view release on metacpan or  search on metacpan

redland/raptor/src/raptor_sax2.c  view on Meta::CPAN

#endif
#ifdef RAPTOR_XML_LIBXML
    xmlParseChunk(xc, (char*)buffer, 0, 1);
#endif
    return 0;
  }


#ifdef RAPTOR_XML_EXPAT
  rc=XML_Parse(xp, (char*)buffer, len, is_end);
  if(!rc) /* expat: 0 is failure */
    goto handle_error;
  if(is_end)
    return 0;
#endif

#ifdef RAPTOR_XML_LIBXML

  /* This works around some libxml versions that fail to work
   * if the buffer size is larger than the entire file
   * and thus the entire parsing is done in one operation.
   *
   * The code below:
   *   2.4.19 (oldest tested) to 2.4.24 - required
   *   2.4.25                           - works with or without it
   *   2.4.26 or later                  - fails with this code
   */

#if LIBXML_VERSION < 20425
  if(sax2->first_read && is_end) {
    /* parse all but the last character */
    rc=xmlParseChunk(xc, (char*)buffer, len-1, 0);
    if(rc)
      goto handle_error;
    /* last character */
    rc=xmlParseChunk(xc, (char*)buffer + (len-1), 1, 0);
    if(rc)
      goto handle_error;
    /* end */
    xmlParseChunk(xc, (char*)buffer, 0, 1);
    return 0;
  }
#endif

#if LIBXML_VERSION < 20425
  sax2->first_read=0;
#endif
    
  rc=xmlParseChunk(xc, (char*)buffer, len, is_end);
  if(rc) /* libxml: non 0 is failure */
    goto handle_error;
  if(is_end)
    return 0;
#endif

  return 0;

  handle_error:

#ifdef RAPTOR_XML_EXPAT
#ifdef EXPAT_UTF8_BOM_CRASH
  if(sax2->tokens_count) {
#endif
    /* Work around a bug with the expat 1.95.1 shipped with RedHat 7.2
     * which dies here if the error is before <?xml?...
     * The expat 1.95.1 source release version works fine.
     */
    if(sax2->locator)
      raptor_sax2_update_document_locator(sax2, sax2->locator);
#ifdef EXPAT_UTF8_BOM_CRASH
  }
#endif
#endif /* EXPAT */

#if RAPTOR_XML_EXPAT
  if(1) {
    const char *error_prefix="XML Parsing failed - "; /* 21 chars */
    #define ERROR_PREFIX_LEN 21
    const char *error_message=XML_ErrorString(XML_GetErrorCode(xp));
    size_t error_length;
    char *error_buffer;

    error_length=strlen(error_message);
    error_buffer=(char*)RAPTOR_MALLOC(cstring, 
                                      ERROR_PREFIX_LEN + error_length+1);
    if(error_buffer) {
      strncpy(error_buffer, error_prefix, ERROR_PREFIX_LEN);
      strncpy(error_buffer+ERROR_PREFIX_LEN, error_message, error_length+1);

      sax2->error_handler(sax2->error_data, sax2->locator, error_buffer);
      RAPTOR_FREE(cstring, error_buffer);
    } else
      sax2->error_handler(sax2->error_data, sax2->locator, "XML Parsing failed");
  }
#endif
#ifdef RAPTOR_XML_LIBXML
  sax2->error_handler(sax2->error_data, sax2->locator, "XML Parsing failed");
#endif

  return 1;
}


/**
 * raptor_sax2_set_feature:
 * @sax2: #raptor_sax2 SAX2 object
 * @feature: feature to set from enumerated #raptor_feature values
 * @value: integer feature value (0 or larger)
 *
 * Set various SAX2 features.
 * 
 * The allowed features are available via raptor_sax2_features_enumerate().
 *
 * Return value: non 0 on failure or if the feature is unknown
 **/
int
raptor_sax2_set_feature(raptor_sax2 *sax2, raptor_feature feature, int value)
{
  if(value < 0)
    return -1;
  
  switch(feature) {
    case RAPTOR_FEATURE_NORMALIZE_LANGUAGE:
      sax2->feature_normalize_language=value;
      break;

    case RAPTOR_FEATURE_NO_NET:
      sax2->feature_no_net=value;
      break;

