PRIVATE int HTRule_flush (HTStream * me)
{
if (me) {
char * flush = HTChunk_data(me->buffer);
if (flush) HTRule_parseLine(rules, flush);
HTChunk_clear(me->buffer);
}
return HT_OK;
}
/*
** Folding is either of CF LWS, LF LWS, CRLF LWS
*/
PRIVATE int HTRule_put_block (HTStream * me, const char * b, int l)
{
while (l > 0) {
if (me->EOLstate == EOL_FCR) {
if (*b == LF) /* CRLF */
me->EOLstate = EOL_FLF;
else if (isspace((int) *b)) /* Folding: CR SP */
me->EOLstate = EOL_DOT;
else { /* New line */
HTRule_parseLine(rules, HTChunk_data(me->buffer));
me->EOLstate = EOL_BEGIN;
HTChunk_clear(me->buffer);
continue;
}
} else if (me->EOLstate == EOL_FLF) {
if (isspace((int) *b)) /* Folding: LF SP or CR LF SP */
me->EOLstate = EOL_DOT;
else { /* New line */
HTRule_parseLine(rules, HTChunk_data(me->buffer));
me->EOLstate = EOL_BEGIN;
HTChunk_clear(me->buffer);
continue;
}
} else if (me->EOLstate == EOL_DOT) {
if (isspace((int) *b)) {
me->EOLstate = EOL_BEGIN;
HTChunk_putc(me->buffer, ' ');
} else {
HTRule_parseLine(rules, HTChunk_data(me->buffer));
me->EOLstate = EOL_BEGIN;
HTChunk_clear(me->buffer);
continue;
}
} else if (*b == CR) {
me->EOLstate = EOL_FCR;
} else if (*b == LF) {
me->EOLstate = EOL_FLF; /* Line found */
} else
HTChunk_putc(me->buffer, *b);
l--; b++;
}
return HT_OK;
}
/*
** Searches for HTTP Request Line before going into transparent mode
*/
PRIVATE int HTTPReceive_put_block (HTStream * me, const char * b, int l)
{
if (!me->transparent) {
const char *p=b;
while (l>0 && *p!=CR && *p!=LF) l--, p++;
HTChunk_putb(me->buffer, b, p-b);
if (*p==CR || *p==LF) {
int status = ParseRequest(me);
HTChunk_clear(me->buffer);
if (status != HT_OK) return status;
me->transparent = YES;
b=p;
}
}
if (l > 0) {
int status = PUTBLOCK(b, l);
if (status == HT_LOADED) me->transparent = NO;
return status;
}
return HT_OK;
}
PRIVATE int SGML_write (HTStream * context, const char * b, int l)
{
const SGML_dtd *dtd = context->dtd;
HTChunk *string = context->string;
const char *text = b;
int count = 0;
while (l-- > 0)
{
char c = *b++;
switch(context->state)
{
got_element_open:
/*
** The label is jumped when the '>' of a the element
** start tag has been detected. This DOES NOT FALL TO
** THE CODE S_after_open, only processes the tag and
** sets the state (c should still contain the
** terminating character of the tag ('>'))
*/
if (context->current_tag && context->current_tag->name)
start_element(context);
context->state = S_after_open;
break;
case S_after_open:
/*
** State S_after_open is entered only for single
** character after the element opening tag to test
** against newline. Strip one trainling newline only
** after opening nonempty element. - SGML: Ugh!
*/
text = b;
count = 0;
if (c == '\n' && (context->contents != SGML_EMPTY))
{
context->state = S_text;
break;
}
--text;
goto S_text;
S_text:
context->state = S_text;
case S_text:
#ifdef ISO_2022_JP
if (c == '\033')
{
context->state = S_esc;
++count;
break;
}
#endif /* ISO_2022_JP */
if (c == '&')
{
if (count > 0)
PUTB(text, count);
count = 0;
HTChunk_clear(string);
context->state = S_ero;
}
else if (c == '<')
{
if (count > 0)
PUTB(text, count);
count = 0;
HTChunk_clear(string);
/* should scrap LITERAL, and use CDATA and
RCDATA -- msa */
context->state =
(context->contents == SGML_LITERAL) ?
