]> git.saurik.com Git - wxWidgets.git/blobdiff - src/html/htmlpars.cpp
correction to last commit: don't test unsetenv() return value, it's void under Darwin
[wxWidgets.git] / src / html / htmlpars.cpp
index 737389ce1182580c3e4805f1b6cdd8e2917b65fb..9d3bf3cb8e0f61275e9ffc617d6c263b84e48c13 100644 (file)
@@ -15,7 +15,7 @@
 
 #if wxUSE_HTML && wxUSE_STREAMS
 
-#ifndef WXPRECOMP
+#ifndef WX_PRECOMP
     #include "wx/dynarray.h"
     #include "wx/log.h"
     #include "wx/intl.h"
@@ -519,17 +519,6 @@ wxString wxHtmlEntitiesParser::Parse(const wxString& input)
     return output;
 }
 
-struct wxHtmlEntityInfo
-{
-    const wxChar *name;
-    unsigned code;
-};
-
-extern "C" int LINKAGEMODE wxHtmlEntityCompare(const void *key, const void *item)
-{
-    return wxStrcmp((wxChar*)key, ((wxHtmlEntityInfo*)item)->name);
-}
-
 #if !wxUSE_UNICODE
 wxChar wxHtmlEntitiesParser::GetCharForCode(unsigned code)
 {
@@ -548,6 +537,21 @@ wxChar wxHtmlEntitiesParser::GetCharForCode(unsigned code)
 }
 #endif
 
+struct wxHtmlEntityInfo
+{
+    const wxStringCharType *name;
+    unsigned code;
+};
+
+extern "C" int LINKAGEMODE wxHtmlEntityCompare(const void *key, const void *item)
+{
+#if wxUSE_UNICODE_UTF8
+    return strcmp((char*)key, ((wxHtmlEntityInfo*)item)->name);
+#else
+    return wxStrcmp((wxChar*)key, ((wxHtmlEntityInfo*)item)->name);
+#endif
+}
+
 wxChar wxHtmlEntitiesParser::GetEntityChar(const wxString& entity)
 {
     unsigned code = 0;
@@ -571,260 +575,265 @@ wxChar wxHtmlEntitiesParser::GetEntityChar(const wxString& entity)
     }
     else
     {
+        // store the literals in wx's internal representation (either char*
+        // in UTF-8 or wchar_t*) for best performance:
+        #define ENTITY(name, code) { wxSTRING_TEXT(name), code }
+
         static wxHtmlEntityInfo substitutions[] = {
-            { wxT("AElig"),198 },
-            { wxT("Aacute"),193 },
-            { wxT("Acirc"),194 },
-            { wxT("Agrave"),192 },
-            { wxT("Alpha"),913 },
-            { wxT("Aring"),197 },
-            { wxT("Atilde"),195 },
-            { wxT("Auml"),196 },
-            { wxT("Beta"),914 },
-            { wxT("Ccedil"),199 },
-            { wxT("Chi"),935 },
-            { wxT("Dagger"),8225 },
-            { wxT("Delta"),916 },
-            { wxT("ETH"),208 },
-            { wxT("Eacute"),201 },
-            { wxT("Ecirc"),202 },
-            { wxT("Egrave"),200 },
-            { wxT("Epsilon"),917 },
-            { wxT("Eta"),919 },
-            { wxT("Euml"),203 },
-            { wxT("Gamma"),915 },
-            { wxT("Iacute"),205 },
-            { wxT("Icirc"),206 },
-            { wxT("Igrave"),204 },
-            { wxT("Iota"),921 },
-            { wxT("Iuml"),207 },
-            { wxT("Kappa"),922 },
-            { wxT("Lambda"),923 },
-            { wxT("Mu"),924 },
-            { wxT("Ntilde"),209 },
-            { wxT("Nu"),925 },
-            { wxT("OElig"),338 },
-            { wxT("Oacute"),211 },
-            { wxT("Ocirc"),212 },
-            { wxT("Ograve"),210 },
-            { wxT("Omega"),937 },
-            { wxT("Omicron"),927 },
-            { wxT("Oslash"),216 },
-            { wxT("Otilde"),213 },
-            { wxT("Ouml"),214 },
-            { wxT("Phi"),934 },
-            { wxT("Pi"),928 },
-            { wxT("Prime"),8243 },
-            { wxT("Psi"),936 },
-            { wxT("Rho"),929 },
