src/entity.c

   1 /*
   2  * Claws Mail -- a GTK+ based, lightweight, and fast e-mail client
   3  * Copyright (C) 2017 Ricardo Mones and the Claws Mail team
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 3 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  17  */
  18 #ifdef HAVE_CONFIG_H
  19 #include "config.h"
  20 #include "claws-features.h"
  21 #endif
  22
  23 #include "defs.h"
  24 #include "utils.h"
  25 #include "entity.h"
  26
  27 #define ENTITY_MAX_LEN 8
  28 #define DECODED_MAX_LEN 6
  29
  30 static GHashTable *symbol_table = NULL;
  31
  32 typedef struct _EntitySymbol EntitySymbol;
  33
  34 struct _EntitySymbol
  35 {
  36         gchar *const key;
  37         gchar *const value;
  38 };
  39
  40 /* in alphabetical order with upper-case version first */
  41 static EntitySymbol symbolic_entities[] = {
  42         /* A */
  43         {"Aacute", "Á"},
  44         {"aacute", "á"},
  45         {"Acirc", "Â"},
  46         {"acirc", "â"},
  47         {"acute", "´"},
  48         {"AElig", "Æ"},
  49         {"aelig", "æ"},
  50         {"Agrave", "À"},
  51         {"agrave", "à"},
  52         {"alefsym", "ℵ"},
  53         {"Alpha", "Α"},
  54         {"alpha", "α"},
  55         {"amp", "&"},
  56         {"and", "∧"},
  57         {"ang", "∠"},
  58         {"apos", "'"},
  59         {"Aring", "Å"},
  60         {"aring", "å"},
  61         {"asymp", "≈"},
  62         {"Atilde", "Ã"},
  63         {"atilde", "ã"},
  64         {"Auml", "Ä"},
  65         {"auml", "ä"},
  66         /* B */
  67         {"bdquo", "„"},
  68         {"Beta", "Β"},
  69         {"beta", "β"},
  70         {"brvbar", "¦"},
  71         {"bull", "•"},
  72         /* C */
  73         {"cap", "∩"},
  74         {"Ccedil", "Ç"},
  75         {"ccedil", "ç"},
  76         {"cedil", "¸"},
  77         {"cent", "¢"},
  78         {"Chi", "Χ"},
  79         {"chi", "χ"},
  80         {"circ", "ˆ"},
  81         {"clubs", "♣"},
  82         {"cong", "≅"},
  83         {"copy", "©"},
  84         {"crarr", "↵"},
  85         {"cup", "∪"},
  86         {"curren", "¤"},
  87         /* D */
  88         {"dagger", "†"},
  89         {"Dagger", "‡"},
  90         {"dArr", "⇓"},
  91         {"darr", "↓"},
  92         {"deg", "°"},
  93         {"Delta", "Δ"},
  94         {"delta", "δ"},
  95         {"diams", "♦"},
  96         {"divide", "÷"},
  97         /* E */
  98         {"Eacute", "É"},
  99         {"eacute", "é"},
 100         {"Ecirc", "Ê"},
 101         {"ecirc", "ê"},
 102         {"Egrave", "È"},
 103         {"egrave", "è"},
 104         {"empty", "∅"},
 105         {"emsp", "\xE2\x80\x83"},
 106         {"ensp", "\xE2\x80\x82"},
 107         {"Epsilon", "Ε"},
 108         {"epsilon", "ε"},
 109         {"equiv", "≡"},
 110         {"Eta", "Η"},
 111         {"eta", "η"},
 112         {"ETH", "Ð"},
 113         {"eth", "ð"},
 114         {"Euml", "Ë"},
 115         {"euml", "ë"},
 116         {"euro", "€"},
 117         {"exist", "∃"},
 118         /* F */
 119         {"fnof", "ƒ"},
 120         {"forall", "∀"},
 121         {"frac12", "½"},
 122         {"frac14", "¼"},
 123         {"frac34", "¾"},
 124         {"frasl", "⁄"},
 125         /* G */
 126         {"Gamma", "Γ"},
 127         {"gamma", "γ"},
 128         {"ge", "≥"},
 129         {"gt", ">"},
 130         /* H */
 131         {"hArr", "⇔"},
 132         {"harr", "↔"},
 133         {"hearts", "♥"},
 134         {"hellip", "…"},
 135         /* I */
 136         {"Iacute", "Í"},
 137         {"iacute", "í"},
 138         {"IArr", "⇐"},
 139         {"Icirc", "Î"},
 140         {"icirc", "î"},
 141         {"iexcl", "¡"},
 142         {"Igrave", "Ì"},
 143         {"igrave", "ì"},