S_literal : S_tag;
}
else if (c == '\n')
/* Newline - ignore if before end tag! */
context->state = S_nl;
else
++count;
break;
case S_nl:
if (c == '<')
{
if (count > 0)
PUTB(text, count);
count = 0;
HTChunk_clear(string);
context->state =
(context->contents == SGML_LITERAL) ?
S_literal : S_nl_tago;
}
else
{
++count;
goto S_text;
}
break;
case S_nl_tago: /* Had newline and tag opener */
if (c != '/')
PUTC('\n'); /* Only ignore newline before </ */
context->state = S_tag;
goto handle_S_tag;
#ifdef ISO_2022_JP
case S_esc:
if (c=='$')
context->state = S_dollar;
else if (c=='(')
context->state = S_paren;
else
context->state = S_text;
++count;
break;
case S_dollar:
if (c=='@' || c=='B')
context->state = S_nonascii_text;
else
context->state = S_text;
++count;
break;
case S_paren:
if (c=='B' || c=='J')
context->state = S_text;
else
context->state = S_text;
++count;
break;
case S_nonascii_text:
if (c == '\033')
context->state = S_esc;
++count;
break;
#endif /* ISO_2022_JP */
/* In literal mode, waits only for specific end tag!
** Only foir compatibility with old servers.
*/
case S_literal:
HTChunk_putc(string, c);
if ( TOUPPER(c) !=
((HTChunk_size(string) == 1) ? '/'
: context->current_tag->name[HTChunk_size(string)-2]))
{
/* If complete match, end literal */
if ((c == '>') &&
(!context->current_tag->name[HTChunk_size(string)-2]))
{
end_element
(context,context->current_tag);
/*
...setting SGML_MIXED below is a
bit of kludge, but a good guess that
currently works, anything other than
SGML_LITERAL would work... -- msa */
context->contents = SGML_MIXED;
}
else
{
/* If Mismatch: recover string. */
PUTC( '<');
PUTB(HTChunk_data(string), HTChunk_size(string));
}
context->state = S_text;
text = b;
count = 0;
}
break;
/*
** Character reference or Entity
*/
case S_ero:
if (c == '#')
{
/* &# is Char Ref Open */
context->state = S_cro;
break;
}
context->state = S_entity;
/** FALL THROUGH TO S_entity !! ***/
/*
** Handle Entities
*/
case S_entity:
if (isalnum((int) c))
HTChunk_putc(string, c);
else
{
HTChunk_terminate(string);
handle_entity(context);
text = b;
count = 0;
if (c != ';')
{
--text;
goto S_text;
}
context->state = S_text;
}
break;
/* Character reference
*/
case S_cro:
if (isalnum((int)c))
/* accumulate a character NUMBER */
HTChunk_putc(string, c);
else
{
int value;
HTChunk_terminate(string);
if (sscanf(HTChunk_data(string), "%d", &value)==1)
PUTC((char)value);
else
{
PUTB("&#", 2);
PUTB(HTChunk_data(string), HTChunk_size(string)-1);
}
text = b;
count = 0;
if (c != ';')
{
--text;
goto S_text;
}
context->state = S_text;
}
break;
case S_tag: /* new tag */
handle_S_tag:
if (isalnum((int)c))
HTChunk_putc(string, c);
else { /* End of tag name */
int i;
if (c == '/') {
if (HTChunk_size(string) > 0)
HTTRACE(SGML_TRACE, "`<%s/' found!\n" _ HTChunk_data(string));
context->state = S_end;
break;
} else if (c == '!') {
if (HTChunk_size(string) > 0)
HTTRACE(SGML_TRACE, " `<%s!' found!\n" _ HTChunk_data(string));
context->state = S_md;
break;
}
HTChunk_terminate(string);
context->current_tag = SGMLFindTag(dtd, HTChunk_data(string));
if (context->current_tag == NULL) {
HTTRACE(SGML_TRACE, "*** Unknown element %s\n" _ HTChunk_data(string));
(*context->actions->unparsed_begin_element)
(context->target, HTChunk_data(string), HTChunk_size(string));
} else {
for (i=0; i<context->current_tag->number_of_attributes; i++) {
context->present[i] = NO;
context->value[i] = -1;
}
}
context->token = 0;
HTChunk_clear(string);
context->current_attribute_number = INVALID;
goto S_tag_gap;
}
break;
S_tag_gap:
context->state = S_tag_gap;
case S_tag_gap: /* Expecting attribute or > */
if (isspace((int) c))
break; /* Gap between attributes */
if (c == '>')
goto got_element_open;
else
goto S_attr;
S_attr:
