convert LaTeX sectioning commands to embedded TEI divs equivalents

author: Robert Alessi <alessi@robertalessi.net> 2020-04-26 19:47:27 +0200
committer: Robert Alessi <alessi@robertalessi.net> 2020-04-26 19:47:27 +0200
commit: b3d7ebd008cea5f2c9fc532316ac9051650e67d8 (patch)
tree: aabfb72e495543e5c6ae8bfb6d0aaf63409848f3 /ekdosis.dtx
parent: 4145508cdf18110cc2d1dc47d9a9d86e3e60469d (diff)
download: ekdosis-b3d7ebd008cea5f2c9fc532316ac9051650e67d8.tar.gz
1 files changed, 80 insertions, 5 deletions
diff --git a/ekdosis.dtx b/ekdosis.dtx
index 63b3b75..5c142e2 100644
--- a/ekdosis.dtx
+++ b/ekdosis.dtx
@@ -1193,6 +1193,10 @@ local note = lpeg.Cs("note")
 local lnbrk = lpeg.Cs("\\\\")
 local poemline = lpeg.Cs(lnbrk * bsqbrackets^-1)
 local endpoem = lpeg.Cs(lnbrk * lpeg.S("*!") * bsqbrackets^-1)
+local sections = lpeg.Cs(lpeg.P("book") + lpeg.P("part") + lpeg.P("chapter")
+                            + lpeg.P("section") + lpeg.P("subsection")
+                            + lpeg.P("subsubsection") + lpeg.P("paragraph")
+                            + lpeg.P("subparagraph"))
 -- Bind to local variables
 local next = next
@@ -1661,12 +1665,12 @@ local function envtotei(str)
            then
               str = gsub(str, (lpeg.P("\\par") + lpeg.P("<p>")) * spcenc^-1 * lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")
                             * lpeg.Cs(envtotags[i].a) * lpeg.P("}")
-                             * bsqbracketsii * bcbracesii * spcenc^-1,
+                             * bcbracesii * spcenc^-1,
                          "\n<"..envtotags[i].b..envtotags[i].c..">")
            else
               str = gsub(str, (lpeg.P("\\par") + lpeg.P("<p>")) * spcenc^-1 * lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")
                             * lpeg.Cs(envtotags[i].a) * lpeg.P("}")
-                             * bsqbracketsii * bcbracesii * spcenc^-1,
+                             * bcbracesii * spcenc^-1,
                          "</p>\n<"..envtotags[i].b..envtotags[i].c..">")
            end
            str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")
@@ -1674,7 +1678,7 @@ local function envtotei(str)
                       "</"..envtotags[i].b..">\n<p>")
            str = gsub(str, lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")
                          * lpeg.Cs(envtotags[i].a) * lpeg.P("}")
-                          * bsqbracketsii * bcbracesii * spcenc^-1,
+                          * bcbracesii * spcenc^-1,
                       "<"..envtotags[i].b..envtotags[i].c..">")
            str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")
                          * lpeg.Cs(envtotags[i].a) * lpeg.P("}"),
@@ -1682,7 +1686,7 @@ local function envtotei(str)
         else
            str = gsub(str, lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")
                          * lpeg.Cs(envtotags[i].a) * lpeg.P("}")
-                          * bsqbracketsii * bcbracesii * spcenc^-1,
+                          * bcbracesii * spcenc^-1,
                       "<"..envtotags[i].b..envtotags[i].c..">")
            str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")
                          * lpeg.Cs(envtotags[i].a) * lpeg.P("}"),
@@ -1691,7 +1695,7 @@ local function envtotei(str)
      else
         str = gsub(str, lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")
                       * lpeg.Cs(envtotags[i].a) * lpeg.P("}")
-                       * bsqbracketsii * bcbracesii * spcenc^-1,
+                       * bcbracesii * spcenc^-1,
                    "")
         str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")
