beta 2011.11.22 16:49

1 files changed, 73 insertions, 2 deletions

diff --git a/tex/context/base/lxml-aux.lua b/tex/context/base/lxml-aux.lua
index 4798fe06a..560ef557d 100644
--- a/tex/context/base/lxml-aux.lua
+++ b/tex/context/base/lxml-aux.lua
@@ -20,8 +20,9 @@ local xmlinheritedconvert = xml.inheritedconvert
 local xmlapplylpath = xml.applylpath
 
 local type, setmetatable, getmetatable = type, setmetatable, getmetatable
-local insert, remove, fastcopy = table.insert, table.remove, table.fastcopy
-local gmatch, gsub = string.gmatch, string.gsub
+local insert, remove, fastcopy, concat = table.insert, table.remove, table.fastcopy, table.concat
+local gmatch, gsub, format = string.gmatch, string.gsub, string.format
+local utfbyte = utf.byte
 
 local function report(what,pattern,c,e)
     report_xml("%s element '%s' (root: '%s', position: %s, index: %s, pattern: %s)",what,xmlname(e),xmlname(e.__p__),c,e.ni,pattern)
@@ -516,6 +517,75 @@ function xml.remapname(root, pattern, newtg, newns, newrn)
 end
 
 --[[ldx--
+<p>Helper (for q2p).</p>
+--ldx]]--
+
+function xml.cdatatotext(e)
+    local dt = e.dt
+    if #dt == 1 then
+        local first = dt[1]
+        if first.tg == "@cd@" then
+            e.dt = first.dt
+        end
+    else
+        -- maybe option
+    end
+end
+
+xml.builtinentities = table.tohash { "amp", "quot", "apos", "lt", "gt" } -- used often so share
+
+local entities        = characters and characters.entities or nil
+local builtinentities = xml.builtinentities
+
+function xml.addentitiesdoctype(root,option) -- we could also have a 'resolve' i.e. inline hex
+    if not entities then
+        require("char-ent")
+        entities = characters.entities
+    end
+    if entities and root and root.tg == "@rt@" and root.statistics then
+        local list = { }
+        local hexify = option == "hexadecimal"
+        for k, v in table.sortedhash(root.statistics.entities.names) do
+            if not builtinentities[k] then
+                local e = entities[k]
+                if not e then
+                    e = format("[%s]",k)
+                elseif hexify then
+                    e = format("&#%05X;",utfbyte(k))
+                end
+                list[#list+1] = format("  <!ENTITY %s %q >",k,e)
+            end
+        end
+        local dt = root.dt
+        local n = dt[1].tg == "@pi@" and 2 or 1
+        if #list > 0 then
+            insert(dt, n, { "\n" })
+            insert(dt, n, {
+               tg      = "@dt@", -- beware, doctype is unparsed
+               dt      = { format("Something [\n%s\n] ",concat(list)) },
+               ns      = "",
+               special = true,
+            })
+            insert(dt, n, { "\n\n" })
+        else
+         -- insert(dt, n, { table.serialize(root.statistics) })
+        end
+    end
+end
+
+-- local str = [==[
+-- <?xml version='1.0' standalone='yes' ?>
+-- <root>
+-- <a>test &nbsp; test &#123; test</a>
+-- <b><![CDATA[oeps]]></b>
+-- </root>
+-- ]==]
+--
+-- local x = xml.convert(str)
+-- xml.addentitiesdoctype(x,"hexadecimal")
+-- print(x)
+
+--[[ldx--
 <p>Here are a few synonyms.</p>
 --ldx]]--
 
@@ -547,3 +617,4 @@ xml.inject_element             = xml.inject                obsolete.inject_eleme
 xml.remap_tag                  = xml.remaptag              obsolete.remap_tag             = xml.remaptag
 xml.remap_name                 = xml.remapname             obsolete.remap_name            = xml.remapname
 xml.remap_namespace            = xml.remapnamespace        obsolete.remap_namespace       = xml.remapnamespace
+