/* Tokenize a header. tokens may be NULL, in which case the number of tokens
are counted, allowing the caller to allocate enough room */
static void tokenize(const char *header, php_rfc822_token_t *tokens,
int *ntokens, int report_errors) {
register const char *p, *q, *start;
int in_bracket = 0;
/* NB: parser assumes that the header has two bytes of NUL terminator */
YYCURSOR = header;
YYLIMIT = YYCURSOR + strlen(YYCURSOR) + 1;
*ntokens = 0;
state_ground:
start = YYCURSOR;
#if DEBUG_RFC822_SCANNER
printf("ground: start=%p limit=%p cursor=%p: [%d] %s\n",
start, YYLIMIT, YYCURSOR, *YYCURSOR, YYCURSOR);
#endif
{
YYCTYPE yych;
static const unsigned char yybm[] = {
0, 192, 192, 192, 192, 192, 192, 192,
192, 96, 96, 192, 192, 96, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
96, 64, 0, 192, 192, 64, 192, 192,
64, 64, 192, 192, 64, 192, 64, 64,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 64, 64, 64, 64, 64, 64,
64, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 64, 192, 64, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
192, 192, 192, 192, 192, 192, 192, 192,
};
if ((YYLIMIT - YYCURSOR) < 2) YYFILL(2);
yych = *YYCURSOR;
if (yybm[0+yych] & 32) {
goto yy4;
}
if (yych <= '-') {
if (yych <= '%') {
if (yych <= '!') {
if (yych <= 0x00) goto yy2;
if (yych <= ' ') goto yy21;
goto yy19;
} else {
if (yych <= '"') goto yy12;
if (yych <= '$') goto yy21;
goto yy19;
}
} else {
if (yych <= ')') {
if (yych <= '\'') goto yy21;
if (yych <= '(') goto yy10;
goto yy7;
} else {
if (yych == ',') goto yy19;
goto yy21;
}
}
} else {
if (yych <= '>') {
if (yych <= ';') {
if (yych <= '/') goto yy19;
if (yych <= '9') goto yy21;
goto yy19;
} else {
if (yych <= '<') goto yy15;
if (yych <= '=') goto yy19;
goto yy17;
}
} else {
if (yych <= '[') {
if (yych <= '@') goto yy19;
if (yych <= 'Z') goto yy21;
goto yy19;
} else {
if (yych <= '\\') goto yy9;
if (yych <= ']') goto yy19;
goto yy21;
}
}
}
yy2:
++YYCURSOR;
{ goto stop; }
yy4:
++YYCURSOR;
if (YYLIMIT <= YYCURSOR) YYFILL(1);
yych = *YYCURSOR;
if (yybm[0+yych] & 32) {
goto yy4;
}
{ DBG_STATE("SPACE"); goto state_ground; }
yy7:
++YYCURSOR;
yy8:
{ REPORT_ERR("token not valid in ground state"); goto state_ground; }
yy9:
yych = *++YYCURSOR;
if (yybm[0+yych] & 128) {
goto yy21;
}
goto yy8;
yy10:
++YYCURSOR;
{ DBG_STATE("START COMMENT");
if (tokens) {
tokens->token = '(';
tokens->value = start;
tokens->valuelen = 0;
}
goto state_comment;
}
yy12:
++YYCURSOR;
if (YYLIMIT <= YYCURSOR) YYFILL(1);
yych = *YYCURSOR;
if (yybm[0+yych] & 64) {
goto yy12;
}
if (yych >= 0x01) goto yy26;
yy15:
++YYCURSOR;
if ((yych = *YYCURSOR) == '>') goto yy24;
{ DBG_STATE("LANGLE");
if (in_bracket) {
REPORT_ERR("already in < bracket");
goto state_ground;
}
in_bracket = 1;
ADD_ATOM_TOKEN();
goto state_ground;
}
yy17:
++YYCURSOR;
{ DBG_STATE("RANGLE");
if (!in_bracket) {
REPORT_ERR("not in < bracket");
goto state_ground;
}
in_bracket = 0;
ADD_ATOM_TOKEN();
goto state_ground;
}
yy19:
++YYCURSOR;
{ DBG_STATE("ATOM"); ADD_ATOM_TOKEN(); goto state_ground; }
yy21:
++YYCURSOR;
if (YYLIMIT <= YYCURSOR) YYFILL(1);
yych = *YYCURSOR;
if (yybm[0+yych] & 128) {
goto yy21;
}
{ DBG_STATE("ANY");
if (tokens) {
tokens->token = 0;
tokens->valuelen = YYCURSOR - start;
tokens->value = start;
tokens++;
}
++*ntokens;
goto state_ground;
}
yy24:
++YYCURSOR;
{ DBG_STATE("NULL <>");
