src/plugins/rssyl/libfeed/parser.c

   1 /*
   2  * Copyright (C) 2006 Andrej Kacian <andrej@kacian.sk>
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public License as
   6  * published by the Free Software Foundation; either version 2 of the
   7  * License, or (at your option) any later version.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12  * General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public
  15  * License along with this program; if not, write to the
  16  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  17  * Boston, MA 02111-1307, USA.
  18  */
  19
  20 #ifdef HAVE_CONFIG_H
  21 # include <config.h>
  22 #endif
  23
  24 #include <glib.h>
  25 #include <curl/curl.h>
  26 #include <expat.h>
  27 #include <string.h>
  28 #include <ctype.h>
  29 #include <errno.h>
  30
  31 #include <codeconv.h>
  32
  33 #include "feed.h"
  34
  35 #include "parser.h"
  36
  37 enum {
  38         FEED_TYPE_NONE,
  39         FEED_TYPE_RDF,
  40         FEED_TYPE_RSS_20,
  41         FEED_TYPE_ATOM_03,
  42         FEED_TYPE_ATOM_10,
  43         FEED_TYPE_OPML
  44 } FeedTypes;
  45
  46 static void _handler_set(XML_Parser parser, guint type)
  47 {
  48         if( parser == NULL )
  49                 return;
  50
  51         switch(type) {
  52                 case FEED_TYPE_RSS_20:
  53                         XML_SetElementHandler(parser,
  54                                         feed_parser_rss20_start,
  55                                         feed_parser_rss20_end);
  56                         break;
  57
  58                 case FEED_TYPE_RDF:
  59                         XML_SetElementHandler(parser,
  60                                         feed_parser_rdf_start,
  61                                         feed_parser_rdf_end);
  62                         break;
  63
  64                 case FEED_TYPE_ATOM_10:
  65                         XML_SetElementHandler(parser,
  66                                         feed_parser_atom10_start,
  67                                         feed_parser_atom10_end);
  68                         break;
  69         }
  70 }
  71
  72 static void _elparse_start_chooser(void *data,
  73                 const gchar *el, const gchar **attr)
  74 {
  75         FeedParserCtx *ctx = (FeedParserCtx *)data;
  76         guint feedtype = FEED_TYPE_NONE;
  77         gchar *version;
  78
  79         if( ctx->depth == 0 ) {
  80
  81                 /* RSS 2.0 detected */
  82                 if( !strcmp(el, "rss") ) {
  83                         feedtype = FEED_TYPE_RSS_20;
  84                 } else if( !strcmp(el, "rdf:RDF") ) {
  85                         feedtype = FEED_TYPE_RDF;
  86                 } else if( !strcmp(el, "feed") ) {
  87
  88                         /* ATOM feed detected, let's check version */
  89                         version = feed_parser_get_attribute_value(attr, "xmlns");
  90                         if( !strcmp(version, "http://www.w3.org/2005/Atom") ||
  91                                         !strcmp(version, "https://www.w3.org/2005/Atom") )
  92                                 feedtype = FEED_TYPE_ATOM_10;
  93                         else
  94                                 feedtype = FEED_TYPE_ATOM_03;
  95                 }
  96         }
  97
  98         _handler_set(ctx->parser, feedtype);
  99
 100         ctx->depth++;
 101 }
 102
 103 static void _elparse_end_dummy(void *data, const gchar *el)
 104 {
 105         FeedParserCtx *ctx = (FeedParserCtx *)data;
 106
 107         if( ctx->str != NULL ) {
 108                 g_string_free(ctx->str, TRUE);
 109                 ctx->str = NULL;
 110         }
 111
 112         ctx->depth--;
 113 }
 114
 115 void libfeed_expat_chparse(void *data, const gchar *s, gint len)
 116 {
 117         FeedParserCtx *ctx = (FeedParserCtx *)data;
 118         gchar *buf = NULL;
 119         gint i, xblank = 1;
 120
 121         buf = malloc(len+1);
 122         strncpy(buf, s, len);
 123         buf[len] = '\0';
 124
 125         /* check if the string is blank, ... */
