fix: improve resolving unquoted strings and missing colon

docs/index.html

@@ -20,9 +20,9 @@ <h1>jsonrepair</h1>
       <div class="column">
         <label for="input-text">Input (invalid JSON)</label>
         <textarea id="input-text" autocomplete="off" autocapitalize="off" spellcheck="false">{
-  'firstName' 'John'
+  'firstName': 'John'
 
-  lastName: “Smith”
+  lastName “Smith”
 
   fullName: John Smith,
 

src/jsonrepair.test.ts

@@ -107,6 +107,12 @@ describe('jsonRepair', () => {
       strictEqual(jsonrepair("{a:'foo',b:'bar'}"), '{"a":"foo","b":"bar"}')
     })
 
+    it('should replace special quotes with double quotes', () => {
+      strictEqual(jsonrepair('{“a”:“b”}'), '{"a":"b"}')
+      strictEqual(jsonrepair('{‘a’:‘b’}'), '{"a":"b"}')
+      strictEqual(jsonrepair('{`a´:`b´}'), '{"a":"b"}')
+    })
+
     it('should leave string content untouched', () => {
       strictEqual(jsonrepair('"{a:b}"'), '"{a:b}"')
     })
@@ -341,7 +347,13 @@ describe('jsonRepair', () => {
 
     it('should repair missing colon between object key and value', () => {
       strictEqual(jsonrepair('{"a" "b"}'), '{"a": "b"}')
+      strictEqual(jsonrepair('{"a" 2}'), '{"a": 2}')
       strictEqual(jsonrepair('{\n"a" "b"\n}'), '{\n"a": "b"\n}')
+      strictEqual(jsonrepair('{"a" \'b\'}'), '{"a": "b"}')
+      strictEqual(jsonrepair("{'a' 'b'}"), '{"a": "b"}')
+      strictEqual(jsonrepair('{“a” “b”}'), '{"a": "b"}')
+      strictEqual(jsonrepair("{a 'b'}"), '{"a": "b"}')
+      strictEqual(jsonrepair('{a “b”}'), '{"a": "b"}')
     })
 
     it('should repair missing a combination of comma, quotes and brackets', () => {

src/jsonrepair.ts

@@ -561,8 +561,6 @@ export function jsonrepair(text: string): string {
     }
 
     if (i > start) {
-      const symbol = text.slice(start, i)
-
       if (text.charCodeAt(i) === codeOpenParenthesis) {
         // repair a MongoDB function call like NumberLong("2")
         // repair a JSONP function call like callback({...});
@@ -582,6 +580,13 @@ export function jsonrepair(text: string): string {
         return true
       } else {
         // repair unquoted string
+
+        // first, go back to prevent getting trailing whitespaces in the string
+        while (isWhitespace(text.charCodeAt(i - 1)) && i > 0) {
+          i--
+        }
+
+        const symbol = text.slice(start, i)
         output += JSON.stringify(symbol)
 
         return true

src/stringUtils.ts

@@ -37,12 +37,12 @@ const codeHairSpace = 0x200a
 const codeNarrowNoBreakSpace = 0x202f
 const codeMediumMathematicalSpace = 0x205f
 const codeIdeographicSpace = 0x3000
-const codeDoubleQuoteLeft = 0x201c
-const codeDoubleQuoteRight = 0x201d
-const codeQuoteLeft = 0x2018
-const codeQuoteRight = 0x2019
-const codeGraveAccent = 0x0060
-const codeAcuteAccent = 0x00b4
+const codeDoubleQuoteLeft = 0x201c // “
+const codeDoubleQuoteRight = 0x201d // ”
+const codeQuoteLeft = 0x2018 // ‘
+const codeQuoteRight = 0x2019 // ’
+const codeGraveAccent = 0x0060 // `
+const codeAcuteAccent = 0x00b4 // ´
 
 export function isHex(code: number): boolean {
   return (
@@ -65,16 +65,17 @@ export function isValidStringCharacter(code: number): boolean {
 }
 
 export function isDelimiter(char: string): boolean {
-  return regexDelimiter.test(char)
+  return regexDelimiter.test(char) || (char && isQuote(char.charCodeAt(0)))
 }
 
-const regexDelimiter = /^[,:[\]{}()\n"]$/
+const regexDelimiter = /^[,:[\]{}()\n]$/
 
 export function isStartOfValue(char: string): boolean {
-  return regexStartOfValue.test(char)
+  return regexStartOfValue.test(char) || (char && isQuote(char.charCodeAt(0)))
 }
 
-const regexStartOfValue = /^[[{\w"-_]$/
+// alpha, number, minus, or opening bracket or brace
+const regexStartOfValue = /^[[{\w-]$/
 
 export function isControlCharacter(code: number) {
   return (
@@ -146,11 +147,11 @@ export function isSingleQuote(code: number): boolean {
 export function stripLastOccurrence(
   text: string,
   textToStrip: string,
-  stripWhitespace = false
+  stripRemainingText = false
 ): string {
   const index = text.lastIndexOf(textToStrip)
   return index !== -1
-    ? text.substring(0, index) + (stripWhitespace ? '' : text.substring(index + 1))
+    ? text.substring(0, index) + (stripRemainingText ? '' : text.substring(index + 1))
     : text
 }