-            { wxT("Scaron"),352 },
-            { wxT("Sigma"),931 },
-            { wxT("THORN"),222 },
-            { wxT("Tau"),932 },
-            { wxT("Theta"),920 },
-            { wxT("Uacute"),218 },
-            { wxT("Ucirc"),219 },
-            { wxT("Ugrave"),217 },
-            { wxT("Upsilon"),933 },
-            { wxT("Uuml"),220 },
-            { wxT("Xi"),926 },
-            { wxT("Yacute"),221 },
-            { wxT("Yuml"),376 },
-            { wxT("Zeta"),918 },
-            { wxT("aacute"),225 },
-            { wxT("acirc"),226 },
-            { wxT("acute"),180 },
-            { wxT("aelig"),230 },
-            { wxT("agrave"),224 },
-            { wxT("alefsym"),8501 },
-            { wxT("alpha"),945 },
-            { wxT("amp"),38 },
-            { wxT("and"),8743 },
-            { wxT("ang"),8736 },
-            { wxT("aring"),229 },
-            { wxT("asymp"),8776 },
-            { wxT("atilde"),227 },
-            { wxT("auml"),228 },
-            { wxT("bdquo"),8222 },
-            { wxT("beta"),946 },
-            { wxT("brvbar"),166 },
-            { wxT("bull"),8226 },
-            { wxT("cap"),8745 },
-            { wxT("ccedil"),231 },
-            { wxT("cedil"),184 },
-            { wxT("cent"),162 },
-            { wxT("chi"),967 },
-            { wxT("circ"),710 },
-            { wxT("clubs"),9827 },
-            { wxT("cong"),8773 },
-            { wxT("copy"),169 },
-            { wxT("crarr"),8629 },
-            { wxT("cup"),8746 },
-            { wxT("curren"),164 },
-            { wxT("dArr"),8659 },
-            { wxT("dagger"),8224 },
-            { wxT("darr"),8595 },
-            { wxT("deg"),176 },
-            { wxT("delta"),948 },
-            { wxT("diams"),9830 },
-            { wxT("divide"),247 },
-            { wxT("eacute"),233 },
-            { wxT("ecirc"),234 },
-            { wxT("egrave"),232 },
-            { wxT("empty"),8709 },
-            { wxT("emsp"),8195 },
-            { wxT("ensp"),8194 },
-            { wxT("epsilon"),949 },
-            { wxT("equiv"),8801 },
-            { wxT("eta"),951 },
-            { wxT("eth"),240 },
-            { wxT("euml"),235 },
-            { wxT("euro"),8364 },
-            { wxT("exist"),8707 },
-            { wxT("fnof"),402 },
-            { wxT("forall"),8704 },
-            { wxT("frac12"),189 },
-            { wxT("frac14"),188 },
-            { wxT("frac34"),190 },
-            { wxT("frasl"),8260 },
-            { wxT("gamma"),947 },
-            { wxT("ge"),8805 },
-            { wxT("gt"),62 },
-            { wxT("hArr"),8660 },
-            { wxT("harr"),8596 },
-            { wxT("hearts"),9829 },
-            { wxT("hellip"),8230 },
-            { wxT("iacute"),237 },
-            { wxT("icirc"),238 },
-            { wxT("iexcl"),161 },
-            { wxT("igrave"),236 },
-            { wxT("image"),8465 },
-            { wxT("infin"),8734 },
-            { wxT("int"),8747 },
-            { wxT("iota"),953 },
-            { wxT("iquest"),191 },
-            { wxT("isin"),8712 },
-            { wxT("iuml"),239 },
-            { wxT("kappa"),954 },
-            { wxT("lArr"),8656 },
-            { wxT("lambda"),955 },
-            { wxT("lang"),9001 },
-            { wxT("laquo"),171 },
-            { wxT("larr"),8592 },
-            { wxT("lceil"),8968 },
-            { wxT("ldquo"),8220 },
