tex4ebook (29jul15)

git-svn-id: svn://tug.org/texlive/trunk@37751 c570f23f-e606-0410-a88d-b1316a301751

5 files changed, 576 insertions, 0 deletions

diff --git a/Master/texmf-dist/scripts/tex4ebook/config-t4e.lua b/Master/texmf-dist/scripts/tex4ebook/config-t4e.lua
new file mode 100755
index 00000000000..9cc6681efec
--- /dev/null
+++ b/Master/texmf-dist/scripts/tex4ebook/config-t4e.lua
@@ -0,0 +1,19 @@
+-- Config file for tex4ebook
+
+-- Font config
+
+Font{
+  name   = "cmr",
+  file   = "lmroman10-regular.otf",
+  family = "Latin Modern",
+  weight = "normal",
+  style  = "normal"
+}
+
+Font{
+  name   = "cmmi",
+  file   = "lmroman10-italic.otf",
+  family = "Latin Modern",
+  weight = "normal",
+  style  = "italic"
+}
diff --git a/Master/texmf-dist/scripts/tex4ebook/exec_epub.lua b/Master/texmf-dist/scripts/tex4ebook/exec_epub.lua
new file mode 100755
index 00000000000..45977c2ee2f
--- /dev/null
+++ b/Master/texmf-dist/scripts/tex4ebook/exec_epub.lua
@@ -0,0 +1,300 @@
+module("exec_epub",package.seeall)
+local lfs = require("lfs")
+local os = require("os")
+local io = require("io")
+--local ebookutils = require("ebookutils")
+local ebookutils = require "mkutils"
+-- font loading doesn't work, font database format changes  often
+-- and it is different between TL and Miktex
+-- local load_font = require("list-fonts")
+local outputdir_name="OEBPS"
+local metadir_name = "META-INF"
+local mimetype_name="mimetype"
+outputdir=""
+outputfile=""
+outputfilename=""
+basedir = ""
+tidy = false
+local include_fonts = false
+local metadir=""
+
+
+
+function prepare(params)
+	local randname=tostring(math.random(12000))
+	-- if not ebookutils.file_exists("tex4ht.env") then
+	--  local env_file = kpse.find_file("epub2.env")
+	--print("Local env file: "..env_file)
+	--  ebookutils.copy(env_file,"tex4ht.env")
+	-- end
+	local makedir= function(path)
+		local current = lfs.currentdir()
+		local dir = ebookutils.prepare_path(path .. "/")
+		if type(dir) == "table" then
+			local parts,msg =  ebookutils.find_directories(dir)
+			if parts then 
+			 ebookutils.mkdirectories(parts)
+		  end
+		end
+		lfs.chdir(current)
+	end
+	basedir = params.input.."-".. params.format
+	outputdir= basedir.."/"..outputdir_name --"outdir-"..randname --os.tmpdir()
+	makedir(outputdir)
+	-- lfs.mkdir(outputdir)
+	--ebookutils.mkdirectories(ebookutils.prepare_path(outputdir.."/"))
+	metadir = basedir .."/" .. metadir_name --"metadir-"..randname
+	makedir(metadir)
+	--local dd = ebookutils.prepare_path(metadir.."/")
+	--for _,d in pairs(dd) do print("metadir path: "..d) end
+	-- lfs.mkdir(metadir)
+	--local status, msg = ebookutils.mkdirectories(ebookutils.prepare_path(metadir.."/"))
+	--if not status then print("make mmetadir error:" ..msg) end
+	mimetype= basedir .. "/" ..mimetype_name --os.tmpname()
+	print(outputdir)
+	print(mimetype)
+	tidy = params.tidy
+	include_fonts = params.include_fonts
+	params["t4ht_par"] = params["t4ht_par"] -- + "-d"..string.format(params["t4ht_dir_format"],outputdir)
+	return(params)
+end
+
+function run(out,params)
+	--local currentdir=
+	outputfilename=out
+	outputfile = outputfilename..".epub"
+	print("Output file: "..outputfile)
+	--lfs.chdir(metadir)
+	local m= io.open(metadir.."/container.xml","w")
+	m:write([[
+<?xml version="1.0"?>
+<container version="1.0" xmlns="urn:oasis:names:tc:opendocument:xmlns:container">
+<rootfiles>
+<rootfile full-path="OEBPS/content.opf"
+media-type="application/oebps-package+xml"/>
+</rootfiles>
+</container>
+	]])
+	m:close()
+	--lfs.chdir("..")
