new HTML tags parser and entities substitution code

git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@10744 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
This commit is contained in:
Václav Slavík
2001-07-01 15:09:35 +00:00
parent ec7c3e898a
commit daa616fca0
6 changed files with 619 additions and 479 deletions

View File

@@ -157,160 +157,12 @@ bool HP_TagHandler::HandleTag(const wxHtmlTag& tag)
}
else
{ // "PARAM"
if (m_Name == wxEmptyString && tag.GetParam(wxT("NAME")) == wxT("Name"))
{
if (m_Name == wxEmptyString && tag.GetParam(wxT("NAME")) == wxT("Name"))
m_Name = tag.GetParam(wxT("VALUE"));
if (m_Name.Find(wxT('&')) != -1)
{
#define ESCSEQ(escape, subst) \
{ _T("&") _T(escape) _T(";"), _T("&") _T(escape) _T(" "), _T("&") _T(escape), _T(subst) }
static wxChar* substitutions[][4] =
{
ESCSEQ("quot", "\""),
ESCSEQ("#34", "\""),
ESCSEQ("#8220", "\""),
ESCSEQ("#8221", "\""),
ESCSEQ("lt", "<"),
ESCSEQ("#60", "<"),
ESCSEQ("gt", ">"),
ESCSEQ("#62", ">"),
ESCSEQ("#94", "^"), /* ^ */
ESCSEQ("nbsp", " "),
ESCSEQ("#32", " "),
ESCSEQ("iexcl", "!"),
ESCSEQ("#33", "!"),
ESCSEQ("cent", "<EFBFBD>"/* <20> */),
ESCSEQ("#162", "<EFBFBD>"/* <20> */),
ESCSEQ("trade", "(TM)"),
ESCSEQ("#153", "(TM)"),
ESCSEQ("#8482", "(TM)"),
ESCSEQ("yen", "<EFBFBD>"),
ESCSEQ("#165", "<EFBFBD>"),
ESCSEQ("brkbar", "<EFBFBD>"),
ESCSEQ("#166", "<EFBFBD>"),
ESCSEQ("sect", "<EFBFBD>"),
ESCSEQ("#167", "<EFBFBD>"),
ESCSEQ("uml", "<EFBFBD>"),
ESCSEQ("#168", "<EFBFBD>"),
ESCSEQ("copy", "<EFBFBD>"), /* <20> */
ESCSEQ("#169", "<EFBFBD>"),
ESCSEQ("ordf", "<EFBFBD>"),
ESCSEQ("#170", "<EFBFBD>"),
ESCSEQ("laquo", "<EFBFBD>"), /* <20> */
ESCSEQ("#171", "<EFBFBD>"),
ESCSEQ("not", "<EFBFBD>"),
ESCSEQ("#172", "<EFBFBD>"),
ESCSEQ("reg", "<EFBFBD>"), /* <20> */
ESCSEQ("#174", "<EFBFBD>"),
ESCSEQ("deg", "<EFBFBD>"), /* <20> */
ESCSEQ("#176", "<EFBFBD>"),
ESCSEQ("plusm", "<EFBFBD>"), /* <20> */
ESCSEQ("#177", "<EFBFBD>"),
ESCSEQ("acute", "<EFBFBD>"),
ESCSEQ("#180", "<EFBFBD>"),
ESCSEQ("macron", "<EFBFBD>"),
ESCSEQ("#175", "<EFBFBD>"),
ESCSEQ("micro", "<EFBFBD>"), /* <20> */
ESCSEQ("#181", "<EFBFBD>"),
ESCSEQ("para", "<EFBFBD>"), /* <20> */
ESCSEQ("#182", "<EFBFBD>"),
ESCSEQ("ordm", "<EFBFBD>"), /* <20> */
ESCSEQ("#186", "<EFBFBD>"),
ESCSEQ("raquo", "<EFBFBD>"), /* <20> */
ESCSEQ("#187", "<EFBFBD>"),
ESCSEQ("iquest", "<EFBFBD>"), /* <20> */
ESCSEQ("#191", "<EFBFBD>"),
ESCSEQ("Agrave", "\300"/* <20> */),
ESCSEQ("#193", "\300"/* <20> */),