                       * lpeg.Cs(envtotags[i].a) * lpeg.P("}"),
@@ -1770,6 +1774,8 @@ local function partotei(str)
   str = string.gsub(str, "<p>%s-</p>", "")
   str = string.gsub(str, "(%<p%>)%s-(%</div%>)$", "%2")
   str = string.gsub(str, "(%<p%>)%s-$", "")
+   str = string.gsub(str, "(<p>)%s-(<div.->)", "%2%1")
+   str = string.gsub(str, "(%<p%>)%s-(<head>.-</head>)", "%2%1")
   return str
 end
@@ -1812,6 +1818,72 @@ local function self_close_tags(str)
   return str
 end
+local divdepth = {
+   book = 0,
+   part = 1,
+   chapter = 2,
+   section = 3,
+   subsection = 4,
+   subsubsection = 5,
+   paragraph = 6,
+   subparagraph = 7
+}
+local function section_totei(str)
+   str = gsub(str, dblbkslash * sections * spce^-1 * bcbraces, "%1%2%3[]%4")
+   str = gsub(str, dblbkslash * sections * spce^-1 * bsqbrackets * bcbraces,
+              function(bkslash, secname, space, opt, arg)
+                 ctr = divdepth[secname]
+                 arg = string.sub(arg, 2, -2)
+                 -- return string.format("\\par <div%s type=\"%s\" depth=\"%s\"><head>%s</head>",
+                 --                   ctr, secname, ctr, arg)
+                 return string.format("\\par <div%s type=\"%s\"><head>%s</head>",
+                                      ctr, secname, arg)
+   end)
+   return str
+end
+local function close_divs_at_end(str)
+   local isdiv = false
+   if string.find(str, "</div>$")
+   then
+      isdiv = true
+      str = string.gsub(str, "(.*)(</div>)$", "%1")
+   else
+   end
+   firstdiv = string.match(str, "<div%d") or ""
+   firstdiv = tonumber(string.match(firstdiv, "%d")) or ""
+   lastdiv = string.match(string.reverse(str), "%dvid<") or ""
+   lastdiv = tonumber(string.match(lastdiv, "%d")) or ""
+   local closedivs = ""
+   if tonumber(firstdiv)
+   then
+      while lastdiv >= firstdiv
+      do
+         closedivs = closedivs.."</div"..lastdiv..">"
+         lastdiv = lastdiv - 1
+      end
+   end
+--   str = string.gsub(str, "(<div)(%d)(.->)", "%1%3")
+   if isdiv
+   then
+      return str..closedivs.."</div>"
+   else
+      return str..closedivs
+   end
+end
+local function close_divs_in_between(str)
+   for i = 0, 9
+   do
+      str = string.gsub(str, "(<div"..i..".->)(.-)(<div"..i..".->)", function(divi, between, divii)
+                           between = close_divs_in_between(between)
+                           return string.format("%s%s</div%s>%s", divi, between, i, divii)
+      end)
+   end
+   return str
+end
 local function textotei(str)
   str = xml_entities(str)
   str = texpatttotei(str)
@@ -1821,10 +1893,13 @@ local function textotei(str)
   str = relocate_notes(str)
   str = versetotei(str)
   str = envtotei(str)
+   str = section_totei(str)
   str = cmdtotei(str)
   str = self_close_tags(str)
   str = partotei(str)
   str = checkpars(str)
+   str = close_divs_at_end(str)
+   str = close_divs_in_between(str)
   return str
 end
author	Robert Alessi <alessi@robertalessi.net>	2020-04-26 19:47:27 +0200
committer	Robert Alessi <alessi@robertalessi.net>	2020-04-26 19:47:27 +0200
commit	b3d7ebd008cea5f2c9fc532316ac9051650e67d8 (patch)
tree	aabfb72e495543e5c6ae8bfb6d0aaf63409848f3 /ekdosis.dtx
parent	4145508cdf18110cc2d1dc47d9a9d86e3e60469d (diff)
download	ekdosis-b3d7ebd008cea5f2c9fc532316ac9051650e67d8.tar.gz

diff --git a/ekdosis.dtx b/ekdosis.dtx index 63b3b75..5c142e2 100644 --- a/ekdosis.dtx +++ b/ekdosis.dtx
@@ -1193,6 +1193,10 @@ local note = lpeg.Cs("note")
1193	local lnbrk = lpeg.Cs("\\\\")	1193	local lnbrk = lpeg.Cs("\\\\")
1194	local poemline = lpeg.Cs(lnbrk * bsqbrackets^-1)	1194	local poemline = lpeg.Cs(lnbrk * bsqbrackets^-1)
