How to convert partially invalid JSON to a valid one?

Question

I am using php to scrape a webpage and get this string:

'[{endTime:"2019-06-05T17:15:00.000+10:00",startTime:"2019-06-05T17:00:00.000+10:00"}]'

which is not valid json, the key names are encapsulated ...

I use preg_replace to create valid json:

$x = '[{endTime:"2019-06-05T17:15:00.000+10:00",startTime:"2019-06-05T17:00:00.000+10:00"}]'
$j = preg_replace('/(\w+)\s{0,1}:/', '"\1":', $x);

and get this value:

'[{"endTime":"2019-06-"05T17":"15":00.000+"10":00","startTime":"2019-06-"05T17":"00":00.000+"10":00"}]'

but I want this value:

'[{"endTime":"2019-06-05T17:15:00.000+10:00","startTime":"2019-06-05T17:00:00.000+10:00"}]'

How do I solve this problem?

Possible duplicate of [Invalid JSON parsing using PHP](https://stackoverflow.com/questions/1575198/invalid-json-parsing-using-php) — Konstantin Svintsov, Jun 03 '19 at 04:33
Thanks @KonstantinSvintsov - no that does not solve the question, because I don't have a problem with single/double quotes - it actually points out a problem with that answer — David Bray, Jun 03 '19 at 04:49
Find a parser that can parse *Javascript literals*. Don’t try to “fix” arbitrary strings with regexen. — deceze, Jun 03 '19 at 05:09

Emma · Accepted Answer · 2019-06-03T04:56:56.343

RegEx 1

Your original expression seems to be find, we would just slightly modify that to:

([{,])(\w+)(\s+)?:

and it might work, we are adding a left boundary:

([{,])

and a right boundary:

and our key attribute is in this capturing group:

(\w+)

RegEx 2

We can expand our first expression to:

([{,])(\s+)?(\w+)(\s+)?:

in case, we might be having spaces before the key attribute:

Demo

Test 1

$re = '/([{,])(\w+)(\s+)?:/m';
$x = '[{endTime:"2019-06-05T17:15:00.000+10:00",startTime:"2019-06-05T17:00:00.000+10:00"}]';
$subst = '$1"$2":';

$result = preg_replace($re, $subst, $x);

echo $result;

Test 2

$re = '/([{,])(\s+)?(\w+)(\s+)?:/m';
$x = '[{endTime:"2019-06-05T17:15:00.000+10:00",startTime:"2019-06-05T17:00:00.000+10:00"}]';
$subst = '$1"$3":';

$result = preg_replace($re, $subst, $x);

echo $result;

Output

[{"endTime":"2019-06-05T17:15:00.000+10:00","startTime":"2019-06-05T17:00:00.000+10:00"}]

Demo

RegEx Circuit

jex.im visualizes regular expressions:

my output to the test is :["{":"2019-06-05T17:15:00.000+10:00"",":"2019-06-05T17:00:00.000+10:00"}] — David Bray, Jun 03 '19 at 04:55

score 0 · Answer 2 · answered Jun 03 '19 at 04:23

0

use this pattern :

([{,])([^:]+):

it will find all texts which are following by { or ,

and use this for replacement:

$1"$2":

It will add a doublequote on both sides of your word.

answered Jun 03 '19 at 04:23

Hamed Ghasempour

435
3
12

this is close, but it's just too greedy, it works on that example, but if the value contains a comma, it's to greedy on the comma - eg value "21 Way street, suburb" – David Bray Jun 03 '19 at 04:51
Your example seems that it is always a dateTime, maybe you should ask your question more clear than this :-) – Hamed Ghasempour Jun 03 '19 at 06:02
I agree - but it was where I was stumbling - thanks for your input – David Bray Jun 03 '19 at 06:12

How to convert partially invalid JSON to a valid one?

2 Answers2

RegEx 1

RegEx 2

Demo

Test 1

Test 2

Output

Demo

RegEx Circuit