From d74898029f38c59a316716e734eb5115c5a4614c Mon Sep 17 00:00:00 2001 From: Rainer Gerhards Date: Thu, 21 Feb 2008 17:39:36 +0000 Subject: - modified parser and tokenizer to support slight ABNF modifications from yesterday - change in ABNF was wrong - made a slightly different change - fixed bugs in tokenizer - expression compiler finished (except bugs, of course ;)) --- ctok.c | 6 +++- debug.c | 4 +-- doc/expression.html | 2 +- expr.c | 92 ++++++++++++++++++++++++++++++++++++++--------------- var.c | 6 ++++ vmop.c | 5 ++- vmop.h | 1 + 7 files changed, 86 insertions(+), 30 deletions(-) diff --git a/ctok.c b/ctok.c index db5609ec..8b88e1a2 100644 --- a/ctok.c +++ b/ctok.c @@ -224,6 +224,10 @@ ctokGetNumber(ctok_t *pThis, ctok_token_t *pToken) CHKiRet(ctokGetCharFromStream(pThis, &c)); c = tolower(c); } + + /* we need to unget the character that made the loop terminate */ + CHKiRet(ctokUngetCharFromStream(pThis, c)); + CHKiRet(var.SetInt64(pToken->pVar, n)); dbgprintf("number, number is: '%lld'\n", n); @@ -518,7 +522,7 @@ ctokGetToken(ctok_t *pThis, ctok_token_t **ppToken) } while(bRetry); /* warning: do ... while()! */ *ppToken = pToken; -RUNLOG_VAR("%d", pToken->tok); + dbgoprint((obj_t*) pToken, "token: %d\n", pToken->tok); finalize_it: if(iRet != RS_RET_OK) { diff --git a/debug.c b/debug.c index b88e2ec6..00b7916b 100644 --- a/debug.c +++ b/debug.c @@ -930,7 +930,7 @@ int dbgEntrFunc(dbgFuncDB_t *pFuncDB, int line) /* when we reach this point, we have a fully-initialized FuncDB! */ //if(bLogFuncFlow) /* quick debug hack... select the best for you! */ - if(bLogFuncFlow && !strcmp((char*)pFuncDB->file, "omlibdbi.c")) /* quick debug hack... select the best for you! */ + if(bLogFuncFlow && !strcmp((char*)pFuncDB->file, "expr.c")) /* quick debug hack... select the best for you! */ dbgprintf("%s:%d: %s: enter\n", pFuncDB->file, pFuncDB->line, pFuncDB->func); if(pThrd->stackPtr >= (int) (sizeof(pThrd->callStack) / sizeof(dbgFuncDB_t*))) { dbgprintf("%s:%d: %s: debug module: call stack for this thread full, suspending call tracking\n", @@ -960,7 +960,7 @@ void dbgExitFunc(dbgFuncDB_t *pFuncDB, int iStackPtrRestore) dbgFuncDBPrintActiveMutexes(pFuncDB, "WARNING: mutex still owned by us as we exit function, mutex: ", pthread_self()); //if(bLogFuncFlow) /* quick debug hack... select the best for you! */ - if(bLogFuncFlow && !strcmp((char*)pFuncDB->file, "omlibdbi.c")) /* quick debug hack... select the best for you! */ + if(bLogFuncFlow && !strcmp((char*)pFuncDB->file, "expr.c")) /* quick debug hack... select the best for you! */ dbgprintf("%s:%d: %s: exit\n", pFuncDB->file, pFuncDB->line, pFuncDB->func); pThrd->stackPtr = iStackPtrRestore; if(pThrd->stackPtr < 0) { diff --git a/doc/expression.html b/doc/expression.html index b2ed7a06..b10b5c3d 100644 --- a/doc/expression.html +++ b/doc/expression.html @@ -9,7 +9,7 @@ far, they are supported for filtering messages.

