json_decode_nice + keep linebreaks:
function json_decode_nice($json, $assoc = TRUE){
$json = str_replace(array("\n","\r"),"\\n",$json);
$json = preg_replace('/([{,]+)(\s*)([^"]+?)\s*:/','$1"$3":',$json);
$json = preg_replace('/(,)\s*}$/','}',$json);
return json_decode($json,$assoc);
}
by phpdoc at badassawesome dot com, I just changed line 2.
If you want to keep the linebreaks just escape the slash.
json_decode
(PHP 5 >= 5.2.0, PECL json >= 1.2.0)
json_decode — Decodes a JSON string
Description
$json
[, bool $assoc = false
[, int $depth = 512
[, int $options = 0
]]] )Takes a JSON encoded string and converts it into a PHP variable.
Parameters
-
json -
The
jsonstring being decoded.This function only works with UTF-8 encoded data.
-
assoc -
When
TRUE, returned objects will be converted into associative arrays. -
depth -
User specified recursion depth.
-
options -
Bitmask of JSON decode options. Currently only
JSON_BIGINT_AS_STRINGis supported (default is to cast large integers as floats)
Return Values
Returns the value encoded in json in appropriate
PHP type. Values true, false and
null (case-insensitive) are returned as TRUE, FALSE
and NULL respectively. NULL is returned if the
json cannot be decoded or if the encoded
data is deeper than the recursion limit.
Examples
Example #1 json_decode() examples
<?php
$json = '{"a":1,"b":2,"c":3,"d":4,"e":5}';
var_dump(json_decode($json));
var_dump(json_decode($json, true));
?>
The above example will output:
object(stdClass)#1 (5) {
["a"] => int(1)
["b"] => int(2)
["c"] => int(3)
["d"] => int(4)
["e"] => int(5)
}
array(5) {
["a"] => int(1)
["b"] => int(2)
["c"] => int(3)
["d"] => int(4)
["e"] => int(5)
}
Example #2 Accessing invalid object properties
Accessing elements within an object that contain characters not permitted under PHP's naming convention (e.g. the hyphen) can be accomplished by encapsulating the element name within braces and the apostrophe.
<?php
$json = '{"foo-bar": 12345}';
$obj = json_decode($json);
print $obj->{'foo-bar'}; // 12345
?>
Example #3 common mistakes using json_decode()
<?php
// the following strings are valid JavaScript but not valid JSON
// the name and value must be enclosed in double quotes
// single quotes are not valid
$bad_json = "{ 'bar': 'baz' }";
json_decode($bad_json); // null
// the name must be enclosed in double quotes
$bad_json = '{ bar: "baz" }';
json_decode($bad_json); // null
// trailing commas are not allowed
$bad_json = '{ bar: "baz", }';
json_decode($bad_json); // null
?>
Example #4 depth errors
<?php
// Encode the data.
$json = json_encode(
array(
1 => array(
'English' => array(
'One',
'January'
),
'French' => array(
'Une',
'Janvier'
)
)
)
);
// Define the errors.
$constants = get_defined_constants(true);
$json_errors = array();
foreach ($constants["json"] as $name => $value) {
if (!strncmp($name, "JSON_ERROR_", 11)) {
$json_errors[$value] = $name;
}
}
// Show the errors for different depths.
foreach (range(4, 3, -1) as $depth) {
var_dump(json_decode($json, true, $depth));
echo 'Last error: ', $json_errors[json_last_error()], PHP_EOL, PHP_EOL;
}
?>
The above example will output:
array(1) {
[1]=>
array(2) {
["English"]=>
array(2) {
[0]=>
string(3) "One"
[1]=>
string(7) "January"
}
["French"]=>
array(2) {
[0]=>
string(3) "Une"
[1]=>
string(7) "Janvier"
}
}
}
Last error: JSON_ERROR_NONE
NULL
Last error: JSON_ERROR_DEPTH
Example #5 json_decode() of large integers
<?php
$json = '{"number": 12345678901234567890}';
var_dump(json_decode($json));
var_dump(json_decode($json, false, 512, JSON_BIGINT_AS_STRING));
?>
The above example will output:
object(stdClass)#1 (1) {
["number"]=>
float(1.2345678901235E+19)
}
object(stdClass)#1 (1) {
["number"]=>
string(20) "12345678901234567890"
}
Notes
Note:
The JSON spec is not JavaScript, but a subset of JavaScript.
Note:
In the event of a failure to decode, json_last_error() can be used to determine the exact nature of the error.
Changelog
| Version | Description |
|---|---|
| 5.4.0 |
The options parameter was added.
|
| 5.3.0 | Added the optional depth. The default recursion depth was increased from 128 to 512 |
| 5.2.3 | The nesting limit was increased from 20 to 128 |
| 5.2.1 | Added support for JSON decoding of basic types. |
See Also
- json_encode() - Returns the JSON representation of a value
- json_last_error() - Returns the last error occurred
I added a 3rd regex to the json_decode_nice function by "colin.mollenhour.com" to handle a trailing comma in json definition.
