File

teal-src/util/datamapper.tl @ 11455:a5050e21ab08

util.datamapper: Separate extraction of xml from coercion to target type Now it gets the text, attribute or name first, then turns it into whatever the schema wants. This should be easier to further factor out into preparation for array support.
author Kim Alvefur <zash@zash.se>
date Sun, 14 Mar 2021 03:06:37 +0100
parent 11454:1d9c1893cc5e
child 11456:4e376a43fe40
line wrap: on
line source

-- Copyright (C) 2021 Kim Alvefur
--
-- This project is MIT/X11 licensed. Please see the
-- COPYING file in the source package for more information.
--
-- Based on
-- https://json-schema.org/draft/2020-12/json-schema-core.html
-- https://json-schema.org/draft/2020-12/json-schema-validation.html
-- http://spec.openapis.org/oas/v3.0.1#xmlObject
-- https://github.com/OAI/OpenAPI-Specification/issues/630 (text:true)
--
-- XML Object Extensions:
-- text to refer to the text content at the same time as attributes
-- x_name_is_value for enum fields where the <tag-name/> is the value
-- x_single_attribute for <tag attr="this"/>
--
-- TODO arrays
-- TODO pointers
-- TODO cleanup / refactor
--

local st = require "util.stanza";
local js = require "util.jsonschema"

local function toboolean ( s : string ) : boolean
	if s == "true" or s == "1" then
		return true
	elseif s == "false" or s == "0" then
		return false
	elseif s then
		return true
	end
end

local function totype(t : js.schema_t.type_e, s : string) : any
	if t == "string" then
		return s;
	elseif t == "boolean" then
		return toboolean(s)
	elseif t == "number" or t == "integer" then
		return tonumber(s)
	end
end

local enum value_goes
	"in_tag_name"
	"in_text"
	"in_text_tag"
	"in_attribute"
	"in_single_attribute"
	"in_children"
end

local function parse_object (schema : js.schema_t, s : st.stanza_t) : table
	local out : { string : any } = {}
	if schema.properties then
		for prop, propschema in pairs(schema.properties) do
			-- TODO factor out, if it's generic enough
			local name = prop
			local namespace = s.attr.xmlns;
			local prefix : string = nil
			local value_where : value_goes = "in_text_tag"
			local single_attribute : string
			local enums : { any }

			local proptype : js.schema_t.type_e
			if propschema is js.schema_t then
				proptype = propschema.type
			elseif propschema is js.schema_t.type_e then
				proptype = propschema
			end

			if proptype == "object" or proptype == "array" then
				value_where = "in_children"
			end

			if propschema is js.schema_t and propschema.xml then
				if propschema.xml.name then
					name = propschema.xml.name
				end
				if propschema.xml.namespace then
					namespace = propschema.xml.namespace
				end
				if propschema.xml.prefix then
					prefix = propschema.xml.prefix
				end
				if propschema.xml.attribute then
					value_where = "in_attribute"
				elseif propschema.xml.text then
					-- XXX Not yet in OpenAPI
					value_where = "in_text"
				elseif propschema.xml.x_name_is_value then
					-- XXX Custom extension
					value_where = "in_tag_name"
				elseif propschema.xml.x_single_attribute then
					-- XXX Custom extension
					single_attribute = propschema.xml.x_single_attribute
					value_where = "in_single_attribute"
				end
				if propschema["const"] then
					enums = { propschema["const"] }
				elseif propschema["enum"] then
					enums = propschema["enum"]
				end
			end

			local value : string
			if value_where == "in_tag_name" then
				local c : st.stanza_t
				if proptype == "boolean" then
					c = s:get_child(name, namespace);
				elseif enums and proptype == "string" then
					-- XXX O(n²) ?
					-- Probably better to flip the table and loop over :childtags(nil, ns), should be 2xO(n)
					-- BUT works first, optimize later
					for i = 1, #enums do
						c = s:get_child(enums[i] as string, namespace);
						if c then break end
					end
				else
					c = s:get_child(nil, namespace);
				end
				value = c.name;
			elseif value_where == "in_attribute" then
				local attr = name
				if prefix then
					attr = prefix .. ':' .. name
				elseif namespace ~= s.attr.xmlns then
					attr = namespace .. "\1" .. name
				end
				value = s.attr[attr]

			elseif value_where == "in_text" then
				value = s:get_text()