+	m=io.open(mimetype,"w")
+	m:write("application/epub+zip")
+	m:close()
+	local htlatex_run = "${htlatex} ${input} \"${config}${tex4ht_sty_par}\" \"${tex4ht_par}\" \"${t4ht_par}\" \"${latex_par}\"" % params
+	print("Make4ht run")
+	print("-------------------")
+	params.config_file.Make.params = params
+  local mode = params.mode
+	if params.config_file.Make:length() < 1 then
+    if mode == "draft" then
+      params.config_file.Make:htlatex()
+    else
+      params.config_file.Make:htlatex()
+      params.config_file.Make:htlatex()
+      params.config_file.Make:htlatex() 
+    end
+	end
+	if #params.config_file.Make.image_patterns > 0 then
+		params["t4ht_par"] = params["t4ht_par"] .." -p"
+	end
+	params.config_file.Make:tex4ht()
+	params.config_file.Make:t4ht()
+	params.config_file.Make:run()
+	print("-------------------")
+	--[[for k,v in pairs(params.config_file.Make) do
+		print(k.. " : "..type(v))
+	end--]]
+  --print(os.execute(htlatex_run))
+end
+
+local mimetypes = {
+	css = "text/css",
+	png = "image/png", 
+	jpg = "image/jpeg",
+	gif = "image/gif",
+	svg = "image/svg+xml",
+	html= "application/xhtml+xml",
+	xhtml= "application/xhtml+xml",
+	ncx = "application/x-dtbncx+xml",
+	otf = "application/opentype",
+	ttf = "application/truetype",
+	woff = "application/font-woff",
+  js = "text/javascript"
+}
+
+function make_opf()
+	-- Join files content.opf and content-part2.opf
+	-- make item record for every converted image
+	local lg_item = function(item)
+		-- Find mimetype and make item tag for each converted file in the lg file
+		local fname,ext = item:match("([%a%d%_%-]*)%p([%a%d]*)$")
+		local mimetype = mimetypes[ext] or ""
+		if mimetype == "" then print("Mimetype for "..ext.." is not registered"); return nil end
+		local dir_part = item:split("/")
+		table.remove(dir_part,#dir_part)
+		local id=table.concat(dir_part,"-")..fname.."_"..ext
+		return "<item id='"..id .. "' href='"..item.."' media-type='"..mimetype.."' />",id
+	end
+	local find_all_files= function(s,r)
+		local r = r or "([%a%d%_%-]*)%.([x]?html)"
+		local files = {}
+		for i, ext in s:gmatch(r) do
+			--local i, ext = s:match(r)-- do
+			ext = ext or "true"
+			files[i] = ext 
+		end 
+		return files
+	end
+	local tidyconf = nil
+	if tidy then 
+		tidyconf = kpse.find_file("tidyconf.conf")
+	end
+	--local opf_first_part = outputdir .. "/content.opf" 
+	local opf_first_part =   "content.opf" 
+	local opf_second_part =  "content-part2.opf"
+	--local opf_second_part = outputdir .. "/content-part2.opf"
+	if 
+		ebookutils.file_exists(opf_first_part) and ebookutils.file_exists(opf_second_part) 
+		then
+			local h_first  = io.open(opf_first_part,"r")
+			local h_second = io.open(opf_second_part,"r")
+			local opf_complete = {}
+			table.insert(opf_complete,h_first:read("*all"))
+			local used_html = find_all_files(opf_complete[1])
+			local lg_file = ebookutils.parse_lg(outputfilename..".lg")
+			local used_files = lg_file["files"]
+			--[[for f in lfs.dir("./OEBPS") do
+			--table.insert(used_files,f)
+			--used_files[f] = true
+			end--]]
+			local all_html = find_all_files(table.concat(used_files,"\n"))
+			local outside_spine = {}
+			-- This was duplicated code
+			--[[for i, ext in pairs(all_html) do
+				if not used_html[i] then
