From 0431089dd2bd506171d5cc22183d53c9b81f611e Mon Sep 17 00:00:00 2001
From: ubq323 <ubq323@ubq323.website>
Date: Tue, 5 Nov 2024 18:55:43 +0000
Subject: initial

---
 conlex.lua | 116 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 116 insertions(+)
 create mode 100755 conlex.lua

(limited to 'conlex.lua')

diff --git a/conlex.lua b/conlex.lua
new file mode 100755
index 0000000..afee980
--- /dev/null
+++ b/conlex.lua
@@ -0,0 +1,116 @@
+#!/usr/bin/env lua5.3
+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 items
+end
+
+local function engloss(langue, text)
+	local items = glossors[langue](text)
+	return template([[
+	<section>%text%</section>
+	<section id=place></section>]], {text=table.concat(items, "\n")})
+end
+
+local function enform()
+	return [[
+<section>
+<form>
+	<textarea name=q></textarea>
+	<button>perform <i>conlation</i></button>
+</form>
+</section>]]
+end
+
+local function response(content)
+	print("Content-Type: text/html")
+	print()
+	print(template(document_template, {content=content}))
+end
+
+local function main()
+	local text = nil
+	local qs = os.getenv"QUERY_STRING"
+	if qs then text = parse_qs(qs).q end
+
+	if not text then
+		response(enform())
+	else
+		response(engloss('sanila', text))
+	end
+end
+
+main()
+
-- 
cgit v1.2.3