    case RAPTOR_FEATURE_SCANNING:
    case RAPTOR_FEATURE_ASSUME_IS_RDF:
    case RAPTOR_FEATURE_ALLOW_NON_NS_ATTRIBUTES:
    case RAPTOR_FEATURE_ALLOW_OTHER_PARSETYPES:
    case RAPTOR_FEATURE_ALLOW_BAGID:
    case RAPTOR_FEATURE_ALLOW_RDF_TYPE_RDF_LIST:
    case RAPTOR_FEATURE_NON_NFC_FATAL:
    case RAPTOR_FEATURE_WARN_OTHER_PARSETYPES:
    case RAPTOR_FEATURE_CHECK_RDF_ID:
    case RAPTOR_FEATURE_RELATIVE_URIS:
    case RAPTOR_FEATURE_START_URI:
    case RAPTOR_FEATURE_WRITER_AUTO_INDENT:
    case RAPTOR_FEATURE_WRITER_AUTO_EMPTY:
    case RAPTOR_FEATURE_WRITER_INDENT_WIDTH:
    case RAPTOR_FEATURE_WRITER_XML_VERSION:
    case RAPTOR_FEATURE_WRITER_XML_DECLARATION:
    default:
      return -1;
      break;
  }

  return 0;
}


void
raptor_sax2_update_document_locator(raptor_sax2* sax2, 
                                    raptor_locator* locator)
{
#ifdef RAPTOR_XML_EXPAT
  raptor_expat_update_document_locator(sax2, locator);
#endif
#ifdef RAPTOR_XML_LIBXML
  raptor_libxml_update_document_locator(sax2, locator);
#endif
}


/* start of an element */
void 
raptor_sax2_start_element(void* user_data, const unsigned char *name,
                          const unsigned char **atts)
{
  raptor_sax2* sax2=(raptor_sax2*)user_data;
  raptor_qname* el_name;
  unsigned char **xml_atts_copy=NULL;
  size_t xml_atts_size=0;
  int all_atts_count=0;
  int ns_attributes_count=0;
  raptor_qname** named_attrs=NULL;
  raptor_xml_element* xml_element=NULL;
  unsigned char *xml_language=NULL;
  raptor_uri *xml_base=NULL;

#ifdef RAPTOR_XML_EXPAT
#ifdef EXPAT_UTF8_BOM_CRASH
  sax2->tokens_count++;
#endif
#endif

#ifdef RAPTOR_XML_LIBXML
  if(atts) {
    int i;
    
    /* Do XML attribute value normalization */
    for (i = 0; atts[i]; i+=2) {
      unsigned char *value=(unsigned char*)atts[i+1];
      unsigned char *src = value;
      unsigned char *dst = xmlStrdup(value);

      if (!dst) {
        sax2->error_handler(sax2->error_data, sax2->locator, "Out of memory");
	return;
      }

      atts[i+1]=dst;

      while (*src == 0x20 || *src == 0x0d || *src == 0x0a || *src == 0x09) 
        src++;
      while (*src) {
	if (*src == 0x20 || *src == 0x0d || *src == 0x0a || *src == 0x09) {
          while (*src == 0x20 || *src == 0x0d || *src == 0x0a || *src == 0x09)
            src++;
          if (*src)
            *dst++ = 0x20;
	} else {
          *dst++ = *src++;
	}
      }
      *dst = '\0';
      xmlFree(value);
    }
  }
#endif

  raptor_sax2_inc_depth(sax2);

  if(atts) {
    int i;

    /* Save passed in XML attributes pointers so we can 
     * NULL the pointers when they get handled below (various atts[i]=NULL)
     */
    for (i = 0; atts[i]; i++);
    xml_atts_size=sizeof(unsigned char*) * i;
    if(xml_atts_size) {
      xml_atts_copy=(unsigned char**)RAPTOR_MALLOC(cstringpointer,xml_atts_size);
      memcpy(xml_atts_copy, atts, xml_atts_size);
    }

    /* XML attributes processing:
     *   xmlns*   - XML namespaces (Namespaces in XML REC)
     *     Deleted and used to synthesise namespaces declarations
     *   xml:lang - XML language (XML REC)
     *     Deleted and optionally normalised to lowercase
     *   xml:base - XML Base (XML Base REC)

redland/raptor/src/raptor_sax2.c  view on Meta::CPAN

                                              sizeof(raptor_qname*));
    if(!named_attrs) {
      sax2->fatal_error_handler(sax2->fatal_error_data, sax2->locator, "Out of memory");
      RAPTOR_FREE(raptor_xml_element, xml_element);
      raptor_free_qname(raptor_xml_element_get_name(xml_element));
      return;
    }

    for (i = 0; i < all_atts_count; i++) {
      raptor_qname* attr;

