DocBook: ensure that <info>-related insets in the abstract are not generated in the abstract.

This helps generate more conformant DocBook files.

Also implement wrapper tags for InsetText.

autotests/export/docbook/svglo.lyx

@@ -0,0 +1,215 @@
+#LyX 2.4 created this file. For more info see https://www.lyx.org/
+\lyxformat 599
+\begin_document
+\begin_header
+\save_transient_properties true
+\origin unavailable
+\textclass svglobal3
+\begin_preamble
+\RequirePackage{fix-cm}
+
+\smartqed  % flush right qed marks, e.g. at end of proof
+\end_preamble
+\use_default_options true
+\maintain_unincluded_children no
+\language english
+\language_package default
+\inputencoding utf8
+\fontencoding auto
+\font_roman "default" "default"
+\font_sans "default" "default"
+\font_typewriter "default" "default"
+\font_math "auto" "auto"
+\font_default_family default
+\use_non_tex_fonts false
+\font_sc false
+\font_roman_osf false
+\font_sans_osf false
+\font_typewriter_osf false
+\font_sf_scale 100 100
+\font_tt_scale 100 100
+\use_microtype false
+\use_dash_ligatures false
+\graphics default
+\default_output_format default
+\output_sync 0
+\bibtex_command bibtex
+\index_command default
+\paperfontsize default
+\spacing single
+\use_hyperref false
+\papersize default
+\use_geometry false
+\use_package amsmath 1
+\use_package amssymb 1
+\use_package cancel 1
+\use_package esint 1
+\use_package mathdots 1
+\use_package mathtools 1
+\use_package mhchem 1
+\use_package stackrel 1
+\use_package stmaryrd 1
+\use_package undertilde 1
+\cite_engine basic
+\cite_engine_type default
+\biblio_style plain
+\use_bibtopic false
+\use_indices false
+\paperorientation portrait
+\suppress_date false
+\justification true
+\use_refstyle 0
+\use_minted 0
+\use_lineno 0
+\index Index
+\shortcut idx
+\color #008000
+\end_index
+\secnumdepth 3
+\tocdepth 3
+\paragraph_separation indent
+\paragraph_indentation default
+\is_math_indent 0
+\math_numbering_side default
+\quotes_style english
+\dynamic_quotes 0
+\papercolumns 1
+\papersides 1
+\paperpagestyle default
+\tablestyle default
+\tracking_changes false
+\output_changes false
+\change_bars false
+\postpone_fragile_content false
+\html_math_output 0
+\html_css_as_file 0
+\html_be_strict false
+\docbook_table_output 0
+\end_header
+
+\begin_body
+
+\begin_layout Title
+Title
+\end_layout
+
+\begin_layout Abstract
+Abstract text.
+ 
+\begin_inset Flex Keywords
+status open
+
+\begin_layout Plain Layout
+First keyword 
+\begin_inset ERT
+status collapsed
+
+\begin_layout Plain Layout
+
+
+\backslash
+and 
+\end_layout
+
+\end_inset
+
+Second keyword 
+\begin_inset ERT
+status collapsed
+
+\begin_layout Plain Layout
+
+
+\backslash
+and 
+\end_layout
+
+\end_inset
+
+More
+\end_layout
+
+\end_inset
+
+ 
+\begin_inset Flex PACS
+status open
+
+\begin_layout Plain Layout
+PACS code1 
+\begin_inset ERT
+status collapsed
+
+\begin_layout Plain Layout
+
+
+\backslash
+and 
+\end_layout
+
+\end_inset
+
+PACS code2 
+\begin_inset ERT
+status collapsed
+
+\begin_layout Plain Layout
+
+
+\backslash
+and 
+\end_layout
+
+\end_inset
+
+more
+\end_layout
+
+\end_inset
+
+ 
+\begin_inset Flex Subclass
+status open
+
+\begin_layout Plain Layout
+MSC code1 
+\begin_inset ERT
+status collapsed
+
+\begin_layout Plain Layout
+
+
+\backslash
+and 
+\end_layout
+
+\end_inset
+
+MSC code2 
+\begin_inset ERT
+status collapsed
+
+\begin_layout Plain Layout
+
+
+\backslash
+and 
+\end_layout
+
+\end_inset
+
+more
+\end_layout
+
+\end_inset
+
+
+\end_layout
+
+\begin_layout Standard
+Your text comes here.
+ Separate text sections with
+\end_layout
+
+\end_body
+\end_document

