anne/lyx_mirror
1
0
Fork 0
mirror of https://git.lyx.org/repos/lyx.git synced 2024-12-22 05:16:21 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

MathML: Convert HTML entities to XML entities.

Browse Source
This commit is contained in:
Thibaut Cuvelier 2020-06-08 04:07:47 +02:00 committed by Pavel Sanda
parent 054bdc5d18
commit 718eede94b
17 changed files with 958 additions and 915 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1664
lib/symbols
View File

File diff suppressed because it is too large Load Diff

4
src/mathed/InsetMathBig.cpp
View File

@ -139,7 +139,7 @@ void InsetMathBig::mathmlize(MathStream & ms) const
{
ms << "<" << from_ascii(ms.namespacedTag("mo"))
<< " form='prefix' fence='true' stretchy='true' symmetric='true'>"
<< convertDelimToXMLEscape(delim_)
<< convertDelimToXMLEscape(delim_, ms.xmlMode())
<< "</" << from_ascii(ms.namespacedTag("mo")) << ">";
}
@ -154,7 +154,7 @@ void InsetMathBig::htmlize(HtmlStream & os) const
default: name = "big"; break;
}
os << MTag("span", "class='" + name + "symbol'")
<< convertDelimToXMLEscape(delim_)
<< convertDelimToXMLEscape(delim_, false)
<< ETag("span");
}

5
src/mathed/InsetMathChar.cpp
View File

@ -224,7 +224,10 @@ void InsetMathChar::mathmlize(MathStream & ms) const
case '>': entity = "&gt;"; break;
case '&': entity = "&amp;"; break;
case ' ': {
ms << from_ascii("&nbsp;");
if (ms.xmlMode())
ms << from_ascii("&#0160;");
else
ms << from_ascii("&nbsp;");
return;
}
default: break;

