redland/raptor/raptor-1.4.18.entities.patch

   1 --- misc/raptor-1.4.18/src/raptor.h.old 2008-06-20 07:47:38.000000000 +0200
   2 +++ misc/build/raptor-1.4.18/src/raptor.h       2012-02-15 16:54:21.000000000 +0100
   3 @@ -376,6 +376,7 @@ typedef struct {
   4   * @RAPTOR_FEATURE_JSON_EXTRA_DATA: JSON serializer extra top-level data
   5   * @RAPTOR_FEATURE_RSS_TRIPLES: Atom/RSS serializer writes extra RDF triples it finds (none, rdf-xml, atom-triples)
   6   * @RAPTOR_FEATURE_ATOM_ENTRY_URI: Atom entry URI.  If given, generate an Atom Entry Document with the item having the given URI, otherwise generate an Atom Feed Document with any items found.
   7 + * @RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES: When reading XML, load external entities.
   8   * @RAPTOR_FEATURE_LAST: Internal
   9   *
  10   * Raptor parser, serializer or XML writer features.
  11 @@ -416,7 +417,8 @@ typedef enum {
  12    RAPTOR_FEATURE_JSON_EXTRA_DATA,
  13    RAPTOR_FEATURE_RSS_TRIPLES,
  14    RAPTOR_FEATURE_ATOM_ENTRY_URI,
  15 -  RAPTOR_FEATURE_LAST=RAPTOR_FEATURE_ATOM_ENTRY_URI
  16 +  RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES,
  17 +  RAPTOR_FEATURE_LAST=RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES
  18  } raptor_feature;
  19
  20
  21 --- misc/raptor-1.4.18/src/raptor_feature.c.old 2008-06-05 08:54:16.000000000 +0200
  22 +++ misc/build/raptor-1.4.18/src/raptor_feature.c       2012-02-15 16:55:09.000000000 +0100
  23 @@ -89,7 +89,8 @@ static const struct
  24    { RAPTOR_FEATURE_JSON_CALLBACK     , 6,  "jsonCallback", "JSON serializer callback" },
  25    { RAPTOR_FEATURE_JSON_EXTRA_DATA   , 6,  "jsonExtraData", "JSON serializer extra data" },
  26    { RAPTOR_FEATURE_RSS_TRIPLES       , 6,  "rssTriples", "Atom/RSS serializer writes extra RDF triples" },
  27 -  { RAPTOR_FEATURE_ATOM_ENTRY_URI    , 6,  "atomEntryUri", "Atom serializer Entry URI" }
  28 +  { RAPTOR_FEATURE_ATOM_ENTRY_URI    , 6,  "atomEntryUri", "Atom serializer Entry URI" },
  29 +  { RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES, 1, "loadExternalEntities", "Load external XML entities." }
  30  };
  31
  32
  33 --- misc/raptor-1.4.18/src/raptor_internal.h.old        2008-06-03 07:04:09.000000000 +0200
  34 +++ misc/build/raptor-1.4.18/src/raptor_internal.h      2012-02-15 16:52:08.000000000 +0100
  35 @@ -983,6 +983,14 @@ struct raptor_sax2_s {
  36
  37    /* base URI for resolving relative URIs or xml:base URIs */
  38    raptor_uri* base_uri;
  39 +
  40 +  /* call SAX2 handlers if non-0 */
  41 +  int enabled;
  42 +
  43 +  /* FEATURE:
  44 +   * non 0 if XML entities should be loaded
  45 +   */
  46 +  int feature_load_external_entities;
  47  };
  48
  49  int raptor_sax2_init(void);
  50 --- misc/raptor-1.4.18/src/raptor_libxml.c.old  2008-06-14 05:35:27.000000000 +0200
  51 +++ misc/build/raptor-1.4.18/src/raptor_libxml.c        2012-02-15 16:52:08.000000000 +0100
  52 @@ -142,18 +142,126 @@ raptor_libxml_hasExternalSubset (void* u
  53
  54  static xmlParserInputPtr
  55  raptor_libxml_resolveEntity(void* user_data,
  56 -                            const xmlChar *publicId, const xmlChar *systemId) {
  57 -  raptor_sax2* sax2=(raptor_sax2*)user_data;
  58 -  return libxml2_resolveEntity(sax2->xc, publicId, systemId);
