tex4ebook (7sep20)

git-svn-id: svn://tug.org/texlive/trunk@56290 c570f23f-e606-0410-a88d-b1316a301751

10 files changed, 256 insertions, 16 deletions

diff --git a/Master/texmf-dist/doc/support/tex4ebook/README b/Master/texmf-dist/doc/support/tex4ebook/README
index 796445674a2..3634974ab82 100644
--- a/Master/texmf-dist/doc/support/tex4ebook/README
+++ b/Master/texmf-dist/doc/support/tex4ebook/README
@@ -81,7 +81,7 @@ But it is optional. You shouldn't need to modify your \TeX\ files
   
 `-f,--format (default epub)`
 
-:    Output format. Possible values are `epub`, `epub3` and `mobi`.
+:    Output format. Possible values are `epub`, `epub3`, `mobi`, `azw` and `azw3`.
 
 `-j,--jobname`
 
@@ -304,7 +304,7 @@ Type
 -  WARNING: filename.epub: item (OEBPS/foo.boo) exists in the zip file, but is
 not declared in the OPF file
 
-  Delete the `filename-(epub|epub3|mobi)` folder and `filename.epub`. Then
+  Delete the `filename-(epub|epub3|mobi|azw|azw3)` folder and `filename.epub`. Then
   run `tex4ebook` again.
 
 - WARNING(ACC-009): hsmmt10t.epub/OEBPS/hsmmt10tch17.xhtml(235,15): 
diff --git a/Master/texmf-dist/doc/support/tex4ebook/changelog.tex b/Master/texmf-dist/doc/support/tex4ebook/changelog.tex
index 4cb5c8c7ca9..f77423e85b7 100644
--- a/Master/texmf-dist/doc/support/tex4ebook/changelog.tex
+++ b/Master/texmf-dist/doc/support/tex4ebook/changelog.tex
@@ -3,6 +3,60 @@
 
 \begin{itemize}
 \item
+  2020/09/07
+
+  \begin{itemize}
+  \tightlist
+  \item
+    released version \texttt{0.3b}
+  \end{itemize}
+\item
+  2020/08/26
+
+  \begin{itemize}
+  \tightlist
+  \item
+    fixed hiearchical structure in NCX TOC for chapters in backmatter
+    and appendix
+  \item
+    load \texttt{common\_domfilters} extension by default.
+  \end{itemize}
+\item
+  2020/07/09
+
+  \begin{itemize}
+  \tightlist
+  \item
+    addded \texttt{AZW} and \texttt{AZW3} format support.
+  \end{itemize}
+\item
+  2020/06/21
+
+  \begin{itemize}
+  \tightlist
+  \item
+    save \texttt{\textbackslash{}title} element
+  \item
+    save contents of \texttt{\textbackslash{}author} in macro directly
+  \end{itemize}
+\item
+  2020/06/15
+
+  \begin{itemize}
+  \tightlist
+  \item
+    remove child elements from elements that don't allow them in the OPF
+    and NCX file.
+  \end{itemize}
+\item
+  2020/03/14
+
+  \begin{itemize}
+  \tightlist
+  \item
+    explicitly list supported section types in the NCX table
+  \end{itemize}
+\item
   2019/11/01
 
   \begin{itemize}
diff --git a/Master/texmf-dist/doc/support/tex4ebook/readme.tex b/Master/texmf-dist/doc/support/tex4ebook/readme.tex
index 13b064bf31d..0fdb16c1c91 100644
--- a/Master/texmf-dist/doc/support/tex4ebook/readme.tex
+++ b/Master/texmf-dist/doc/support/tex4ebook/readme.tex
@@ -92,8 +92,8 @@ Specify make4ht build file\footnote{\url{https://github.com/michal-h21/make4ht\#
 Defaulf build file filename is \texttt{filename.mk4}, use this option if
 you use different filename.
 \item[\texttt{-f,-\/-format\ (default\ epub)}]
-Output format. Possible values are \texttt{epub}, \texttt{epub3} and
-\texttt{mobi}.
+Output format. Possible values are \texttt{epub}, \texttt{epub3},
+\texttt{mobi}, \texttt{azw} and \texttt{azw3}.
 \item[\texttt{-j,-\/-jobname}]
 Specify the output file name, without file extension.
 \item[\texttt{-l,-\/-lua}]
@@ -371,8 +371,9 @@ issues:}\label{common-validation-issues}}
   but is not declared in the OPF file
 \end{itemize}
 
-Delete the \texttt{filename-(epub\textbar{}epub3\textbar{}mobi)} folder
-and \texttt{filename.epub}. Then run \texttt{tex4ebook} again.
+Delete the
+\texttt{filename-(epub\textbar{}epub3\textbar{}mobi\textbar{}azw\textbar{}azw3)}
+folder and \texttt{filename.epub}. Then run \texttt{tex4ebook} again.
 