64
src/mathed/InsetMathDecoration.cpp
View File

@ -169,45 +169,46 @@ void InsetMathDecoration::infoize(odocstream & os) const
namespace {
struct Attributes {
Attributes() : over(false) {}
Attributes(bool o, string const & t)
: over(o), tag(t) {}
Attributes(bool o, string const & t, string const & entity)
: over(o), tag(t), entity(entity) {}
bool over;
string tag;
string entity;
};
typedef map<string, Attributes> TranslationMap;
void buildTranslationMap(TranslationMap & t) {
// the decorations we need to support are listed in lib/symbols
t["acute"] = Attributes(true, "&acute;");
t["bar"] = Attributes(true, "&OverBar;");
t["breve"] = Attributes(true, "&breve;");
t["check"] = Attributes(true, "&caron;");
t["ddddot"] = Attributes(true, "&DotDot;");
t["dddot"] = Attributes(true, "&TripleDot;");
t["ddot"] = Attributes(true, "&Dot;");
t["dot"] = Attributes(true, "&dot;");
t["grave"] = Attributes(true, "&grave;");
t["hat"] = Attributes(true, "&circ;");
t["mathring"] = Attributes(true, "&ring;");
t["overbrace"] = Attributes(true, "&OverBrace;");
t["overleftarrow"] = Attributes(true, "&xlarr;");
t["overleftrightarrow"] = Attributes(true, "&xharr;");
t["overline"] = Attributes(true, "&macr;");
t["overrightarrow"] = Attributes(true, "&xrarr;");
t["tilde"] = Attributes(true, "&tilde;");
t["underbar"] = Attributes(false, "&UnderBar;");
t["underbrace"] = Attributes(false, "&UnderBrace;");
t["underleftarrow"] = Attributes(false, "&xlarr;");
t["underleftrightarrow"] = Attributes(false, "&xharr;");
t["acute"] = Attributes(true, "&acute;", "&#x00B4;");
t["bar"] = Attributes(true, "&OverBar;", "&#x00AF;");
t["breve"] = Attributes(true, "&breve;", "&#x02D8;");
t["check"] = Attributes(true, "&caron;", "&#x02C7;");
t["ddddot"] = Attributes(true, "&DotDot;", "&#x20DC;");
t["dddot"] = Attributes(true, "&TripleDot;", "&#x20DB;");
t["ddot"] = Attributes(true, "&Dot;", "&#x00A8;");
t["dot"] = Attributes(true, "&dot;", "&#x02D9;");
t["grave"] = Attributes(true, "&grave;", "&#x0060;");
t["hat"] = Attributes(true, "&circ;", "&#x02C6;");
t["mathring"] = Attributes(true, "&ring;", "&#x02DA;");
t["overbrace"] = Attributes(true, "&OverBrace;", "&#xFE37;");
t["overleftarrow"] = Attributes(true, "&xlarr;", "&#x27F5;");
t["overleftrightarrow"] = Attributes(true, "&xharr;", "&#x27F7;");
t["overline"] = Attributes(true, "&macr;", "&#x00AF;");
t["overrightarrow"] = Attributes(true, "&xrarr;", "&#x27F6;");
t["tilde"] = Attributes(true, "&tilde;", "&#x02DC;");
t["underbar"] = Attributes(false, "&UnderBar;", "&#x0332;");
t["underbrace"] = Attributes(false, "&UnderBrace;", "&#xFE38;");
t["underleftarrow"] = Attributes(false, "&xlarr;", "&#x27F5;");
t["underleftrightarrow"] = Attributes(false, "&xharr;", "&#x27F7;");
// this is the macron, again, but it works
t["underline"] = Attributes(false, "&macr;");
t["underrightarrow"] = Attributes(false, "&xrarr;");
t["undertilde"] = Attributes(false, "&Tilde;");
t["utilde"] = Attributes(false, "&Tilde;");
t["vec"] = Attributes(true, "&rarr;");
t["widehat"] = Attributes(true, "&Hat;");
t["widetilde"] = Attributes(true, "&Tilde;");
t["underline"] = Attributes(false, "&macr;", "&#x00AF;");
t["underrightarrow"] = Attributes(false, "&xrarr;", "&#x27F6;");
t["undertilde"] = Attributes(false, "&Tilde;", "&#x223C;");
t["utilde"] = Attributes(false, "&Tilde;", "&#x223C;");
t["vec"] = Attributes(true, "&rarr;", "&#x2192;");
t["widehat"] = Attributes(true, "&Hat;", "&#x005E;");
t["widetilde"] = Attributes(true, "&Tilde;", "&#x223C;");
}
TranslationMap const & translationMap() {
@ -224,10 +225,11 @@ void InsetMathDecoration::mathmlize(MathStream & ms) const
TranslationMap::const_iterator cur = t.find(to_utf8(key_->name));
LASSERT(cur != t.end(), return);
char const * const outag = cur->second.over ? "mover" : "munder";
std::string decoration = ms.xmlMode() ? cur->second.entity : cur->second.tag;
ms << MTag(outag)
<< MTag("mrow") << cell(0) << ETag("mrow")
<< "<" << from_ascii(ms.namespacedTag("mo")) << " stretchy=\"true\">"
<< from_ascii(cur->second.tag)
<< from_ascii(decoration)
<< "</" << from_ascii(ms.namespacedTag("mo")) << ">"
<< ETag(outag);
}

10
src/mathed/InsetMathDelim.cpp
View File

@ -170,7 +170,7 @@ void InsetMathDelim::mathematica(MathematicaStream & os) const
{
if (isAbs()) {
if (cell(0).size() == 1 && cell(0).front()->asMatrixInset())
os << "Det" << cell(0) << ']';
os << "Det[" << cell(0) << ']';
else
os << "Abs[" << cell(0) << ']';
}
@ -183,7 +183,7 @@ void InsetMathDelim::mathmlize(MathStream & ms) const
{
ms << "<" << from_ascii(ms.namespacedTag("mo")) << " form='prefix' fence='true' stretchy='true' symmetric='true'>"
<< "<" << from_ascii(ms.namespacedTag("mrow")) << ">"
<< convertDelimToXMLEscape(left_)
<< convertDelimToXMLEscape(left_, ms.xmlMode())
<< "</" << from_ascii(ms.namespacedTag("mrow")) << ">"
<< "</" << from_ascii(ms.namespacedTag("mo")) << ">"
<< "\n"
@ -191,7 +191,7 @@ void InsetMathDelim::mathmlize(MathStream & ms) const
<< "\n"
<< "<" << from_ascii(ms.namespacedTag("mo")) << " form='postfix' fence='true' stretchy='true' symmetric='true'>"
<< "<" << from_ascii(ms.namespacedTag("mrow")) << ">"
<< convertDelimToXMLEscape(right_)
<< convertDelimToXMLEscape(right_, ms.xmlMode())
<< "</" << from_ascii(ms.namespacedTag("mrow")) << ">"
<< "</" << from_ascii(ms.namespacedTag("mo")) << ">"
<< "\n";
@ -200,9 +200,9 @@ void InsetMathDelim::mathmlize(MathStream & ms) const
void InsetMathDelim::htmlize(HtmlStream & os) const
{
os << convertDelimToXMLEscape(left_)
os << convertDelimToXMLEscape(left_, false)
<< cell(0)
<< convertDelimToXMLEscape(right_);
<< convertDelimToXMLEscape(right_, false);
}