  59 +                            const xmlChar *publicId, const xmlChar *systemId)
  60 +{
  61 +  raptor_sax2* sax2 = (raptor_sax2*)user_data;
  62 +  xmlParserCtxtPtr ctxt = sax2->xc;
  63 +  const unsigned char *uri_string = NULL;
  64 +  xmlParserInputPtr entity_input;
  65 +  int load_entity = 0;
  66 +
  67 +  if(ctxt->input)
  68 +    uri_string = (const unsigned char *)ctxt->input->filename;
  69 +
  70 +  if(!uri_string)
  71 +    uri_string = (const unsigned char *)ctxt->directory;
  72 +
  73 +  load_entity = sax2->feature_load_external_entities;
  74 +
  75 +  if(load_entity) {
  76 +    entity_input = xmlLoadExternalEntity((const char*)uri_string,
  77 +                                         (const char*)publicId,
  78 +                                         ctxt);
  79 +  } else {
  80 +    RAPTOR_DEBUG4("Not loading entity URI %s by policy for publicId '%s' systemId '%s'\n", uri_string, publicId, systemId);
  81 +  }
  82 +
  83 +  return entity_input;
  84  }
  85
  86
  87  static xmlEntityPtr
  88 -raptor_libxml_getEntity(void* user_data, const xmlChar *name) {
  89 -  raptor_sax2* sax2=(raptor_sax2*)user_data;
  90 -  return libxml2_getEntity(sax2->xc, name);
  91 -}
  92 +raptor_libxml_getEntity(void* user_data, const xmlChar *name)
  93 +{
  94 +  raptor_sax2* sax2 = (raptor_sax2*)user_data;
  95 +  xmlParserCtxtPtr xc = sax2->xc;
  96 +  xmlEntityPtr ret = NULL;
  97 +
  98 +  if(!xc)
  99 +    return NULL;
 100 +
 101 +  if(!xc->inSubset) {
 102 +    /* looks for hardcoded set of entity names - lt, gt etc. */
 103 +    ret = xmlGetPredefinedEntity(name);
 104 +    if(ret) {
 105 +      RAPTOR_DEBUG2("Entity '%s' found in predefined set\n", name);
 106 +      return ret;
 107 +    }
 108 +  }
 109
 110 +  /* This section uses xmlGetDocEntity which looks for entities in
 111 +   * memory only, never from a file or URI
 112 +   */
 113 +  if(xc->myDoc && (xc->myDoc->standalone == 1)) {
 114 +    RAPTOR_DEBUG2("Entity '%s' document is standalone\n", name);
 115 +    /* Document is standalone: no entities are required to interpret doc */
 116 +    if(xc->inSubset == 2) {
 117 +      xc->myDoc->standalone = 0;
 118 +      ret = xmlGetDocEntity(xc->myDoc, name);
 119 +      xc->myDoc->standalone = 1;
 120 +    } else {
 121 +      ret = xmlGetDocEntity(xc->myDoc, name);
 122 +      if(!ret) {
 123 +        xc->myDoc->standalone = 0;
 124 +        ret = xmlGetDocEntity(xc->myDoc, name);
 125 +        xc->myDoc->standalone = 1;
 126 +      }
 127 +    }
 128 +  } else {
 129 +    ret = xmlGetDocEntity(xc->myDoc, name);
 130 +  }
 131 +
 132 +  if(ret && !ret->children &&
 133 +    (ret->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) {
 134 +    /* Entity is an external general parsed entity. It may be in a
 135 +     * catalog file, user file or user URI
 136 +     */
 137 +    int val = 0;
 138 +    xmlNodePtr children;
 139 +    int load_entity = 0;
 140 +
 141 +    load_entity = sax2->feature_load_external_entities;
 142 +
 143 +    if(!load_entity) {
 144 +      RAPTOR_DEBUG2("Not getting entity URI %s by policy\n", ret->URI);
 145 +      children = xmlNewText((const xmlChar*)"");
 146 +    } else {
 147 +      /* Disable SAX2 handlers so that the SAX2 events do not all get
 148 +       * sent to callbacks during dealing with the entity parsing.