+					print("Prvni insert: ".. i .."."..ext)
+					local item, id = lg_item(i.."."..ext) 
+					table.insert(opf_complete,item)
+					table.insert(outside_spine,id)
+				end
+			end--]]
+			local all_used_files = find_all_files(opf_complete[1],"([%a%d%-%_]+%.[%a%d]+)")
+			local used_paths = {}
+			for _,k in ipairs(lg_file["files"]) do
+				local ext = k:match("%.([%a%d]*)$")
+				local parts = k:split "/"
+				local fn = parts[#parts]
+				local allow_in_spine =  {html="",xhtml = "", xml = ""}
+				table.remove(parts,#parts)
+				--table.insert(parts,1,"OEBPS")
+				table.insert(parts,1,outputdir)
+				--print("SSSSS "..fn.." ext .." .. ext)
+				--if string.find("jpg gif png", ext) and not all_used_files[k] then
+				local item,id = lg_item(k) 
+				if item then
+					local path = table.concat(parts)
+					if not used_paths[path] then
+						ebookutils.mkdirectories(parts)
+						used_paths[path]=true
+					end
+					if allow_in_spine[ext] and tidy then 
+						if tidyconf then
+							print("Tidy: "..k)
+							local run ="tidy -c  -w 200 -q -utf8 -m -config " .. tidyconf .." " .. k
+							os.execute(run) 
+						else
+							print "Tidy: Cannot load tidyconf.conf"
+						end
+					end
+					ebookutils.copy(k, outputdir .. "/"..k)
+					if not all_used_files[fn] then
+						table.insert(opf_complete,item)
+						if allow_in_spine[ext] then 
+						table.insert(outside_spine,id)
+						end
+					end
+				end
+			end
+			for _,f in ipairs(lg_file["images"]) do
+				local f = f.output
+				local p = lg_item(f)
+				ebookutils.copy(f, outputdir .. "/"..f)
+				table.insert(opf_complete,p)
+			end
+			local end_opf = h_second:read("*all")
+			local spine_items = {}
+			for _,i in ipairs(outside_spine) do
+				table.insert(spine_items,
+				'<itemref idref="${idref}" linear="no" />' % {idref=i})
+			end
+			table.insert(opf_complete,end_opf % {spine = table.concat(spine_items,"\n")})
+			h_first:close()
+			h_second:close()
+			h_first = io.open(opf_first_part,"w")
+      local opf_completed = table.concat(opf_complete,"\n")
+      -- poor man's tidy: remove trailing whitespace befora xml tags
+      opf_completed = opf_completed:gsub("[ ]*<","<")
+			h_first:write(opf_completed)
+			h_first:close()
+			os.remove(opf_second_part)
+			--ebookutils.copy(outputfilename ..".css",outputdir.."/")
+			ebookutils.copy(opf_first_part,outputdir.."/"..opf_first_part)
+			--for c,v in pairs(lg_file["fonts"]) do
+			--	print(c, table.concat(v,", "))
+			--end
+			--print(table.concat(opf_complete,"\n"))
+		else
+			print("Missing opf file")
+		end
+	end
+  local function find_zip()
+    if io.popen("zip -v","r"):close() then
+      return "zip"
+    elseif io.popen("miktex-zip -v","r"):close() then
+      return "miktex-zip"
+    end
+    print "It appears you don't have zip command installed. I can't pack the ebook"
+    return "zip"
+  end
+
+	function pack_container()
+		if os.execute("tidy -v") > 0 then
+			print("Warning:\n  tidy command seems missing, you need to install it" ..
+			" in order\n  to make valid epub file") 
+		else
+		  print("Tidy ncx "..
+		  os.execute("tidy -xml -i -q -utf8 -m " .. 
+			outputdir .. "/" .. outputfilename .. ".ncx"))
+			print("Tidy opf "..
+			os.execute("tidy -xml -i -q -utf8 -m " .. 
