// ----------------------------------------------------------------------
void SAXInterruptibleReader::
parse()
throw(runtime_error)
{
char buf[16384];
int len;
stop_flag_ = false;
if(!initialized_)
open_file();
while(!cache_.empty() && !stop_flag_)
{
struct CacheData* cd = cache_.front();
switch(cd->tag_type)
{
case SIR_OPENTAG:
start_element(cd->name, cd->atts);
break;
case SIR_TEXTTAG:
text_element(cd->name);
break;
case SIR_CLOSETAG:
end_element(cd->name);
break;
}
cache_.pop_front();
delete cd;
}
//Read the file until the end of file
while( !is_->eof() && !stop_flag_ )
{
is_->read( buf, sizeof(buf) );
len = is_->gcount();
if (XML_Parse(parser, buf, len, is_->eof()) == XML_STATUS_ERROR)
{
std::cerr << XML_ErrorString(XML_GetErrorCode(parser)) << "at line " << XML_GetCurrentLineNumber(parser) << std::endl;
reset();
throw std::runtime_error("Error in parsing XML input");
}
}
//Done -> Close the file and free all associated memory
if( !stop_flag_ )
{
parsing_done();
reset();
}
}
// ----------------------------------------------------------------------
void SAXInterruptibleReader::
handle_start_element(string name, AttList& atts)
throw(runtime_error)
{
string _name = string(name);
if( stop_flag_ )
{
CacheData* cd = new CacheData;
cd->tag_type = SIR_OPENTAG;
cd->name = _name ;
cd->atts = atts;
cache_.push_back(cd);
}
else
{
start_element(_name, atts);
}
}
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;
}
static void raptor_rdfa_start_element(void *user_data,
raptor_xml_element *xml_element)
{
raptor_qname* qname = raptor_xml_element_get_name(xml_element);
int nb_attributes = raptor_xml_element_get_attributes_count(xml_element);
raptor_qname** attrs = raptor_xml_element_get_attributes(xml_element);
unsigned char* localname = raptor_qname_to_counted_name(qname, NULL);
const raptor_namespace* qname_ns = raptor_qname_get_namespace(qname);
int nb_namespaces = 0;
const char** namespaces = NULL;
int nb_defaulted = 0;
char** attr = NULL;
int i;
const char* ns_name = NULL;
const char* ns_uri = NULL;
if(nb_attributes > 0) {
/* Everything written into 'attr' is a shared pointer into
* xml_element or contained objects - qnames, namespaces, uris
* and values
*/
attr = (char**)malloc(sizeof(char*) * (1 + (nb_attributes * 5)));
for(i = 0; i < nb_attributes; i++) {
const raptor_namespace* attr_ns = attrs[i]->nspace;
char** attri = &attr[5 * i];
/* 5 tuple: (localname, prefix, URI, value, end) */
attri[0] = (char*)attrs[i]->local_name;
attri[1] = attr_ns ? (char*)attr_ns->prefix : NULL;
attri[2] = attr_ns ? (char*)raptor_uri_as_string(attr_ns->uri) : NULL;
attri[3] = (char*)attrs[i]->value;
attri[4] = attri[3] + attrs[i]->value_length;
}
attr[5 * i] = NULL;
}
/*
* @ctx: the user data (XML parser context)
* @localname: the local name of the element
* @prefix: the element namespace prefix if available
* @URI: the element namespace name if available
* @nb_namespaces: number of namespace definitions on that node
* @namespaces: pointer to the array of prefix/URI pairs namespace definitions
* @nb_attributes: the number of attributes on that node
* @nb_defaulted: the number of defaulted attributes. The defaulted
* ones are at the end of the array
* @attributes: pointer to the array of (localname/prefix/URI/value/end)
* attribute values.
*/
if(qname_ns) {
ns_name = (const char*)raptor_namespace_get_prefix(qname_ns);
ns_uri = (const char*)raptor_uri_as_string(qname_ns->uri);
}
start_element(user_data, (const char*)localname,
ns_name,
ns_uri,
nb_namespaces,
(const char**)namespaces,
nb_attributes,
nb_defaulted,
(const char**)attr);
if(attr)
free(attr);
raptor_free_memory(localname);
}