 149 +       */
 150 +      sax2->enabled = 0;
 151 +      val = xmlParseCtxtExternalEntity(xc, ret->URI, ret->ExternalID, &children);
 152 +      sax2->enabled = 1;
 153 +    }
 154 +
 155 +    if(!val) {
 156 +      xmlAddChildList((xmlNodePtr)ret, children);
 157 +    } else {
 158 +      xc->validate = 0;
 159 +      return NULL;
 160 +    }
 161 +
 162 +    ret->owner = 1;
 163 +
 164 +/* ret->checked was added with commit a37a6ad91a61d168ecc4b29263def3363fff4da6
 165 +   in libxml2 before 2.6.27 it does not exist and ret->children != 0 will be
 166 +   tested instead, which is true due to xmlAddChildList above */
 167 +#if LIBXML_VERSION >= 20627 || !defined(__APPLE__)
 168 +    /* Mark this entity as having been checked - never do this again */
 169 +    if(!ret->checked)
 170 +      ret->checked = 1;
 171 +#else
 172 +    if (atoi(xmlParserVersion) >= 20627) {
 173 +        int *const pChecked = (&ret->owner) + 1;
 174 +        if (!*pChecked) /* owner precedes checked and is also of type int */
 175 +          *pChecked = 1;
 176 +    }
 177 +#endif
 178 +  }
 179 +
 180 +  return ret;
 181 +}
 182 +
 183
 184  static xmlEntityPtr
 185  raptor_libxml_getParameterEntity(void* user_data, const xmlChar *name) {
 186 --- misc/raptor-1.4.18/src/raptor_parse.c.old   2008-06-15 09:18:50.000000000 +0200
 187 +++ misc/build/raptor-1.4.18/src/raptor_parse.c 2012-02-15 16:52:08.000000000 +0100
 188 @@ -1294,6 +1294,7 @@ raptor_set_feature(raptor_parser *parser
 189      case RAPTOR_FEATURE_MICROFORMATS:
 190      case RAPTOR_FEATURE_HTML_LINK:
 191      case RAPTOR_FEATURE_WWW_TIMEOUT:
 192 +    case RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES:
 193        parser->features[(int)feature]=value;
 194        break;
 195
 196 @@ -1414,6 +1415,7 @@ raptor_get_feature(raptor_parser *parser
 197      case RAPTOR_FEATURE_MICROFORMATS:
 198      case RAPTOR_FEATURE_HTML_LINK:
 199      case RAPTOR_FEATURE_WWW_TIMEOUT:
 200 +    case RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES:
 201        result=(parser->features[(int)feature] != 0);
 202        break;
 203
 204 --- misc/raptor-1.4.18/src/raptor_rdfxml.c.old  2008-06-15 10:12:06.000000000 +0200
 205 +++ misc/build/raptor-1.4.18/src/raptor_rdfxml.c        2012-02-15 16:52:08.000000000 +0100
 206 @@ -1124,6 +1124,9 @@ raptor_rdfxml_parse_start(raptor_parser*
 207    raptor_sax2_set_feature(rdf_xml_parser->sax2,
 208                            RAPTOR_FEATURE_NO_NET,
 209                            rdf_parser->features[RAPTOR_FEATURE_NO_NET]);
 210 +  raptor_sax2_set_feature(rdf_xml_parser->sax2,
 211 +                          RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES,
 212 +                          rdf_parser->features[RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES]);
 213
 214    raptor_sax2_parse_start(rdf_xml_parser->sax2, uri);
 215
 216 --- misc/raptor-1.4.18/src/raptor_rss.c.old     2008-05-21 22:25:57.000000000 +0200
 217 +++ misc/build/raptor-1.4.18/src/raptor_rss.c   2012-02-15 16:52:08.000000000 +0100
 218 @@ -251,6 +251,9 @@ raptor_rss_parse_start(raptor_parser *rd
 219    raptor_sax2_set_feature(rss_parser->sax2,
 220                            RAPTOR_FEATURE_NO_NET,
 221                            rdf_parser->features[RAPTOR_FEATURE_NO_NET]);