-            { wxT("le"),8804 },
-            { wxT("lfloor"),8970 },
-            { wxT("lowast"),8727 },
-            { wxT("loz"),9674 },
-            { wxT("lrm"),8206 },
-            { wxT("lsaquo"),8249 },
-            { wxT("lsquo"),8216 },
-            { wxT("lt"),60 },
-            { wxT("macr"),175 },
-            { wxT("mdash"),8212 },
-            { wxT("micro"),181 },
-            { wxT("middot"),183 },
-            { wxT("minus"),8722 },
-            { wxT("mu"),956 },
-            { wxT("nabla"),8711 },
-            { wxT("nbsp"),160 },
-            { wxT("ndash"),8211 },
-            { wxT("ne"),8800 },
-            { wxT("ni"),8715 },
-            { wxT("not"),172 },
-            { wxT("notin"),8713 },
-            { wxT("nsub"),8836 },
-            { wxT("ntilde"),241 },
-            { wxT("nu"),957 },
-            { wxT("oacute"),243 },
-            { wxT("ocirc"),244 },
-            { wxT("oelig"),339 },
-            { wxT("ograve"),242 },
-            { wxT("oline"),8254 },
-            { wxT("omega"),969 },
-            { wxT("omicron"),959 },
-            { wxT("oplus"),8853 },
-            { wxT("or"),8744 },
-            { wxT("ordf"),170 },
-            { wxT("ordm"),186 },
-            { wxT("oslash"),248 },
-            { wxT("otilde"),245 },
-            { wxT("otimes"),8855 },
-            { wxT("ouml"),246 },
-            { wxT("para"),182 },
-            { wxT("part"),8706 },
-            { wxT("permil"),8240 },
-            { wxT("perp"),8869 },
-            { wxT("phi"),966 },
-            { wxT("pi"),960 },
-            { wxT("piv"),982 },
-            { wxT("plusmn"),177 },
-            { wxT("pound"),163 },
-            { wxT("prime"),8242 },
-            { wxT("prod"),8719 },
-            { wxT("prop"),8733 },
-            { wxT("psi"),968 },
-            { wxT("quot"),34 },
-            { wxT("rArr"),8658 },
-            { wxT("radic"),8730 },
-            { wxT("rang"),9002 },
-            { wxT("raquo"),187 },
-            { wxT("rarr"),8594 },
-            { wxT("rceil"),8969 },
-            { wxT("rdquo"),8221 },
-            { wxT("real"),8476 },
-            { wxT("reg"),174 },
-            { wxT("rfloor"),8971 },
-            { wxT("rho"),961 },
-            { wxT("rlm"),8207 },
-            { wxT("rsaquo"),8250 },
-            { wxT("rsquo"),8217 },
-            { wxT("sbquo"),8218 },
-            { wxT("scaron"),353 },
-            { wxT("sdot"),8901 },
-            { wxT("sect"),167 },
-            { wxT("shy"),173 },
-            { wxT("sigma"),963 },
-            { wxT("sigmaf"),962 },
-            { wxT("sim"),8764 },
-            { wxT("spades"),9824 },
-            { wxT("sub"),8834 },
-            { wxT("sube"),8838 },
-            { wxT("sum"),8721 },
-            { wxT("sup"),8835 },
-            { wxT("sup1"),185 },
-            { wxT("sup2"),178 },
-            { wxT("sup3"),179 },
-            { wxT("supe"),8839 },
-            { wxT("szlig"),223 },
-            { wxT("tau"),964 },
-            { wxT("there4"),8756 },
-            { wxT("theta"),952 },
-            { wxT("thetasym"),977 },
-            { wxT("thinsp"),8201 },
-            { wxT("thorn"),254 },
-            { wxT("tilde"),732 },
-            { wxT("times"),215 },
-            { wxT("trade"),8482 },
-            { wxT("uArr"),8657 },
-            { wxT("uacute"),250 },
-            { wxT("uarr"),8593 },