 \begin{itemize}
 \item
diff --git a/Master/texmf-dist/doc/support/tex4ebook/tex4ebook-doc.pdf b/Master/texmf-dist/doc/support/tex4ebook/tex4ebook-doc.pdf
index 1f956367135..942a92df29c 100644
--- a/Master/texmf-dist/doc/support/tex4ebook/tex4ebook-doc.pdf
+++ b/Master/texmf-dist/doc/support/tex4ebook/tex4ebook-doc.pdf
diff --git a/Master/texmf-dist/scripts/tex4ebook/tex4ebook b/Master/texmf-dist/scripts/tex4ebook/tex4ebook
index ebd7485c2d1..dc602c4e72b 100755
--- a/Master/texmf-dist/scripts/tex4ebook/tex4ebook
+++ b/Master/texmf-dist/scripts/tex4ebook/tex4ebook
@@ -23,7 +23,7 @@ local tex4ht_sty_par=""
 local tex4ht_par=""
 local t4ht_par=""
 local latex_par=""
-local output_formats={epub=true,mobi=true,epub3=true}
+local output_formats={epub=true,mobi=true,epub3=true,azw=true, azw3=true}
 local executor=nil
 local tidy = false
 local include_fonts = false
@@ -67,7 +67,7 @@ else
 end
 
 if args.version then
-  print "tex4ebook v0.3a"
+  print "tex4ebook v0.3b"
   return 
 end
 
@@ -129,7 +129,9 @@ log:status("Input file: ".. params.tex_file)
 local output_format = params.output_format 
 -- use epub as default output_format
 output_format = output_format or "epub"
-local extensions = ebookutils.load_extensions(params.extensions, output_format)
+-- load common_domfilters extension by default
+local extensions = ebookutils.add_extensions("+common_domfilters", params.extensions)
+extensions = ebookutils.load_extensions(extensions, output_format)
 -- but also support tex4ebook!s own parameters
 local oldparams = {
   -- htlatex=latex_cmd
diff --git a/Master/texmf-dist/scripts/tex4ebook/tex4ebook-exec_azw.lua b/Master/texmf-dist/scripts/tex4ebook/tex4ebook-exec_azw.lua
new file mode 100755
index 00000000000..397292b648b
--- /dev/null
+++ b/Master/texmf-dist/scripts/tex4ebook/tex4ebook-exec_azw.lua
@@ -0,0 +1,35 @@
+module(...,package.seeall)
+local eb = require("tex4ebook-exec_epub")
+local ebookutils = require("mkutils")
+local log = logging.new "exec_azw"
+
+function prepare(params)
+	return eb.prepare(params)
+end
+
+function run(out,params)
+	return eb.run(out, params)
+end
+
+function writeContainer()
+	local ret =  eb.writeContainer()
+  -- convert the epub file to azw
+  local epubpath = eb.basedir .. "/" .. eb.outputfile
+
+  -- find the azw filename 
+  local azwfile = eb.outputfile:gsub("epub$", "azw")
+  local azwdist = eb.destdir ..  azwfile
+  local command = "kindlegen " .. epubpath .. " -o " .. azwfile
+	log:info("Pack azw ".. command)
+  local status, output = ebookutils.execute(command)
+  -- copy the azw file to the destination directory
+  -- the destination directory will be created by the epub writer, so it is possible to use
+  -- the cp function which doesn't try to create directory
+  ebookutils.cp(eb.basedir .. "/" .. azwfile, azwdist)
+
+	return ret
+end
+
+function clean()
+	return eb.clean()
+end
diff --git a/Master/texmf-dist/scripts/tex4ebook/tex4ebook-exec_azw3.lua b/Master/texmf-dist/scripts/tex4ebook/tex4ebook-exec_azw3.lua
new file mode 100755
index 00000000000..33e8a1182c5
--- /dev/null
+++ b/Master/texmf-dist/scripts/tex4ebook/tex4ebook-exec_azw3.lua
@@ -0,0 +1,35 @@
+module(...,package.seeall)
+local eb = require("tex4ebook-exec_epub")
+local ebookutils = require("mkutils")
+local log = logging.new "exec_azw"
+
+function prepare(params)
+	return eb.prepare(params)
+end
+
+function run(out,params)
+	return eb.run(out, params)
+end
+
+function writeContainer()
+	local ret =  eb.writeContainer()
+  -- convert the epub file to azw
+  local epubpath = eb.basedir .. "/" .. eb.outputfile
+
+  -- find the azw filename 
+  local azwfile = eb.outputfile:gsub("epub$", "azw3")
+  local azwdist = eb.destdir ..  azwfile
+  local command = "kindlegen " .. epubpath .. " -o " .. azwfile
+	log:info("Pack azw ".. command)
+  local status, output = ebookutils.execute(command)
+  -- copy the azw file to the destination directory
+  -- the destination directory will be created by the epub writer, so it is possible to use
+  -- the cp function which doesn't try to create directory
+  ebookutils.cp(eb.basedir .. "/" .. azwfile, azwdist)
+
+	return ret
+end
+
+function clean()
+	return eb.clean()
+end
diff --git a/Master/texmf-dist/scripts/tex4ebook/tex4ebook-exec_epub.lua b/Master/texmf-dist/scripts/tex4ebook/tex4ebook-exec_epub.lua
index 52e2ab6134d..b0ccee62027 100755
--- a/Master/texmf-dist/scripts/tex4ebook/tex4ebook-exec_epub.lua
+++ b/Master/texmf-dist/scripts/tex4ebook/tex4ebook-exec_epub.lua
@@ -5,6 +5,7 @@ local io = require("io")
 local log = logging.new("exec_epub")
 --local ebookutils = require("ebookutils")
 local ebookutils = require "mkutils"
+local dom = require("luaxml-domobject")
 local outputdir_name="OEBPS"
 local metadir_name = "META-INF"
 local mimetype_name="mimetype"
@@ -319,11 +320,123 @@ function pack_container()
   ebookutils.cp(basedir .."/"..outputfile, destdir .. outputfile)
 end
 