37
src/mathed/InsetMathDots.cpp
View File

@ -91,18 +91,31 @@ void InsetMathDots::mathmlize(MathStream & ms) const
// lib/symbols as generating a dots inset
docstring const & n = key_->name;
std::string ent;
if (n == "dots" || n == "dotsc" || n == "dotso" || n == "ldots")
ent = "&hellip;";
else if (n == "adots" || n == "iddots")
ent = "&utdot;";
else if (n == "cdots" || n == "dotsb" || n == "dotsi" || n == "dotsm")
ent = "&ctdot;";
else if (n == "ddots")
ent = "&dtdot;";
else if (n == "vdots")
ent = "&vellip;";
else
LASSERT(false, ent = "&hellip;");
if (ms.xmlMode()) {
if (n == "dots" || n == "dotsc" || n == "dotso" || n == "ldots")
ent = "&#x2026;";
else if (n == "adots" || n == "iddots")
ent = "&#x22F0;";
else if (n == "cdots" || n == "dotsb" || n == "dotsi" || n == "dotsm")
ent = "&#x22EF;";
else if (n == "ddots")
ent = "&#x22F1;";
else if (n == "vdots")
ent = "&#x22EE;";
else LASSERT(false, ent = "&#x2026;");
} else {
if (n == "dots" || n == "dotsc" || n == "dotso" || n == "ldots")
ent = "&hellip;";
else if (n == "adots" || n == "iddots")
ent = "&utdot;";
else if (n == "cdots" || n == "dotsb" || n == "dotsi" || n == "dotsm")
ent = "&ctdot;";
else if (n == "ddots")
ent = "&dtdot;";
else if (n == "vdots")
ent = "&vellip;";
else LASSERT(false, ent = "&hellip;");
}
ms << MTag("mi") << from_ascii(ent) << ETag("mi");
}

15
src/mathed/InsetMathHull.cpp
View File

@ -2406,8 +2406,13 @@ int InsetMathHull::plaintext(odocstringstream & os,
int InsetMathHull::docbook(odocstream & os, OutputParams const & runparams) const
{
MathStream ms(os);
// With DocBook 5, MathML must be within its own namespace; defined in Buffer.cpp::writeDocBookSource as "m".
// Output everything in a separate stream so that this does not interfere with the standard flow of DocBook tags.
odocstringstream osmath;
MathStream ms(osmath, "m", true);
int res = 0;
// Choose the tag around the MathML equation.
docstring name;
if (getType() == hullSimple)
name = from_ascii("inlineequation");
@ -2415,8 +2420,12 @@ int InsetMathHull::docbook(odocstream & os, OutputParams const & runparams) cons
name = from_ascii("informalequation");
docstring bname = name;
if (!label(0).empty())
bname += " id='" + xml::cleanID(label(0)) + "'";
for (row_type i = 0; i < nrows(); ++i) {
if (!label(i).empty()) {
bname += " xml:id=\"" + xml::cleanID(label(i)) + "\"";
break;
}
}
++ms.tab(); ms.cr(); ms.os() << '<' << bname << '>';

4
src/mathed/InsetMathMacro.cpp
View File

@ -1181,7 +1181,7 @@ void InsetMathMacro::mathmlize(MathStream & ms) const
// macro_ is 0 if this is an unknown macro
LATTEST(d->macro_ || d->displayMode_ != DISPLAY_NORMAL);
if (d->macro_) {
docstring const xmlname = d->macro_->xmlname();
docstring const xmlname = (ms.xmlMode()) ? d->macro_->xmlname() : d->macro_->htmlname();
if (!xmlname.empty()) {
char const * type = d->macro_->MathMLtype();
ms << "<" << from_ascii(ms.namespacedTag(type)) << ">"
@ -1203,7 +1203,7 @@ void InsetMathMacro::htmlize(HtmlStream & os) const
// macro_ is 0 if this is an unknown macro
LATTEST(d->macro_ || d->displayMode_ != DISPLAY_NORMAL);
if (d->macro_) {
docstring const xmlname = d->macro_->xmlname();
docstring const xmlname = d->macro_->htmlname();
if (!xmlname.empty()) {
os << ' ' << xmlname << ' ';
return;