-            { wxT("ucirc"),251 },
-            { wxT("ugrave"),249 },
-            { wxT("uml"),168 },
-            { wxT("upsih"),978 },
-            { wxT("upsilon"),965 },
-            { wxT("uuml"),252 },
-            { wxT("weierp"),8472 },
-            { wxT("xi"),958 },
-            { wxT("yacute"),253 },
-            { wxT("yen"),165 },
-            { wxT("yuml"),255 },
-            { wxT("zeta"),950 },
-            { wxT("zwj"),8205 },
-            { wxT("zwnj"),8204 },
+            ENTITY("AElig", 198),
+            ENTITY("Aacute", 193),
+            ENTITY("Acirc", 194),
+            ENTITY("Agrave", 192),
+            ENTITY("Alpha", 913),
+            ENTITY("Aring", 197),
+            ENTITY("Atilde", 195),
+            ENTITY("Auml", 196),
+            ENTITY("Beta", 914),
+            ENTITY("Ccedil", 199),
+            ENTITY("Chi", 935),
+            ENTITY("Dagger", 8225),
+            ENTITY("Delta", 916),
+            ENTITY("ETH", 208),
+            ENTITY("Eacute", 201),
+            ENTITY("Ecirc", 202),
+            ENTITY("Egrave", 200),
+            ENTITY("Epsilon", 917),
+            ENTITY("Eta", 919),
+            ENTITY("Euml", 203),
+            ENTITY("Gamma", 915),
+            ENTITY("Iacute", 205),
+            ENTITY("Icirc", 206),
+            ENTITY("Igrave", 204),
+            ENTITY("Iota", 921),
+            ENTITY("Iuml", 207),
+            ENTITY("Kappa", 922),
+            ENTITY("Lambda", 923),
+            ENTITY("Mu", 924),
+            ENTITY("Ntilde", 209),
+            ENTITY("Nu", 925),
+            ENTITY("OElig", 338),
+            ENTITY("Oacute", 211),
+            ENTITY("Ocirc", 212),
+            ENTITY("Ograve", 210),
+            ENTITY("Omega", 937),
+            ENTITY("Omicron", 927),
+            ENTITY("Oslash", 216),
+            ENTITY("Otilde", 213),
+            ENTITY("Ouml", 214),
+            ENTITY("Phi", 934),
+            ENTITY("Pi", 928),
+            ENTITY("Prime", 8243),
+            ENTITY("Psi", 936),
+            ENTITY("Rho", 929),
+            ENTITY("Scaron", 352),
+            ENTITY("Sigma", 931),
+            ENTITY("THORN", 222),
+            ENTITY("Tau", 932),
+            ENTITY("Theta", 920),
+            ENTITY("Uacute", 218),
+            ENTITY("Ucirc", 219),
+            ENTITY("Ugrave", 217),
+            ENTITY("Upsilon", 933),
+            ENTITY("Uuml", 220),
+            ENTITY("Xi", 926),
+            ENTITY("Yacute", 221),
+            ENTITY("Yuml", 376),
+            ENTITY("Zeta", 918),
+            ENTITY("aacute", 225),
+            ENTITY("acirc", 226),
+            ENTITY("acute", 180),
+            ENTITY("aelig", 230),
+            ENTITY("agrave", 224),
+            ENTITY("alefsym", 8501),
+            ENTITY("alpha", 945),
+            ENTITY("amp", 38),
+            ENTITY("and", 8743),
+            ENTITY("ang", 8736),
+            ENTITY("aring", 229),
+            ENTITY("asymp", 8776),
+            ENTITY("atilde", 227),
+            ENTITY("auml", 228),
+            ENTITY("bdquo", 8222),
+            ENTITY("beta", 946),
+            ENTITY("brvbar", 166),
+            ENTITY("bull", 8226),
+            ENTITY("cap", 8745),
+            ENTITY("ccedil", 231),
+            ENTITY("cedil", 184),
+            ENTITY("cent", 162),
+            ENTITY("chi", 967),
+            ENTITY("circ", 710),
