|
1 -- Copyright (C) 2021 Kim Alvefur |
|
2 -- |
|
3 -- This project is MIT/X11 licensed. Please see the |
|
4 -- COPYING file in the source package for more information. |
|
5 -- |
|
6 -- Based on |
|
7 -- https://json-schema.org/draft/2020-12/json-schema-core.html |
|
8 -- https://json-schema.org/draft/2020-12/json-schema-validation.html |
|
9 -- http://spec.openapis.org/oas/v3.0.1#xmlObject |
|
10 -- https://github.com/OAI/OpenAPI-Specification/issues/630 (text:true) |
|
11 -- |
|
12 -- XML Object Extensions: |
|
13 -- text to refer to the text content at the same time as attributes |
|
14 -- x_name_is_value for enum fields where the <tag-name/> is the value |
|
15 -- x_single_attribute for <tag attr="this"/> |
|
16 -- |
|
17 -- TODO pointers |
|
18 -- TODO cleanup / refactor |
|
19 -- TODO s/number/integer/ once we have appropriate math.type() compat |
|
20 -- |
|
21 |
|
22 if not math.type then require "prosody.util.mathcompat" end |
|
23 |
|
24 local st = require "prosody.util.stanza"; |
|
25 local json = require "prosody.util.json" |
|
26 local pointer = require "prosody.util.jsonpointer"; |
|
27 |
|
28 local json_type_name = json.json_type_name; |
|
29 local json_schema_object = require "prosody.util.jsonschema" |
|
30 local type schema_t = boolean | json_schema_object |
|
31 |
|
32 local function toboolean ( s : string ) : boolean |
|
33 if s == "true" or s == "1" then |
|
34 return true |
|
35 elseif s == "false" or s == "0" then |
|
36 return false |
|
37 elseif s then |
|
38 return true |
|
39 end |
|
40 end |
|
41 |
|
42 local function totype(t : json_type_name, s : string) : any |
|
43 if not s then return nil end |
|
44 if t == "string" then |
|
45 return s; |
|
46 elseif t == "boolean" then |
|
47 return toboolean(s) |
|
48 elseif t == "number" or t == "integer" then |
|
49 return tonumber(s) |
|
50 end |
|
51 end |
|
52 |
|
53 local enum value_goes |
|
54 "in_tag_name" |
|
55 "in_text" |
|
56 "in_text_tag" |
|
57 "in_attribute" |
|
58 "in_single_attribute" |
|
59 "in_children" |
|
60 "in_wrapper" |
|
61 end |
|
62 |
|
63 local function resolve_schema(schema : schema_t, root : json_schema_object) : schema_t |
|
64 if schema is json_schema_object then |
|
65 if schema["$ref"] and schema["$ref"]:sub(1, 1) == "#" then |
|
66 return pointer.resolve(root as table, schema["$ref"]:sub(2)) as schema_t; |
|
67 end |
|
68 end |
|
69 return schema; |
|
70 end |
|
71 |
|
72 local function guess_schema_type(schema : json_schema_object) : json_type_name |
|
73 local schema_types = schema.type |
|
74 if schema_types is json_type_name then |
|
75 return schema_types |
|
76 elseif schema_types ~= nil then |
|
77 error "schema has unsupported 'type' property" |
|
78 elseif schema.properties then |
|
79 return "object" |
|
80 elseif schema.items then |
|
81 return "array" |
|
82 end |
|
83 return "string" -- default assumption |
|
84 end |
|
85 |
|
86 local function unpack_propschema( propschema : schema_t, propname : string, current_ns : string ) |
|
87 : json_type_name, value_goes, string, string, string, string, { any } |
|
88 local proptype : json_type_name = "string" |
|
89 local value_where : value_goes = propname and "in_text_tag" or "in_text" |
|
90 local name = propname |
|
91 local namespace : string |
|
92 local prefix : string |
|
93 local single_attribute : string |
|
94 local enums : { any } |
|
95 |
|
96 if propschema is json_schema_object then |
|
97 proptype = guess_schema_type(propschema); |
|
98 elseif propschema is string then -- Teal says this can never be a string, but it could before so best be sure |
|