autotests/export/docbook/svglo.xml

@@ -0,0 +1,19 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!-- This DocBook file was created by LyX 2.4.0dev
+  See http://www.lyx.org/ for more information -->
+<article xml:lang="en_US" xmlns="http://docbook.org/ns/docbook" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:m="http://www.w3.org/1998/Math/MathML" xmlns:xi="http://www.w3.org/2001/XInclude" version="5.2">
+<info>
+<title>Title</title>
+<subjectset role='pacs'><subject>PACS code1 <!-- \and  -->
+PACS code2 <!-- \and  -->
+more</subject></subjectset><subjectset role='mcs'><subject>MSC code1 <!-- \and  -->
+MSC code2 <!-- \and  -->
+more</subject></subjectset><keywordset><keyword>First keyword <!-- \and  -->
+Second keyword <!-- \and  -->
+More</keyword></keywordset><abstract>
+<para>Abstract text.   </para>
+</abstract>
+
+</info>
+<para>Your text comes here. Separate text sections with</para>
+</article>

lib/layouts/svglobal3.layout

@@ -83,6 +83,11 @@ InsetLayout Flex:Subclass
 	CopyStyle	Flex:Keywords
 	LatexName	subclass
 	LabelString	"Mathematics Subject Classification"
+	DocBookTag          subject
+	DocBookTagType      paragraph
+	DocBookWrapperTag   subjectset
+	DocBookWrapperAttr  role='mcs'
+	DocBookInInfo       always
 End
 
 InsetLayout Flex:CRSC

src/OutputParams.h

@@ -376,6 +376,9 @@ public:
 	/// Is the current context a table?
 	bool docbook_in_table = false;
 
+	/// Should the layouts that should/must go into <info> be generated?
+	bool docbook_generate_info = true;
+
 	/// Are we generating this material for inclusion in a TOC-like entity?
 	bool for_toc = false;
 

src/Paragraph.cpp

@@ -3365,7 +3365,7 @@ std::vector<docstring> Paragraph::simpleDocBookOnePar(Buffer const & buf,
 
 		// If this is an InsetNewline, generate a new paragraph. Also reset the fonts, so that tags are closed in
 		// this paragraph.
-		if (getInset(i) != nullptr && getInset(i)->lyxCode() == NEWLINE_CODE) {
+		if (getInset(i) && getInset(i)->lyxCode() == NEWLINE_CODE) {
 			if (!ignore_fonts)
 				xs->closeFontTags();
 

src/insets/InsetLayout.cpp

@@ -91,9 +91,10 @@ bool InsetLayout::read(Lexer & lex, TextClass const & tclass,
 		IL_HTMLSTYLE,
 		IL_HTMLPREAMBLE,
 		IL_DOCBOOKTAG,
-		IL_DOCBOOKTAGTYPE,
 		IL_DOCBOOKATTR,
+		IL_DOCBOOKTAGTYPE,
 		IL_DOCBOOKSECTION,
+		IL_DOCBOOKININFO,
 		IL_DOCBOOKWRAPPERTAG,
 		IL_DOCBOOKWRAPPERTAGTYPE,
 		IL_DOCBOOKWRAPPERATTR,
@@ -142,6 +143,7 @@ bool InsetLayout::read(Lexer & lex, TextClass const & tclass,
 		{ "decoration", IL_DECORATION },
 		{ "display", IL_DISPLAY },
 		{ "docbookattr", IL_DOCBOOKATTR },
+		{ "docbookininfo", IL_DOCBOOKININFO },
 		{ "docbooksection", IL_DOCBOOKSECTION },
 		{ "docbooktag", IL_DOCBOOKTAG },
 		{ "docbooktagtype", IL_DOCBOOKTAGTYPE },
@@ -491,11 +493,14 @@ bool InsetLayout::read(Lexer & lex, TextClass const & tclass,
 		case IL_DOCBOOKTAG:
 			lex >> docbooktag_;
 			break;
+		case IL_DOCBOOKATTR:
+			lex >> docbookattr_;
+			break;
 		case IL_DOCBOOKTAGTYPE:
 			lex >> docbooktagtype_;
 			break;
-		case IL_DOCBOOKATTR:
-			lex >> docbookattr_;
+		case IL_DOCBOOKININFO:
+			lex >> docbookininfo_;
 			break;
 		case IL_DOCBOOKSECTION:
 			lex >> docbooksection_;
@@ -638,6 +643,17 @@ docstring InsetLayout::htmlstyle() const
 	return retval;
 }
 