/*
** Start collecting the attribute name and collect
** it in S_attr.
*/
context->state = S_attr;
HTChunk_truncate(string, context->token);
case S_attr:
if (isspace((int) c) || c == '>' || c == '=')
goto got_attribute_name;
else
HTChunk_putc(string, c);
break;
got_attribute_name:
/*
** This label is entered when attribute name has been
** collected. Process it and enter S_attr_gap for
** potential value or start of the next attribute.
*/
HTChunk_terminate(string) ;
handle_attribute_name
(context, HTChunk_data(string) + context->token);
HTChunk_truncate(string, context->token);
context->state = S_attr_gap;
case S_attr_gap: /* Expecting attribute or = or > */
if (isspace((int) c))
break; /* Gap after attribute */
if (c == '>')
goto got_element_open;
else if (c == '=')
context->state = S_equals;
else
goto S_attr; /* Get next attribute */
break;
case S_equals: /* After attr = */
if (isspace((int) c))
break; /* Before attribute value */
if (c == '>')
{ /* End of tag */
HTTRACE(SGML_TRACE, "found = but no value\n");
goto got_element_open;
}
else if (c == '\'')
context->state = S_squoted;
else if (c == '"')
context->state = S_dquoted;
else
goto S_value;
break;
S_value:
context->state = S_value;
HTChunk_truncate(string, context->token);
case S_value:
if (isspace((int) c) || c == '>')
{
HTChunk_terminate(string);
handle_attribute_value(context);
context->token = HTChunk_size(string);
goto S_tag_gap;
}
else
HTChunk_putc(string, c);
break;
case S_squoted: /* Quoted attribute value */
if (c == '\'')
{
HTChunk_terminate(string);
handle_attribute_value(context);
context->token = HTChunk_size(string);
context->state = S_tag_gap;
}
else if (c && c != '\n' && c != '\r')
HTChunk_putc(string, c);
break;
case S_dquoted: /* Quoted attribute value */
if (c == '"')
{
HTChunk_terminate(string);
handle_attribute_value(context);
context->token = HTChunk_size(string);
context->state = S_tag_gap;
}
else if (c && c != '\n' && c != '\r')
HTChunk_putc(string, c);
break;
case S_end: /* </ */
if (isalnum((int) c))
HTChunk_putc(string, c);
else
{ /* End of end tag name */
HTTag *t;
char * first;
HTChunk_terminate(string);
if ((first=HTChunk_data(string))!=NULL && *first != '\0')
t = SGMLFindTag(dtd, HTChunk_data(string));
else
/* Empty end tag */
/* Original code popped here one
from the stack. If this feature
is required, I have to put the
stack back... -- msa */
t = NULL;
if (!t) {
HTTRACE(SGML_TRACE, "Unknown end tag </%s>\n" _ HTChunk_data(string));
(*context->actions->unparsed_end_element)
(context->target, HTChunk_data(string), HTChunk_size(string));
} else {
context->current_tag = NULL;
end_element(context, t);
}
HTChunk_clear(string);
context->current_attribute_number = INVALID;
if (c != '>')
{
if (!isspace((int) c))
HTTRACE(SGML_TRACE, "`</%s%c' found!\n" _ HTChunk_data(string) _ c);
context->state = S_junk_tag;
}
else
{
text = b;
