conlex.lua -> conglossor.cgi

1 files changed, 165 insertions, 0 deletions

diff --git a/conglossor.cgi b/conglossor.cgi
new file mode 100755
index 0000000..bf3ff8c
--- /dev/null
+++ b/conglossor.cgi
@@ -0,0 +1,165 @@
+#!/usr/bin/env lua5.3
+local lpeg = require'lpeg'
+
+local function template(str, args)
+	return (str:gsub("%%(%w+)%%", args))
+end
+
+local function urldecode(s)
+	return s:gsub('+',' ')
+		:gsub('%%(%x%x)',function(x)
+			return string.char(tonumber(x,16))
+		end)
+end
+local function parse_qs(s)
+	local ix = 1
+	local res = {}
+	while true do
+		local _,next_ix,ek,ev = s:find("([^=]+)=([^&]*)&?",ix)
+		if not ek then break end
+		local k,v = urldecode(ek), urldecode(ev)
+		res[k] = v
+		ix = next_ix + 1
+	end
+	return res
+end
+
+local function eprint(...)
+	local s = {}
+	for i = 1, select('#',...) do
+		s[i] = tostring(select(i, ...))
+	end
+	io.stderr:write(table.concat(s, "\t"), "\n")
+end
+
+local document_template
+do
+	local template_file = assert(io.open"template.html")
+	document_template = template_file:read"a"
+	template_file:close()
+end
+
+local function load_dict(filename)
+	local res = {}
+	for line in io.lines(filename) do
+		local word, pos, def = line:match "^(.-)\t(.-)\t(.-)$"
+		if word then
+			res[word] = {pos=pos, def=def, word=word}
+		end
+	end
+	return res
+end
+
+local function format_def(found_word, entry)
+	local defc
+	if entry then
+		defc = template("<b>%word%</b> <i>%pos%</i>. %def%", entry)
+	else
+		defc = "????"
+	end
+	return template(
+		"<word><def>%defc%</def><wi>%wi%</wi></word>",
+		{defc=defc, wi=found_word}
+	)
+end
+
+local glossors = {}
+
+function glossors.sanila(text)
+	local words = load_dict "sanila.tsv"
+
+	local items = {}
+	-- sanila morphology is easy enough
+	for word in text:gmatch("%S+") do
+		local stripped = word:match("^(.-)[.,]*$")
+		local entry = words[stripped]
+		table.insert(items, format_def(word, entry))
+	end
+
+	return table.concat(items, "\n")
+end
+
+glossors["vötgil"] = function(text)
+	local words = load_dict "vötgil.tsv"
+	local function norm(s)
+		return s:lower():gsub("Ð","ð"):gsub("Ö","ö"):gsub("0","ö"):gsub("9","ð")
+	end
+
+	local letter = lpeg.R("az","AZ") + lpeg.S"09" + "ö" + "Ö" + 'ð' + 'Ð'
+	local word = (letter*letter*letter) / function (w)
+		return format_def(w, words[norm(w)])
+	end * lpeg.Cp()
+
+	local i = 1
+	local out = ""
+	while i < #text do
+		local item, ni = lpeg.match(word, text, i)
+		if item then
+			out = out .. item
+			i = ni
+		else
+			out = out .. text:sub(i,i)
+			i = i + 1
+		end
+	end
+	return out
+end
+
+local function engloss(langue, text)
+	local text = glossors[langue](text)
+	return template([[
+	<section>%text%</section>
+	<section id=place></section>
+	<section><a href="%path%">&lt; return</a></section>
+	]], {text=text,path=os.getenv"SCRIPT_NAME"})
+end
+
+local function enform()
+	local options = {}
+	for k in pairs(glossors) do
+		table.insert(options, template('<option value="%x%" >%x%</option>', {x=k}))
+	end
+	table.sort(options)
+	return template([[
+<section>
+<form>
+	<textarea name=q></textarea>
+	<label for=langue>select con-langue:</label>
+	<select name=langue id=langue>
+		%options%
+	</select>
+	<br><button>perform <i>conglossatio</i></button>
+</form>
+</section>
+]], {
+		options=table.concat(options, "\n"),
+
+	})
+end
+
+local function response(content)
+	print("Content-Type: text/html")
+	print()
+	print(template(document_template, {content=content}))
+end
+
+local function main()
+	local langue, text
+	local qs = os.getenv"QUERY_STRING"
+	if qs then
+		local pqs = parse_qs(qs)
+		langue = pqs.langue
+		text = pqs.q
+	end
+
+	if not text then
+		response(enform())
+	elseif not glossors[langue] then
+		response("unknown langue "..langue)
+	else
+		response(engloss(langue, text))
+	end
+end
+
+main()
+