1
src/mathed/InsetMathMacroTemplate.h
View File

@ -57,7 +57,6 @@ public:
int plaintext(odocstringstream &, OutputParams const &, size_t) const;
///
bool inheritFont() const { return false; }
///
docstring name() const;
///

4
src/mathed/InsetMathMatrix.cpp
View File

@ -95,7 +95,7 @@ void InsetMathMatrix::mathematica(MathematicaStream & os) const
void InsetMathMatrix::mathmlize(MathStream & ms) const
{
ms << "<" << from_ascii(ms.namespacedTag("mo")) << " form='prefix' fence='true' stretchy='true' symmetric='true' lspace='thinmathspace'>"
<< convertDelimToXMLEscape(left_)
<< convertDelimToXMLEscape(left_, ms.xmlMode())
<< "</" << from_ascii(ms.namespacedTag("mo")) << ">"
<< MTag("mtable");
for (row_type row = 0; row < nrows(); ++row) {
@ -114,7 +114,7 @@ void InsetMathMatrix::mathmlize(MathStream & ms) const
}
ms << ETag("mtable")
<< "<" << from_ascii(ms.namespacedTag("mo")) << " form='postfix' fence='true' stretchy='true' symmetric='true' lspace='thinmathspace'>"
<< convertDelimToXMLEscape(right_)
<< convertDelimToXMLEscape(right_, ms.xmlMode())
<< "</" << from_ascii(ms.namespacedTag("mo")) << ">";
}

12
src/mathed/InsetMathSymbol.cpp
View File

@ -168,11 +168,13 @@ void InsetMathSymbol::mathmlize(MathStream & ms) const
// with MathMLtype.
docstring tag = from_ascii(ms.namespacedTag(sym_->MathMLtype()));
ms << '<' << tag << ">";
if (sym_->xmlname == "x")
if ((ms.xmlMode() && sym_->xmlname == "x") || (!ms.xmlMode() && sym_->htmlname == "x"))
// unknown so far
ms << name();
else
else if (ms.xmlMode())
ms << sym_->xmlname;
else
ms << sym_->htmlname;
ms << "</" << tag << '>';
}
@ -184,13 +186,13 @@ void InsetMathSymbol::htmlize(HtmlStream & os, bool spacing) const
char const * type = sym_->MathMLtype();
bool op = (std::string(type) == "mo");
if (sym_->xmlname == "x")
if (sym_->htmlname == "x")
// unknown so far
os << ' ' << name() << ' ';
else if (op && spacing)
os << ' ' << sym_->xmlname << ' ';
os << ' ' << sym_->htmlname << ' ';
else
os << sym_->xmlname;
os << sym_->htmlname;
}

8
src/mathed/MacroTable.cpp
View File

@ -126,6 +126,14 @@ bool MacroData::hidden() const
}
docstring const MacroData::htmlname() const
{
if (sym_)
return sym_->htmlname;
return docstring();
}
docstring const MacroData::xmlname() const
{
if (sym_)

2
src/mathed/MacroTable.h
View File

@ -64,6 +64,8 @@ public:
///
bool hidden() const;
///
docstring const htmlname() const;
///
docstring const xmlname() const;
///
char const * MathMLtype() const;

23
src/mathed/MathFactory.cpp
View File

@ -186,23 +186,23 @@ void initSymbols()
// or
// \def\macroname{definition} requires
// or
// \def\macroname{definition} extra xmlname requires
// \def\macroname{definition} extra htmlname requires
istringstream is(line);
string macro;
string required;
string extra;
string xmlname;
string htmlname;
bool hidden = false;
is >> setw(65536) >> macro >> required;
if ((is >> xmlname)) {
if ((is >> htmlname)) {
extra = required;
if (!(is >> required))
required = "";
} else
xmlname = "";
htmlname = "";
MacroTable::iterator it = MacroTable::globalMacros().insert(
0, from_utf8(macro));
if (!extra.empty() || !xmlname.empty() || !required.empty()) {
if (!extra.empty() || !htmlname.empty() || !required.empty()) {
MathWordList::iterator wit = theMathWordList.find(it->first);
if (wit != theMathWordList.end())
LYXERR(Debug::MATHED, "readSymbols: inset "
@ -212,7 +212,7 @@ void initSymbols()
tmp.inset = "macro";
tmp.name = it->first;
tmp.extra = from_utf8(extra);
tmp.xmlname = from_utf8(xmlname);
tmp.htmlname = from_utf8(htmlname);
if (required == "hiddensymbol") {
required = "";
tmp.hidden = hidden = true;
@ -226,10 +226,10 @@ void initSymbols()
// If you change the following output, please adjust
// development/tools/generate_symbols_images.py.
LYXERR(Debug::MATHED, "read symbol '" << to_utf8(it->first)
<< " inset: macro"
<< " draw: 0"
<< " extra: " << extra
<< " xml: " << xmlname
<< " inset: macro"
<< " draw: 0"
<< " extra: " << extra
<< " html: " << htmlname
<< " requires: " << required
<< " hidden: " << hidden << '\'');
continue;
@ -241,7 +241,7 @@ void initSymbols()
is >> tmp.name >> help;
tmp.inset = to_ascii(help);
if (isFontName(tmp.inset))
is >> charid >> fallbackid >> tmp.extra >> tmp.xmlname;
is >> charid >> fallbackid >> tmp.extra >> tmp.htmlname >> tmp.xmlname;
else
is >> tmp.extra;
// requires is optional
@ -327,6 +327,7 @@ void initSymbols()
<< " inset: " << tmp.inset
<< " draw: " << int(tmp.draw.empty() ? 0 : tmp.draw[0])
<< " extra: " << to_utf8(tmp.extra)
<< " html: " << to_utf8(tmp.htmlname)
<< " xml: " << to_utf8(tmp.xmlname)
<< " requires: " << tmp.required
<< " hidden: " << tmp.hidden << '\'');

2
src/mathed/MathParser.h
View File

@ -54,6 +54,8 @@ public:
docstring draw;
/// operator/..., fontname e
docstring extra;
/// how is this called as HTML entity in MathML?
docstring htmlname;
/// how is this called as XML entity in MathML?
docstring xmlname;
/// required LaTeXFeatures

10
src/mathed/MathStream.cpp
View File

@ -269,8 +269,8 @@ WriteStream & operator<<(WriteStream & ws, unsigned int i)
//////////////////////////////////////////////////////////////////////
MathStream::MathStream(odocstream & os, std::string xmlns)
: os_(os), tab_(0), line_(0), in_text_(false), xmlns_(xmlns)
MathStream::MathStream(odocstream & os, std::string xmlns, bool xmlMode)
: os_(os), tab_(0), line_(0), in_text_(false), xmlns_(xmlns), xml_mode_(xmlMode)
{}
@ -709,8 +709,9 @@ OctaveStream & operator<<(OctaveStream & os, string const & s)
}
docstring convertDelimToXMLEscape(docstring const & name)
docstring convertDelimToXMLEscape(docstring const & name, bool xmlmode)
{
// For the basic symbols, no difference between XML and HTML.
if (name.size() == 1) {
char_type const c = name[0];
if (c == '<')
@ -729,7 +730,8 @@ docstring convertDelimToXMLEscape(docstring const & name)
MathWordList const & words = mathedWordList();
MathWordList::const_iterator it = words.find(name);
if (it != words.end()) {
docstring const escape = it->second.xmlname;
// Only difference between XML and HTML, based on the contents read by MathFactory.
docstring const escape = xmlmode ? it->second.xmlname : it->second.htmlname;
return escape;
}
LYXERR0("Unable to find `" << name <<"' in the mathWordList.");

8
src/mathed/MathStream.h
View File

@ -340,7 +340,7 @@ class MathExportException : public std::exception {};
class MathStream {
public:
/// Builds a stream proxy for os; the MathML namespace is given by xmlns (supposed to be already defined elsewhere in the document).
explicit MathStream(odocstream & os, std::string xmlns="");
explicit MathStream(odocstream & os, std::string xmlns="", bool xmlMode=false);
///
void cr();
///
@ -361,6 +361,8 @@ public:
bool inText() const { return in_text_; }
///
std::string xmlns() const { return xmlns_; }
///
bool xmlMode() const { return xml_mode_; }
/// Returns the tag name prefixed by the name space if needed.
std::string namespacedTag(std::string tag) const { return ((xmlns().empty()) ? "" : xmlns() + ":") + tag; }
private:
@ -379,6 +381,8 @@ private:
///
std::string xmlns_;
///
bool xml_mode_;
///
friend class SetMode;
};
@ -646,7 +650,7 @@ OctaveStream & operator<<(OctaveStream &, char);
OctaveStream & operator<<(OctaveStream &, int);
docstring convertDelimToXMLEscape(docstring const & name);
docstring convertDelimToXMLEscape(docstring const & name, bool xmlmode);
} // namespace lyx