count = 0;
context->state = S_text;
}
}
break;
case S_junk_tag:
if (c == '>')
{
text = b;
count = 0;
context->state = S_text;
}
break;
/*
** Scanning (actually skipping) declarations
*/
case S_md:
if (c == '-')
context->state = S_com_1;
else if (c == '"')
context->state = S_md_dqs;
else if (c == '\'')
context->state = S_md_sqs;
else if (c == '>')
{
text = b;
count = 0;
context->state = S_text;
}
break;
case S_md_dqs: /* Skip double quoted string */
if (c == '"')
context->state = S_md;
else if (c == '>')
{
text = b;
count = 0;
context->state = S_text;
}
break;
case S_md_sqs: /* Skip single quoted string */
if (c == '\'')
context->state = S_md;
else if (c == '>')
{
text = b;
count = 0;
context->state = S_text;
}
break;
case S_com_1: /* Starting a comment? */
context->state = (c == '-') ? S_com : S_md;
if (c == '>')
{
text = b;
count = 0;
context->state = S_text;
}
break;
case S_com: /* ..within comment */
if (c == '-')
context->state = S_com_2;
break;
case S_com_2: /* Ending a comment ? */
context->state = (c == '-') ? S_com_2a : S_com;
break;
case S_com_2a:
if (c == '>') {
text = b;
count = 0;
context->state = S_text;
} else if (c == '-') {
context->state = S_com_2a;
} else
context->state = S_com;
break;
}
}
if (count > 0)
PUTB(text, count);
return HT_OK;
}
PUBLIC NowIn_t CSParse_targetParser(CSParse_t * pCSParse, char demark, void * pVoid)
{
/* ParseContext_t * pParseContext = pCSParse->pParseContext; */
TargetObject_t * pTargetObject = pCSParse->pTargetObject;
BOOL failedOnPunct = NO;
char * token = 0;
StateRet_t ret = StateRet_OK;
int i;
static NowIn_t lastRet = NowIn_END;
if (HTChunk_size(pCSParse->token)) {
HTChunk_terminate(pCSParse->token);
token = HTChunk_data(pCSParse->token);
}
for (i = 0; i < pTargetObject->stateTokenCount; i++) {
StateToken_t * pStateToken = pTargetObject->stateTokens + i;
pCSParse->pStateToken = pStateToken;
if (!(pCSParse->currentSubState & pStateToken->validSubStates))
continue;
if (pStateToken->pCheck) { /* use check function */
StateRet_t checkRes;
checkRes = (*pStateToken->pCheck)(pCSParse, pStateToken, token, demark);
switch (checkRes) {
case StateRet_WARN_BAD_PUNCT:
failedOnPunct = YES;
case StateRet_WARN_NO_MATCH:
continue;
case StateRet_ERROR_BAD_CHAR:
(*pCSParse->pParseContext->pParseErrorHandler)(pCSParse, token, demark, StateRet_ERROR_BAD_CHAR);
/* if (pTargetObject->pDestroy)
(*pTargetObject->pDestroy)(pCSParse); */
return NowIn_ERROR;
default:
break;
}
} else { /* or match by name[s] */
if (!(pStateToken->command & Command_MATCHANY)) {
if (token && pStateToken->name1) {
if (strcasecomp(token, pStateToken->name1) && (!pStateToken->name2 || strcasecomp(token, pStateToken->name2)))
continue;
} else {
if (token != pStateToken->name1)
continue;
}
}
if (Punct_badDemark(pStateToken->validPunctuation, demark)) {
failedOnPunct = YES;
continue;
}
}