 126         for( i = 0; i < strlen(buf); i++ )
 127                 if( !isspace(buf[i]) )
 128                         xblank = 0;
 129
 130         /* ...because we do not want the blanks if we're just starting new GString */
 131         if( xblank > 0 && ctx->str == NULL ) {
 132                 g_free(buf);
 133                 return;
 134         }
 135
 136         if( ctx->str == NULL ) {
 137                 ctx->str = g_string_sized_new(len + 1);
 138         }
 139
 140         g_string_append(ctx->str, buf);
 141         g_free(buf);
 142 }
 143
 144
 145 void feed_parser_set_expat_handlers(FeedParserCtx *ctx)
 146 {
 147         XML_SetUserData(ctx->parser, (void *)ctx);
 148
 149         XML_SetElementHandler(ctx->parser,
 150                         _elparse_start_chooser,
 151                         _elparse_end_dummy);
 152
 153         XML_SetCharacterDataHandler(ctx->parser,
 154                 libfeed_expat_chparse);
 155
 156         XML_SetUnknownEncodingHandler(ctx->parser, feed_parser_unknown_encoding_handler,
 157                         NULL);
 158 }
 159
 160 size_t feed_writefunc(void *ptr, size_t size, size_t nmemb, void *data)
 161 {
 162         gint len = size * nmemb;
 163         FeedParserCtx *ctx = (FeedParserCtx *)data;
 164         gint status, err;
 165
 166         status = XML_Parse(ctx->parser, ptr, len, FALSE);
 167
 168         if( status == XML_STATUS_ERROR ) {
 169                 err = XML_GetErrorCode(ctx->parser);
 170                 printf("\nExpat: --- %s\n\n", XML_ErrorString(err));
 171         }
 172
 173         return len;
 174 }
 175
 176 gchar *feed_parser_get_attribute_value(const gchar **attr, const gchar *name)
 177 {
 178         guint i;
 179
 180         if( attr == NULL || name == NULL )
 181                 return NULL;
 182
 183         for( i = 0; attr[i] != NULL && attr[i+1] != NULL; i += 2 ) {
 184                 if( !strcmp( attr[i], name) )
 185                         return (gchar *)attr[i+1];
 186         }
 187
 188         /* We haven't found anything. */
 189         return NULL;
 190 }
 191
 192 #define CHARSIZEUTF32   4
 193
 194 enum {
 195         LEP_ICONV_OK,
 196         LEP_ICONV_FAILED,
 197         LEP_ICONV_ILSEQ,
 198         LEP_ICONV_INVAL,
 199         LEP_ICONV_UNKNOWN
 200 };
 201
 202 static gint giconv_utf32_char(GIConv cd, const gchar *inbuf, size_t insize,
 203                 guint32 *p_value)
 204 {
 205 #ifdef HAVE_ICONV
 206         size_t outsize;
 207         guchar outbuf[CHARSIZEUTF32];
 208         gchar *outbufp;
 209         gint r, errno;
 210
 211         outsize = sizeof(outbuf);
 212         outbufp = (gchar *)outbuf;
 213 #ifdef HAVE_ICONV_PROTO_CONST
 214         r = g_iconv(cd, (const gchar **)&inbuf, &insize,
 215                         &outbufp, &outsize);
 216 #else
 217         r = g_iconv(cd, (gchar **)&inbuf, &insize,
 218                         &outbufp, &outsize);
 219 #endif
 220         if( r == -1 ) {
 221                 g_iconv(cd, 0, 0, 0, 0);
 222                 switch(errno) {
 223                 case EILSEQ:
 224                         return LEP_ICONV_ILSEQ;
 225                 case EINVAL:
 226                         return LEP_ICONV_INVAL;
 227                 default:
 228                         return LEP_ICONV_UNKNOWN;
 229                 }
 230         } else {
 231                 guint32 value;
 232                 guint i;
 233
 234                 if( (insize > 0) || (outsize > 0) )
 235                         return LEP_ICONV_FAILED;
 236
 237                 value = 0;
 238                 for( i = 0; i < sizeof(outbuf); i++ ) {
 239                         value = (value << 8) + outbuf[i];
 240                 }
 241                 *p_value = value;
 242                 return LEP_ICONV_OK;
 243         }
 244 #else
 245         return LEP_ICONV_FAILED;
 246 #endif
 247 }
 248
 249 static gint feed_parser_setup_unknown_encoding(const gchar *charset,
 250                 XML_Encoding *info)
 251 {
 252         GIConv cd;
 253         gint flag, r;
 254         gchar buf[4];
 255         guint i, j, k;