			elseif value_where == "in_single_attribute" then
				local c = s:get_child(name, namespace)
				value = c and c.attr[single_attribute]
			elseif value_where == "in_text_tag" then
				value = s:get_child_text(name, namespace)
			elseif value_where == "in_children" and propschema is js.schema_t then
				if proptype == "object" then
					local c = s:get_child(name, namespace)
					if c then
						out[prop] = parse_object(propschema, c);
					end
				-- else TODO
				end
			end
			if value_where ~= "in_children" then
				out[prop] = totype(proptype, value)
			end
		end
	end

	return out
end

local function parse (schema : js.schema_t, s : st.stanza_t) : table
	if schema.type == "object" then
		return parse_object(schema, s)
	end
end

local function unparse ( schema : js.schema_t, t : table, current_name : string, current_ns : string ) : st.stanza_t
	if schema.type == "object" then

		if schema.xml then
			if schema.xml.name then
				current_name = schema.xml.name
			end
			if schema.xml.namespace then
				current_ns = schema.xml.namespace
			end
			-- TODO prefix?
		end

		local out = st.stanza(current_name, { xmlns = current_ns })

		for prop, propschema in pairs(schema.properties) do
			local v = t[prop]

			if v ~= nil then
				local proptype : js.schema_t.type_e
				if propschema is js.schema_t then
					proptype = propschema.type
				elseif propschema is js.schema_t.type_e then
					proptype = propschema
				end

				local name = prop
				local namespace = current_ns
				local prefix : string = nil
				local value_where : value_goes = "in_text_tag"
				local single_attribute : string

				if propschema is js.schema_t and propschema.xml then

					if propschema.xml.name then
						name = propschema.xml.name
					end
					if propschema.xml.namespace then
						namespace = propschema.xml.namespace
					end

					if propschema.xml.prefix then
						prefix = propschema.xml.prefix
					end

					if propschema.xml.attribute then
						value_where = "in_attribute"
					elseif propschema.xml.text then
						value_where = "in_text"
					elseif propschema.xml.x_name_is_value then
						value_where = "in_tag_name"
					elseif propschema.xml.x_single_attribute then
						single_attribute = propschema.xml.x_single_attribute
						value_where = "in_single_attribute"
					end
				end

				if value_where == "in_attribute" then
					local attr = name
					if prefix then
						attr = prefix .. ':' .. name
					elseif namespace ~= current_ns then
						attr = namespace .. "\1" .. name
					end

					if proptype == "string" and v is string then
						out.attr[attr] = v
					elseif proptype == "number" and v is number then
						out.attr[attr] = string.format("%g", v)
					elseif proptype == "integer" and v is number then
						out.attr[attr] = string.format("%d", v)
					elseif proptype == "boolean" then
						out.attr[attr] = v and "1" or "0"
					end
				elseif value_where == "in_text" then
					if v is string then
						out:text(v)
					end
				elseif value_where == "in_single_attribute" then
					local propattr : { string : string } = {}

					if namespace ~= current_ns then
						propattr.xmlns = namespace
					end

					if proptype == "string" and v is string then
						propattr[single_attribute] = v
					elseif proptype == "number" and v is number then
						propattr[single_attribute] = string.format("%g", v)
					elseif proptype == "integer" and v is number then
						propattr[single_attribute] = string.format("%d", v)
					elseif proptype == "boolean" and v is boolean then
						propattr[single_attribute] = v and "1" or "0"
					end
					out:tag(name, propattr):up();

				else
					local propattr : { string : string }
					if namespace ~= current_ns then
						propattr = { xmlns = namespace }
					end
					if value_where == "in_tag_name" then
						if proptype == "string" and v is string then
							out:tag(v, propattr):up();
						elseif proptype == "boolean" and v == true then
							out:tag(name, propattr):up();
						end
					elseif proptype == "string" and v is string then
						out:text_tag(name, v, propattr)
					elseif proptype == "number" and v is number then
						out:text_tag(name, string.format("%g", v), propattr)
					elseif proptype == "integer" and v is number then
						out:text_tag(name, string.format("%d", v), propattr)
					elseif proptype == "boolean" and v is boolean then
						out:text_tag(name, v and "1" or "0", propattr)
					elseif proptype == "object" and propschema is js.schema_t and v is table then
						local c = unparse(propschema, v, name, namespace);
						if c then
							out:add_direct_child(c);
						end
					-- else TODO
					end
				end
			end
		end
		return out;

	end
end

return {
	parse = parse,
	unparse = unparse,
}