1195	local endpoem = lpeg.Cs(lnbrk * lpeg.S("!") bsqbrackets^-1)	1195	local endpoem = lpeg.Cs(lnbrk * lpeg.S("!") bsqbrackets^-1)
		1196	local sections = lpeg.Cs(lpeg.P("book") + lpeg.P("part") + lpeg.P("chapter")
		1197	+ lpeg.P("section") + lpeg.P("subsection")
		1198	+ lpeg.P("subsubsection") + lpeg.P("paragraph")
		1199	+ lpeg.P("subparagraph"))
1196		1200
1197	-- Bind to local variables	1201	-- Bind to local variables
1198	local next = next	1202	local next = next
@@ -1661,12 +1665,12 @@ local function envtotei(str)
1661	then	1665	then
1662	str = gsub(str, (lpeg.P("\\par") + lpeg.P("<p>")) * spcenc^-1 * lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")	1666	str = gsub(str, (lpeg.P("\\par") + lpeg.P("<p>")) * spcenc^-1 * lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")
1663	* lpeg.Cs(envtotags[i].a) * lpeg.P("}")	1667	* lpeg.Cs(envtotags[i].a) * lpeg.P("}")
1664	* bsqbracketsii * bcbracesii * spcenc^-1,	1668	* bcbracesii * spcenc^-1,
1665	"\n<"..envtotags[i].b..envtotags[i].c..">")	1669	"\n<"..envtotags[i].b..envtotags[i].c..">")
1666	else	1670	else
1667	str = gsub(str, (lpeg.P("\\par") + lpeg.P("<p>")) * spcenc^-1 * lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")	1671	str = gsub(str, (lpeg.P("\\par") + lpeg.P("<p>")) * spcenc^-1 * lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")
1668	* lpeg.Cs(envtotags[i].a) * lpeg.P("}")	1672	* lpeg.Cs(envtotags[i].a) * lpeg.P("}")
1669	* bsqbracketsii * bcbracesii * spcenc^-1,	1673	* bcbracesii * spcenc^-1,
1670	"</p>\n<"..envtotags[i].b..envtotags[i].c..">")	1674	"</p>\n<"..envtotags[i].b..envtotags[i].c..">")
1671	end	1675	end
1672	str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")	1676	str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")
@@ -1674,7 +1678,7 @@ local function envtotei(str)
1674	"</"..envtotags[i].b..">\n<p>")	1678	"</"..envtotags[i].b..">\n<p>")
1675	str = gsub(str, lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")	1679	str = gsub(str, lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")
1676	* lpeg.Cs(envtotags[i].a) * lpeg.P("}")	1680	* lpeg.Cs(envtotags[i].a) * lpeg.P("}")
1677	* bsqbracketsii * bcbracesii * spcenc^-1,	1681	* bcbracesii * spcenc^-1,
1678	"<"..envtotags[i].b..envtotags[i].c..">")	1682	"<"..envtotags[i].b..envtotags[i].c..">")
1679	str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")	1683	str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")
1680	* lpeg.Cs(envtotags[i].a) * lpeg.P("}"),	1684	* lpeg.Cs(envtotags[i].a) * lpeg.P("}"),
@@ -1682,7 +1686,7 @@ local function envtotei(str)
1682	else	1686	else
1683	str = gsub(str, lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")	1687	str = gsub(str, lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")
1684	* lpeg.Cs(envtotags[i].a) * lpeg.P("}")	1688	* lpeg.Cs(envtotags[i].a) * lpeg.P("}")
1685	* bsqbracketsii * bcbracesii * spcenc^-1,	1689	* bcbracesii * spcenc^-1,
1686	"<"..envtotags[i].b..envtotags[i].c..">")	1690	"<"..envtotags[i].b..envtotags[i].c..">")
1687	str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")	1691	str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")
1688	* lpeg.Cs(envtotags[i].a) * lpeg.P("}"),	1692	* lpeg.Cs(envtotags[i].a) * lpeg.P("}"),
@@ -1691,7 +1695,7 @@ local function envtotei(str)
1691	else	1695	else
1692	str = gsub(str, lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")	1696	str = gsub(str, lpeg.P("\\begin") * spcenc^-1 * lpeg.P("{")
1693	* lpeg.Cs(envtotags[i].a) * lpeg.P("}")	1697	* lpeg.Cs(envtotags[i].a) * lpeg.P("}")
1694	* bsqbracketsii * bcbracesii * spcenc^-1,	1698	* bcbracesii * spcenc^-1,
1695	"")	1699	"")