ESCSEQ("Acirc", "\302"/* <20> */),
ESCSEQ("Atilde", "\303"/* <20> */),
ESCSEQ("Auml", "\304"/* <20> */),
ESCSEQ("Aring", " "),
ESCSEQ("AElig", " "),
ESCSEQ("Ccedil", "\347"/* <20> */),
ESCSEQ("Egrave", "\310"/* <20> */),
ESCSEQ("Eacute", "\311"/* <20> */),
ESCSEQ("Ecirc", "\312"/* <20> */),
ESCSEQ("Euml", "\313"/* <20> */),
ESCSEQ("Igrave", "\314"/* <20> */),
ESCSEQ("Icirc", "\316"/* <20> */),
ESCSEQ("Iuml", "\317"/* <20> */),
ESCSEQ("Ntilde", "\321"/* <20> */),
ESCSEQ("Ograve", "\322"/* <20> */),
ESCSEQ("Ocirc", "\324"/* <20> */),
ESCSEQ("Otilde", "\325"/* <20> */),
ESCSEQ("Ouml", "\326"/* <20> */),
ESCSEQ("Oslash", " "),
ESCSEQ("Ugrave", "\331"/* <20> */),
ESCSEQ("Ucirc", " "),
ESCSEQ("Uuml", "\334"/* <20> */),
ESCSEQ("szlig", "\247"/* <20> */),
ESCSEQ("agrave","\340"/* <20> */),
ESCSEQ("aacute", "\341"/* <20> */),
ESCSEQ("acirc", "\342"/* <20> */),
ESCSEQ("atilde", "\343"/* <20> */),
ESCSEQ("auml", "\344"/* <20> */),
ESCSEQ("aring", "a"),
ESCSEQ("aelig", "ae"),
ESCSEQ("ccedil", "\347"/* <20> */),
ESCSEQ("egrave", "\350"/* <20> */),
ESCSEQ("eacute", "\351"/* <20> */),
ESCSEQ("ecirc", "\352"/* <20> */),
ESCSEQ("euml", "\353"/* <20> */),
ESCSEQ("igrave", "\354"/* <20> */),
ESCSEQ("iacute", "\355"/* <20> */),
ESCSEQ("icirc", " "),
ESCSEQ("iuml", "\357"/* <20> */),
ESCSEQ("eth", " "),
ESCSEQ("ntilde", "\361"/* <20> */),
ESCSEQ("ograve", "\362"/* <20> */),
ESCSEQ("oacute", "\363"/* <20> */),
ESCSEQ("ocirc", "\364"/* <20> */),
ESCSEQ("otilde", "\365"/* <20> */),
ESCSEQ("ouml", "\366"/* <20> */),
ESCSEQ("divide", " "),
ESCSEQ("oslash", " "),
ESCSEQ("ugrave", "\371"/* <20> */),
ESCSEQ("uacute", "\372"/* <20> */),
ESCSEQ("ucirc", "\373"/* <20> */),
ESCSEQ("uuml", "\374"/* <20> */),
ESCSEQ("yuml", ""),
/* this one should ALWAYS stay the last one!!! */
ESCSEQ("amp", "&"),
ESCSEQ("#38", "&"),
{ NULL, NULL, NULL }
};
for (int i = 0; substitutions[i][0] != NULL; i++)
{
m_Name.Replace(substitutions[i][0], substitutions[i][3], TRUE);
m_Name.Replace(substitutions[i][1], substitutions[i][3], TRUE);
m_Name.Replace(substitutions[i][2], substitutions[i][3], TRUE);
}
}
}
if (tag.GetParam(wxT("NAME")) == wxT("Local")) m_Page = tag.GetParam(wxT("VALUE"));
if (tag.GetParam(wxT("NAME")) == wxT("ID")) tag.ScanParam(wxT("VALUE"), wxT("%i"), &m_ID);
if (tag.GetParam(wxT("NAME")) == wxT("Local"))
m_Page = tag.GetParam(wxT("VALUE"));
if (tag.GetParam(wxT("NAME")) == wxT("ID"))
tag.ScanParam(wxT("VALUE"), wxT("%i"), &m_ID);
return FALSE;
}
}