+local function update_file(filename, fn)
+  -- update contents of a filename using function
+  local f = io.open(filename, "r")
+  local content = f:read("*all")
+  f:close()
+  local newcontent = fn(content)
+  local f = io.open(filename, "w")
+  f:write(newcontent)
+  f:close()
+end
+
+local function fix_ncx_toc_levels(dom)
+  -- OK, this is a weird hack. The problem is that when \backmatter
+  -- follows \part, the subsequent chapters are listed under part 
+  -- in the NCX TOC
+  -- I've added special element <navmark> to the ncx file to detect mark numbers.
+  -- chapters in backmatter should have empty mark number
+
+  -- get current <navpoint> section type and mark number
+  local get_navpoint_info = function(navpoint)
+    local navmarks = navpoint:query_selector("navmark")
+    if navmarks and #navmarks > 0 then
+      -- we are interested only in the first navmark, because it contains the current navPoint info
+      local navmark = navmarks[1]
+      return navmark:get_attribute("type"), navmark:get_text()
+    end
+    return nil, "Cannot find navLabel"
+  end
+
+  local fix_chapters = function(part)
+    -- move chapters from backmatter and appendix from the current part
+    -- find part position in the element list, it will be place where backmatter will be moved
+    local part_pos = part:find_element_pos() + 1
+    local part_parent = part:get_parent()
+    -- child chapters
+    local children = part:get_children()
+    -- loop over children from back, where backmatter or appendix may be placed
+    for i = #children, 1, -1 do
+      local current = children[i]
+      local sect_type, mark = get_navpoint_info(current)
+      if sect_type then
+        -- remove spaces from mark
+        mark = mark:gsub("%s", "") 
+        if sect_type == "appendix" or (sect_type == "chapter" and mark=="") then
+          -- move chapter at the same level as part, after part
+          -- the correct order will be kept, because we move from the back of the node list
+          -- and every new node is placed before previous added nodes
+          part_parent:add_child_node(current:copy_node(), part_pos)
+          current:remove_node()
+        else
+          -- break processing if we find normal chapter
+          break
+        end
+      end
+    end
+  end
+
+  -- find the last part in the ncx table. it can contain incorrectly nested chapters
+  local parts = {}
+  for _, navpoint in ipairs(dom:query_selector("navMap navPoint")) do
+    -- we are not able to match just direct navPoint ancestors, so we will use this trick
+    if navpoint:get_parent():get_element_name() == "navMap" then
+      local sec_type, navmark = get_navpoint_info(navpoint)
+      if sec_type == "part" then table.insert(parts, navpoint) end
+    end
+  end
+  if #parts > 0 then
+    -- fix chapters in the last part
+    fix_chapters(parts[#parts])
+  end
+  return dom
+end
+
+local function clean_xml_files()
+  local opf_file = outputdir .. "/content.opf"
+  update_file(opf_file, function(content)
+    -- remove wrong elements from the OPF file
+    -- open opf file and create LuaXML DOM
+    local opf_dom = dom.parse(content)
+    -- remove child elements from elements that don't allow them
+    for _, el in ipairs(opf_dom:query_selector("dc|title, dc|creator")) do
+      -- get text content
+      local text = el:get_text()
+      -- replace element text with a new text node containing original text
+      el._children = {el:create_text_node(text)}
+    end
+    return opf_dom:serialize()
+  end)
+  local ncxfilename = outputdir .. "/" .. outputfilename .. ".ncx"
+  update_file(ncxfilename, function(content)
+    -- remove spurious spaces at the beginning
+    content = content:gsub("^%s*","")
+    local ncx_dom = dom.parse(content)
+    fix_ncx_toc_levels(ncx_dom)
+    -- remove child elements from <text> element
+    for _, el in ipairs(ncx_dom:query_selector("text")) do
+      local text = el:get_text()
+      -- replace element text with a new text node containing original text
+      el._children = {el:create_text_node(text)}
+    end
+    for _, el in ipairs(ncx_dom:query_selector("navPoint")) do
+      -- fix attribute names. this issue is caused by a  LuaXML behavior
+      -- that makes all attributes lowercase
+      el._attr["playOrder"] = el._attr["playorder"]
+      el._attr["playorder"] = nil
+    end
+    return ncx_dom:serialize()
+  end)
+
+end
 
