Better errors when the token is a non-printable control char

I had a heck of a time trying to debug a bit of JSON that had an extra
byte-order mark char in it while testing the previous commit.

This makes the token print out with JSON.stringify, and includes the
char code in hex, so it's easier to tell why it's a problem.

Author: isaacs

index.js

@@ -1,17 +1,27 @@
 'use strict'
 
+const hexify = char => {
+  const h = char.charCodeAt(0).toString(16).toUpperCase()
+  return '0x' + (h.length % 2 ? '0' : '') + h
+}
+
 const parseError = (e, txt, context) => {
   if (!txt) {
     return {
       message: e.message + ' while parsing empty string',
       position: 0,
     }
   }
-  const badToken = e.message.match(/^Unexpected token.*position\s+(\d+)/i)
-  const errIdx = badToken ? +badToken[1]
+  const badToken = e.message.match(/^Unexpected token (.) .*position\s+(\d+)/i)
+  const errIdx = badToken ? +badToken[2]
     : e.message.match(/^Unexpected end of JSON.*/i) ? txt.length - 1
     : null
 
+  const msg = badToken ? e.message.replace(/^Unexpected token ./, `Unexpected token ${
+      JSON.stringify(badToken[1])
+    } (${hexify(badToken[1])})`)
+    : e.message
+
   if (errIdx !== null && errIdx !== undefined) {
     const start = errIdx <= context ? 0
       : errIdx - context
@@ -26,12 +36,12 @@ const parseError = (e, txt, context) => {
     const near = txt === slice ? '' : 'near '
 
     return {
-      message: e.message + ` while parsing ${near}'${slice}'`,
+      message: msg + ` while parsing ${near}${JSON.stringify(slice)}`,
       position: errIdx,
     }
   } else {
     return {
-      message: e.message + ` while parsing '${txt.slice(0, context * 2)}'`,
+      message: msg + ` while parsing '${txt.slice(0, context * 2)}'`,
       position: 0,
     }
   }

test/index.js

@@ -29,12 +29,31 @@ t.test('parses JSON if it is a Buffer, removing BOM bytes', t => {
   t.end()
 })
 
+t.test('better errors when faced with \\b and other malarky', t => {
+  const str = JSON.stringify({
+    foo: 1,
+    bar: {
+      baz: [1, 2, 3, 'four']
+    }
+  })
+  const data = Buffer.from(str)
+  const bombom = Buffer.concat([Buffer.from([0xEF, 0xBB, 0xBF, 0xEF, 0xBB, 0xBF]), data])
+  t.throws(() => parseJson(bombom), {
+    message: /\(0xFEFF\) in JSON at position 0/
+  }, 'only strips a single BOM, not multiple')
+  const bs = str + '\b\b\b\b\b\b\b\b\b\b\b\b'
+  t.throws(() => parseJson(bs), {
+    message: /^Unexpected token "\\b" \(0x08\) in JSON at position.*\\b"$/
+  })
+  t.end()
+})
+
 t.test('throws SyntaxError for unexpected token', t => {
   const data = 'foo'
   t.throws(
     () => parseJson(data),
     {
-      message: 'Unexpected token o in JSON at position 1 while parsing \'foo\'',
+      message: 'Unexpected token "o" (0x6F) in JSON at position 1 while parsing "foo"',
       code: 'EJSONPARSE',
       position: 1,
       name: 'JSONParseError',
@@ -49,7 +68,7 @@ t.test('throws SyntaxError for unexpected end of JSON', t => {
   t.throws(
     () => parseJson(data),
     {
-      message: 'Unexpected end of JSON input while parsing \'{"foo: bar}\'',
+      message: 'Unexpected end of JSON input while parsing "{\\\"foo: bar}"',
       code: 'EJSONPARSE',
       position: 10,
       name: 'JSONParseError',
@@ -79,7 +98,7 @@ t.test('SyntaxError with less context (limited start)', t => {
   t.throws(
     () => parseJson(data, null, 3),
     {
-      message: 'Unexpected end of JSON input while parsing near \'...3210\'',
+      message: 'Unexpected end of JSON input while parsing near "...3210"',
       code: 'EJSONPARSE',
       position: 8,
       name: 'JSONParseError',
@@ -93,7 +112,7 @@ t.test('SyntaxError with less context (limited end)', t => {
   t.throws(
     () => parseJson(data, null, 2),
     {
-      message: 'Unexpected token a in JSON at position 0 while parsing near \'ab...\'',
+      message: 'Unexpected token "a" \(0x61\) in JSON at position 0 while parsing near "ab..."',
       code: 'EJSONPARSE',
       position: 0,
       name: 'JSONParseError',