C-like comments (/* some c

Formal Definition

Below is the formal definition of expression format (in ABNF, RFC 2234):
-

if_stmt  := "if" expr "then"  # an aid, not part of expression itself

expr     := e_and *("or" e_and)
e_and    := e_cmp *("and" e_cmp)
e_cmp    := val 0*1(cmp_op val)
val      := ["-"] term *(("+" / "-") term)
term     := factor *(("*" / "/" / "%") factor)
factor   := ["not"] terminal
terminal := var / constant / function / ( "(" expr ")" )
function := name "(" *("," expr) ")"
var      := "$" varname
varname  := msgvar / sysvar
msgvar   := name
sysvar   := "$" name
name     := alpha *(alnum)
constant := string / number
string   := simpstr / tplstr ; tplstr will be implemented in next phase
simpstr  := "'" *char "'"    ; use your imagination for char ;)
tplstr   := '"' template '"' ; not initially implemented
number   := ["-"] 1*digit ; 0nn = octal, 0xnn = hex, nn = decimal
cmp_op   := "==" / "!=" / "<>" / "<" / ">" / "<=" / ">=" / "contains" / "startswith"
digit    := %x30-39
alpha    := "a" ... "z" # all letters
alnum    :* alpha / digit / "_"
+

if_stmt  := "if" expr "then"  # an aid, not part of expression itself

expr     := e_and *("or" e_and)
e_and    := e_cmp *("and" e_cmp)
e_cmp    := val 0*1(cmp_op val)
val      := term *(("+" / "-") term)
term     := factor *(("*" / "/" / "%") factor)
factor   := ["not"] ["-"] terminal
terminal := var / constant / function / ( "(" expr ")" )
function := name "(" *("," expr) ")"
var      := "$" varname
varname  := msgvar / sysvar
msgvar   := name
sysvar   := "$" name
name     := alpha *(alnum)
constant := string / number
string   := simpstr / tplstr ; tplstr will be implemented in next phase
simpstr  := "'" *char "'"    ; use your imagination for char ;)
tplstr   := '"' template '"' ; not initially implemented
number   := ["-"] 1*digit ; 0nn = octal, 0xnn = hex, nn = decimal
cmp_op   := "==" / "!=" / "<>" / "<" / ">" / "<=" / ">=" / "contains" / "startswith"
digit    := %x30-39
alpha    := "a" ... "z" # all letters
alnum    :* alpha / digit / "_"

[rsyslog.conf overview] [manual index] [rsyslog site]