+
+std::string const & InsetLayout::docbookininfo() const
+{
+	// Same as Layout::docbookininfo.
+	// Indeed, a trilean. Only titles should be "maybe": otherwise, metadata is "always", content is "never".
+	if (docbookininfo_.empty() || (docbookininfo_ != "never" && docbookininfo_ != "always" && docbookininfo_ != "maybe"))
+		docbookininfo_ = "never";
+	return docbookininfo_;
+}
+
+
 void InsetLayout::readArgument(Lexer & lex)
 {
 	Layout::latexarg arg;

src/insets/InsetLayout.h

@@ -154,6 +154,8 @@ public:
 	///
 	std::string docbookattr() const { return docbookattr_; }
 	///
+	std::string const & docbookininfo() const;
+	///
 	bool docbooksection() const { return docbooksection_; }
 	///
 	std::string docbookwrappertag() const { return docbookwrappertag_; }
@@ -295,6 +297,8 @@ private:
 	///
 	std::string docbookattr_;
 	///
+	mutable std::string docbookininfo_;
+	///
 	bool docbooksection_ = false;
 	///
 	std::string docbookwrappertag_;

src/insets/InsetText.cpp

@@ -616,16 +616,28 @@ void InsetText::docbook(XMLStream & xs, OutputParams const & rp, XHTMLOptions op
 	}
 
 	InsetLayout const & il = getLayout();
-	if (opts & WriteOuterTag && !il.docbooktag().empty() && il.docbooktag() != "NONE" && il.docbooktag() != "IGNORE") {
-		docstring attrs = docstring();
-		if (!il.docbookattr().empty())
-			attrs += from_ascii(il.docbookattr());
-		if (il.docbooktag() == "link")
-			attrs += from_ascii(" xlink:href=\"") + text_.asString() + from_ascii("\"");
-		xs << xml::StartTag(il.docbooktag(), attrs);
+
+	// Maybe this is an <info> paragraph that should not be generated at all (i.e. right now, its place is somewhere
+	// else, typically outside the current paragraph).
+	if (!rp.docbook_generate_info && il.docbookininfo() != "never")
+		return;
+
+	// Start outputting this inset.
+	if (opts & WriteOuterTag) {
+		if (!il.docbookwrappertag().empty() && il.docbookwrappertag() != "NONE" && il.docbookwrappertag() != "IGNORE")
+			xs << xml::StartTag(il.docbookwrappertag(), il.docbookwrapperattr());
+
+		if (!il.docbooktag().empty() && il.docbooktag() != "NONE" && il.docbooktag() != "IGNORE") {
+			docstring attrs = docstring();
+			if (!il.docbookattr().empty())
+				attrs += from_ascii(il.docbookattr());
+			if (il.docbooktag() == "link")
+				attrs += from_ascii(" xlink:href=\"") + text_.asString() + from_ascii("\"");
+			xs << xml::StartTag(il.docbooktag(), attrs);
+		}
 	}
 
-	// No need for labels that are generated from counters.
+	// No need for labels that are generated from counters. They should be handled by the external DocBook processor.
 
 	// With respect to XHTML, paragraphs are still allowed here.
 	if (!allowMultiPar())
@@ -637,8 +649,13 @@ void InsetText::docbook(XMLStream & xs, OutputParams const & rp, XHTMLOptions op
 	docbookParagraphs(text_, buffer(), xs, runparams);
 	xs.endDivision();
 
-	if (opts & WriteOuterTag && !il.docbooktag().empty() && il.docbooktag() != "NONE" && il.docbooktag() != "IGNORE")
-		xs << xml::EndTag(il.docbooktag());
+	if (opts & WriteOuterTag) {
+		if (!il.docbooktag().empty() && il.docbooktag() != "NONE" && il.docbooktag() != "IGNORE")
+			xs << xml::EndTag(il.docbooktag());
+
+		if (!il.docbookwrappertag().empty() && il.docbookwrappertag() != "NONE" && il.docbookwrappertag() != "IGNORE")
+			xs << xml::EndTag(il.docbookwrappertag());
+	}
 }
 
 

src/output_docbook.cpp

@@ -160,11 +160,10 @@ string fontToAttribute(xml::FontTypes type) {
 	// If there is a role (i.e. nonstandard use of a tag), output the attribute. Otherwise, the sheer tag is sufficient
 	// for the font.
 	string role = fontToRole(type);
-	if (!role.empty()) {
+	if (!role.empty())
 		return "role='" + role + "'";
-	} else {
+	else
 		return "";
-	}
 }
 
