static int
raptor_librdfa_parse_start(raptor_parser* rdf_parser)
{
raptor_locator *locator=&rdf_parser->locator;
raptor_librdfa_parser_context *librdfa_parser=(raptor_librdfa_parser_context*)rdf_parser->context;
int rc;
char* base_uri_string=NULL;
locator->line=1;
locator->column=0;
locator->byte=0;
if(rdf_parser->base_uri)
base_uri_string=(char*)raptor_uri_as_string(rdf_parser->base_uri);
if(librdfa_parser->context)
rdfa_free_context(librdfa_parser->context);
librdfa_parser->context=rdfa_create_context(base_uri_string);
if(!librdfa_parser->context)
return 1;
librdfa_parser->context->namespace_handler=raptor_librdfa_sax2_new_namespace_handler;
librdfa_parser->context->namespace_handler_user_data=rdf_parser;
librdfa_parser->context->error_handlers=&rdf_parser->error_handlers;
librdfa_parser->context->callback_data=rdf_parser;
rdfa_set_triple_handler(librdfa_parser->context,
raptor_librdfa_generate_statement);
rc = rdfa_parse_start(librdfa_parser->context);
if(rc != RDFA_PARSE_SUCCESS)
return 1;
return 0;
}
int main(int argc, char** argv)
{
printf("Speed test...\n");
size_t bytes;
clock_t stime = clock();
rdfacontext* context = rdfa_create_context("http://example.org/speed");
rdfa_set_triple_handler(context, &process_triple);
rdfa_parse_start(context);
while(g_iteration <= MAX_ITERATIONS)
{
bytes = fill_buffer(context);
rdfa_parse_buffer(context, bytes);
}
rdfa_parse_buffer(context, 0);
rdfa_parse_end(context);
rdfa_free_context(context);
clock_t etime = clock();
float delta = etime - stime;
printf("Processed %1.2f triples per second from %lli bytes of data.\n",
(MAX_ITERATIONS / (delta / CLOCKS_PER_SEC)), g_bytes_processed);
return 0;
}
int main(int argc, char** argv)
{
if(argc < 2)
{
printf("%s usage:\n\n"
"%s <input.xhtml>\n", argv[0], argv[0]);
}
else
{
FILE* xhtml_file = fopen(argv[1], "r");
char* filename = rindex(argv[1], '/');
filename++;
if(xhtml_file != NULL)
{
char* base_uri = rdfa_join_string(BASE_URI, filename);
rdfacontext* context = rdfa_create_context(base_uri);
context->callback_data = xhtml_file;
rdfa_set_triple_handler(context, &process_triple);
rdfa_set_buffer_filler(context, &fill_buffer);
rdfa_parse(context);
rdfa_free_context(context);
fclose(xhtml_file);
free(base_uri);
}
else
{
perror("failed to open file:");
}
}
return 0;
}
static void
raptor_librdfa_parse_terminate(raptor_parser* rdf_parser)
{
raptor_librdfa_parser_context *librdfa_parser=(raptor_librdfa_parser_context*)rdf_parser->context;
if(librdfa_parser->context) {
rdfa_parse_end(librdfa_parser->context);
rdfa_free_context(librdfa_parser->context);
librdfa_parser->context=NULL;
}
}
static int
raptor_librdfa_parse_start(raptor_parser* rdf_parser)
{
raptor_locator *locator = &rdf_parser->locator;
raptor_librdfa_parser_context *librdfa_parser;
int rc;
char* base_uri_string = NULL;
librdfa_parser = (raptor_librdfa_parser_context*)rdf_parser->context;
locator->line = -1;
locator->column = -1;
locator->byte = 0;
if(rdf_parser->base_uri)
base_uri_string = (char*)raptor_uri_as_string(rdf_parser->base_uri);
else
/* base URI is required for rdfa - checked in rdfa_create_context() */
return 1;
if(librdfa_parser->context)
rdfa_free_context(librdfa_parser->context);
librdfa_parser->context = rdfa_create_context(base_uri_string);
if(!librdfa_parser->context)
return 1;
librdfa_parser->context->namespace_handler = raptor_librdfa_sax2_new_namespace_handler;
librdfa_parser->context->namespace_handler_user_data = rdf_parser;