+            ENTITY("clubs", 9827),
+            ENTITY("cong", 8773),
+            ENTITY("copy", 169),
+            ENTITY("crarr", 8629),
+            ENTITY("cup", 8746),
+            ENTITY("curren", 164),
+            ENTITY("dArr", 8659),
+            ENTITY("dagger", 8224),
+            ENTITY("darr", 8595),
+            ENTITY("deg", 176),
+            ENTITY("delta", 948),
+            ENTITY("diams", 9830),
+            ENTITY("divide", 247),
+            ENTITY("eacute", 233),
+            ENTITY("ecirc", 234),
+            ENTITY("egrave", 232),
+            ENTITY("empty", 8709),
+            ENTITY("emsp", 8195),
+            ENTITY("ensp", 8194),
+            ENTITY("epsilon", 949),
+            ENTITY("equiv", 8801),
+            ENTITY("eta", 951),
+            ENTITY("eth", 240),
+            ENTITY("euml", 235),
+            ENTITY("euro", 8364),
+            ENTITY("exist", 8707),
+            ENTITY("fnof", 402),
+            ENTITY("forall", 8704),
+            ENTITY("frac12", 189),
+            ENTITY("frac14", 188),
+            ENTITY("frac34", 190),
+            ENTITY("frasl", 8260),
+            ENTITY("gamma", 947),
+            ENTITY("ge", 8805),
+            ENTITY("gt", 62),
+            ENTITY("hArr", 8660),
+            ENTITY("harr", 8596),
+            ENTITY("hearts", 9829),
+            ENTITY("hellip", 8230),
+            ENTITY("iacute", 237),
+            ENTITY("icirc", 238),
+            ENTITY("iexcl", 161),
+            ENTITY("igrave", 236),
+            ENTITY("image", 8465),
+            ENTITY("infin", 8734),
+            ENTITY("int", 8747),
+            ENTITY("iota", 953),
+            ENTITY("iquest", 191),
+            ENTITY("isin", 8712),
+            ENTITY("iuml", 239),
+            ENTITY("kappa", 954),
+            ENTITY("lArr", 8656),
+            ENTITY("lambda", 955),
+            ENTITY("lang", 9001),
+            ENTITY("laquo", 171),
+            ENTITY("larr", 8592),
+            ENTITY("lceil", 8968),
+            ENTITY("ldquo", 8220),
+            ENTITY("le", 8804),
+            ENTITY("lfloor", 8970),
+            ENTITY("lowast", 8727),
+            ENTITY("loz", 9674),
+            ENTITY("lrm", 8206),
+            ENTITY("lsaquo", 8249),
+            ENTITY("lsquo", 8216),
+            ENTITY("lt", 60),
+            ENTITY("macr", 175),
+            ENTITY("mdash", 8212),
+            ENTITY("micro", 181),
+            ENTITY("middot", 183),
+            ENTITY("minus", 8722),
+            ENTITY("mu", 956),
+            ENTITY("nabla", 8711),
+            ENTITY("nbsp", 160),
+            ENTITY("ndash", 8211),
+            ENTITY("ne", 8800),
+            ENTITY("ni", 8715),
+            ENTITY("not", 172),
+            ENTITY("notin", 8713),
+            ENTITY("nsub", 8836),
+            ENTITY("ntilde", 241),
+            ENTITY("nu", 957),
+            ENTITY("oacute", 243),
+            ENTITY("ocirc", 244),
+            ENTITY("oelig", 339),
+            ENTITY("ograve", 242),
+            ENTITY("oline", 8254),
+            ENTITY("omega", 969),
+            ENTITY("omicron", 959),
+            ENTITY("oplus", 8853),
+            ENTITY("or", 8744),
+            ENTITY("ordf", 170),
+            ENTITY("ordm", 186),
+            ENTITY("oslash", 248),
+            ENTITY("otilde", 245),
+            ENTITY("otimes", 8855),
+            ENTITY("ouml", 246),