 222 +  raptor_sax2_set_feature(rss_parser->sax2,
 223 +                          RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES,
 224 +                          rdf_parser->features[RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES]);
 225
 226    raptor_sax2_parse_start(rss_parser->sax2, uri);
 227
 228 --- misc/raptor-1.4.18/src/raptor_sax2.c.old    2008-06-15 10:12:20.000000000 +0200
 229 +++ misc/build/raptor-1.4.18/src/raptor_sax2.c  2012-02-15 16:52:08.000000000 +0100
 230 @@ -96,6 +96,8 @@ raptor_new_sax2(void* user_data, raptor_
 231
 232    sax2->user_data=user_data;
 233
 234 +  sax2->enabled = 1;
 235 +
 236    sax2->locator=error_handlers->locator;
 237
 238    sax2->error_handlers=error_handlers;
 239 @@ -687,6 +689,10 @@ raptor_sax2_set_feature(raptor_sax2 *sax
 240        sax2->feature_no_net=value;
 241        break;
 242
 243 +    case RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES:
 244 +      sax2->feature_load_external_entities=value;
 245 +      break;
 246 +
 247      case RAPTOR_FEATURE_SCANNING:
 248      case RAPTOR_FEATURE_ASSUME_IS_RDF:
 249      case RAPTOR_FEATURE_ALLOW_NON_NS_ATTRIBUTES:
 250 @@ -767,6 +773,9 @@ raptor_sax2_start_element(void* user_dat
 251    unsigned char *xml_language=NULL;
 252    raptor_uri *xml_base=NULL;
 253
 254 +  if(!sax2->enabled)
 255 +    return;
 256 +
 257  #ifdef RAPTOR_XML_EXPAT
 258  #ifdef EXPAT_UTF8_BOM_CRASH
 259    sax2->tokens_count++;
 260 @@ -990,6 +999,9 @@ raptor_sax2_end_element(void* user_data,
 261    raptor_sax2* sax2=(raptor_sax2*)user_data;
 262    raptor_xml_element* xml_element;
 263
 264 +  if(!sax2->enabled)
 265 +    return;
 266 +
 267  #ifdef RAPTOR_XML_EXPAT
 268  #ifdef EXPAT_UTF8_BOM_CRASH
 269    sax2->tokens_count++;
 270 @@ -1025,6 +1037,10 @@ void
 271  raptor_sax2_characters(void* user_data, const unsigned char *s, int len)
 272  {
 273    raptor_sax2* sax2=(raptor_sax2*)user_data;
 274 +
 275 +  if(!sax2->enabled)
 276 +    return;
 277 +
 278    if(sax2->characters_handler)
 279      sax2->characters_handler(sax2->user_data, sax2->current_element, s, len);
 280  }
 281 @@ -1035,6 +1051,10 @@ void
 282  raptor_sax2_cdata(void* user_data, const unsigned char *s, int len)
 283  {
 284    raptor_sax2* sax2=(raptor_sax2*)user_data;
 285 +
 286 +  if(!sax2->enabled)
 287 +    return;
 288 +
 289  #ifdef RAPTOR_XML_EXPAT
 290  #ifdef EXPAT_UTF8_BOM_CRASH
 291    sax2->tokens_count++;
 292 @@ -1051,6 +1071,10 @@ void
 293  raptor_sax2_comment(void* user_data, const unsigned char *s)
 294  {
 295    raptor_sax2* sax2=(raptor_sax2*)user_data;
 296 +
 297 +  if(!sax2->enabled)
 298 +    return;
 299 +
 300    if(sax2->comment_handler)
 301      sax2->comment_handler(sax2->user_data, sax2->current_element, s);
 302  }
 303 @@ -1066,6 +1090,10 @@ raptor_sax2_unparsed_entity_decl(void* u
 304                                   const unsigned char* notationName)
 305  {
 306    raptor_sax2* sax2=(raptor_sax2*)user_data;
 307 +
 308 +  if(!sax2->enabled)
 309 +    return;
 310 +
 311    if(sax2->unparsed_entity_decl_handler)
 312      sax2->unparsed_entity_decl_handler(sax2->user_data,
 313                                         entityName, base, systemId,
 314 @@ -1082,6 +1110,10 @@ raptor_sax2_external_entity_ref(void* us