      /* Skip previously processed attributes */
      if(!atts[i<<1])
        continue;

      /* namespace-name[i] stored in named_attrs[i] */
      attr=raptor_new_qname(&sax2->namespaces,
                            atts[i<<1], atts[(i<<1)+1],
                            (raptor_simple_message_handler)raptor_sax2_simple_error, sax2);
      if(!attr) { /* failed - tidy up and return */
        int j;

        for (j=0; j < i; j++)
          RAPTOR_FREE(raptor_qname, named_attrs[j]);
        RAPTOR_FREE(raptor_qname_array, named_attrs);
        raptor_free_xml_element(xml_element);
        return;
      }

      named_attrs[offset++]=attr;
    }
  } /* end if ns_attributes_count */


  if(named_attrs)
    raptor_xml_element_set_attributes(xml_element,
                                      named_attrs, ns_attributes_count);

  raptor_xml_element_push(sax2, xml_element);

  if(sax2->start_element_handler)
    sax2->start_element_handler(sax2->user_data, xml_element);

  if(xml_atts_copy) {
    /* Restore passed in XML attributes, free the copy */
    memcpy((void*)atts, xml_atts_copy, xml_atts_size);
    RAPTOR_FREE(cstringpointer, xml_atts_copy);
  }

}


/* end of an element */
void
raptor_sax2_end_element(void* user_data, const unsigned char *name)
{
  raptor_sax2* sax2=(raptor_sax2*)user_data;
  raptor_xml_element* xml_element;

#ifdef RAPTOR_XML_EXPAT
#ifdef EXPAT_UTF8_BOM_CRASH
  sax2->tokens_count++;
#endif
#endif

  xml_element=sax2->current_element;
  if(xml_element) {
#ifdef RAPTOR_DEBUG_VERBOSE
    fprintf(stderr, "\nraptor_rdfxml_end_element_handler: End ns-element: ");
    raptor_qname_print(stderr, xml_element->name);
    fputc('\n', stderr);
#endif

    if(sax2->end_element_handler)
      sax2->end_element_handler(sax2->user_data, xml_element);
  }
  
  raptor_namespaces_end_for_depth(&sax2->namespaces, 
                                  raptor_sax2_get_depth(sax2));
  xml_element=raptor_xml_element_pop(sax2);
  if(xml_element)
    raptor_free_xml_element(xml_element);

  raptor_sax2_dec_depth(sax2);
}




/* characters */
void
raptor_sax2_characters(void* user_data, const unsigned char *s, int len)
{
  raptor_sax2* sax2=(raptor_sax2*)user_data;
  if(sax2->characters_handler)
    sax2->characters_handler(sax2->user_data, sax2->current_element, s, len);
}


/* like <![CDATA[...]> */
void
raptor_sax2_cdata(void* user_data, const unsigned char *s, int len)
{
  raptor_sax2* sax2=(raptor_sax2*)user_data;
#ifdef RAPTOR_XML_EXPAT
#ifdef EXPAT_UTF8_BOM_CRASH
  sax2->tokens_count++;
#endif
#endif

  if(sax2->cdata_handler)
    sax2->cdata_handler(sax2->user_data, sax2->current_element, s, len);
}


/* comment */
void
raptor_sax2_comment(void* user_data, const unsigned char *s)
{
  raptor_sax2* sax2=(raptor_sax2*)user_data;
  if(sax2->comment_handler)
    sax2->comment_handler(sax2->user_data, sax2->current_element, s);
}


/* unparsed (NDATA) entity */
void
raptor_sax2_unparsed_entity_decl(void* user_data, 
                                 const unsigned char* entityName, 
                                 const unsigned char* base, 
                                 const unsigned char* systemId, 
                                 const unsigned char* publicId, 
                                 const unsigned char* notationName)
{
  raptor_sax2* sax2=(raptor_sax2*)user_data;
  if(sax2->unparsed_entity_decl_handler)
    sax2->unparsed_entity_decl_handler(sax2->user_data,
                                       entityName, base, systemId, 
                                       publicId, notationName);
}


/* external entity reference */
int
raptor_sax2_external_entity_ref(void* user_data, 
                                const unsigned char* context, 
                                const unsigned char* base, 
                                const unsigned char* systemId, 
                                const unsigned char* publicId)
{
  raptor_sax2* sax2=(raptor_sax2*)user_data;
  if(sax2->external_entity_ref_handler)
    return sax2->external_entity_ref_handler(sax2->user_data,
                                             context, base, systemId, publicId);
  return 0;
}