1696	str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")	1700	str = gsub(str, spcenc^-1 * lpeg.P("\\end") * spcenc^-1 * lpeg.P("{")
1697	* lpeg.Cs(envtotags[i].a) * lpeg.P("}"),	1701	* lpeg.Cs(envtotags[i].a) * lpeg.P("}"),
@@ -1770,6 +1774,8 @@ local function partotei(str)
1770	str = string.gsub(str, "<p>%s-</p>", "")	1774	str = string.gsub(str, "<p>%s-</p>", "")
1771	str = string.gsub(str, "(%<p%>)%s-(%</div%>)$", "%2")	1775	str = string.gsub(str, "(%<p%>)%s-(%</div%>)$", "%2")
1772	str = string.gsub(str, "(%<p%>)%s-$", "")	1776	str = string.gsub(str, "(%<p%>)%s-$", "")
		1777	str = string.gsub(str, "(<p>)%s-(<div.->)", "%2%1")
		1778	str = string.gsub(str, "(%<p%>)%s-(<head>.-</head>)", "%2%1")
1773	return str	1779	return str
1774	end	1780	end
1775		1781
@@ -1812,6 +1818,72 @@ local function self_close_tags(str)
1812	return str	1818	return str
1813	end	1819	end
1814		1820
		1821	local divdepth = {
		1822	book = 0,
		1823	part = 1,
		1824	chapter = 2,
		1825	section = 3,
		1826	subsection = 4,
		1827	subsubsection = 5,
		1828	paragraph = 6,
		1829	subparagraph = 7
		1830	}
		1831
		1832	local function section_totei(str)
		1833	str = gsub(str, dblbkslash * sections * spce^-1 * bcbraces, "%1%2%3[]%4")
		1834	str = gsub(str, dblbkslash * sections * spce^-1 * bsqbrackets * bcbraces,
		1835	function(bkslash, secname, space, opt, arg)
		1836	ctr = divdepth[secname]
		1837	arg = string.sub(arg, 2, -2)
		1838	-- return string.format("\\par <div%s type=\"%s\" depth=\"%s\"><head>%s</head>",
		1839	-- ctr, secname, ctr, arg)
		1840	return string.format("\\par <div%s type=\"%s\"><head>%s</head>",
		1841	ctr, secname, arg)
		1842	end)
		1843	return str
		1844	end
		1845
		1846	local function close_divs_at_end(str)
		1847	local isdiv = false
		1848	if string.find(str, "</div>$")
		1849	then
		1850	isdiv = true
		1851	str = string.gsub(str, "(.*)(</div>)$", "%1")
		1852	else
		1853	end
		1854	firstdiv = string.match(str, "<div%d") or ""
		1855	firstdiv = tonumber(string.match(firstdiv, "%d")) or ""
		1856	lastdiv = string.match(string.reverse(str), "%dvid<") or ""
		1857	lastdiv = tonumber(string.match(lastdiv, "%d")) or ""
		1858	local closedivs = ""
		1859	if tonumber(firstdiv)
		1860	then
		1861	while lastdiv >= firstdiv
		1862	do
		1863	closedivs = closedivs.."</div"..lastdiv..">"
		1864	lastdiv = lastdiv - 1
		1865	end
		1866	end
		1867	-- str = string.gsub(str, "(<div)(%d)(.->)", "%1%3")
		1868	if isdiv
		1869	then
		1870	return str..closedivs.."</div>"
		1871	else
		1872	return str..closedivs
		1873	end
		1874	end
		1875
		1876	local function close_divs_in_between(str)
		1877	for i = 0, 9
		1878	do
		1879	str = string.gsub(str, "(<div"..i..".->)(.-)(<div"..i..".->)", function(divi, between, divii)
		1880	between = close_divs_in_between(between)
		1881	return string.format("%s%s</div%s>%s", divi, between, i, divii)
		1882	end)
		1883	end
		1884	return str
		1885	end
		1886
1815	local function textotei(str)	1887	local function textotei(str)
1816	str = xml_entities(str)	1888	str = xml_entities(str)
1817	str = texpatttotei(str)	1889	str = texpatttotei(str)
@@ -1821,10 +1893,13 @@ local function textotei(str)
1821	str = relocate_notes(str)	1893	str = relocate_notes(str)
1822	str = versetotei(str)	1894	str = versetotei(str)
1823	str = envtotei(str)	1895	str = envtotei(str)
		1896	str = section_totei(str)
1824	str = cmdtotei(str)	1897	str = cmdtotei(str)
1825	str = self_close_tags(str)	1898	str = self_close_tags(str)
1826	str = partotei(str)	1899	str = partotei(str)
1827	str = checkpars(str)	1900	str = checkpars(str)
		1901	str = close_divs_at_end(str)
		1902	str = close_divs_in_between(str)
1828	return str	1903	return str
1829	end	1904	end
1830		1905