 144         {"image", "ℑ"},
 145         {"infin", "∞"},
 146         {"int", "∫"},
 147         {"Iota", "Ι"},
 148         {"iota", "ι"},
 149         {"iquest", "¿"},
 150         {"isin", "∈"},
 151         {"Iuml", "Ï"},
 152         {"iuml", "ï"},
 153         /* K */
 154         {"Kappa", "Κ"},
 155         {"kappa", "κ"},
 156         /* L */
 157         {"Lambda", "Λ"},
 158         {"lambda", "λ"},
 159         {"lang", "〈"},
 160         {"laquo", "«"},
 161         {"larr", "←"},
 162         {"lceil", "⌈"},
 163         {"ldquo", "“"},
 164         {"le", "≤"},
 165         {"lfloor", "⌊"},
 166         {"lowast", "∗"},
 167         {"loz", "◊"},
 168         {"lrm", "\xE2\x80\x8E"},
 169         {"lsaquo", "‹"},
 170         {"lsquo", "‘"},
 171         {"lt", "<"},
 172         /* M */
 173         {"macr", "¯"},
 174         {"mdash", "—"},
 175         {"micro", "µ"},
 176         {"middot", "·"},
 177         {"minus", "−"},
 178         {"Mu", "Μ"},
 179         {"mu", "μ"},
 180         /* N */
 181         {"nabla", "∇"},
 182         {"nbsp", "\xC2\xA0"},
 183         {"ndash", "–"},
 184         {"ne", "≠"},
 185         {"ni", "∋"},
 186         {"not", "¬"},
 187         {"notin", "∉"},
 188         {"nsub", "⊄"},
 189         {"Ntilde", "Ñ"},
 190         {"ntilde", "ñ"},
 191         {"Nu", "Ν"},
 192         {"nu", "ν"},
 193         /* O */
 194         {"Oacute", "Ó"},
 195         {"oacute", "ó"},
 196         {"Ocirc", "Ô"},
 197         {"ocirc", "ô"},
 198         {"OElig", "Œ"},
 199         {"oelig", "œ"},
 200         {"Ograve", "Ò"},
 201         {"ograve", "ò"},
 202         {"oline", "‾"},
 203         {"Omega", "Ω"},
 204         {"omega", "ω"},
 205         {"Omicron", "Ο"},
 206         {"omicron", "ο"},
 207         {"oplus", "⊕"},
 208         {"or", "∨"},
 209         {"ordf", "ª"},
 210         {"ordm", "º"},
 211         {"Oslash", "Ø"},
 212         {"oslash", "ø"},
 213         {"Otilde", "Õ"},
 214         {"otilde", "õ"},
 215         {"otimes", "⊗"},
 216         {"Ouml", "Ö"},
 217         {"ouml", "ö"},
 218         /* P */
 219         {"para", "¶"},
 220         {"part", "∂"},
 221         {"permil", "‰"},
 222         {"perp", "⊥"},
 223         {"Phi", "Φ"},
 224         {"phi", "φ"},
 225         {"Pi", "Π"},
 226         {"pi", "π"},
 227         {"piv", "ϖ"},
 228         {"plusmn", "±"},
 229         {"pound", "£"},
 230         {"Prime", "″"},
 231         {"prime", "′"},
 232         {"prod", "∏"},
 233         {"prop", "∝"},
 234         {"Psi", "Ψ"},
 235         {"psi", "ψ"},
 236         /* Q */
 237         {"quot", "\""},
 238         /* R */
 239         {"radic", "√"},
 240         {"rang", "〉"},
 241         {"raquo", "»"},
 242         {"rArr", "⇒"},
 243         {"rarr", "→"},
 244         {"rceil", "⌉"},
 245         {"rdquo", "”"},
 246         {"real", "ℜ"},
 247         {"reg", "®"},
 248         {"rfloor", "⌋"},
 249         {"Rho", "Ρ"},
 250         {"rho", "ρ"},
 251         {"rlm", "\xE2\x80\x8F"},
 252         {"rsaquo", "›"},
 253         {"rsquo", "’"},
 254         /* S */
 255         {"sbquo", "‚"},
 256         {"Scaron", "Š"},
 257         {"scaron", "š"},
 258         {"sdot", "⋅"},
 259         {"sect", "§"},
 260         {"shy", "\xC2\xAD"},
 261         {"Sigma", "Σ"},
 262         {"sigma", "σ"},
 263         {"sigmaf", "ς"},
 264         {"sim", "∼"},
 265         {"spades", "♠"},
 266         {"sub", "⊂"},
 267         {"sube", "⊆"},
 268         {"sum", "∑"},
 269         {"sup", "⊃"},
 270         {"sup1", "¹"},
 271         {"sup2", "²"},
 272         {"sup3", "³"},
 273         {"supe", "⊇"},
 274         {"szlig", "ß"},
 275         /* T */
 276         {"Tau", "Τ"},
