well formed unicode string (#147) (#151)

* surrogate is detected and defered to JSON.stringify()

* remove console log from test

* for lone and unpair surrogate  mirror test used

Author: srmarjani

index.js

@@ -236,16 +236,22 @@ function $asString (str) {
 // magically escape strings for json
 // relying on their charCodeAt
 // everything below 32 needs JSON.stringify()
+// every string that contain surrogate needs JSON.stringify()
 // 34 and 92 happens all the time, so we
 // have a fast case for them
 function $asStringSmall (str) {
   var result = ''
   var last = 0
   var found = false
+  var surrogateFound = false
   var l = str.length
   var point = 255
   for (var i = 0; i < l && point >= 32; i++) {
     point = str.charCodeAt(i)
+    if (point >= 0xD800 && point <= 0xDFFF) {
+      // The current character is a surrogate.
+      surrogateFound = true
+    }
     if (point === 34 || point === 92) {
       result += str.slice(last, i) + '\\'
       last = i
@@ -258,7 +264,7 @@ function $asStringSmall (str) {
   } else {
     result += str.slice(last)
   }
-  return point < 32 ? JSON.stringify(str) : '"' + result + '"'
+  return ((point < 32) || (surrogateFound === true)) ? JSON.stringify(str) : '"' + result + '"'
 }
 
 function addPatternProperties (schema, externalSchema, fullSchema) {

test/surrogate.test.js

@@ -0,0 +1,67 @@
+'use strict'
+
+const test = require('tap').test
+const validator = require('is-my-json-valid')
+const build = require('..')
+
+test(`render a string with surrogate pairs as JSON:test 1`, (t) => {
+  t.plan(2)
+
+  const schema = {
+    title: 'surrogate',
+    type: 'string'
+  }
+
+  const validate = validator(schema)
+  const stringify = build(schema)
+  const output = stringify('𝌆')
+
+  t.equal(output, '"𝌆"')
+  t.ok(validate(JSON.parse(output)), 'valid schema')
+})
+
+test(`render a string with surrogate pairs as JSON: test 2`, (t) => {
+  t.plan(2)
+
+  const schema = {
+    title: 'long',
+    type: 'string'
+  }
+
+  const validate = validator(schema)
+  const stringify = build(schema)
+  const output = stringify('\uD834\uDF06')
+
+  t.equal(output, '"𝌆"')
+  t.ok(validate(JSON.parse(output)), 'valid schema')
+})
+
+test(`render a string with Unpaired surrogate code as JSON`, (t) => {
+  t.plan(2)
+
+  const schema = {
+    title: 'surrogate',
+    type: 'string'
+  }
+
+  const validate = validator(schema)
+  const stringify = build(schema)
+  const output = stringify('\uDF06\uD834')
+  t.equal(output, JSON.stringify('\uDF06\uD834'))
+  t.ok(validate(JSON.parse(output)), 'valid schema')
+})
+
+test(`render a string with lone surrogate code as JSON`, (t) => {
+  t.plan(2)
+
+  const schema = {
+    title: 'surrogate',
+    type: 'string'
+  }
+
+  const validate = validator(schema)
+  const stringify = build(schema)
+  const output = stringify('\uDEAD')
+  t.equal(output, JSON.stringify('\uDEAD'))
+  t.ok(validate(JSON.parse(output)), 'valid schema')
+})