99 error("schema as string is not supported: "..propschema.." {"..current_ns.."}"..propname) |
|
100 end |
|
101 |
|
102 if proptype == "object" or proptype == "array" then |
|
103 value_where = "in_children" |
|
104 end |
|
105 |
|
106 if propschema is json_schema_object then |
|
107 local xml = propschema.xml |
|
108 if xml then |
|
109 if xml.name then |
|
110 name = xml.name |
|
111 end |
|
112 if xml.namespace and xml.namespace ~= current_ns then |
|
113 namespace = xml.namespace |
|
114 end |
|
115 if xml.prefix then |
|
116 prefix = xml.prefix |
|
117 end |
|
118 if proptype == "array" and xml.wrapped then |
|
119 value_where = "in_wrapper" |
|
120 elseif xml.attribute then |
|
121 value_where = "in_attribute" |
|
122 elseif xml.text then |
|
123 value_where = "in_text" |
|
124 elseif xml.x_name_is_value then |
|
125 value_where = "in_tag_name" |
|
126 elseif xml.x_single_attribute then |
|
127 single_attribute = xml.x_single_attribute |
|
128 value_where = "in_single_attribute" |
|
129 end |
|
130 end |
|
131 if propschema["const"] then |
|
132 enums = { propschema["const"] } |
|
133 elseif propschema["enum"] then |
|
134 enums = propschema["enum"] |
|
135 end |
|
136 end |
|
137 |
|
138 return proptype, value_where, name, namespace, prefix, single_attribute, enums |
|
139 end |
|
140 |
|
141 local parse_object : function (schema : schema_t, s : st.stanza_t, root : json_schema_object) : { string : any } |
|
142 local parse_array : function (schema : schema_t, s : st.stanza_t, root : json_schema_object) : { any } |
|
143 |
|
144 local function extract_value (s : st.stanza_t, value_where : value_goes, proptype : json.json_type_name, name : string, namespace : string, prefix : string, single_attribute : string, enums : { any }) : string |
|
145 if value_where == "in_tag_name" then |
|
146 local c : st.stanza_t |
|
147 if proptype == "boolean" then |
|
148 c = s:get_child(name, namespace); |
|
149 elseif enums and proptype == "string" then |
|
150 -- XXX O(n²) ? |
|
151 -- Probably better to flip the table and loop over :childtags(nil, ns), should be 2xO(n) |
|
152 -- BUT works first, optimize later |
|
153 for i = 1, #enums do |
|
154 c = s:get_child(enums[i] as string, namespace); |
|
155 if c then break end |
|
156 end |
|
157 else |
|
158 c = s:get_child(nil, namespace); |
|
159 end |
|
160 if c then |
|
161 return c.name; |
|
162 end |
|
163 elseif value_where == "in_attribute" then |
|
164 local attr = name |
|
165 if prefix then |
|
166 attr = prefix .. ':' .. name |
|
167 elseif namespace and namespace ~= s.attr.xmlns then |
|
168 attr = namespace .. "\1" .. name |
|
169 end |
|
170 return s.attr[attr] |
|
171 |
|
172 elseif value_where == "in_text" then |
|
173 return s:get_text() |
|
174 |
|
175 elseif value_where == "in_single_attribute" then |
|
176 local c = s:get_child(name, namespace) |
|
177 return c and c.attr[single_attribute] |
|
178 elseif value_where == "in_text_tag" then |
|
179 return s:get_child_text(name, namespace) |
|
180 end |
|
181 end |
|
182 |
|
183 function parse_object (schema : schema_t, s : st.stanza_t, root : json_schema_object) : { string : any } |
|
184 local out : { string : any } = {} |
|
185 schema = resolve_schema(schema, root) |
|
186 if schema is json_schema_object and schema.properties then |
|
187 for prop, propschema in pairs(schema.properties) do |
|
188 propschema = resolve_schema(propschema, root) |
|
189 |
|
190 local proptype, value_where, name, namespace, prefix, single_attribute, enums = unpack_propschema(propschema, prop, s.attr.xmlns) |
|
191 |
|