+			outputdir .. "/" .. "content.opf"))
+		end
+		print(mimetype)
+    local zip = find_zip()
+		print("Pack mimetype " .. os.execute("cd "..basedir.." && "..zip.." -q0X "..outputfile .." ".. mimetype_name))
+		print("Pack metadir "   .. os.execute("cd "..basedir.." && "..zip.." -qXr9D " .. outputfile.." "..metadir_name))
+		print("Pack outputdir " .. os.execute("cd "..basedir.." && "..zip.." -qXr9D " .. outputfile.." "..outputdir_name))
+		print("Copy generated epub ")
+		ebookutils.cp(basedir .."/"..outputfile, outputfile)
+end
+
+
+	function writeContainer()
+		make_opf()
+		pack_container()
+	end
+	local function deldir(path)
+		for entry in lfs.dir(path) do
+			if entry~="." and entry~=".." then  
+				os.remove(path.."/"..entry)
+			end
+		end
+		os.remove(path)
+		--]]
+	end
+
+	function clean()
+		--deldir(outputdir)
+		--deldir(metadir)
+		--os.remove(mimetype)
+	end
diff --git a/Master/texmf-dist/scripts/tex4ebook/exec_epub3.lua b/Master/texmf-dist/scripts/tex4ebook/exec_epub3.lua
new file mode 100755
index 00000000000..55f7f71c309
--- /dev/null
+++ b/Master/texmf-dist/scripts/tex4ebook/exec_epub3.lua
@@ -0,0 +1,92 @@
+module(...,package.seeall)
+local eb = require("exec_epub")
+
+local outputdir = nil
+local input     = nil
+function prepare(params)
+	local basedir = params.input.."-".. params.format
+  local outputdir_name="OEBPS"
+	outputdir= basedir.."/"..outputdir_name
+  input = params.input 
+  params.packages = params.packages .. "\\Configure{ext}{xhtml}"
+	return eb.prepare(params)
+end
+
+function run(out,params)
+	return eb.run(out, params)
+end
+
+
+local function makeTOC(document)
+  local template = [[
+<!DOCTYPE html>
+<html xmlns="http://www.w3.org/1999/xhtml" 
+ xmlns:epub="http://www.idpf.org/2007/ops"
+>
+<head><title>TOC</title></head>
+<body>
+<nav id="pub-toc" epub:type="toc">
+ <h1>Table of contents</h1>
+ <ol class="toc" hidden="hidden">
+  <li>
+   <a href="${document}">Document</a>
+  </li>
+ </ol>
+</nav>
+</body>
+</html>
+]] % {document=document}
+  return template
+end
+
+local function cleanOPF()
+  -- in epub3, there must be table of contents
+	-- if there is no toc in the document, we must add generic one
+	local opf =  "content.opf"
+	local f = io.open(opf,"r")
+	if not f then 
+    print("Cannot open "..opf .. " for toc searching")
+		return nil
+  end
+  local content = f:read("*all")
+	f:close()
+	if content:find "properties[%s]*=[%s]*\"[^\"]*nav" then
+    print "TOC nav found"
+  else
+    print "no TOC, using generic one"
+    local pattern = input.."(%..?html)"
+    local ext = content:match(pattern)
+    local inputfile = input .. ext
+    print("Main file name", inputfile)
+		-- write toc file
+    local toc_name = "generic_toc" ..ext
+		local f = io.open(outputdir .. "/" .. toc_name, "w")
+		f:write(makeTOC(inputfile))
+		f:close()
+		-- add toc file to the conten.opf
+		content = content:gsub("<manifest>","<manifest>\n<item id='htmltoc'" ..
+		  " properties=\"nav\" media-type=\"application/xhtml+xml\" href=\""..
+			toc_name .."\" />\n")
+    content = content:gsub("<spine([^>]*)>", "<spine%1>\n<itemref idref=\"htmltoc\" linear=\"no\"/>\n")
+    -- remove empty guide element
+  end
+  content = content:gsub("<guide>%s*</guide>","")
+  f = io.open(outputdir .. "/" ..opf,"w")
+  f:write(content)
+  f:close()
+  --makeTOC(inputfile)
+end
+
+
+
+function writeContainer()			
+	--local ret =  eb.writeContainer()
+	eb.make_opf()
+	cleanOPF()
+	local ret = eb.pack_container()
+	return ret
+end
+
+function clean()
+	return eb.clean()
+end
diff --git a/Master/texmf-dist/scripts/tex4ebook/exec_mobi.lua b/Master/texmf-dist/scripts/tex4ebook/exec_mobi.lua
new file mode 100755
index 00000000000..be9148b5a10
--- /dev/null
+++ b/Master/texmf-dist/scripts/tex4ebook/exec_mobi.lua
@@ -0,0 +1,20 @@
+module(...,package.seeall)
+local eb = require("exec_epub")
+
+function prepare(params)
+	return eb.prepare(params)
+end
+
+function run(out,params)
+	return eb.run(out, params)
+end
+
+function writeContainer()
+	local ret =  eb.writeContainer()
+	print("Pack mobi "..os.execute("kindlegen " .. eb.outputfile))
+	return ret
+end
+
+function clean()
+	return eb.clean()
+end
diff --git a/Master/texmf-dist/scripts/tex4ebook/list-fonts.lua b/Master/texmf-dist/scripts/tex4ebook/list-fonts.lua
