen] = '\0'; /* terminate it */ + *resstr = pBuf; + } else { + ABORT_FINALIZE(RS_RET_FIELD_NOT_FOUND); } - *bMustFree = 0; - return r->d.estr; +finalize_it: + RETiRet; +} + + +static rsRetVal +doExtractFieldByStr(uchar *str, char *delim, rs_size_t lenDelim, 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 != 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(char))); + /* 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; + 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 %d 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); + } + + if(bMustFree) free(str); + if(r[0].datatype == 'S') es_deleteStr(r[0].d.estr); + if(r[2].datatype == 'S') es_deleteStr(r[2].d.estr); + if(r[3].datatype == 'S') es_deleteStr(r[3].d.estr); +finalize_it: + if(bHadNoMatch) { + cnfexprEval(func->expr[4], &r[4], usrptr); + estr = var2String(&r[4], &bMustFree); + if(r[4].datatype == 'S') es_deleteStr(r[4].d.estr); + } + ret->datatype = 'S'; + ret->d.estr = estr; + return; +} + + /* Perform a function call. This has been moved out of cnfExprEval in order * to keep the code small and easier to maintain. */ @@ -921,8 +1334,13 @@ doFuncCall(struct cnffunc *func, struct var *ret, void* usrptr) 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) { @@ -950,8 +1368,12 @@ doFuncCall(struct cnffunc *func, struct var *ret, void* 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'; - ret->d.estr = es_newStrFromCStr(envvar, strlen(envvar)); if(bMustFree) es_deleteStr(estr); if(r[0].datatype == 'S') es_deleteStr(r[0].d.estr); free(str); @@ -990,8 +1412,7 @@ doFuncCall(struct cnffunc *func, struct var *ret, void* usrptr) break; case CNFFUNC_RE_MATCH: cnfexprEval(func->expr[0], &r[0], usrptr); - estr = var2String(&r[0], &bMustFree); - str = es_str2cstr(estr, NULL); + str = (char*) var2CString(&r[0], &bMustFree); retval = regexp.regexec(func->funcdata, str, 0, NULL, 0); if(retval == 0) ret->d.n = 1; @@ -1002,10 +1423,54 @@ doFuncCall(struct cnffunc *func, struct var *ret, void* usrptr) } } ret->datatype = 'N'; - if(bMustFree) es_deleteStr(estr); - free(str); + if(bMustFree) free(str); if(r[0].datatype == 'S') es_deleteStr(r[0].d.estr); break; + case CNFFUNC_RE_EXTRACT: + doFunc_re_extract(func, ret, 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); + if(r[0].datatype == 'S') es_deleteStr(r[0].d.estr); + if(r[1].datatype == 'S') es_deleteStr(r[1].d.estr); + if(r[2].datatype == 'S') es_deleteStr(r[2].d.estr); + 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; default: if(Debug) { fname = es_str2cstr(func->fname, NULL); @@ -1018,6 +1483,67 @@ doFuncCall(struct cnffunc *func, struct var *ret, void* usrptr) } } +static inline void +evalVar(struct cnfvar *var, void *usrptr, struct var *ret) +{ + rsRetVal localRet; + es_str_t *estr; + struct json_object *json; + + if(var->name[0] == '$' && var->name[1] == '!') { + /* TODO: unify string libs */ + estr = es_newStrFromBuf(var->name+1, strlen(var->name)-1); + localRet = msgGetCEEPropJSON((msg_t*)usrptr, estr, &json); + es_deleteStr(estr); + ret->datatype = 'J'; + ret->d.json = (localRet == RS_RET_OK) ? json : NULL; + } else { + ret->datatype = 'S'; + ret->d.estr = cnfGetVar(var->name, usrptr); + } + +} + +/* 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 *estr_l, struct cnfarray* ar, 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 \ if(r.datatype == 'S') es_deleteStr(r.d.estr); \ if(l.datatype == 'S') es_deleteStr(l.d.estr) @@ -1029,16 +1555,27 @@ doFuncCall(struct