192 if value_where == "in_children" and propschema is json_schema_object then |
|
193 if proptype == "object" then |
|
194 local c = s:get_child(name, namespace) |
|
195 if c then |
|
196 out[prop] = parse_object(propschema, c, root); |
|
197 end |
|
198 elseif proptype == "array" then |
|
199 local a = parse_array(propschema, s, root); |
|
200 if a and a[1] ~= nil then |
|
201 out[prop] = a; |
|
202 end |
|
203 else |
|
204 error "unreachable" |
|
205 end |
|
206 elseif value_where == "in_wrapper" and propschema is json_schema_object and proptype == "array" then |
|
207 local wrapper = s:get_child(name, namespace); |
|
208 if wrapper then |
|
209 out[prop] = parse_array(propschema, wrapper, root); |
|
210 end |
|
211 else |
|
212 local value : string = extract_value (s, value_where, proptype, name, namespace, prefix, single_attribute, enums) |
|
213 |
|
214 out[prop] = totype(proptype, value) |
|
215 end |
|
216 end |
|
217 end |
|
218 |
|
219 return out |
|
220 end |
|
221 |
|
222 function parse_array (schema : json_schema_object, s : st.stanza_t, root : json_schema_object) : { any } |
|
223 local itemschema : schema_t = resolve_schema(schema.items, root); |
|
224 local proptype, value_where, child_name, namespace, prefix, single_attribute, enums = unpack_propschema(itemschema, nil, s.attr.xmlns) |
|
225 local attr_name : string |
|
226 if value_where == "in_single_attribute" then -- FIXME this shouldn't be needed |
|
227 value_where = "in_attribute"; |
|
228 attr_name = single_attribute; |
|
229 end |
|
230 local out : { any } = {} |
|
231 |
|
232 if proptype == "object" then |
|
233 if itemschema is json_schema_object then |
|
234 for c in s:childtags(child_name, namespace) do |
|
235 table.insert(out, parse_object(itemschema, c, root)); |
|
236 end |
|
237 else |
|
238 error "array items must be schema object" |
|
239 end |
|
240 elseif proptype == "array" then |
|
241 if itemschema is json_schema_object then |
|
242 for c in s:childtags(child_name, namespace) do |
|
243 table.insert(out, parse_array(itemschema, c, root)); |
|
244 end |
|
245 end |
|
246 else |
|
247 for c in s:childtags(child_name, namespace) do |
|
248 local value : string = extract_value (c, value_where, proptype, attr_name or child_name, namespace, prefix, single_attribute, enums) |
|
249 |
|
250 table.insert(out, totype(proptype, value)); |
|
251 end |
|
252 end |
|
253 return out; |
|
254 end |
|
255 |
|
256 local function parse (schema : json_schema_object, s : st.stanza_t) : table |
|
257 local s_type = guess_schema_type(schema) |
|
258 if s_type == "object" then |
|
259 return parse_object(schema, s, schema) |
|
260 elseif s_type == "array" then |
|
261 return parse_array(schema, s, schema) |
|
262 else |
|
263 error "top-level scalars unsupported" |
|
264 end |
|
265 end |
|
266 |
|
267 local function toxmlstring(proptype : json_type_name, v : any) : string |
|
268 if proptype == "string" and v is string then |
|
269 return v |
|
270 elseif proptype == "number" and v is number then |
|
271 return string.format("%g", v) |
|
272 elseif proptype == "integer" and v is number then -- TODO is integer |
|
273 return string.format("%d", v) |
|
274 elseif proptype == "boolean" then |
|
275 return v and "1" or "0" |
|
276 end |
|
277 end |
|
278 |
|
279 local unparse : function (json_schema_object, table, string, string, st.stanza_t, json_schema_object) : st.stanza_t |
|
280 |
|
281 local function unparse_property(out : st.stanza_t, v : any, proptype : json_type_name, propschema : schema_t, value_where : value_goes, name : string, namespace : string, current_ns : string, prefix : string, single_attribute : string, root : json_schema_object) |