librdfa_parser->context->world = rdf_parser->world;
librdfa_parser->context->locator = &rdf_parser->locator;
librdfa_parser->context->callback_data = rdf_parser;
/* returns triples */
rdfa_set_default_graph_triple_handler(librdfa_parser->context,
raptor_librdfa_generate_statement);
/* returns RDFa Processing Graph error triples - not used by raptor */
rdfa_set_processor_graph_triple_handler(librdfa_parser->context, NULL);
librdfa_parser->context->raptor_rdfa_version = librdfa_parser->rdfa_version;
rc = rdfa_parse_start(librdfa_parser->context);
if(rc != RDFA_PARSE_SUCCESS)
return 1;
return 0;
}
int main(int argc, char** argv)
{
printf("Speed test...\n");
clock_t stime = clock();
rdfacontext* g_context = rdfa_create_context("http://example.org/speed");
rdfa_set_default_graph_triple_handler(g_context, &process_triple);
rdfa_set_buffer_filler(g_context, &fill_buffer);
rdfa_parse(g_context);
rdfa_free_context(g_context);
clock_t etime = clock();
float delta = etime - stime;
printf("Processed %1.2f triples per second from %lli bytes of data.\n",
(MAX_ITERATIONS / (delta / CLOCKS_PER_SEC)), g_bytes_processed);
return 0;
}
int main(int argc, char** argv)
{
#ifdef LIBRDFA_IN_RAPTOR
raptor_init();
#endif
if(argc < 2)
{
printf("%s usage:\n\n"
"%s <input.xhtml>\n", argv[0], argv[0]);
}
else
{
char* filename;
g_xhtml_file = fopen(argv[1], "r");
filename = rindex(argv[1], '/');
if(filename == NULL)
{
filename = argv[1];
}
else
{
filename++;
}
if(g_xhtml_file != NULL)
{
unsigned int buffer_length = 65535;
char* buffer = malloc(buffer_length);
char* base_uri = rdfa_join_string(BASE_URI, filename);
rdfacontext* context = rdfa_create_context(base_uri);
buffer_status* status = malloc(sizeof(buffer_status));
// get all of the buffer text
fread(buffer, sizeof(char), buffer_length, g_xhtml_file);
fclose(g_xhtml_file);
// initialize the callback data
status->buffer = buffer;
status->current_offset = 0;
status->total_length = strlen(buffer);
context->callback_data = status;
// setup the parser
rdfa_set_default_graph_triple_handler(
context, &default_graph_triple);
rdfa_set_processor_graph_triple_handler(
context, &processor_graph_triple);
rdfa_set_buffer_filler(context, &fill_buffer);
rdfa_parse(context);
rdfa_free_context(context);
free(base_uri);
}
else
{
perror("failed to open file:");
}
}
#ifdef LIBRDFA_IN_RAPTOR
raptor_finish();
#endif
return 0;
}
void run_curie_tests()
{
rdfacontext* context =
rdfa_create_context("http://example.org/");
rdfa_init_context(context);
rdfa_update_mapping(
context->uri_mappings, "dc", "http://purl.org/dc/elements/1.1/");
rdfa_update_mapping(
context->uri_mappings, "dctv", "http://purl.org/dc/dcmitype/");
printf("------------------------ CURIE tests ---------------------\n");
run_test(context, "IRI", "http://www.example.org/iri",
rdfa_resolve_curie, "http://www.example.org/iri",
CURIE_PARSE_HREF_SRC);
run_test(context, "Safe CURIE", "[dc:title]",
rdfa_resolve_curie, "http://purl.org/dc/elements/1.1/title",
CURIE_PARSE_PROPERTY);
run_test(context, "Unsafe CURIE", "dc:title",
rdfa_resolve_curie, "http://purl.org/dc/elements/1.1/title",
CURIE_PARSE_PROPERTY);
run_test(context, "Non-prefixed CURIE", ":nonprefixed",
rdfa_resolve_curie, "http://example.org/nonprefixed",
CURIE_PARSE_PROPERTY);
run_test(context, "Reference-only CURIE", "foobar",
rdfa_resolve_curie, NULL, CURIE_PARSE_PROPERTY);
run_test(context, "Reference-only safe CURIE", "[foobar]",