cnffunc *func, struct var *ret, void* usrptr) ret->d.n = var2Number(&l, &convok_l) x var2Number(&r, &convok_r); \ 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); \ - cnfexprEval(expr->r, &r, usrptr); \ - estr_r = var2String(&r, &bMustFree) + 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(bMustFree2) es_deleteStr(estr_l); \ - FREE_BOTH_RET + if(bMustFree) es_deleteStr(estr_r); \ + if(expr->r->nodetype != 'S' && expr->r->nodetype != 'A' && r.datatype == 'S') es_deleteStr(r.d.estr); \ + if(bMustFree2) es_deleteStr(estr_l); \ + if(l.datatype == 'S') es_deleteStr(l.d.estr) /* evaluate an expression. * Note that we try to avoid malloc whenever possible (because of @@ -1059,29 +1596,63 @@ cnfexprEval(struct cnfexpr *expr, struct var *ret, void* usrptr) int bMustFree, bMustFree2; long long n_r, n_l; - dbgprintf("eval expr %p, type '%c'(%u)\n", expr, expr->nodetype, expr->nodetype); + 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); - 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); /*CMP*/ + 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 { - n_l = var2Number(&l, &convok_l); - if(convok_l) { - ret->d.n = (n_l == r.d.n); /*CMP*/ + cnfexprEval(expr->r, &r, usrptr); + if(r.datatype == 'S') { + ret->d.n = !es_strcmp(l.d.estr, r.d.estr); /*CMP*/ } else { - estr_r = var2String(&r, &bMustFree); - ret->d.n = !es_strcmp(l.d.estr, estr_r); /*CMP*/ - if(bMustFree) es_deleteStr(estr_r); + 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); + } } + if(r.datatype == 'S') es_deleteStr(r.d.estr); } + } 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, &bMustFree); + ret->d.n = !es_strcmp(estr_l, estr_r); /*CMP*/ + if(bMustFree) es_deleteStr(estr_r); + } + } + if(r.datatype == 'S') es_deleteStr(r.d.estr); + } + 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) { @@ -1094,24 +1665,31 @@ cnfexprEval(struct cnfexpr *expr, struct var *ret, void* usrptr) } else { ret->d.n = (l.d.n == r.d.n); /*CMP*/ } + if(r.datatype == 'S') es_deleteStr(r.d.estr); } - FREE_BOTH_RET; + if(l.datatype == 'S') es_deleteStr(l.d.estr); break; case CMP_NE: 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); /*CMP*/ + 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 { - n_l = var2Number(&l, &convok_l); - if(convok_l) { - ret->d.n = (n_l != r.d.n); /*CMP*/ + if(r.datatype == 'S') { + ret->d.n = es_strcmp(l.d.estr, r.d.estr); /*CMP*/ } else { - estr_r = var2String(&r, &bMustFree); - ret->d.n = es_strcmp(l.d.estr, estr_r); /*CMP*/ - if(bMustFree) es_deleteStr(estr_r); + 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 { @@ -1265,25 +1843,45 @@ cnfexprEval(struct cnfexpr *expr, struct var *ret, void* usrptr) case CMP_STARTSWITH: PREP_TWO_STRINGS; ret->datatype = 'N'; - ret->d.n = es_strncmp(estr_l, estr_r, estr_r->lenStr) == 0; + 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'; - ret->d.n = es_strncasecmp(estr_l, estr_r, estr_r->lenStr) == 0; + 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'; - ret->d.n = es_strContains(estr_l, estr_r) != -1; + 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'; - ret->d.n = es_strCaseContains(estr_l, estr_r) != -1; + 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: @@ -1330,9 +1928,27 @@ cnfexprEval(struct cnfexpr *expr, struct var *ret, void* usrptr) 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 = cnfGetVar(((struct cnfvar*)expr)->name, usrptr); + ret->d.estr = es_strdup(estr_l); + es_addStr(&ret->d.estr, estr_r); + FREE_TWO_STRINGS; break; case '+': COMP_NUM_BINOP(+); @@ -1369,6 +1985,16 @@ cnfexprEval(struct cnfexpr *expr, struct var *ret, void* usrptr) //--------------------------------------------------------- +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) { @@ -1380,12 +2006,14 @@ cnffuncDestruct(struct cnffunc *func) /* some functions require special destruction */ switch(func->fID) { case CNFFUNC_RE_MATCH: - regexp.regfree(func->funcdata); - free(func->funcdata); - free(func->fname); + case CNFFUNC_RE_EXTRACT: + if(func->funcdata != NULL) + regexp.regfree(func->funcdata); break; default:break; } + free(func->funcdata); + free(func->fname); } /* Destruct an expression and all sub-expressions contained in it. @@ -1394,7 +2022,13 @@ void cnfexprDestruct(struct cnfexpr *expr) { - dbgprintf("cnfexprDestruct expr %p, type '%c'(%u)\n", expr, expr->nodetype, expr->nodetype); + 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: @@ -1408,6 +2042,7 @@ cnfexprDestruct(struct cnfexpr *expr) case CMP_CONTAINSI: case OR: case AND: + case '&': case '+': case '-': case '*': @@ -1431,6 +2066,9 @@ cnfexprDestruct(struct cnfexpr *expr) case 'F': cnffuncDestruct((struct cnffunc*)expr); break; + case 'A': + cnfarrayContentDestruct((struct cnfarray*)expr); + break; default:break; } free(expr); @@ -1459,13 +2097,39 @@ doIndent(int indent) 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; - //dbgprintf("expr %p, indent %d, type '%c'\n", expr, indent, expr->nodetype); switch(expr->nodetype) { case CMP_EQ: cnfexprPrint(expr->l, indent+1); @@ -1549,6 +2213,9 @@ cnfexprPrint(struct cnfexpr *expr, int 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); @@ -1562,10 +2229,16 @@ cnfexprPrint(struct cnfexpr *expr, int 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 '*': @@ -1579,11 +2252,109 @@ cnfexprPrint(struct cnfexpr *expr, int indent) cnfexprPrint(expr->r, indent+1); break; default: - dbgprintf("error: unknown nodetype %u\n", - (unsigned) expr->nodetype); + 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]\n", cstr); + free(cstr); + break; + case S_ACT: + doIndent(indent); dbgprintf("ACTION %p [%s:%s]\n", stmt->d.act, + 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.propID)); + if(stmt->d.s_propfilt.propName != NULL) { + cstr = es_str2cstr(stmt->d.s_propfilt.propName, NULL); + doIndent(indent); + dbgprintf("\tCEE-Prop.: '%s'\n", cstr); + free(cstr); + } + 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(long long val) @@ -1607,6 +2378,52 @@ cnfstringvalNew(es_str_t *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 *ar, es_str_t *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) { @@ -1618,63 +2435,723 @@ cnfvarNew(char *name) return var; } -struct cnfrule * -cnfruleNew(enum cnfFiltType filttype, struct cnfactlst *actlst) +struct cnfstmt * +cnfstmtNew(unsigned s_type) { - struct cnfrule* cnfrule; - if((cnfrule = malloc(sizeof(struct cnfrule))) != NULL) { - cnfrule->nodetype = 'R'; - cnfrule->filttype = filttype; - cnfrule->actlst = cnfactlstReverse(actlst); + struct cnfstmt* cnfstmt; + if((cnfstmt = malloc(sizeof(struct cnfstmt))) != NULL) { + cnfstmt->nodetype = s_type; + cnfstmt->printable = NULL; + cnfstmt->next = NULL; } - return cnfrule; + 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: + if(stmt->d.s_propfilt.propName != NULL) + es_deleteStr(stmt->d.s_propfilt.propName); + 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 -cnfrulePrint(struct cnfrule *rule) +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; + rsRetVal lRet; + if((cnfstmt = cnfstmtNew(S_PROPFILT)) != NULL) { + cnfstmt->printable = (uchar*)propfilt; + cnfstmt->d.s_propfilt.t_then = t_then; + cnfstmt->d.s_propfilt.propName = NULL; + cnfstmt->d.s_propfilt.regex_cache = NULL; + cnfstmt->d.s_propfilt.pCSCompValue = NULL; + lRet = DecodePropFilter((uchar*)propfilt, cnfstmt); + } + 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 && localRet != RS_RET_OK_WARN) { + 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(!