+            ENTITY("para", 182),
+            ENTITY("part", 8706),
+            ENTITY("permil", 8240),
+            ENTITY("perp", 8869),
+            ENTITY("phi", 966),
+            ENTITY("pi", 960),
+            ENTITY("piv", 982),
+            ENTITY("plusmn", 177),
+            ENTITY("pound", 163),
+            ENTITY("prime", 8242),
+            ENTITY("prod", 8719),
+            ENTITY("prop", 8733),
+            ENTITY("psi", 968),
+            ENTITY("quot", 34),
+            ENTITY("rArr", 8658),
+            ENTITY("radic", 8730),
+            ENTITY("rang", 9002),
+            ENTITY("raquo", 187),
+            ENTITY("rarr", 8594),
+            ENTITY("rceil", 8969),
+            ENTITY("rdquo", 8221),
+            ENTITY("real", 8476),
+            ENTITY("reg", 174),
+            ENTITY("rfloor", 8971),
+            ENTITY("rho", 961),
+            ENTITY("rlm", 8207),
+            ENTITY("rsaquo", 8250),
+            ENTITY("rsquo", 8217),
+            ENTITY("sbquo", 8218),
+            ENTITY("scaron", 353),
+            ENTITY("sdot", 8901),
+            ENTITY("sect", 167),
+            ENTITY("shy", 173),
+            ENTITY("sigma", 963),
+            ENTITY("sigmaf", 962),
+            ENTITY("sim", 8764),
+            ENTITY("spades", 9824),
+            ENTITY("sub", 8834),
+            ENTITY("sube", 8838),
+            ENTITY("sum", 8721),
+            ENTITY("sup", 8835),
+            ENTITY("sup1", 185),
+            ENTITY("sup2", 178),
+            ENTITY("sup3", 179),
+            ENTITY("supe", 8839),
+            ENTITY("szlig", 223),
+            ENTITY("tau", 964),
+            ENTITY("there4", 8756),
+            ENTITY("theta", 952),
+            ENTITY("thetasym", 977),
+            ENTITY("thinsp", 8201),
+            ENTITY("thorn", 254),
+            ENTITY("tilde", 732),
+            ENTITY("times", 215),
+            ENTITY("trade", 8482),
+            ENTITY("uArr", 8657),
+            ENTITY("uacute", 250),
+            ENTITY("uarr", 8593),
+            ENTITY("ucirc", 251),
+            ENTITY("ugrave", 249),
+            ENTITY("uml", 168),
+            ENTITY("upsih", 978),
+            ENTITY("upsilon", 965),
+            ENTITY("uuml", 252),
+            ENTITY("weierp", 8472),
+            ENTITY("xi", 958),
+            ENTITY("yacute", 253),
+            ENTITY("yen", 165),
+            ENTITY("yuml", 255),
+            ENTITY("zeta", 950),
+            ENTITY("zwj", 8205),
+            ENTITY("zwnj", 8204),
             {NULL, 0}};
+        #undef ENTITY
         static size_t substitutions_cnt = 0;
 
         if (substitutions_cnt == 0)
@@ -844,7 +853,7 @@ wxChar wxHtmlEntitiesParser::GetEntityChar(const wxString& entity)
             }
         }
 #else
-        info = (wxHtmlEntityInfo*) bsearch(entity.c_str(), substitutions,
+        info = (wxHtmlEntityInfo*) bsearch(entity.wx_str(), substitutions,
                                            substitutions_cnt,
                                            sizeof(wxHtmlEntityInfo),
                                            wxHtmlEntityCompare);
@@ -879,7 +888,7 @@ public:
     wxObject* GetProduct() { return NULL; }
 
 protected:
-    virtual void AddText(const wxChar* WXUNUSED(txt)) {}
+    virtual void AddText(const wxString& WXUNUSED(txt)) {}
 
     DECLARE_NO_COPY_CLASS(wxMetaTagParser)
 };