rdfa_resolve_curie, NULL, CURIE_PARSE_PROPERTY);
run_test(context, "Empty safe CURIE", "[]",
rdfa_resolve_curie, NULL, CURIE_PARSE_PROPERTY);
run_test(context, "Blank named safe CURIE", "[_:frank]",
rdfa_resolve_curie, "_:frank", CURIE_PARSE_PROPERTY);
rdfalist* dctvlist = rdfa_create_list(2);
rdfa_add_item(
dctvlist, "http://purl.org/dc/dcmitype/Image", RDFALIST_FLAG_NONE);
rdfa_add_item(
dctvlist, "http://purl.org/dc/dcmitype/Sound", RDFALIST_FLAG_NONE);
run_list_test(
context, "XHTML multiple @type_of", "[dctv:Image] [dctv:Sound]",
rdfa_resolve_curie_list, dctvlist, CURIE_PARSE_INSTANCEOF_DATATYPE);
rdfa_free_list(dctvlist);
rdfalist* nllist = rdfa_create_list(2);
rdfa_add_item(
nllist, XHTML_VOCAB_URI "next", RDFALIST_FLAG_NONE);
rdfa_add_item(
nllist, XHTML_VOCAB_URI "license", RDFALIST_FLAG_NONE);
run_list_test(
context, "XHTML multiple @rel/@rev", "next license",
rdfa_resolve_curie_list, nllist, CURIE_PARSE_RELREV);
rdfa_free_list(nllist);
rdfalist* dtlist = rdfa_create_list(2);
rdfa_add_item(
dtlist, XHTML_VOCAB_URI "description", RDFALIST_FLAG_NONE);
rdfa_add_item(
dtlist, XHTML_VOCAB_URI "title", RDFALIST_FLAG_NONE);
run_list_test(
context, "XHTML multiple @property", "description title",
rdfa_resolve_curie_list, dtlist, CURIE_PARSE_PROPERTY);
rdfa_free_list(dtlist);
run_test_set(context, "XHTML @rel/@rev reserved",
my_g_relrev_reserved_words, XHTML_RELREV_RESERVED_WORDS_SIZE,
rdfa_resolve_relrev_curie, XHTML_VOCAB_URI, CURIE_PARSE_RELREV);
printf("---------------------- CURIE test results ---------------------\n"
"%i passed, %i failed\n",
g_test_passes, g_test_fails);
rdfa_free_context(context);
}
static void end_element(void* parser_context, const char* name,
const char* prefix,const xmlChar* URI)
{
/*xmlParserCtxtPtr parser = (xmlParserCtxtPtr)parser_context;*/
rdfalist* context_stack =
(rdfalist*)((rdfacontext*)parser_context)->context_stack;
rdfacontext* context = (rdfacontext*)rdfa_pop_item(context_stack);
rdfacontext* parent_context = (rdfacontext*)
context_stack->items[context_stack->num_items - 1]->data;
/* append the text to the current context's XML literal */
char* buffer = (char*)malloc(strlen(name) + 4);
#if defined(DEBUG) && DEBUG > 0
printf("DEBUG: </%s>\n", name);
printf("context->local_list_mappings (start of end_element): ");
rdfa_print_mapping(context->local_list_mappings,
(print_mapping_value_fp)rdfa_print_triple_list);
#endif
sprintf(buffer, "</%s>", name);
if(context->xml_literal == NULL)
{
context->xml_literal =
rdfa_replace_string(context->xml_literal, buffer);
context->xml_literal_size = strlen(buffer);
}
else
{
context->xml_literal = rdfa_n_append_string(
context->xml_literal, &context->xml_literal_size,
buffer, strlen(buffer));
}
free(buffer);
/* 9. The next step of the iteration is to establish any
* [current object literal]; */
/* generate the complete object literal triples */
if(context->property != NULL)
{
/* save the current xml literal */
char* saved_xml_literal = context->xml_literal;
char* content_start = NULL;
char* content_end = NULL;
/* ensure to mark only the inner-content of the XML node for
* processing the object literal. */
buffer = NULL;
if(context->xml_literal != NULL)
{
/* get the data between the first tag and the last tag */
content_start = strchr(context->xml_literal, '>');
content_end = strrchr(context->xml_literal, '<');
if((content_start != NULL) && (content_end != NULL))
{