 
@@ -412,10 +411,6 @@ void makeParagraph(
 		OutputParams const & runparams,
 		ParagraphList::const_iterator const & par)
 {
-	// If this kind of layout should be ignored, already leave.
-	if (par->layout().docbooktag() == "IGNORE")
-		return;
-
 	// Useful variables.
 	auto const begin = text.paragraphs().begin();
 	auto const end = text.paragraphs().end();
@@ -511,7 +506,7 @@ void makeParagraph(
 	//		or we're not in the last paragraph, anyway.
 	//   (ii) We didn't open it and docbook_in_par is true,
 	//		but we are in the first par, and there is a next par.
-	bool const close_par = open_par && (!runparams.docbook_in_par);
+	bool const close_par = open_par && !runparams.docbook_in_par;
 
 	// Determine if this paragraph has some real content. Things like new pages are not caught
 	// by Paragraph::empty(), even though they do not generate anything useful in DocBook.
@@ -542,10 +537,6 @@ void makeEnvironment(Text const &text,
                      OutputParams const &runparams,
                      ParagraphList::const_iterator const & par)
 {
-	// If this kind of layout should be ignored, already leave.
-	if (par->layout().docbooktag() == "IGNORE")
-		return;
-
 	// Useful variables.
 	auto const end = text.paragraphs().end();
 	auto nextpar = par;
@@ -648,13 +639,6 @@ ParagraphList::const_iterator makeListEnvironment(Text const &text,
 	auto const end = text.paragraphs().end();
 	auto const envend = findEndOfEnvironment(par, end);
 
-	// If this kind of layout should be ignored, already leave.
-	if (begin->layout().docbooktag() == "IGNORE") {
-		auto nextpar = par;
-		++nextpar;
-		return nextpar;
-	}
-
 	// Output the opening tag for this environment.
 	Layout const & envstyle = par->layout();
 	openTag(xs, envstyle.docbookwrappertag(), envstyle.docbookwrapperattr(), envstyle.docbookwrappertagtype());
@@ -741,9 +725,6 @@ void makeCommand(
 		OutputParams const & runparams,
 		ParagraphList::const_iterator const & par)
 {
-	// If this kind of layout should be ignored, already leave.
-	if (par->layout().docbooktag() == "IGNORE")
-		return;
 
 	// Useful variables.
 	// Unlike XHTML, no need for labels, as they are handled by DocBook tags.
@@ -909,30 +890,77 @@ DocBookInfoTag getParagraphsWithInfo(ParagraphList const &paragraphs,
 } // end anonymous namespace
 
 
+std::set<const Inset *> gatherInfo(ParagraphList::const_iterator par)
+{
+	// This function has a structure highly similar to makeAny and its friends. It's only made to be called on what
+	// should become the document's <abstract>.
+	std::set<const Inset *> values;
+
+	// If this kind of layout should be ignored, already leave.
+	if (par->layout().docbooktag() == "IGNORE")
+		return values;
+
+	// If this should go in info, mark it as such. Dive deep into the abstract, as it may hide many things that
+	// DocBook doesn't want to be inside the abstract.
+	for (pos_type i = 0; i < par->size(); ++i) {
+		if (par->getInset(i) && par->getInset(i)->asInsetText()) {
+			InsetText const *inset = par->getInset(i)->asInsetText();
+
+			if (inset->getLayout().docbookininfo() != "never") {
+				values.insert(inset);
+			} else {
+				auto subpar = inset->paragraphs().begin();
+				while (subpar != inset->paragraphs().end()) {
+					values.merge(gatherInfo(subpar));
+					++subpar;
+				}
+			}
+		}
+	}
+
+	return values;
+}
+
+
 ParagraphList::const_iterator makeAny(Text const &text,
                                       Buffer const &buf,
                                       XMLStream &xs,
                                       OutputParams const &runparams,
                                       ParagraphList::const_iterator par)
 {
-	switch (par->layout().latextype) {
-	case LATEX_COMMAND:
-		makeCommand(text, buf, xs, runparams, par);
-		break;
-	case LATEX_ENVIRONMENT:
-		makeEnvironment(text, buf, xs, runparams, par);
-		break;
-	case LATEX_LIST_ENVIRONMENT:
-	case LATEX_ITEM_ENVIRONMENT:
-		// Only case when makeAny() might consume more than one paragraph.
-		return makeListEnvironment(text, buf, xs, runparams, par);
-	case LATEX_PARAGRAPH:
-		makeParagraph(text, buf, xs, runparams, par);
-		break;
-	case LATEX_BIB_ENVIRONMENT:
-		makeBibliography(text, buf, xs, runparams, par);
-		break;
+	bool ignoreParagraph = false;
+
+	// If this kind of layout should be ignored, already leave.
+	ignoreParagraph |= par->layout().docbooktag() == "IGNORE";
+
+	// For things that should go into <info>, check the variable rp.docbook_generate_info. This does not apply to the
+	// abstract itself.
+	bool isAbstract = par->layout().docbookabstract() || par->layout().docbooktag() == "abstract";
+	ignoreParagraph |= !isAbstract && par->layout().docbookininfo() != "never" && !runparams.docbook_generate_info;
+
+	// Switch on the type of paragraph to call the right handler.
+	if (!ignoreParagraph) {
+		switch (par->layout().latextype) {
+		case LATEX_COMMAND:
+			makeCommand(text, buf, xs, runparams, par);
+			break;
+		case LATEX_ENVIRONMENT:
+			makeEnvironment(text, buf, xs, runparams, par);
+			break;
+		case LATEX_LIST_ENVIRONMENT:
+		case LATEX_ITEM_ENVIRONMENT:
+			// Only case when makeAny() might consume more than one paragraph.
+			return makeListEnvironment(text, buf, xs, runparams, par);
+		case LATEX_PARAGRAPH:
+			makeParagraph(text, buf, xs, runparams, par);
+			break;
+		case LATEX_BIB_ENVIRONMENT:
+			makeBibliography(text, buf, xs, runparams, par);
+			break;
+		}
 	}
+
+	// For cases that are not lists, the next paragraph to handle is the next one.
 	++par;
 	return par;
 }
@@ -964,6 +992,9 @@ void outputDocBookInfo(
 	// This check must be performed *before* a decision on whether or not to output <info> is made.
 	bool hasAbstract = !info.abstract.empty();
 	docstring abstract;
+	set<const Inset *> infoInsets; // Paragraphs that should go into <info>, but are hidden in an <abstract>
+	// paragraph. (This happens for quite a few layouts, unfortunately.)
+
 	if (hasAbstract) {
 		// Generate the abstract XML into a string before further checks.
 		// Usually, makeAny only generates one paragraph at a time. However, for the specific case of lists, it might
@@ -971,14 +1002,20 @@ void outputDocBookInfo(
 		odocstringstream os2;
 		XMLStream xs2(os2);
 