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("$syslogfacility", ((struct cnfvar*)expr->l)->name)) { + if(expr->r->nodetype == 'N') { + int fac = (int) ((struct cnfnumval*)expr->r)->val; + if(fac >= 0 && fac <= 24) { + DBGPRINTF("optimizer: change comparison OP to FUNC prifilt()\n"); + func = cnffuncNew_prifilt(0); /* fac is irrelevant, set below... */ + prifiltSetFacility(func->funcdata, fac, expr->nodetype); + cnfexprDestruct(expr); + expr = (struct cnfexpr*) func; + } else { + parser_errmsg("invalid syslogfacility %d, expression will always " + "evaluate to FALSE", fac); + } + } + } + return expr; +} + +/* optimize a comparison with a variable as left-hand operand + * NOTE: Currently support CMP_EQ, CMP_NE only and code NEEDS + * TO BE CHANGED fgr other comparisons! + */ +static inline struct cnfexpr* +cnfexprOptimize_CMP_var(struct cnfexpr *expr) +{ + struct cnffunc *func; + + if(!strcmp("$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 acutally 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("------ start rule %p:\n", rule); - dbgprintf("%s: ", cnfFiltType2str(rule->filttype)); - switch(rule->filttype) { - case CNFFILT_NONE: + DBGPRINTF("optimizer: sorting array for CMP_EQ/NEQ comparison\n"); + 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 CNFFILT_PRI: - case CNFFILT_PROP: - dbgprintf("%s\n", rule->filt.s); + case '-': + if(getConstNumber(expr, &ln, &rn)) { + expr->nodetype = 'N'; + ((struct cnfnumval*)expr)->val = ln - rn; + } break; - case CNFFILT_SCRIPT: - dbgprintf("\n"); - cnfexprPrint(rule->filt.expr, 0); + 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 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->l->nodetype == 'V') { + expr = cnfexprOptimize_CMP_var(expr); + } else if(expr->r->nodetype == 'A') { + cnfexprOptimize_CMPEQ_arr((struct cnfarray *)expr->r); + } + 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; } - cnfactlstPrint(rule->actlst); - dbgprintf("------ end rule %p\n", rule); + return expr; } -/* note: the sysline itself was already freed during processing - * and as such MUST NOT be freed again! +/* removes NOPs from a statement list and returns the + * first non-NOP entry. */ -void -cnfcfsyslinelstDestruct(struct cnfcfsyslinelst *cfslst) +static inline struct cnfstmt * +removeNOPs(struct cnfstmt *root) { - struct cnfcfsyslinelst *toDel; - while(cfslst != NULL) { - toDel = cfslst; - cfslst = cfslst->next; - free(toDel); + 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; } -void -cnfruleDestruct(struct cnfrule *rule) + +static inline void +cnfstmtOptimizeIf(struct cnfstmt *stmt) { - if( rule->filttype == CNFFILT_PRI - || rule->filttype == CNFFILT_PROP) - free(rule->filt.s); - cnfactlstDestruct(rule->actlst); - free(rule); + 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\n", pRuleset, rsName); + 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) { @@ -1735,6 +3212,27 @@ funcName2ID(es_str_t *fname, unsigned short 