/* set content end to null terminator */
context->xml_literal = ++content_start;
*content_end = '\0';
}
}
/* update the plain literal if the XML Literal is an empty string */
if(context->xml_literal != NULL && strlen(context->xml_literal) == 0)
{
context->plain_literal =
rdfa_replace_string(context->plain_literal, "");
}
/* process data between first tag and last tag
* this needs the xml literal to be null terminated */
if(context->rdfa_version == RDFA_VERSION_1_0)
{
rdfa_complete_object_literal_triples(context);
}
else
{
rdfa_complete_current_property_value_triples(context);
}
if(content_end != NULL)
{
/* set content end back */
*content_end = '<';
}
if(saved_xml_literal != NULL)
{
/* restore xml literal */
context->xml_literal = saved_xml_literal;
}
}
/*printf(context->plain_literal);*/
if(parent_context != NULL) {
/* append the XML literal and plain text literals to the parent
* literals */
if(context->xml_literal != NULL)
{
if(parent_context->xml_literal == NULL)
{
parent_context->xml_literal =
rdfa_replace_string(
parent_context->xml_literal, context->xml_literal);
parent_context->xml_literal_size = context->xml_literal_size;
}
else
{
parent_context->xml_literal =
rdfa_n_append_string(
parent_context->xml_literal,
&parent_context->xml_literal_size,
context->xml_literal, context->xml_literal_size);
}
/* if there is an XML literal, there is probably a plain literal */
if(context->plain_literal != NULL)
{
if(parent_context->plain_literal == NULL)
{
parent_context->plain_literal =
rdfa_replace_string(
parent_context->plain_literal, context->plain_literal);
parent_context->plain_literal_size =
context->plain_literal_size;
}
else
{
parent_context->plain_literal =
rdfa_n_append_string(
parent_context->plain_literal,
&parent_context->plain_literal_size,
context->plain_literal,
context->plain_literal_size);
}
}
}
/* preserve the bnode count by copying it to the parent_context */
parent_context->bnode_count = context->bnode_count;
parent_context->underscore_colon_bnode_name = \
rdfa_replace_string(parent_context->underscore_colon_bnode_name,
context->underscore_colon_bnode_name);
}
/* 10. If the [ skip element ] flag is 'false', and [ new subject ]
* was set to a non-null value, then any [ incomplete triple ]s
* within the current context should be completed: */
if((context->skip_element == 0) && (context->new_subject != NULL))
{
rdfa_complete_incomplete_triples(context);
}
/* 14. Once all the child elements have been traversed, list triples are
* generated, if necessary. */
if(context->rdfa_version == RDFA_VERSION_1_1 && (context->new_subject != NULL))
{
rdfa_complete_list_triples(context);
if(parent_context != NULL)
{
/* copy the current mapping to the parent mapping */
rdfa_free_mapping(parent_context->local_list_mappings,
(free_mapping_value_fp)rdfa_free_list);
parent_context->local_list_mappings = rdfa_copy_mapping(
(void**)context->local_list_mappings,
(copy_mapping_value_fp)rdfa_replace_list);
#if defined(DEBUG) && DEBUG > 0
printf("parent_context->local_list_mappings (after copy): ");
rdfa_print_mapping(context->local_list_mappings,
(print_mapping_value_fp)rdfa_print_triple_list);
#endif
rdfa_free_mapping(context->local_list_mappings,
(free_mapping_value_fp)rdfa_free_list);
context->local_list_mappings = NULL;
}
}
/* free the context */
rdfa_free_context(context);
#if defined(DEBUG) && DEBUG > 0
printf("-------------------------------------------------------------\n");
#endif
}