/* open or close and do the appropriate callbacks */
if (lastRet != NowIn_CHAIN)
ParseTrace("%30s %c ", token ? token : "", demark);
ParseTrace("%10s - %s:%10s => ", pCSParse->pTargetObject->note, CSParse_subState2str(pCSParse->currentSubState), pStateToken->note);
if (pStateToken->command & Command_NOTOKEN) {
HTChunk_clear(pCSParse->token);
token = 0;
}
if (pStateToken->command & Command_OPEN && pTargetObject->pOpen)
if ((*pTargetObject->pOpen)(pCSParse, token, demark) == StateRet_ERROR)
return NowIn_ERROR;
if (pStateToken->command & (Command_OPEN|Command_CLOSE) && pCSParse->pParseContext->pTargetChangeCallback) {
ParseTrace("%3d", pStateToken->command & Command_CLOSE ? -pTargetObject->targetChange : pTargetObject->targetChange);
if ((*pCSParse->pParseContext->pTargetChangeCallback)(pCSParse, pTargetObject, pTargetObject->targetChange,
(BOOL)(pStateToken->command & Command_CLOSE), pVoid) == StateRet_ERROR)
return NowIn_ERROR;
} else
ParseTrace(" ");
if (pStateToken->command & Command_CLOSE && pTargetObject->pClose)
ret = (*pTargetObject->pClose)(pCSParse, token, demark);
if (pStateToken->pPrep && ret != NowIn_ERROR)
ret = (*pStateToken->pPrep)(pCSParse, token, demark);
if (pStateToken->pNextTargetObject)
pCSParse->pTargetObject = pStateToken->pNextTargetObject;
if (pStateToken->nextSubState != SubState_X)
pCSParse->currentSubState = pStateToken->nextSubState;
ParseTrace("%10s - %s", pCSParse->pTargetObject->note, CSParse_subState2str(pCSParse->currentSubState));
if (pStateToken->command & Command_CHAIN) {
ParseTrace(" -O-O-");
return lastRet = NowIn_CHAIN;
}
ParseTrace("\n");
return lastRet = ret == StateRet_ERROR_BAD_CHAR ? NowIn_ERROR : ret == StateRet_DONE ? NowIn_END : NowIn_ENGINE;
}
(*pCSParse->pParseContext->pParseErrorHandler)(pCSParse, token, demark, failedOnPunct ? StateRet_WARN_BAD_PUNCT : StateRet_WARN_NO_MATCH);
if (pTargetObject->pDestroy)
(*pTargetObject->pDestroy)(pCSParse);
return NowIn_ERROR;
}
/* CSParse_parseChunk - elemental parse engine for all pics nowIns. This passes
* tokenized data into the handler functions in the CSParse_t.handlersOf. These
* handlers are responsibel for placing the data in the appropriate target.
* The text is broken into nowIns and passed a SubParser based on the current
* nowIn which is one of:
* NowIn_NEEDOPEN - get paren and go to NowIn_ENGINE, text is an error
* NowIn_ENGINE - in a containing structure, text goes to engineOf_
* NowIn_NEEDCLOSE - get paren and go to NowIn_ENGINE, text is an error
* NowIn_END - expect no more text or parens
* NowIn_ERROR -
*/
PUBLIC CSDoMore_t CSParse_parseChunk (CSParse_t * pCSParse, const char * ptr, int len, void * pVoid)
{
int i;
if (!len || !ptr)
return CSDoMore_error;
for (i = 0; i < len; i++) {
pCSParse->offset++;
if (pCSParse->quoteState) {
if (pCSParse->quoteState == ptr[i]) {
pCSParse->quoteState = 0;
pCSParse->demark = ' ';
}
else