|
282 |
|
283 if value_where == "in_attribute" then |
|
284 local attr = name |
|
285 if prefix then |
|
286 attr = prefix .. ':' .. name |
|
287 elseif namespace and namespace ~= current_ns then |
|
288 attr = namespace .. "\1" .. name |
|
289 end |
|
290 |
|
291 out.attr[attr] = toxmlstring(proptype, v) |
|
292 elseif value_where == "in_text" then |
|
293 out:text(toxmlstring(proptype, v)) |
|
294 elseif value_where == "in_single_attribute" then |
|
295 assert(single_attribute) |
|
296 local propattr : { string : string } = {} |
|
297 |
|
298 if namespace and namespace ~= current_ns then |
|
299 propattr.xmlns = namespace |
|
300 end |
|
301 |
|
302 propattr[single_attribute] = toxmlstring(proptype, v) |
|
303 out:tag(name, propattr):up(); |
|
304 |
|
305 else |
|
306 local propattr : { string : string } |
|
307 if namespace ~= current_ns then |
|
308 propattr = { xmlns = namespace } |
|
309 end |
|
310 if value_where == "in_tag_name" then |
|
311 if proptype == "string" and v is string then |
|
312 out:tag(v, propattr):up(); |
|
313 elseif proptype == "boolean" and v == true then |
|
314 out:tag(name, propattr):up(); |
|
315 end |
|
316 elseif proptype == "object" and propschema is json_schema_object and v is table then |
|
317 local c = unparse(propschema, v, name, namespace, nil, root); |
|
318 if c then |
|
319 out:add_direct_child(c); |
|
320 end |
|
321 elseif proptype == "array" and propschema is json_schema_object and v is table then |
|
322 if value_where == "in_wrapper" then |
|
323 local c = unparse(propschema, v, name, namespace, nil, root); |
|
324 if c then |
|
325 out:add_direct_child(c); |
|
326 end |
|
327 else |
|
328 unparse(propschema, v, name, namespace, out, root); |
|
329 end |
|
330 else |
|
331 out:text_tag(name, toxmlstring(proptype, v), propattr) |
|
332 end |
|
333 end |
|
334 end |
|
335 |
|
336 function unparse ( schema : json_schema_object, t : table, current_name : string, current_ns : string, ctx : st.stanza_t, root : json_schema_object ) : st.stanza_t |
|
337 |
|
338 if root == nil then root = schema end |
|
339 |
|
340 if schema.xml then |
|
341 if schema.xml.name then |
|
342 current_name = schema.xml.name |
|
343 end |
|
344 if schema.xml.namespace then |
|
345 current_ns = schema.xml.namespace |
|
346 end |
|
347 -- TODO prefix? |
|
348 end |
|
349 |
|
350 local out = ctx or st.stanza(current_name, { xmlns = current_ns }) |
|
351 |
|
352 local s_type = guess_schema_type(schema) |
|
353 if s_type == "object" then |
|
354 |
|
355 for prop, propschema in pairs(schema.properties) do |
|
356 propschema = resolve_schema(propschema, root) |
|
357 local v = t[prop] |
|
358 |
|
359 if v ~= nil then |
|
360 local proptype, value_where, name, namespace, prefix, single_attribute = unpack_propschema(propschema, prop, current_ns) |
|
361 unparse_property(out, v, proptype, propschema, value_where, name, namespace, current_ns, prefix, single_attribute, root) |
|
362 end |
|
363 end |
|
364 return out; |
|
365 |
|
366 elseif s_type == "array" then |
|
367 local itemschema = resolve_schema(schema.items, root) |
|
368 local proptype, value_where, name, namespace, prefix, single_attribute = unpack_propschema(itemschema, current_name, current_ns) |
|
369 for _, item in ipairs(t as { string }) do |
|
370 unparse_property(out, item, proptype, itemschema, value_where, name, namespace, current_ns, prefix, single_attribute, root) |
|
371 end |
|
372 return out; |
|
373 end |
|
374 end |
|
375 |
|
376 return { |
|
377 parse = parse, |
|
378 unparse = unparse, |
|
379 } |