MathML: let the user change the MathML version.

Discussed in https://www.lyx.org/trac/ticket/13058.

development/FORMAT

@@ -7,7 +7,12 @@ changes happened in particular if possible. A good example would be
 
 -----------------------
 
-2024-08-25 Jürgen Spitzmüller <spitz@lyx.org> 
+2024-09-29 Thibaut Cuvelier <tcuvelier@lyx.org>
+    * Format incremented to 631: Add support for MathML versions, which
+      are stored separately for DocBook and XHTML (\docbook_mathml_version,
+      which is new, and \html_math_output, which receives a new value).
+
+2024-08-25 Jürgen Spitzmüller <spitz@lyx.org>
 	* Format incremented to 630: Add support for a number of languages in
 	  babel and polyglossia:
 	  Polyglossia: Chinese (Simplified), Chinese (Traditional), Japanese, Kurdish (Sorani),

lib/lyx2lyx/lyx_2_5.py

@@ -1015,6 +1015,48 @@ def revert_new_babel_languages(document):
         if document.language == "hebrew" or find_token(document.body, "\\lang oldrussian", 0) != -1:
             add_to_preamble(document, ["\\PassOptionsToPackage{provide*=*}{babel}"])
 
+
+def convert_mathml_version(document):
+    """Add MathML version header for DocBook to use MathML 3 preferably.
+
+    For cleanliness, add this header close to other DocBook headers if present.
+
+    Leave XHTML alone, as the default value is still probably what the user wants (MathML Core)."""
+
+    i = find_token(document.header, "\\docbook", 0)
+    if i == -1:
+        document.header += ["\\docbook_mathml_version 0"]
+    else:
+        document.header.insert(i + 1, "\\docbook_mathml_version 0")
+
+
+def revert_mathml_version(document):
+    """Remove MathML version header.
+
+    For XHTML, only remove the value 4 for \html_math_output (MathML 3) and replace it with 0
+    (MathML Core with format 631+, MathML for 630-).
+
+    For DocBook, totally remove the header (the default with 630- is MathML)."""
+
+    while True:
+        i = find_token(document.header, "\\html_math_output", 0)
+        if i == -1:
+            # nothing to do
+            break
+
+        # remove XHTML header if using the new value, leave alone otherwise.
+        if "4" in document.header:
+            document.header[i] = "\\html_math_output 0"
+
+    while True:
+        i = find_token(document.header, "\\docbook_mathml_version", 0)
+        if i == -1:
+            # nothing to do
+            return
+
+        # remove header
+        del document.header[i]
+
 ##
 # Conversion hub
 #
@@ -1030,11 +1072,13 @@ convert = [
     [627, [convert_nomencl, convert_index_sc]],
     [628, []],
     [629, []],
-    [630, []]
+    [630, []],
+    [631, [convert_mathml_version]]
 ]
 
 
 revert = [
+    [630, [revert_mathml_version]],
     [629, [revert_new_polyglossia_languages, revert_new_babel_languages]],
     [628, [revert_langopts]],
     [627, [revert_nomentbl]],

src/Buffer.cpp

@@ -4472,8 +4472,11 @@ void Buffer::setMathFlavor(OutputParams & op) const
 	// In particular, this function has no impact on the DocBook code, as it
 	// uses another mechanism to handle math flavours.
 	switch (params().html_math_output) {
-	case BufferParams::MathML:
-		op.math_flavor = OutputParams::MathAsMathML;
+	case BufferParams::MathMLCore:
+		op.math_flavor = OutputParams::MathAsMathMLCore;
+		break;
+	case BufferParams::MathML3:
+		op.math_flavor = OutputParams::MathAsMathML3;
 		break;
 	case BufferParams::HTML:
 		op.math_flavor = OutputParams::MathAsHTML;