 function writeContainer()
   make_opf()
+  clean_xml_files()
   pack_container()
 end
+
 local function deldir(path)
   for entry in lfs.dir(path) do
     if entry~="." and entry~=".." then  
diff --git a/Master/texmf-dist/tex/latex/tex4ebook/tex4ebook.4ht b/Master/texmf-dist/tex/latex/tex4ebook/tex4ebook.4ht
index e623ac8bc24..6ee87ed3aeb 100644
--- a/Master/texmf-dist/tex/latex/tex4ebook/tex4ebook.4ht
+++ b/Master/texmf-dist/tex/latex/tex4ebook/tex4ebook.4ht
@@ -157,7 +157,7 @@ href="#1" media-type="\a:CoverMimeType" />}}
   "http://www.daisy.org/z3986/2005/ncx-2005-1.dtd">\Hnewline}}
 
 %Get sanitized value of \title
-%\def\Title{\Ref{TITLE+}}
+\ifTag{TITLE+}{\def\Title{\LikeRef{TITLE+}}}{}
 % If title contains macros, SafeTitle should be configured 
 % in the custom cfg file
 \NewConfigure{SafeTitle}[1]{\gdef\Title{#1}}
@@ -255,7 +255,7 @@ href="#1" media-type="\a:CoverMimeType" />}}
 \def\navsection#1#2{
   \ConfigureToc{#1}%
   {\expandafter\closelevels\expandafter{#2}
-  \a:NavSection\Ncx:Mark}
+  \def\curr:sect:type{#1}\a:NavSection\Ncx:Mark}
   {\c:NavSection}
   {}
   {\b:NavSection%
@@ -361,16 +361,16 @@ href="#1" media-type="\a:CoverMimeType" />}}
  \Configure{toTocLink}{}{}%
  \Configure{NavSection}{\booltrue{tocnoempty}\HCode{\Hnewline<navPoint id="navPoint-}%
   \stepnavpoint\HCode{" playOrder="}%
-  \the\navpoint\HCode{">\Hnewline<navLabel>\Hnewline<text>}%
+  \the\navpoint\HCode{">\Hnewline<navLabel>\Hnewline<text><navmark type="\curr:sect:type">}%
 }{\HCode{</text>\Hnewline%
   </navLabel>\Hnewline}%
  \HCode{<content src="\navmapsrc" />}%
-}{ }{\HCode{</navPoint>\Hnewline}}
+}{\HCode{</navmark>}}{\HCode{</navPoint>\Hnewline}}
 % Meta inf
 \ncx:head
 % Book title
 \ncx:title
-\tableofcontents%[part,chapter,likechapter,section,likesection,subsection,likesubsection]%
+\tableofcontents[part,appendix,chapter,likechapter,appendixsec,section,likesection,appendixsubsec,subsection,likesubsection]%
 %Hack to get close tag working
 \HCode{</ncx>}
 \EndNoFonts
diff --git a/Master/texmf-dist/tex/latex/tex4ebook/tex4ebook.sty b/Master/texmf-dist/tex/latex/tex4ebook/tex4ebook.sty
index 27a46b756ec..1b0114748c9 100644
--- a/Master/texmf-dist/tex/latex/tex4ebook/tex4ebook.sty
+++ b/Master/texmf-dist/tex/latex/tex4ebook/tex4ebook.sty
@@ -144,12 +144,12 @@
 \let\tf@orig@author\author
 \newcommand\tf@author[1]{%
   \tf@orig@author{#1}%
-  \let\Author\@author%
+  \def\Author{#1}%
 }
 
 \newcommand\tf@optauthor[2][]{%
   \tf@orig@author[#1]{#2}%
-  \let\Author\@author%
+  \def\Author{#2}%
 }
 
 % support optiona argument for \author as well