ADD_ATOM_TOKEN();
if (tokens) {
tokens->token = 0;
tokens->value = "";
tokens->valuelen = 0;
tokens++;
}
++*ntokens;
start++;
ADD_ATOM_TOKEN();
goto state_ground;
}
yy26:
++YYCURSOR;
{ DBG_STATE("QUOTE STRING");
if (tokens) {
tokens->token = '"';
tokens->value = start + 1;
tokens->valuelen = YYCURSOR - start - 2;
tokens++;
}
++*ntokens;
goto state_ground;
}
}
state_comment:
{
int comment_depth = 1;
while (1) {
if (*YYCURSOR == 0) {
/* unexpected end of header */
REPORT_ERR("unexpected end of header");
/* fake a quoted string for this last token */
if (tokens)
tokens->token = '"';
++*ntokens;
return;
} else if (*YYCURSOR == '(') {
comment_depth++;
} else if (*YYCURSOR == ')' && --comment_depth == 0) {
/* end of nested comment sequence */
YYCURSOR++;
if (tokens)
tokens->valuelen++;
break;
} else if (*YYCURSOR == '\\' && YYCURSOR[1]) {
YYCURSOR++;
if (tokens)
tokens->valuelen++;
}
YYCURSOR++;
}
if (tokens) {
tokens->valuelen = YYCURSOR - tokens->value;
tokens++;
}
++*ntokens;
goto state_ground;
}
stop:
#if DEBUG_RFC822_SCANNER
printf("STOPing parser ntokens=%d YYCURSOR=%p YYLIMIT=%p start=%p "
"cursor=[%d] %s start=%s\n", *ntokens,
YYCURSOR, YYLIMIT, start, *YYCURSOR, YYCURSOR, start);
#else
;
#endif
}
int STM_Execute1( STM_StateMachine *sm, STM_StateMachine **currSm )
{
int action; /* Action to execute. */
STM_Input input; /* Input being processed. */
int nextState; /* Next state to change to. */
uint8_t *stateData; /* Ptr to state data for current state. */
STM_StateMachine *svSm; /* Saved value of STM_CurrStateMachine */
ASSERT( sm != NULL );
ASSERT( sm->magic == STM_MAGIC );
if (( input = GetInput( sm )) == STM_INPUT_NONE )
{
/*
* No input, indicate that we didn't do anything
*/
return FALSE;
}
/*
* Set the global state machine variable, so that actions which use it
* can access this state machine. Save the old value off so that action
* routines can execute other state machines.
*/
if ( currSm != NULL )
{
svSm = *currSm;
*currSm = sm;
}
/*
* We've got an input, run it through the state machine
*/
DBG_INPUT( "STM %s: processing input %s (%d)\n",
sm->name, sm->currInputStr, sm->currInput );
stateData = &sm->stateData[ sm->currState * STM_BytesPerState( sm )];
nextState = stateData[ STM_NextStateOffset( input )];
/*
* Exit the old state
*/
if ( nextState != sm->currState )
{
int exitAction;
/*
* We're leaving the old state, call the exit routine if one is
* defined
*/
exitAction = stateData[ STM_ExitOffset() ];
ExecuteAction( sm, exitAction, "exit " );
}
/*
* Call the action routine
*/
action = stateData[ STM_ActionOffset( input )];
ExecuteAction( sm, action, "" );
if ( nextState != sm->currState )
{
int enterAction;
/*
* Enter the new state
*/
DBG_STATE( "STM %s: Changing states from %s (%d) to %s (%d)\n",
sm->name,
sm->stateStr[ sm->currState ], sm->currState,
sm->stateStr[ nextState ], nextState );
sm->currState = nextState;
stateData = &sm->stateData[ sm->currState * STM_BytesPerState( sm )];
/*
* We're entering the new state, call the enter routine if one is
* defined
*/
enterAction = stateData[ STM_EnterOffset() ];
ExecuteAction( sm, enterAction, "enter " );
}
if ( currSm != NULL )
{
*currSm = svSm;
}
/*
* We processed an input, tell the caller
*/
sm->currInput = STM_INPUT_NONE;
#if defined( STM_DEBUG )
sm->currInputStr = "- None -";
#endif
return TRUE;
} // STM_Execute1