src/BufferParams.cpp

@@ -480,11 +480,12 @@ BufferParams::BufferParams()
 	// default index
 	indiceslist().addDefault(B_("Index"));
 	html_be_strict = false;
-	html_math_output = MathML;
+	html_math_output = MathMLCore;
 	html_math_img_scale = 1.0;
 	html_css_as_file = false;
 	docbook_table_output = HTMLTable;
 	docbook_mathml_prefix = MPrefix;
+	docbook_mathml_version = MathMLStream::mathml3;
 	display_pixel_ratio = 1.0;
 
 	shell_escape = false;
@@ -1226,6 +1227,10 @@ string BufferParams::readToken(Lexer & lex, string const & token,
 		int temp;
 		lex >> temp;
 		docbook_mathml_prefix = static_cast<MathMLNameSpacePrefix>(temp);
+	} else if (token == "\\docbook_mathml_version") {
+		int temp;
+		lex >> temp;
+		docbook_mathml_version = static_cast<MathMLStream::MathMLVersion>(temp);
 	} else if (token == "\\output_sync") {
 		lex >> output_sync;
 	} else if (token == "\\output_sync_macro") {
@@ -1621,6 +1626,7 @@ void BufferParams::writeFile(ostream & os, Buffer const * buf) const
 
 	os << "\\docbook_table_output " << docbook_table_output << '\n';
 	os << "\\docbook_mathml_prefix " << docbook_mathml_prefix << '\n';
+	os << "\\docbook_mathml_version " << docbook_mathml_version << '\n';
 
 	if (html_math_img_scale != 1.0)
 		os << "\\html_math_img_scale " << convert<string>(html_math_img_scale) << '\n';

src/BufferParams.h

@@ -28,6 +28,8 @@
 #include <map>
 #include <vector>
 
+#include "mathed/MathStream.h"
+
 namespace lyx {
 
 namespace support {
@@ -587,7 +589,8 @@ public:
 
 	// do not change these values. we rely upon them.
 	enum MathOutput {
-		MathML = 0,
+		MathMLCore = 0,
+		MathML3 = 4,
 		HTML = 1,
 		Images = 2,
 		LaTeX = 3
@@ -625,6 +628,9 @@ public:
 	/// what prefix to use when outputting MathML. present choices are above
 	MathMLNameSpacePrefix docbook_mathml_prefix;
 
+	/// what version of MathML to use for DocBook output (likely different from the version used for XHTML)
+	MathMLStream::MathMLVersion docbook_mathml_version;
+
 	/// allow the LaTeX backend to run external programs
 	bool shell_escape;
 	/// generate output usable for reverse/forward search

src/CutAndPaste.cpp

@@ -658,7 +658,7 @@ void putClipboard(ParagraphList const & paragraphs,
 	// We don't want to produce images that are not used. Therefore,
 	// output formulas as MathML. Even if this is not understood by all
 	// applications, the number that can parse it should go up in the future.
-	buffer->params().html_math_output = BufferParams::MathML;
+	buffer->params().html_math_output = BufferParams::MathMLCore;
 
 	// Copy authors to the params. We need those pointers.
 	for (Author const & a : bp.authors())

src/OutputParams.h

@@ -54,7 +54,8 @@ class OutputParams {
 public:
 	enum MathFlavor {
 		NotApplicable,
-		MathAsMathML,
+		MathAsMathMLCore,
+		MathAsMathML3,
 		MathAsHTML,
 		MathAsImages,
 		MathAsLaTeX

src/frontends/qt/GuiDocument.cpp

@@ -952,6 +952,8 @@ GuiDocument::GuiDocument(GuiView & lv)
 		this, SLOT(change_adaptor()));
 	connect(outputModule->mathmlprefixCB, SIGNAL(currentIndexChanged(int)),
 		this, SLOT(change_adaptor()));
+	connect(outputModule->mathmlverCB, SIGNAL(currentIndexChanged(int)),
+		this, SLOT(change_adaptor()));
 
 	connect(outputModule->shellescapeCB, SIGNAL(stateChanged(int)),
 		this, SLOT(shellescapeChanged()));
@@ -4076,6 +4078,12 @@ void GuiDocument::applyView()
 	auto const mp = static_cast<BufferParams::MathMLNameSpacePrefix>(mathmlprefix);
 	bp_.docbook_mathml_prefix = mp;
 