 315                                  const unsigned char* publicId)
 316  {
 317    raptor_sax2* sax2=(raptor_sax2*)user_data;
 318 +
 319 +  if(!sax2->enabled)
 320 +    return 0;
 321 +
 322    if(sax2->external_entity_ref_handler)
 323      return sax2->external_entity_ref_handler(sax2->user_data,
 324                                               context, base, systemId, publicId);
 325 --- misc/raptor-1.4.18/src/raptor_serialize.c.old       2008-06-20 02:55:31.000000000 +0200
 326 +++ misc/build/raptor-1.4.18/src/raptor_serialize.c     2012-02-15 16:52:08.000000000 +0100
 327 @@ -859,6 +859,7 @@ raptor_serializer_set_feature(raptor_ser
 328
 329      /* Shared */
 330      case RAPTOR_FEATURE_NO_NET:
 331 +    case RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES:
 332
 333      /* XML writer features */
 334      case RAPTOR_FEATURE_WRITER_AUTO_INDENT:
 335 @@ -965,6 +966,7 @@ raptor_serializer_set_feature_string(rap
 336
 337      /* Shared */
 338      case RAPTOR_FEATURE_NO_NET:
 339 +    case RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES:
 340
 341      /* XML writer features */
 342      case RAPTOR_FEATURE_WRITER_AUTO_INDENT:
 343 @@ -1102,6 +1104,7 @@ raptor_serializer_get_feature(raptor_ser
 344
 345      /* Shared */
 346      case RAPTOR_FEATURE_NO_NET:
 347 +    case RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES:
 348
 349      /* XML writer features */
 350      case RAPTOR_FEATURE_WRITER_AUTO_INDENT:
 351 @@ -1201,6 +1204,7 @@ raptor_serializer_get_feature_string(rap
 352
 353      /* Shared */
 354      case RAPTOR_FEATURE_NO_NET:
 355 +    case RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES:
 356
 357      /* XML writer features */
 358      case RAPTOR_FEATURE_WRITER_AUTO_INDENT:
 359 --- misc/raptor-1.4.18/src/raptor_turtle_writer.c.old   2008-06-20 07:47:48.000000000 +0200
 360 +++ misc/build/raptor-1.4.18/src/raptor_turtle_writer.c 2012-02-15 16:52:08.000000000 +0100
 361 @@ -723,6 +723,7 @@ raptor_turtle_writer_set_feature(raptor_
 362
 363      /* Shared */
 364      case RAPTOR_FEATURE_NO_NET:
 365 +    case RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES:
 366
 367      /* XML writer features */
 368      case RAPTOR_FEATURE_RELATIVE_URIS:
 369 @@ -836,6 +837,7 @@ raptor_turtle_writer_get_feature(raptor_
 370
 371      /* Shared */
 372      case RAPTOR_FEATURE_NO_NET:
 373 +    case RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES:
 374
 375      /* XML writer features */
 376      case RAPTOR_FEATURE_RELATIVE_URIS:
 377 --- misc/raptor-1.4.18/src/raptor_xml_writer.c.old      2008-06-03 07:05:56.000000000 +0200
 378 +++ misc/build/raptor-1.4.18/src/raptor_xml_writer.c    2012-02-15 16:52:08.000000000 +0100
 379 @@ -906,6 +906,7 @@ raptor_xml_writer_set_feature(raptor_xml
 380
 381      /* Shared */
 382      case RAPTOR_FEATURE_NO_NET:
 383 +    case RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES:
 384
 385      /* XML writer features */
 386      case RAPTOR_FEATURE_RELATIVE_URIS:
 387 @@ -1026,6 +1027,7 @@ raptor_xml_writer_get_feature(raptor_xml
 388
 389      /* Shared */
 390      case RAPTOR_FEATURE_NO_NET:
 391 +    case RAPTOR_FEATURE_LOAD_EXTERNAL_ENTITIES:
 392
 393      /* XML writer features */
 394      case RAPTOR_FEATURE_RELATIVE_URIS: