diff options
Diffstat (limited to 'teal-src/prosody/util/jsonschema.tl')
| -rw-r--r-- | teal-src/prosody/util/jsonschema.tl | 505 |
1 files changed, 505 insertions, 0 deletions
diff --git a/teal-src/prosody/util/jsonschema.tl b/teal-src/prosody/util/jsonschema.tl new file mode 100644 index 00000000..1cc2314f --- /dev/null +++ b/teal-src/prosody/util/jsonschema.tl @@ -0,0 +1,505 @@ +-- Copyright (C) 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 +-- + +if not math.type then require "prosody.util.mathcompat" end + + +local utf8_enc = rawget(_G, "utf8") or require"prosody.util.encodings".utf8; +local utf8_len = utf8_enc.len or function(s : string) : integer + local _, count = s:gsub("[%z\001-\127\194-\253][\128-\191]*", ""); + return count; +end; + +local json = require "prosody.util.json" +local null = json.null; + +local pointer = require "prosody.util.jsonpointer" + +local type json_type_name = json.json_type_name + +-- json_type_name here is non-standard +local type schema_t = boolean | json_schema_object + +local record json_schema_object + type json_type_name = json.json_type_name + type schema_object = json_schema_object + + -- json-schema-core meta stuff + ["$schema"] : string + ["$vocabulary"] : { string : boolean } + ["$id"] : string + ["$comment"] : string + ["$defs"] : { string : schema_t } + ["$anchor"] : string -- NYI + ["$dynamicAnchor"] : string -- NYI + ["$ref"] : string + ["$dynamicRef"] : string -- NYI + + -- combinations + allOf : { schema_t } + anyOf : { schema_t } + oneOf : { schema_t } + + -- conditional logic + ["not"] : schema_t + ["if"] : schema_t + ["then"] : schema_t + ["else"] : schema_t + + dependentRequired : { string : { string } } + + -- arrays + prefixItems : { schema_t } + items : schema_t + contains : schema_t + + -- objects + properties : { string : schema_t } + patternProperties: { string : schema_t } -- NYI + additionalProperties: schema_t + propertyNames : schema_t + + -- unevaluated + unevaluatedItems : schema_t -- NYI + unevaluatedProperties : schema_t -- NYI + + -- json-schema-validation + type : json_type_name | { json_type_name } + enum : { any } + const : any + + -- numbers + multipleOf : number + maximum : number + exclusiveMaximum : number + minimum : number + exclusiveMinimum : number + + -- strings + maxLength : integer + minLength : integer + pattern : string -- NYI + + -- arrays + maxItems : integer + minItems : integer + uniqueItems : boolean + maxContains : integer + minContains : integer + + -- objects + maxProperties : integer -- NYI + minProperties : integer -- NYI + required : { string } + dependentSchemas : { string : schema_t } + + -- semantic format + format : string + + -- for Lua + luaPatternProperties: { string : schema_t } + luaPattern : string + + -- xml + record xml_t + name : string + namespace : string + prefix : string + attribute : boolean + wrapped : boolean + + -- nonstantard, maybe in the future + text : boolean + x_name_is_value : boolean + x_single_attribute : string + end + + xml : xml_t + + -- descriptive + title : string + description : string + deprecated : boolean + readOnly : boolean + writeOnly : boolean + + -- methods + validate : function (schema : schema_t, data : any, root : json_schema_object, sloc : string, iloc : string, errs:errors) : boolean, errors +end + +-- TODO validator function per schema property + +local function simple_validate(schema : json_type_name | { json_type_name }, data : any) : boolean + if schema == nil then + return true + elseif schema == "object" and data is table then + return type(data) == "table" and (next(data)==nil or type((next(data, nil))) == "string") + elseif schema == "array" and data is table then + return type(data) == "table" and (next(data)==nil or type((next(data, nil))) == "number") + elseif schema == "integer" then + return math.type(data) == schema + elseif schema == "null" then + return data == null + elseif schema is { json_type_name } then + for _, one in ipairs(schema as { json_type_name }) do + if simple_validate(one, data) then + return true + end + end + return false + else + return type(data) == schema + end +end + +local record validation_error + instanceLocation : string + schemaLocation : string + error : string +end +local type errors = { validation_error } +local function mkerr(sloc:string,iloc:string,err:string) : validation_error + return { schemaLocation = sloc; instanceLocation = iloc; error = err }; +end + +local function validate (schema : schema_t, data : any, root : json_schema_object, sloc : string, iloc : string, errs:errors) : boolean, errors + if schema is boolean then + return schema + end + + if root == nil then + root = schema as json_schema_object + iloc = "" + sloc = "" + errs = {}; + end + + if schema["$ref"] and schema["$ref"]:sub(1,1) == "#" then + local referenced = pointer.resolve(root as table, schema["$ref"]:sub(2)) as schema_t + if referenced ~= nil and referenced ~= root and referenced ~= schema then + if not validate(referenced, data, root, schema["$ref"], iloc, errs) then + table.insert(errs, mkerr(sloc.."