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*)"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 { return CNFFUNC_INVALID; } @@ -1751,7 +3249,7 @@ initFunc_re_match(struct cnffunc *func) func->funcdata = NULL; if(func->expr[1]->nodetype != 'S') { - parser_errmsg("param 2 of re_match() must be a constant string"); + parser_errmsg("param 2 of re_match/extract() must be a constant string"); FINALIZE; } @@ -1775,6 +3273,30 @@ finalize_it: 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; +} + + struct cnffunc * cnffuncNew(es_str_t *fname, struct cnffparamlst* paramlst) { @@ -1792,6 +3314,7 @@ cnffuncNew(es_str_t *fname, struct cnffparamlst* paramlst) 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; @@ -1804,25 +3327,54 @@ cnffuncNew(es_str_t *fname, struct cnffparamlst* paramlst) /* 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; default:break; } } return func; } + +/* A special function to create a prifilt() expression during optimization + * phase. + */ +struct cnffunc * +cnffuncNew_prifilt(int fac) +{ + struct cnffunc* func; + + if((func = malloc(sizeof(struct cnffunc))) != NULL) { + func->nodetype = 'F'; + func->fname = es_newStrFromCStr("prifilt", sizeof("prifilt")-1); + func->nParams = 0; + func->fID = CNFFUNC_PRIFILT; + func->funcdata = calloc(1, sizeof(struct funcData_prifilt)); + ((struct funcData_prifilt *)func->funcdata)->pmask[fac >> 3] = 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; - unsigned i; + 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) { @@ -1833,21 +3385,46 @@ cnfDoInclude(char *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_NOSPACE || result == GLOB_ABORTED || cfgFiles.gl_pathc == 0) { + 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)); - parser_errmsg("error accessing config file or directory '%s': %s", - finalName, 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; } - for(i = 0; i < cfgFiles.gl_pathc; i++) { + /* 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) - continue; /* continue with the next file if we can't stat() the file */ + 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); @@ -1867,16 +3444,28 @@ cnfDoInclude(char *name) void varDelete(struct var *v) { - if(v->datatype == 'S') + 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) { - varDelete(¶mvals[i].val); + if(paramvals[i].bUsed) { + varDelete(¶mvals[i].val); + } } free(paramvals); } @@ -1909,6 +3498,15 @@ cstrPrint(char *text, es_str_t *estr) 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) @@ -2056,3 +3654,52 @@ unescapeStr(uchar *s, int len) 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"; + } +} diff --git a/grammar/rainerscript.h b/grammar/rainerscript.h index f5627597..d00cc4c3 100644 --- a/grammar/rainerscript.h +++ b/grammar/rainerscript.h @@ -3,7 +3,11 @@ #include <stdio.h> #include <libestr.h> #include <typedefs.h> +#include <sys/types.h> +#include <regex.h> + +#define LOG_NFACILITIES 24 /* current number of syslog facilities */ #define CNFFUNC_MAX_ARGS 32 /**< maximum number of arguments that any function can have (among * others, this is used to size data structures). @@ -13,6 +17,7 @@ extern int Debug; /* 1 if in debug mode, 0 otherwise -- to be enhanced */ enum cnfobjType { CNFOBJ_ACTION, + CNFOBJ_RULESET, CNFOBJ_GLOBAL, CNFOBJ_INPUT, CNFOBJ_MODULE, @@ -29,6 +34,9 @@ cnfobjType2str(enum cnfobjType ot) case CNFOBJ_ACTION: return "action"; break; + case CNFOBJ_RULESET: + return "ruleset"; + break; case CNFOBJ_GLOBAL: return "global"; break; @@ -63,16 +71,20 @@ enum cnfactType { CNFACT_V2, CNFACT_LEGACY }; struct var { union { es_str_t *estr; - struct cnfexpr *expr; + struct cnfarray *ar; long long n; + struct json_object *json; } d; - char datatype; /* 'N' number, 'S' string, 'E' expression */ + char datatype; /* 'N' number, 'S' string, 'J' JSON, 'A' array + * Note: 'A' is only supported during config phase + */ }; struct cnfobj { enum cnfobjType objType; struct nvlst *nvlst; struct objlst *subobjs; + struct cnfstmt *script; }; struct objlst { @@ -91,23 +103,6 @@ struct nvlst { */ }; -struct cnfcfsyslinelst { - struct cnfcfsyslinelst *next; - char *line; -}; - -struct cnfactlst { - struct cnfactlst *next; - struct cnfcfsyslinelst *syslines; - enum cnfactType actType; - union { - struct nvlst *lst; - char *legActLine; - } data; - char *cnfFile; - int lineno; -}; - /* the following structures support expressions, and may (very much later * be the sole foundation for the AST. * @@ -118,7 +113,19 @@ struct cnfactlst { * R - rule * S - string * V - var + * A - (string) array + * ... plus the S_* #define's below: */ +#define S_STOP 4000 +#define S_PRIFILT 4001 +#define S_PROPFILT 4002 +#define S_IF 4003 +#define S_ACT 4004 +#define S_NOP 4005 /* usually used to disable some statement */ +#define S_SET 4006 +#define S_UNSET 4007 +#define S_CALL 4008 + enum cnfFiltType { CNFFILT_NONE, CNFFILT_PRI, CNFFILT_PROP, CNFFILT_SCRIPT }; static inline char* cnfFiltType2str(enum cnfFiltType filttype) @@ -137,14 +144,44 @@ cnfFiltType2str(enum cnfFiltType filttype) } -struct cnfrule { +struct cnfstmt { unsigned nodetype; - enum cnfFiltType filttype; + struct cnfstmt *next; + uchar *printable; /* printable text for debugging */ union { - char *s; - struct cnfexpr *expr; - } filt; - struct cnfactlst *actlst; + struct { + struct cnfexpr *expr; + struct cnfstmt *t_then; + struct cnfstmt *t_else; + } s_if; + struct { + uchar *varname; + struct cnfexpr *expr; + } s_set; + struct { + uchar *varname; + } s_unset; + struct { + es_str_t *name; + struct cnfstmt *stmt; + } s_call; + struct { + uchar pmask[LOG_NFACILITIES+1]; /* priority mask */ + struct cnfstmt *t_then; + struct cnfstmt *t_else; + } s_prifilt; + struct { + fiop_t operation; + regex_t *regex_cache;/* cache for compiled REs, if used */ + struct cstr_s *pCSCompValue;/* value to "compare" against */ + sbool isNegated; + uintTiny propID;/* ID of the requested property */ + es_str_t *propName;/* name of property for CEE-based filters */ + struct cnfstmt *t_then; + struct cnfstmt *t_else; + } s_propfilt; + struct action_s *act; + } d; }; struct cnfexpr { @@ -168,6 +205,12 @@ struct cnfvar { char *name; }; +struct cnfarray { + unsigned nodetype; + int nmemb; + es_str_t **arr; +}; + struct cnffparamlst { unsigned nodetype; /* P */ struct cnffparamlst *next; @@ -182,7 +225,10 @@ enum cnffuncid { CNFFUNC_TOLOWER, CNFFUNC_CSTR, CNFFUNC_CNUM, - CNFFUNC_RE_MATCH + CNFFUNC_RE_MATCH, + CNFFUNC_RE_EXTRACT, + CNFFUNC_FIELD, + CNFFUNC_PRIFILT }; struct cnffunc { @@ -233,13 +279,19 @@ struct cnfparamvals { /* the values we obtained for param descr. */ unsigned char bUsed; }; +struct funcData_prifilt { + uchar pmask[LOG_NFACILITIES+1]; /* priority mask */ +}; + int cnfParseBuffer(char *buf, unsigned lenBuf); void readConfFile(FILE *fp, es_str_t **str); struct objlst* objlstNew(struct cnfobj *obj); void objlstDestruct(struct objlst *lst); void objlstPrint(struct objlst *lst); -struct nvlst* nvlstNew(es_str_t *name, es_str_t *value); +struct