-		set<pit_type> doneParas;
+		auto rp = runparams;
+		rp.docbook_generate_info = false;
+
+		set<pit_type> doneParas; // Paragraphs that have already been converted (mostly to deal with lists).
 		for (auto const & p : info.abstract) {
 			if (doneParas.find(p) == doneParas.end()) {
 				auto oldPar = paragraphs.iterator_at(p);
-				auto newPar = makeAny(text, buf, xs2, runparams, oldPar);
+				auto newPar = makeAny(text, buf, xs2, rp, oldPar);
+
+				infoInsets.merge(gatherInfo(oldPar));
 
 				// Insert the indices of all the paragraphs that were just generated (typically, one).
 				// **Make the hypothesis that, when an abstract has a list, all its items are consecutive.**
+				// Otherwise, makeAny and makeListEnvironment would have to be adapted too.
 				pit_type id = p;
 				while (oldPar != newPar) {
 					doneParas.emplace(id);
@@ -1009,13 +1046,11 @@ void outputDocBookInfo(
 		xs << xml::CR();
 	}
 
-	// Output the elements that should go in <info>, before and after the abstract.
+	// Output the elements that should go in <info>.
+	// - First, the title.
 	for (auto pit : info.shouldBeInInfo) // Typically, the title: these elements are so important and ubiquitous
 		// that mandating a wrapper like <info> would repel users. Thus, generate them first.
 		makeAny(text, buf, xs, runparams, paragraphs.iterator_at(pit));
-	for (auto pit : info.mustBeInInfo)
-		makeAny(text, buf, xs, runparams, paragraphs.iterator_at(pit));
-
 	// If there is no title, generate one (required for the document to be valid).
 	// This code is called for the main document, for table cells, etc., so be precise in this condition.
 	if (text.isMainText() && info.shouldBeInInfo.empty() && !runparams.inInclude) {
@@ -1025,8 +1060,14 @@ void outputDocBookInfo(
 		xs << xml::CR();
 	}
 
-	// Always output the abstract as the last item of the <info>, as it requires special treatment (especially if
-	// it contains several paragraphs that are empty).
+	// - Then, other metadata.
+	for (auto pit : info.mustBeInInfo)
+		makeAny(text, buf, xs, runparams, paragraphs.iterator_at(pit));
+	for (auto const * inset : infoInsets)
+		inset->docbook(xs, runparams);
+
+	// - Finally, always output the abstract as the last item of the <info>, as it requires special treatment
+	// (especially if it contains several paragraphs that are empty).
 	if (hasAbstract) {
 		if (info.abstractLayout) {
 			xs << XMLStream::ESCAPE_NONE << abstract;