int32
PA_TagIndex(char *str)
{
return((int32)pa_tokenize_tag(str));
}
PRIVATE int net_ColorHTMLWrite (NET_StreamClass *stream, CONST char *s, int32 l)
{
int32 i;
int32 last_output_point;
char *new_markup=0;
char *tmp_markup=0;
char tiny_buf[4];
CONST char *cp;
int status;
DataObject *obj=stream->data_object;
last_output_point = 0;
for(i = 0, cp = s; i < l; i++, cp++)
{
switch(obj->state)
{
case IN_CONTENT:
/* do nothing until you find a '<' "<!--" or '&' */
if(*cp == '<')
{
/* XXX we can miss a comment spanning a block boundary */
if(i+4 <= l && !XP_STRNCMP(cp, "<!--", 4))
{
StrAllocCopy(new_markup, BEGIN_COMMENT_MARKUP);
StrAllocCat(new_markup, "<");
obj->state = IN_COMMENT;
}
else
{
new_markup = net_BeginColorHTMLTag(obj);
}
}
else if(*cp == '&')
{
StrAllocCopy(new_markup, BEGIN_AMPERSAND_THINGY_MARKUP);
StrAllocCat(new_markup, "&");
obj->state = IN_AMPERSAND_THINGY;
}
break;
case IN_SCRIPT:
/* do nothing until you find '</SCRIPT>' */
if(*cp == '<')
{
/* XXX we can miss a </SCRIPT> spanning a block boundary */
if(i+8 <= l && !XP_STRNCASECMP(cp, "</SCRIPT", 8))
{
new_markup = net_BeginColorHTMLTag(obj);
}
}
break;
case ABOUT_TO_BEGIN_TAG:
/* we have seen the first '<'
* once we see a non-whitespace character
* we will be in the tag identifier
*/
if(*cp == '>')
{
StrAllocCopy(new_markup, END_TAG_NAME_MARKUP);
tmp_markup = net_EndColorHTMLTag(obj);
StrAllocCat(new_markup, tmp_markup);
FREE_AND_CLEAR(tmp_markup);
}
else if(!XP_IS_SPACE(*cp))
{
obj->state = IN_BEGIN_TAG;
obj->tag_index = 0;
obj->tag[obj->tag_index++] = *cp;
if(*cp == '<')
StrAllocCopy(new_markup, "<");
}
break;
case IN_BEGIN_TAG:
/* go to the IN_TAG state when we see
* the first whitespace
*/
if(XP_IS_SPACE(*cp))
{
StrAllocCopy(new_markup, END_TAG_NAME_MARKUP);
XP_SPRINTF(tiny_buf, "%c", *cp);
StrAllocCat(new_markup, tiny_buf);
obj->state = IN_TAG;
obj->tag[obj->tag_index] = '\0';
obj->tag_type = pa_tokenize_tag(obj->tag);
}
else if(*cp == '>')
{
StrAllocCopy(new_markup, END_TAG_NAME_MARKUP);
tmp_markup = net_EndColorHTMLTag(obj);
StrAllocCat(new_markup, tmp_markup);
FREE_AND_CLEAR(tmp_markup);
}
else if(*cp == '<')
{
/* protect ourselves from markup */
if(!obj->in_broken_html)
{
obj->in_broken_html = TRUE;
StrAllocCopy(new_markup, BEGIN_BROKEN_ATTRIBUTE_MARKUP);
StrAllocCat(new_markup, "<");
}
else
{
StrAllocCopy(new_markup, "<");
}
}
else
{
if (obj->tag_index < MAXTAGLEN)
obj->tag[obj->tag_index++] = *cp;
}
break;
case IN_TAG:
/* do nothing until you find a opening '=' or end '>' */
if(*cp == '=')
{
StrAllocCopy(new_markup, "=");
StrAllocCat(new_markup, BEGIN_ATTRIBUTE_VALUE_MARKUP);
obj->state = BEGIN_ATTRIBUTE_VALUE;
}
else if(*cp == '>')
{
new_markup = net_EndColorHTMLTag(obj);
}
else if(*cp == '<')
{
/* protect ourselves from markup */
StrAllocCopy(new_markup, "<");
}
break;
case BEGIN_ATTRIBUTE_VALUE:
/* when we reach the first non-whitespace
* we will enter the UNQUOTED or the QUOTED
* ATTRIBUTE state
*/
if(!XP_IS_SPACE(*cp))
{
if(*cp == '"')
{
obj->state = IN_QUOTED_ATTRIBUTE_VALUE;
/* no need to jump to the quoted attr handler
* since this char can't be a dangerous char
*/
}
else
{