nvlst* nvlstNewArray(struct cnfarray *ar); +struct nvlst* nvlstNewStr(es_str_t *value); +struct nvlst* nvlstSetName(struct nvlst *lst, es_str_t *name); void nvlstDestruct(struct nvlst *lst); void nvlstPrint(struct nvlst *lst); void nvlstChkUnused(struct nvlst *lst); @@ -247,11 +299,6 @@ struct nvlst* nvlstFindName(struct nvlst *lst, es_str_t *name); struct cnfobj* cnfobjNew(enum cnfobjType objType, struct nvlst *lst); void cnfobjDestruct(struct cnfobj *o); void cnfobjPrint(struct cnfobj *o); -struct cnfactlst* cnfactlstNew(enum cnfactType actType, struct nvlst *lst, char *actLine); -void cnfactlstDestruct(struct cnfactlst *actlst); -void cnfactlstPrint(struct cnfactlst *actlst); -struct cnfactlst* cnfactlstAddSysline(struct cnfactlst* actlst, char *line); -struct cnfactlst* cnfactlstReverse(struct cnfactlst *actlst); struct cnfexpr* cnfexprNew(unsigned nodetype, struct cnfexpr *l, struct cnfexpr *r); void cnfexprPrint(struct cnfexpr *expr, int indent); void cnfexprEval(struct cnfexpr *expr, struct var *ret, void *pusr); @@ -259,9 +306,6 @@ int cnfexprEvalBool(struct cnfexpr *expr, void *usrptr); void cnfexprDestruct(struct cnfexpr *expr); struct cnfnumval* cnfnumvalNew(long long val); struct cnfstringval* cnfstringvalNew(es_str_t *estr); -struct cnfrule * cnfruleNew(enum cnfFiltType filttype, struct cnfactlst *actlst); -void cnfruleDestruct(struct cnfrule *rule); -void cnfrulePrint(struct cnfrule *rule); struct cnfvar* cnfvarNew(char *name); struct cnffunc * cnffuncNew(es_str_t *fname, struct cnffparamlst* paramlst); struct cnffparamlst * cnffparamlstNew(struct cnfexpr *expr, struct cnffparamlst *next); @@ -270,11 +314,33 @@ int cnfparamGetIdx(struct cnfparamblk *params, char *name); struct cnfparamvals* nvlstGetParams(struct nvlst *lst, struct cnfparamblk *params, struct cnfparamvals *vals); void cnfparamsPrint(struct cnfparamblk *params, struct cnfparamvals *vals); +int cnfparamvalsIsSet(struct cnfparamblk *params, struct cnfparamvals *vals); void varDelete(struct var *v); void cnfparamvalsDestruct(struct cnfparamvals *paramvals, struct cnfparamblk *blk); -void cnfcfsyslinelstDestruct(struct cnfcfsyslinelst *cfslst); +struct cnfstmt * cnfstmtNew(unsigned s_type); +void cnfstmtPrintOnly(struct cnfstmt *stmt, int indent, sbool subtree); +void cnfstmtPrint(struct cnfstmt *stmt, int indent); +struct cnfstmt* scriptAddStmt(struct cnfstmt *root, struct cnfstmt *s); +struct objlst* objlstAdd(struct objlst *root, struct cnfobj *o); +char *rmLeadingSpace(char *s); +struct cnfstmt * cnfstmtNewPRIFILT(char *prifilt, struct cnfstmt *t_then); +struct cnfstmt * cnfstmtNewPROPFILT(char *propfilt, struct cnfstmt *t_then); +struct cnfstmt * cnfstmtNewAct(struct nvlst *lst); +struct cnfstmt * cnfstmtNewLegaAct(char *actline); +struct cnfstmt * cnfstmtNewSet(char *var, struct cnfexpr *expr); +struct cnfstmt * cnfstmtNewUnset(char *var); +struct cnfstmt * cnfstmtNewCall(es_str_t *name); +struct cnfstmt * cnfstmtNewContinue(void); +void cnfstmtDestructLst(struct cnfstmt *root); +void cnfstmtOptimize(struct cnfstmt *root); +struct cnfarray* cnfarrayNew(es_str_t *val); +struct cnfarray* cnfarrayDup(struct cnfarray *old); +struct cnfarray* cnfarrayAdd(struct cnfarray *ar, es_str_t *val); +void cnfarrayContentDestruct(struct cnfarray *ar); +char* getFIOPName(unsigned iFIOP); rsRetVal initRainerscript(void); void unescapeStr(uchar *s, int len); +char * tokenval2str(int tok); /* debug helper */ void cstrPrint(char *text, es_str_t *estr); |