/$ref", iloc, "Subschema failed validation")) + return false, errs; + end + end + end + + if not simple_validate(schema.type, data) then + table.insert(errs, mkerr(sloc.."/type", iloc, "unexpected type")); + return false, errs; + end + + if schema.type == "object" then + if data is table then + -- just check that there the keys are all strings + for k in pairs(data) do + if not k is string then + table.insert(errs, mkerr(sloc.."/type", iloc, "'object' had non-string keys")); + return false, errs; + end + end + end + end + + if schema.type == "array" then + if data is table then + -- just check that there the keys are all numbers + for i in pairs(data) do + if not i is integer then + table.insert(errs, mkerr(sloc.."/type", iloc, "'array' had non-integer keys")); + return false, errs; + end + end + end + end + + if schema["enum"] ~= nil then + local match = false + for _, v in ipairs(schema["enum"]) do + if v == data then + -- FIXME supposed to do deep-compare + match = true + break + end + end + if not match then + table.insert(errs, mkerr(sloc.."/enum", iloc, "not one of the enumerated values")); + return false, errs; + end + end + + -- XXX this is measured in byte, while JSON measures in ... bork + -- TODO use utf8.len? + if data is string then + if schema.maxLength and utf8_len(data) > schema.maxLength then + table.insert(errs, mkerr(sloc.."/maxLength", iloc, "string too long")) + return false, errs; + end + if schema.minLength and utf8_len(data) < schema.minLength then + table.insert(errs, mkerr(sloc.."/maxLength", iloc, "string too short")) + return false, errs; + end + if schema.luaPattern and not data:match(schema.luaPattern) then + table.insert(errs, mkerr(sloc.."/luaPattern", iloc, "string does not match pattern")) + return false, errs; + end + end + + if data is number then + if schema.multipleOf and (data == 0 or data % schema.multipleOf ~= 0) then + table.insert(errs, mkerr(sloc.."/luaPattern", iloc, "not a multiple")) + return false, errs; + end + + if schema.maximum and not ( data <= schema.maximum ) then + table.insert(errs, mkerr(sloc.."/maximum", iloc, "number exceeds maximum")) + return false, errs; + end + + if schema.exclusiveMaximum and not ( data < schema.exclusiveMaximum ) then + table.insert(errs, mkerr(sloc.."/exclusiveMaximum", iloc, "number exceeds exclusive maximum")) + return false, errs; + end + + if schema.minimum and not ( data >= schema.minimum ) then + table.insert(errs, mkerr(sloc.."/minimum", iloc, "number below minimum")) + return false, errs; + end + + if schema.exclusiveMinimum and not ( data > schema.exclusiveMinimum ) then + table.insert(errs, mkerr(sloc.."/exclusiveMinimum", iloc, "number below exclusive minimum")) + return false, errs; + end + end + + if schema.allOf then + for i, sub in ipairs(schema.allOf) do + if not validate(sub, data, root, sloc.."/allOf/"..i, iloc, errs) then + table.insert(errs, mkerr(sloc.."/allOf", iloc, "did not match all subschemas")) + return false, errs; + end + end + end + + if schema.oneOf then + local valid = 0 + for i, sub in ipairs(schema.oneOf) do + if validate(sub, data, root, sloc.."/oneOf"..i, iloc, errs) then + valid = valid + 1 + end + end + if valid ~= 1 then + table.insert(errs, mkerr(sloc.."/oneOf", iloc, "did not match exactly one subschema")) + return false, errs; + end + end + + if schema.anyOf then + local match = false + for i, sub in ipairs(schema.anyOf) do + if validate(sub, data, root, sloc.."/anyOf/"..i, iloc, errs) then + match = true + break + end + end + if not match then + table.insert(errs, mkerr(sloc.."/anyOf", iloc, "did not match any subschema")) + return false, errs; + end + end + + if schema["not"] then + if validate(schema["not"], data, root, sloc.."/not", iloc, errs) then + table.insert(errs, mkerr(sloc.."/not", iloc, "did match subschema")) + return false, errs; + end + end + + if schema["if"] ~= nil then + if validate(schema["if"], data, root, sloc.."/if", iloc, errs) then + if schema["then"] then + if not validate(schema["then"], data, root, sloc.."/then", iloc, errs) then + table.insert(errs, mkerr(sloc.."/then", iloc, "did not match subschema")) + return false, errs; + end + end + else + if schema["else"] then + if not validate(schema["else"], data, root, sloc.."/else", iloc, errs) then + table.insert(errs, mkerr(sloc.."/else", iloc, "did not match subschema")) + return false, errs; + end + end + end + end + + if schema.const ~= nil and schema.const ~= data then + table.insert(errs, mkerr(sloc.."/const", iloc, "did not match constant value")) + return false, errs; + end + + if data is table then + -- tables combine object and array behavior, thus we do both kinds of + -- validations in this block, which could be useful for validating Lua + -- tables + + if schema.maxItems and #(data as {any}) > schema.maxItems then + table.insert(errs, mkerr(sloc.."