obj->state = IN_UNQUOTED_ATTRIBUTE_VALUE;
/* need to jump to the unquoted attr handler
* since this char can be a dangerous character
*/
goto unquoted_attribute_jump_point;
}
}
else if(*cp == '>')
{
StrAllocCopy(new_markup, END_ATTRIBUTE_VALUE_MARKUP);
tmp_markup = net_EndColorHTMLTag(obj);
StrAllocCat(new_markup, tmp_markup);
FREE_AND_CLEAR(tmp_markup);
}
else if(*cp == '<')
{
/* protect ourselves from markup */
StrAllocCopy(new_markup, "<");
}
break;
case IN_UNQUOTED_ATTRIBUTE_VALUE:
unquoted_attribute_jump_point:
/* do nothing until you find a whitespace */
if(XP_IS_SPACE(*cp))
{
StrAllocCopy(new_markup, END_ATTRIBUTE_VALUE_MARKUP);
XP_SPRINTF(tiny_buf, "%c", *cp);
StrAllocCat(new_markup, tiny_buf);
obj->state = IN_TAG;
}
else if(*cp == '>')
{
StrAllocCopy(new_markup, END_ATTRIBUTE_VALUE_MARKUP);
tmp_markup = net_EndColorHTMLTag(obj);
StrAllocCat(new_markup, tmp_markup);
FREE_AND_CLEAR(tmp_markup);
}
else if(*cp == '<')
{
/* protect ourselves from markup */
StrAllocCopy(new_markup, "<");
}
else if(*cp == '&')
{
/* protect ourselves from markup */
StrAllocCopy(new_markup, "&");
}
break;
case IN_QUOTED_ATTRIBUTE_VALUE:
/* do nothing until you find a closing '"' */
if(*cp == '\"')
{
if(obj->in_broken_html)
{
StrAllocCopy(new_markup, END_BROKEN_ATTRIBUTE_MARKUP);
obj->in_broken_html = FALSE;
}
StrAllocCat(new_markup, "\"");
StrAllocCat(new_markup, END_ATTRIBUTE_VALUE_MARKUP);
obj->state = IN_TAG;
}
else if(*cp == '<')
{
/* protect ourselves from markup */
StrAllocCopy(new_markup, "<");
}
else if(*cp == '&')
{
/* protect ourselves from markup */
StrAllocCopy(new_markup, "&");
}
else if(*cp == '>')
{
/* probably a broken attribute value */
if(!obj->in_broken_html)
{
obj->in_broken_html = TRUE;
StrAllocCopy(new_markup, BEGIN_BROKEN_ATTRIBUTE_MARKUP);
StrAllocCat(new_markup, ">");
}
}
break;
case IN_COMMENT:
/* do nothing until you find a closing '-->' */
if(!XP_STRNCMP(cp, "-->", 3))
{
StrAllocCopy(new_markup, ">");
cp += 2;
i += 2;
StrAllocCat(new_markup, END_COMMENT_MARKUP);
obj->state = IN_CONTENT;
}
else if(*cp == '<')
{
/* protect ourselves from markup */
StrAllocCopy(new_markup, "<");
}
break;
case IN_AMPERSAND_THINGY:
/* do nothing until you find a ';' or space */
if(*cp == ';' || XP_IS_SPACE(*cp))
{
XP_SPRINTF(tiny_buf, "%c", *cp);
StrAllocCopy(new_markup, tiny_buf);
StrAllocCat(new_markup, END_AMPERSAND_THINGY_MARKUP);
obj->state = IN_CONTENT;
}
else if(*cp == '<')
{
/* protect ourselves from markup */
StrAllocCopy(new_markup, "<");
}
break;
default:
XP_ASSERT(0);
break;
}
if(new_markup)
{
/* push all the way up to but not including *cp */
status = (*obj->next_stream->put_block)
(obj->next_stream,
&s[last_output_point],
i-last_output_point);
last_output_point = i+1;
if(status < 0)
{
FREE(new_markup);
return(status);
}
/* add new markup */
status = (*obj->next_stream->put_block)
(obj->next_stream,
new_markup, XP_STRLEN(new_markup));
if(status < 0)
{
FREE(new_markup);
return(status);
}
FREE_AND_CLEAR(new_markup);
}
}
if(last_output_point < l)
return((*obj->next_stream->put_block)(obj->next_stream,
&s[last_output_point],
(l-last_output_point)));
else
return(0);
}