 277         {"tau", "τ"},
 278         {"there4", "∴"},
 279         {"Theta", "Θ"},
 280         {"theta", "θ"},
 281         {"thetasym", "ϑ"},
 282         {"thinsp", "\xE2\x80\x89"},
 283         {"THORN", "Þ"},
 284         {"thorn", "þ"},
 285         {"tilde", "˜"},
 286         {"times", "×"},
 287         {"trade", "™"},
 288         /* U */
 289         {"Uacute", "Ú"},
 290         {"uacute", "ú"},
 291         {"uArr", "⇑"},
 292         {"uarr", "↑"},
 293         {"Ucirc", "Û"},
 294         {"ucirc", "û"},
 295         {"Ugrave", "Ù"},
 296         {"ugrave", "ù"},
 297         {"uml", "¨"},
 298         {"upsih", "ϒ"},
 299         {"Upsilon", "Υ"},
 300         {"upsilon", "υ"},
 301         {"Uuml", "Ü"},
 302         {"uuml", "ü"},
 303         /* W */
 304         {"weierp", "℘"},
 305         /* X */
 306         {"Xi", "Ξ"},
 307         {"xi", "ξ"},
 308         /* Y */
 309         {"Yacute", "Ý"},
 310         {"yacute", "ý"},
 311         {"yen", "¥"},
 312         {"Yuml", "Ÿ"},
 313         {"yuml", "ÿ"},
 314         /* Z */
 315         {"Zeta", "Ζ"},
 316         {"zeta", "ζ"},
 317         {"zwj", "\xE2\x80\x8D"},
 318         {"zwnj", "\xE2\x80\x8C"},
 319         {NULL, NULL}
 320 };
 321
 322 static gchar* entity_extract_to_buffer(gchar *p, gchar b[])
 323 {
 324         gint i = 0;
 325
 326         while (*p != '\0' && *p != ';' && i < ENTITY_MAX_LEN) {
 327                 b[i] = *p;
 328                 ++i, ++p;
 329         }
 330         if (*p != ';' || i == ENTITY_MAX_LEN)
 331                 return NULL;
 332         b[i] = '\0';
 333
 334         return b;
 335 }
 336
 337 static gchar *entity_decode_numeric(gchar *str)
 338 {
 339         gchar b[ENTITY_MAX_LEN];
 340         gchar *p = str, *res;
 341         gboolean hex = FALSE;
 342         gunichar c;
 343
 344         ++p;
 345         if (*p == '\0')
 346                 return NULL;
 347
 348         if (*p == 'x') {
 349                 hex = TRUE;
 350                 ++p;
 351                 if (*p == '\0')
 352                         return NULL;
 353         }
 354
 355         if (entity_extract_to_buffer (p, b) == NULL)
 356                 return NULL;
 357
 358         c = g_ascii_strtoll (b, NULL, (hex? 16: 10));
 359         res = g_malloc0 (DECODED_MAX_LEN + 1);
 360         g_unichar_to_utf8 (c, res);
 361
 362         return res;
 363 }
 364
 365 static gchar *entity_decode_symbol(gchar *str)
 366 {
 367         gchar b[ENTITY_MAX_LEN];
 368         gchar *decoded;
 369
 370         if (entity_extract_to_buffer (str, b) == NULL)
 371                 return NULL;
 372
 373         if (symbol_table == NULL) {
 374                 gint i;
 375
 376                 symbol_table = g_hash_table_new (g_str_hash, g_str_equal);
 377                 for (i = 0; symbolic_entities[i].key != NULL; ++i) {
 378                         g_hash_table_insert (symbol_table,
 379                                 symbolic_entities[i].key, symbolic_entities[i].value);
 380                 }
 381                 debug_print("initialized entities table with %d symbols\n", i);
 382         }
 383
 384         decoded = g_hash_table_lookup (symbol_table, b);
 385         if (decoded != NULL)
 386                 return g_strdup (decoded);
 387
 388         return NULL;
 389 }
 390
 391 gchar *entity_decode(gchar *str)
 392 {
 393         gchar *p = str;
 394         if (p == NULL || *p != '&')
 395                 return NULL;
 396         ++p;
 397         if (*p == '\0')
 398                 return NULL;
 399         if (*p == '#')
 400                 return entity_decode_numeric(p);
 401         else
 402                 return entity_decode_symbol(p);
 403 }