+	int mathmlversion = outputModule->mathmlverCB->currentIndex();
+	if (mathmlversion == -1)
+		mathmlversion = 0;
+	auto const mv = static_cast<MathMLStream::MathMLVersion>(mathmlversion);
+	bp_.docbook_mathml_version = mv;
+
 	bp_.save_transient_properties =
 		outputModule->saveTransientPropertiesCB->isChecked();
 	bp_.postpone_fragile_content =

src/mathed/InsetMathBox.cpp

@@ -119,7 +119,8 @@ void InsetMathBox::validate(LaTeXFeatures & features) const
 	// FIXME XHTML
 	// It'd be better to be able to get this from an InsetLayout, but at present
 	// InsetLayouts do not seem really to work for things that aren't InsetTexts.
-	if (features.runparams().math_flavor == OutputParams::MathAsMathML)
+	if (features.runparams().math_flavor == OutputParams::MathAsMathMLCore ||
+			features.runparams().math_flavor == OutputParams::MathAsMathML3)
 		features.addCSSSnippet("mtext.mathbox { font-style: normal; }");
 	else if (features.runparams().math_flavor == OutputParams::MathAsHTML)
 		features.addCSSSnippet("span.mathbox { font-style: normal; }");
@@ -204,7 +205,8 @@ void InsetMathFBox::validate(LaTeXFeatures & features) const
 	// FIXME XHTML
 	// It'd be better to be able to get this from an InsetLayout, but at present
 	// InsetLayouts do not seem really to work for things that aren't InsetTexts.
-	if (features.runparams().math_flavor == OutputParams::MathAsMathML)
+	if (features.runparams().math_flavor == OutputParams::MathAsMathMLCore ||
+			features.runparams().math_flavor == OutputParams::MathAsMathML3)
 		features.addCSSSnippet(
 			"mtext.fbox { border: 1px solid black; font-style: normal; padding: 0.5ex; }");
 	else if (features.runparams().math_flavor == OutputParams::MathAsHTML)
@@ -343,7 +345,8 @@ void InsetMathMakebox::validate(LaTeXFeatures & features) const
 	// FIXME XHTML
 	// It'd be better to be able to get this from an InsetLayout, but at present
 	// InsetLayouts do not seem really to work for things that aren't InsetTexts.
-	if (features.runparams().math_flavor == OutputParams::MathAsMathML)
+	if (features.runparams().math_flavor == OutputParams::MathAsMathMLCore ||
+			features.runparams().math_flavor == OutputParams::MathAsMathML3)
 		features.addCSSSnippet("mtext.framebox { border: 1px solid black; }");
 	else if (features.runparams().math_flavor == OutputParams::MathAsHTML)
 		features.addCSSSnippet("span.framebox { border: 1px solid black; }");
@@ -423,7 +426,8 @@ void InsetMathBoxed::validate(LaTeXFeatures & features) const
 	// FIXME XHTML
 	// It'd be better to be able to get this from an InsetLayout, but at present
 	// InsetLayouts do not seem really to work for things that aren't InsetTexts.
-	if (features.runparams().math_flavor == OutputParams::MathAsMathML)
+	if (features.runparams().math_flavor == OutputParams::MathAsMathMLCore ||
+			features.runparams().math_flavor == OutputParams::MathAsMathML3)
 		features.addCSSSnippet("mtext.boxed { border: 1px solid black; }");
 	else if (features.runparams().math_flavor == OutputParams::MathAsHTML)
 		features.addCSSSnippet("span.boxed { border: 1px solid black; }");