This documentation is part of the diff --git a/expr.c b/expr.c index e84eab17..156621c1 100644 --- a/expr.c +++ b/expr.c @@ -62,13 +62,14 @@ static rsRetVal terminal(expr_t *pThis, ctok_t *tok) { DEFiRet; - ctok_token_t *pToken; + ctok_token_t *pToken = NULL; var_t *pVar; ISOBJ_TYPE_assert(pThis, expr); ISOBJ_TYPE_assert(tok, ctok); CHKiRet(ctok.GetToken(tok, &pToken)); + /* note: pToken is destructed in finalize_it */ switch(pToken->tok) { case ctok_SIMPSTR: @@ -78,8 +79,8 @@ terminal(expr_t *pThis, ctok_t *tok) break; case ctok_NUMBER: dbgoprint((obj_t*) pThis, "number\n"); - CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, opcode_PUSHCONSTANT, NULL)); /* add to program */ - // push val + CHKiRet(ctok_token.UnlinkVar(pToken, &pVar)); + CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, opcode_PUSHCONSTANT, pVar)); /* add to program */ break; case ctok_FUNCTION: dbgoprint((obj_t*) pThis, "function\n"); @@ -88,13 +89,13 @@ terminal(expr_t *pThis, ctok_t *tok) break; case ctok_MSGVAR: dbgoprint((obj_t*) pThis, "MSGVAR\n"); - CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, opcode_PUSHMSGVAR, NULL)); /* add to program */ - // push val + CHKiRet(ctok_token.UnlinkVar(pToken, &pVar)); + CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, opcode_PUSHMSGVAR, pVar)); /* add to program */ break; case ctok_SYSVAR: dbgoprint((obj_t*) pThis, "SYSVAR\n"); - CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, opcode_PUSHSYSVAR, NULL)); /* add to program */ - // push val + CHKiRet(ctok_token.UnlinkVar(pToken, &pVar)); + CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, opcode_PUSHSYSVAR, pVar)); /* add to program */ break; case ctok_LPAREN: dbgoprint((obj_t*) pThis, "expr\n"); @@ -103,8 +104,6 @@ terminal(expr_t *pThis, ctok_t *tok) CHKiRet(ctok.GetToken(tok, &pToken)); /* get next one */ if(pToken->tok != ctok_RPAREN) ABORT_FINALIZE(RS_RET_SYNTAX_ERROR); - CHKiRet(ctok_token.Destruct(&pToken)); /* "eat" processed token */ - dbgoprint((obj_t*) pThis, "end expr, rparen eaten\n"); break; default: dbgoprint((obj_t*) pThis, "invalid token %d\n", pToken->tok); @@ -113,6 +112,10 @@ terminal(expr_t *pThis, ctok_t *tok) } finalize_it: + if(pToken != NULL) { + CHKiRet(ctok_token.Destruct(&pToken)); /* "eat" processed token */ + } + RETiRet; } @@ -121,6 +124,8 @@ factor(expr_t *pThis, ctok_t *tok) { DEFiRet; ctok_token_t *pToken; + int bWasNot; + int bWasUnaryMinus; ISOBJ_TYPE_assert(pThis, expr); ISOBJ_TYPE_assert(tok, ctok); @@ -128,14 +133,35 @@ factor(expr_t *pThis, ctok_t *tok) CHKiRet(ctok.GetToken(tok, &pToken)); if(pToken->tok == ctok_NOT) { dbgprintf("not\n"); + bWasNot = 1; + CHKiRet(ctok_token.Destruct(&pToken)); /* no longer needed */ + CHKiRet(ctok.GetToken(tok, &pToken)); /* get new one for next check */ + } else { + bWasNot = 0; + } + + if(pToken->tok == ctok_MINUS) { + dbgprintf("unary minus\n"); + bWasUnaryMinus = 1; CHKiRet(ctok_token.Destruct(&pToken)); /* no longer needed */ - CHKiRet(ctok.GetToken(tok, &pToken)); /* get next one */ } else { + bWasUnaryMinus = 0; /* we could not process the token, so push it back */ CHKiRet(ctok.UngetToken(tok, pToken)); } + CHKiRet(terminal(pThis, tok)); - // vm: not + + /* warning: the order if the two following ifs is important. Do not change them, this + * would change the semantics of the expression! + */ + if(bWasUnaryMinus) { + CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, opcode_UNARY_MINUS, NULL)); /* add to program */ + } + + if(bWasNot == 1) { + CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, opcode_NOT, NULL)); /* add to program */ + } finalize_it: RETiRet; @@ -146,12 +172,27 @@ static rsRetVal term(expr_t *pThis, ctok_t *tok) { DEFiRet; + ctok_token_t *pToken; ISOBJ_TYPE_assert(pThis, expr); ISOBJ_TYPE_assert(tok, ctok); CHKiRet(factor(pThis, tok)); - // vm: +/- + + /* *(("*" / "/" / "%") factor) part */ + CHKiRet(ctok.GetToken(tok, &pToken)); + while(pToken->tok == ctok_TIMES || pToken->tok == ctok_DIV || pToken->tok == ctok_MOD) { + dbgoprint((obj_t*) pThis, "/,*,%%\n"); + CHKiRet(factor(pThis, tok)); + CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, (opcode_t) pToken->tok, NULL)); /* add to program */ + CHKiRet(ctok_token.Destruct(&pToken)); /* no longer needed */ + CHKiRet(ctok.GetToken(tok, &pToken)); + } + + /* unget the token that made us exit the loop - it's obviously not one + * we can process. + */ + CHKiRet(ctok.UngetToken(tok, pToken)); finalize_it: RETiRet; @@ -166,19 +207,22 @@ val(expr_t *pThis, ctok_t *tok) ISOBJ_TYPE_assert(pThis, expr); ISOBJ_TYPE_assert(tok, ctok); + CHKiRet(term(pThis, tok)); + + /* *(("+" / "-") term) part */ CHKiRet(ctok.GetToken(tok, &pToken)); - if(pToken->tok == ctok_PLUS || pToken->tok == ctok_MINUS) { - /* TODO: this must be a loop! */ - dbgprintf("plus/minus\n"); + while(pToken->tok == ctok_PLUS || pToken->tok == ctok_MINUS) { + dbgoprint((obj_t*) pThis, "+/-\n"); + CHKiRet(term(pThis, tok)); + CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, (opcode_t) pToken->tok, NULL)); /* add to program */ CHKiRet(ctok_token.Destruct(&pToken)); /* no longer needed */ - // vm: +/-??? - CHKiRet(ctok.GetToken(tok, &pToken)); /* get next one */ - } else { - /* we could not process the token, so push it back */ - CHKiRet(ctok.UngetToken(tok, pToken)); + CHKiRet(ctok.GetToken(tok, &pToken)); } - CHKiRet(term(pThis, tok)); + /* unget the token that made us exit the loop - it's obviously not one + * we can process. + */ + CHKiRet(ctok.UngetToken(tok, pToken)); finalize_it: RETiRet; @@ -229,10 +273,9 @@ e_and(expr_t *pThis, ctok_t *tok) CHKiRet(ctok.GetToken(tok, &pToken)); while(pToken->tok == ctok_AND) { dbgoprint((obj_t*) pThis, "and\n"); - /* fill structure */ - CHKiRet(ctok_token.Destruct(&pToken)); /* no longer needed */ CHKiRet(e_cmp(pThis, tok)); CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, opcode_AND, NULL)); /* add to program */ + CHKiRet(ctok_token.Destruct(&pToken)); /* no longer needed */ CHKiRet(ctok.GetToken(tok, &pToken)); } @@ -260,11 +303,10 @@ expr(expr_t *pThis, ctok_t *tok) /* *("or" e_and) part */ CHKiRet(ctok.GetToken(tok, &pToken)); while(pToken->tok == ctok_OR) { - /* fill structure */ dbgoprint((obj_t*) pThis, "found OR\n"); - CHKiRet(ctok_token.Destruct(&pToken)); /* no longer needed */ CHKiRet(e_and(pThis, tok)); CHKiRet(vmprg.AddVarOperation(pThis->pVmprg, opcode_OR, NULL)); /* add to program */ + CHKiRet(ctok_token.Destruct(&pToken)); /* no longer needed */ CHKiRet(ctok.GetToken(tok, &pToken)); } diff --git a/var.c b/var.c index e199a0ee..9947f159 100644 --- a/var.c +++ b/var.c @@ -78,6 +78,12 @@ CODESTARTobjDebugPrint(var) case VARTYPE_CSTR: dbgoprint((obj_t*) pThis, "type: cstr, val '%s'\n", rsCStrGetSzStr(pThis->val.vpCStr)); break; + case VARTYPE_INT64: + dbgoprint((obj_t*) pThis, "type: int64, val %lld\n", pThis->val.vInt64); + break; + case VARTYPE_INT: + dbgoprint((obj_t*) pThis, "type: int64, val %d\n", pThis->val.vInt); + break; default: dbgoprint((obj_t*) pThis, "type %d currently not suppored in debug output\n", pThis->varType); break; diff --git a/vmop.c b/vmop.c index 34d70fba..56112be5 100644 --- a/vmop.c +++ b/vmop.c @@ -168,7 +168,10 @@ vmopOpcode2Str(vmop_t *pThis, uchar **ppName) *ppName = (uchar*) "PUSHCONSTANT"; break; case opcode_POP: - *ppName = (uchar*) ""; + *ppName = (uchar*) "POP"; + break; + case opcode_UNARY_MINUS: + *ppName = (uchar*) "UNARY_MINUS"; break; default: *ppName = (uchar*) "INVALID opcode"; diff --git a/vmop.h b/vmop.h index dea54dbd..1ea531b3 100644 --- a/vmop.h +++ b/vmop.h @@ -51,6 +51,7 @@ typedef enum { /* do NOT start at 0 to detect uninitialized types after calloc( opcode_PUSHSYSVAR = 1001, /* requires var operand */ opcode_PUSHMSGVAR = 1002, /* requires var operand */ opcode_PUSHCONSTANT = 1003, /* requires var operand */ + opcode_UNARY_MINUS = 1010 } opcode_t; -- cgit