HTChunk_putb(pCSParse->token, ptr+i, 1);
continue;
}
if (ptr[i] == SQUOTE || ptr[i] == DQUOTE) {
if (pCSParse->demark) {
while ((pCSParse->nowIn = (*pCSParse->pParseContext->engineOf)(pCSParse, ' ', pVoid)) == NowIn_CHAIN);
HTChunk_clear(pCSParse->token);
pCSParse->demark = 0;
} else if (HTChunk_size(pCSParse->token) &&
/* && warn(pCSParse, message_UNEXPECTED_CHARACTER, ptr[i])) */
callErrorHandler(pCSParse, ptr+i, ptr[i],
StateRet_ERROR_BAD_CHAR) !=StateRet_OK)
pCSParse->nowIn = NowIn_ERROR;
pCSParse->quoteState = ptr[i];
pCSParse->pParseContext->observedQuotes = YES;
continue;
}
switch (pCSParse->nowIn) {
case NowIn_NEEDOPEN:
if (ptr[i] == LPAREN) {
pCSParse->nowIn = NowIn_ENGINE;
continue;
}
if (isspace((int) ptr[i]))
continue;
/* if (warn(pCSParse, message_UNEXPECTED_CHARACTER, ptr[i])) pCSParse->nowIn = NowIn_ERROR; */
if (callErrorHandler(pCSParse, ptr+i, ptr[i],
StateRet_ERROR_BAD_CHAR) !=StateRet_OK)
pCSParse->nowIn = NowIn_ERROR;
continue;
case NowIn_ENGINE:
if (isspace((int) ptr[i])) {
if (HTChunk_size(pCSParse->token))
pCSParse->demark = ' ';
continue;
}
if (ptr[i] == LPAREN || ptr[i] == RPAREN || pCSParse->demark) {
/* parens override space demarkation */
if (ptr[i] == LPAREN) pCSParse->demark = LPAREN;
if (ptr[i] == RPAREN) pCSParse->demark = RPAREN;
/* call the engine as long as it wants re-entrance */
while ((pCSParse->nowIn = (*pCSParse->pParseContext->engineOf)(pCSParse, pCSParse->demark, pVoid)) == NowIn_CHAIN);
HTChunk_clear(pCSParse->token);
pCSParse->demark = 0;
if (ptr[i] == LPAREN || ptr[i] == RPAREN)
continue;
/* continue with next token */
}
HTChunk_putb(pCSParse->token, ptr+i, 1);
continue;
case NowIn_NEEDCLOSE:
if (ptr[i] == RPAREN) {
pCSParse->nowIn = NowIn_ENGINE;
continue;
}
if (isspace((int) ptr[i]))
continue;
if (callErrorHandler(pCSParse, ptr+i, ptr[i],
StateRet_ERROR_BAD_CHAR) !=StateRet_OK)
pCSParse->nowIn = NowIn_ERROR;
/* if (warn(pCSParse, message_UNEXPECTED_CHARACTER, ptr[i])) pCSParse->nowIn = NowIn_ERROR; */
continue;
case NowIn_END:
#if 0 /* enable this to tell the parser to check the remainder of
the stream after the parsed object thinks it is done */
if (isspace(ptr[i]))
continue;
/* if (warn(pCSParse, message_UNEXPECTED_CHARACTER, ptr[i])) pCSParse->nowIn = NowIn_ERROR; */
if (callErrorHandler(pCSParse, ptr+i, ptr[i],
StateRet_ERROR_BAD_CHAR) !=StateRet_OK)
pCSParse->nowIn = NowIn_ERROR;
continue;
#else
return CSDoMore_done;
#endif
case NowIn_MATCHCLOSE:
if (ptr[i] == RPAREN) {
if (!pCSParse->depth)
pCSParse->nowIn = NowIn_ENGINE;
else
pCSParse->depth--;
}
if (ptr[i] == LPAREN)
pCSParse->depth++;
continue;
case NowIn_ERROR:
return CSDoMore_error;
break;
default:
HTTRACE(PICS_TRACE, "PICS: Internal error in parser - bad nowIn:%d.\n" _
pCSParse->nowIn);
return CSDoMore_error;
}
}
/* check completion */
return pCSParse->nowIn == NowIn_END ? CSDoMore_done : CSDoMore_more;
}