 256         guint32 value;
 257
 258         cd = g_iconv_open("UTF-32BE", charset);
 259         if( cd == (GIConv) -1 )
 260                 return -1;
 261
 262         flag = 0;
 263         for( i = 0; i < 256; i++ ) {
 264                 /* first char */
 265                 buf[0] = i;
 266                 info->map[i] = 0;
 267                 r = giconv_utf32_char(cd, buf, 1, &value);
 268                 if( r == LEP_ICONV_OK) {
 269                         info->map[i] = value;
 270                 } else if( r != LEP_ICONV_INVAL ) {
 271                 } else {
 272                         for( j = 0; j < 256; j++ ) {
 273                                 /* second char */
 274                                 buf[1] = j;
 275                                 r = giconv_utf32_char(cd, buf, 2, &value);
 276                                 if( r == LEP_ICONV_OK ) {
 277                                         flag = 1;
 278                                         info->map[i] = -2;
 279                                 } else if( r != LEP_ICONV_INVAL ) {
 280                                 } else {
 281                                         for( k = 0; k < 256; k++ ) {
 282                                                 /* third char */
 283                                                 buf[2] = k;
 284                                                 r = giconv_utf32_char(cd, buf, 3, &value);
 285                                                 if( r == LEP_ICONV_OK) {
 286                                                         info->map[i] = -3;
 287                                                 }
 288                                         }
 289                                 }
 290                         }
 291                 }
 292         }
 293
 294         g_iconv_close(cd);
 295
 296         return flag;
 297 }
 298
 299 struct FeedParserUnknownEncoding {
 300         gchar *charset;
 301         GIConv cd;
 302 };
 303
 304 static gint feed_parser_unknown_encoding_convert(void *data, const gchar *s)
 305 {
 306         gint r;
 307         struct FeedParserUnknownEncoding *enc_data;
 308         size_t insize;
 309         guint32 value;
 310
 311         enc_data = data;
 312         insize = 4;
 313
 314         if( s == NULL )
 315                 return -1;
 316
 317         r = giconv_utf32_char(enc_data->cd, s, insize, &value);
 318         if( r != LEP_ICONV_OK )
 319                 return -1;
 320
 321         return 0;
 322 }
 323
 324 static void feed_parser_unknown_encoding_data_free(void *data)
 325 {
 326         struct FeedParserUnknownEncoding *enc_data;
 327
 328         enc_data = data;
 329         free(enc_data->charset);
 330         g_iconv_close(enc_data->cd);
 331         free(enc_data);
 332 }
 333
 334 int feed_parser_unknown_encoding_handler(void *encdata, const XML_Char *name,
 335                 XML_Encoding *info)
 336 {
 337         GIConv cd;
 338         struct FeedParserUnknownEncoding *data;
 339         int result;
 340
 341         result = feed_parser_setup_unknown_encoding(name, info);
 342         if( result == 0 ) {
 343                 info->data = NULL;
 344                 info->convert = NULL;
 345                 info->release = NULL;
 346                 return XML_STATUS_OK;
 347         }
 348
 349         cd = g_iconv_open("UTF-32BE", name);
 350         if( cd == (GIConv)-1 )
 351                 return XML_STATUS_ERROR;
 352
 353         data = malloc( sizeof(*data) );
 354         if( data == NULL ) {
 355                 g_iconv_close(cd);
 356                 return XML_STATUS_ERROR;
 357         }
 358
 359         data->charset = strdup(name);
 360         if( data->charset == NULL ) {
 361                 free(data);
 362                 g_iconv_close(cd);
 363                 return XML_STATUS_ERROR;
 364         }
 365
 366         data->cd = cd;
 367         info->data = data;
 368         info->convert = feed_parser_unknown_encoding_convert;
 369         info->release = feed_parser_unknown_encoding_data_free;
 370
 371         return XML_STATUS_OK;
 372 }