src/mathed/InsetMathEnsureMath.cpp

@@ -98,7 +98,8 @@ void InsetMathEnsureMath::validate(LaTeXFeatures & features) const
 	// FIXME XHTML
 	// It'd be better to be able to get this from an InsetLayout, but at present
 	// InsetLayouts do not seem really to work for things that aren't InsetTexts.
-	if (features.runparams().math_flavor == OutputParams::MathAsMathML)
+	if (features.runparams().math_flavor == OutputParams::MathAsMathMLCore ||
+			features.runparams().math_flavor == OutputParams::MathAsMathML3)
 		features.addCSSSnippet("mstyle.math { font-style: italic; }");
 	else if (features.runparams().math_flavor == OutputParams::MathAsHTML)
 		features.addCSSSnippet("span.mathbox { font-style: italic; }");

src/mathed/InsetMathHull.cpp

@@ -2655,9 +2655,13 @@ docstring InsetMathHull::xhtml(XMLStream & xs, OutputParams const & op) const
 	}
 
 	// FIXME Eventually we would like to do this inset by inset.
-	if (mathtype == BufferParams::MathML) {
+	if (mathtype == BufferParams::MathML3 || mathtype == BufferParams::MathMLCore) {
+		MathMLStream::MathMLVersion mathml_version = MathMLStream::mathmlCore;
+		if (mathtype == BufferParams::MathML3)
+			mathml_version = MathMLStream::mathml3;
+
 		odocstringstream os;
-		MathMLStream ms(os, "", MathMLStream::mathmlCore);
+		MathMLStream ms(os, "", mathml_version);
 		try {
 			mathmlize(ms);
 			success = true;

src/tex2lyx/Preamble.cpp

@@ -552,7 +552,7 @@ Preamble::Preamble() : one_language(true), explicit_babel(false),
 	// h_font_roman_opts;
 	// h_font_sans_opts;
 	// h_font_typewriter_opts;
-	//h_font_cjk
+	// h_font_cjk
 	h_is_mathindent           = "0";
 	h_math_numbering_side     = "default";
 	h_graphics                = "default";
@@ -562,6 +562,7 @@ Preamble::Preamble() : one_language(true), explicit_babel(false),
 	h_html_math_output        = "0";
 	h_docbook_table_output    = "0";
 	h_docbook_mathml_prefix   = "1";
+	h_docbook_mathml_version  = "0";
 	h_index[0]                = "Index";
 	h_index_command           = "default";
 	h_inputencoding           = "auto-legacy";
@@ -2168,6 +2169,7 @@ bool Preamble::writeLyXHeader(ostream & os, bool subdoc, string const & outfiled
 	   << "\\html_be_strict " << h_html_be_strict << "\n"
 	   << "\\docbook_table_output " << h_docbook_table_output << "\n"
 	   << "\\docbook_mathml_prefix " << h_docbook_mathml_prefix << "\n"
+	   << "\\docbook_mathml_version " << h_docbook_mathml_version << "\n"
 	   << authors_
 	   << "\\end_header\n\n"
 	   << "\\begin_body\n";

src/tex2lyx/Preamble.h

@@ -199,6 +199,7 @@ private:
 	std::string h_html_math_output;
 	std::string h_docbook_table_output;
 	std::string h_docbook_mathml_prefix;
+	std::string h_docbook_mathml_version;
 	std::string h_index[99];
 	std::string h_index_command;
 	std::string h_inputencoding;

src/version.h

@@ -32,8 +32,8 @@ extern char const * const lyx_version_info;
 
 // Do not remove the comment below, so we get merge conflict in
 // independent branches. Instead add your own.
-#define LYX_FORMAT_LYX 630 // spitz: language updates
-#define LYX_FORMAT_TEX2LYX 630
+#define LYX_FORMAT_LYX 631 // tcuvelier: add MathML version
+#define LYX_FORMAT_TEX2LYX 631
 
 #if LYX_FORMAT_TEX2LYX != LYX_FORMAT_LYX
 #ifndef _MSC_VER