new file mode 100755
index 00000000000..9acd1665f60
--- /dev/null
+++ b/Master/texmf-dist/scripts/tex4ebook/list-fonts.lua
@@ -0,0 +1,145 @@
+kpse.set_program_name("luatex")
+local  function exists(fn)
+	local f = io.open(fn,"r")
+	if f~=nil then io.close(f) return true else return false end
+end
+local function load_fontbase(pathtable)
+	-- path should point to possible location of luaotfload font database
+	-- pathtable should be array of strings
+	-- each path should not end with lua
+
+	local path = pathtable
+	if type(pathtable) =="string" then path = {pathtable} end
+	if #path == 0 then return nil end
+	local currentpath = table.remove(path,1)
+	if exists(currentpath ..".lua") then return require(currentpath) 
+	else return load_fontbase(path) end
+end
+
+--local fontbase  = require(kpse.expand_var("$TEXMFSYSVAR")  .. "/luatex-cache/generic/names/luaotfload-names.lua")
+local sysvar = kpse.expand_var("$TEXMFSYSVAR")
+local var =  kpse.expand_var("$TEXMFVAR")
+local namespath = "/luatex-cache/generic/names/"
+local oldnames = "otfl-names"
+local newnames = "luaotfload-names"
+local compose_path = function(path, name) return path .. namespath .. name end
+local fontbase_path = {
+	compose_path(sysvar, newnames), 
+	compose_path(var, newnames),
+	compose_path(sysvar, oldnames),
+	compose_path(var,oldnames)
+}
+
+local fontbase = load_fontbase(fontbase_path) or {}
+fontbase.mappings = fontbase.mappings or {}
+-- assert(fontbase, "Cannot load font names")
+
+
+-- [[
+-- This is code for querying the database
+--
+local search = arg[1] or ""
+search = unicode.utf8.lower(search)
+--]]
+local fonts = {}
+for _,record in pairs(fontbase.mappings) do
+	local familyname = record.familyname
+	local styles = fonts[familyname] or {}  
+	local sanitized = record.sanitized or record.names
+	local subfamily = sanitized.subfamily --record.sanitized.subfamily or record.names.subfamily
+	styles[subfamily] = record -- We add subfamily as key to prevent duplicates
+	fonts[familyname] = styles
+end
+
+local FontLoader = {}
+
+FontLoader.__index = FontLoader
+
+FontLoader.load_family = function(familyname)
+	local self = setmetatable({},FontLoader)
+	local family = fonts[familyname]
+	--[[ for _, record in pairs(fontbase.mappings) do
+		if record.familyname == familyname then 
+			local style = record.sanitized.subfamily
+			family[style] = record
+		end
+	end
+	--]]
+	self.family = family
+	return self
+end
+FontLoader.get_path =function(self,style)
+	local rec = self.family[style]
+	local filename = rec.filename
+	if type(filename) == "table" then
+		filename = filename[1]
+	end
+	return kpse.find_file(filename,'opentype fonts') or kpse.find_file(filename,'truetype fonts')
+end
+
+
+FontLoader.list = function(self)
+	for style,rec in pairs(self.family) do
+		print(style, self:get_path(style))
+	end
+end
+
+
+local FontNames = {}
+FontNames.__index = FontNames
+
+FontNames.new = function()
+	self = setmetatable({},FontNames)
+	self.names = {}
+	return self
+end
+
+FontNames.add_family = function(self,properties) 
+	local properties = properties or {}
+	return function(names)
+		local names = names or {}
+		for _,n in pairs(names) do
+			self.names[n] = properties
+		end
+	end
+end
+
+FontNames.list = function(self)
+	for name,prop in pairs(self.names) do
+		local p = {}
+		for k,v in pairs(prop) do
+			table.insert(p,k..'='..v)
+		end
+		print(name, table.concat(p))
+	end
+end
+
+--[[
+local fontnames = {}
+for fontname,_ in pairs(fonts) do
+	local lowername = unicode.utf8.lower(fontname)
+	if lowername:find(search) then 
+		table.insert(fontnames, fontname)
+	end
+end
+
+table.sort(fontnames)
+--for fontname, styles in pairs (fonts) do
+for _, name in pairs(fontnames) do
+	local fontname = name
+	local styles = fonts[fontname]
+	local t = {} 
+	for n,_ in pairs(styles) do table.insert(t,n) end
+	print(fontname, table.concat(t,', '))
+end
+--]]
+
+local testfont = FontLoader.load_family("math")
+--testfont:list()
+
+local n= FontNames.new()
+n:add_family {type = "serif", name = "LMMono10"} {"cmr"}
+n:add_family {type = "serif", name = "TeXGyreTermes"} {"ts1-qtmr", "ec-qtmr","ec-qtmri"}
+--n:list()
+
+return {FontLoader = FontLoader, FontNames =  n}