/maxItems", iloc, "too many items")) + return false, errs; + end + + if schema.minItems and #(data as {any}) < schema.minItems then + table.insert(errs, mkerr(sloc.."/minItems", iloc, "too few items")) + return false, errs; + end + + if schema.required then + for _, k in ipairs(schema.required) do + if data[k] == nil then + table.insert(errs, mkerr(sloc.."/required", iloc.."/"..tostring(k), "missing required property")) + return false, errs; + end + end + end + + if schema.dependentRequired then + for k, reqs in pairs(schema.dependentRequired) do + if data[k] ~= nil then + for _, req in ipairs(reqs) do + if data[req] == nil then + table.insert(errs, mkerr(sloc.."/dependentRequired", iloc, "missing dependent required property")) + return false, errs; + end + end + end + end + end + + if schema.propertyNames ~= nil then + -- could be used to validate non-string keys of Lua tables + for k in pairs(data) do + if not validate(schema.propertyNames, k, root, sloc.."/propertyNames", iloc.."/"..tostring(k), errs) then + table.insert(errs, mkerr(sloc.."/propertyNames", iloc.."/"..tostring(k), "a property name did not match subschema")) + return false, errs; + end + end + end + + -- additionalProperties applies to properties not validated by properties + -- or patternProperties, so we must keep track of properties validated by + -- the later + local seen_properties : { string : boolean } = {} + + if schema.properties then + for k, sub in pairs(schema.properties) do + if data[k] ~= nil and not validate(sub, data[k], root, sloc.."/"..tostring(k), iloc.."/"..tostring(k), errs) then + table.insert(errs, mkerr(sloc.."/"..tostring(k), iloc.."/"..tostring(k), "a property did not match subschema")) + return false, errs; + end + seen_properties[k] = true + end + end + + if schema.luaPatternProperties then + -- like patternProperties, but Lua patterns + for pattern, sub in pairs(schema.luaPatternProperties) do + for k in pairs(data) do + if k is string and k:match(pattern) then + if not validate(sub, data[k], root, sloc.."/luaPatternProperties", iloc, errs) then + table.insert(errs, mkerr(sloc.."/luaPatternProperties/"..pattern, iloc.."/"..tostring(k), "a property did not match subschema")) + return false, errs; + end + seen_properties[k] = true + end + end + end + end + + if schema.additionalProperties ~= nil then + for k, v in pairs(data) do + if not seen_properties[k as string] then + if not validate(schema.additionalProperties, v, root, sloc.."/additionalProperties", iloc.."/"..tostring(k), errs) then + table.insert(errs, mkerr(sloc.."/additionalProperties", iloc.."/"..tostring(k), "additional property did not match subschema")) + return false, errs; + end + end + end + end + + if schema.dependentSchemas then + for k, sub in pairs(schema.dependentSchemas) do + if data[k] ~= nil and not validate(sub, data, root, sloc.."/dependentSchemas/"..k, iloc, errs) then + table.insert(errs, mkerr(sloc.."/dependentSchemas", iloc.."/"..tostring(k), "did not match dependent subschema")) + return false, errs; + end + end + end + + if schema.uniqueItems then + -- only works for scalars, would need to deep-compare for objects/arrays/tables + local values : { any : boolean } = {} + for _, v in pairs(data) do + if values[v] then + table.insert(errs, mkerr(sloc.."/uniqueItems", iloc, "had duplicate items")) + return false, errs; + end + values[v] = true + end + end + + local p = 0 + if schema.prefixItems ~= nil then + for i, s in ipairs(schema.prefixItems) do + if data[i] == nil then + break + elseif validate(s, data[i], root, sloc.."/prefixItems/"..i, iloc.."/"..i, errs) then + p = i + else + table.insert(errs, mkerr(sloc.."/prefixItems/"..i, iloc.."/"..tostring(i), "did not match subschema")) + return false, errs; + end + end + end + + if schema.items ~= nil then + for i = p+1, #(data as {any}) do + if not validate(schema.items, data[i], root, sloc, iloc.."/"..i, errs) then + table.insert(errs, mkerr(sloc.."/prefixItems/"..i, iloc.."/"..i, "did not match subschema")) + return false, errs; + end + end + end + + if schema.contains ~= nil then + local found = 0 + for i = 1, #(data as {any}) do + if validate(schema.contains, data[i], root, sloc.."/contains", iloc.."/"..i, errs) then + found = found + 1 + else + table.insert(errs, mkerr(sloc.."/contains", iloc.."/"..i, "did not match subschema")) + end + end + if found < (schema.minContains or 1) then + table.insert(errs, mkerr(sloc.."/minContains", iloc, "too few matches")) + return false, errs; + elseif found > (schema.maxContains or math.huge) then + table.insert(errs, mkerr(sloc.."/maxContains", iloc, "too many matches")) + return false, errs; + end + end + end + + return true; +end + + +json_schema_object.validate = validate; + +return json_schema_object; |