<?php
// http://www.php.net/manual/en/function.json-decode.php#95782
function json_decode_nice($json, $assoc = FALSE){
$json = str_replace(array("\n","\r"),"",$json);
$json = preg_replace('/([{,]+)(\s*)([^"]+?)\s*:/','$1"$3":',$json);
$json = preg_replace('/(,)\s*}$/','}',$json);
return json_decode($json,$assoc);
}
?>
Example:
<?php
$dat_json = <<<EOF
{
"foo" : "bam",
"bar" : "baz",
}
EOF;
$dat_array = json_decode_nice( $dat_json );
var_dump ( $dat_json, $dat_array );
/* RESULTS:
string(35) "{
"foo" : "bam",
"bar" : "baz",
}"
array(2) {
["foo"]=>
string(3) "bam"
["bar"]=>
string(3) "baz"
}
*/
?>
it seems, that some of the people are not aware, that if you are using json_decode to decode a string it HAS to be a propper json string:
<?php
var_dump(json_encode('Hello'));
var_dump(json_decode('Hello')); // wrong
var_dump(json_decode("Hello")); // wrong
var_dump(json_decode('"Hello"')); // correct
var_dump(json_decode("'Hello'")); // wrong
result:
string(7) ""Hello""
NULL
NULL
string(5) "Hello"
NULL
Sometime, i need to allow comments in json file. So i wrote a small func to clean comments in a json string before decoding it:
<?php
/**
* Clean comments of json content and decode it with json_decode().
* Work like the original php json_decode() function with the same params
*
* @param string $json The json string being decoded
* @param bool $assoc When TRUE, returned objects will be converted into associative arrays.
* @param integer $depth User specified recursion depth. (>=5.3)
* @param integer $options Bitmask of JSON decode options. (>=5.4)
* @return string
*/
function json_clean_decode($json, $assoc = false, $depth = 512, $options = 0) {
// search and remove comments like /* */ and //
$json = preg_replace("#(/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+/)|([\s\t](//).*)#", '', $json);
if(version_compare(phpversion(), '5.4.0', '>=')) {
$json = json_decode($json, $assoc, $depth, $options);
}
elseif(version_compare(phpversion(), '5.3.0', '>=')) {
$json = json_decode($json, $assoc, $depth);
}
else {
$json = json_decode($json, $assoc);
}
return $json;
}
?>
Make sure you pass in utf8 content, or json_decode may error out and just return a null value. For a particular web service I was using, I had to do the following:
<?php
$contents = file_get_contents($url);
$contents = utf8_encode($contents);
$results = json_decode($contents);
?>
Hope this helps!
When decoding strings from the database, make sure the input was encoded with the correct charset when it was input to the database.
I was using a form to create records in the DB which had a content field that was valid JSON, but it included curly apostrophes. If the page with the form did not have
<meta http-equiv="Content-Type" content="text/html;charset=utf-8">
in the head, then the data was sent to the database with the wrong encoding. Then, when json_decode tried to convert the string to an object, it failed every time.
with two lines you can convert your string from JavaScript toSource() (see http://www.w3schools.com/jsref/jsref_toSource.asp) output format to JSON accepted format. this works with subobjects too!
note: toSource() is part of JavaScript 1.3 but only implemented in Mozilla based javascript engines (not Opera/IE/Safari/Chrome).
<?php
$str = '({strvar:"string", number:40, boolvar:true, subobject:{substrvar:"sub string", subsubobj:{deep:"deeply nested"}, strnum:"56"}, false_val:false, false_str:"false"})'; // example javascript object toSource() output
$str = substr($str, 1, strlen($str) - 2); // remove outer ( and )
$str = preg_replace("/([a-zA-Z0-9_]+?):/" , "\"$1\":", $str); // fix variable names
$output = json_decode($str, true);
var_dump($output);
?>
var_dump output:
array(6) {
["strvar"]=>
string(6) "string"
["number"]=>
int(40)
["boolvar"]=>
bool(true)
["subobject"]=>
array(3) {
["substrvar"]=>
string(10) "sub string"
["subsubobj"]=>
array(1) {
["deep"]=>
string(13) "deeply nested"
}
["strnum"]=>
string(2) "56"
}
["false_val"]=>
bool(false)
["false_str"]=>
string(5) "false"
}
hope this saves someone some time.
For those of you wanting json_decode to be a little more lenient (more like Javascript), here is a wrapper:
<?php
function json_decode_nice($json, $assoc = FALSE){
$json = str_replace(array("\n","\r"),"",$json);
$json = preg_replace('/([{,]+)(\s*)([^"]+?)\s*:/','$1"$3":',$json);
return json_decode($json,$assoc);
}
?>
Some examples of accepted syntax:
<?php
$json = '{a:{b:"c",d:["e","f",0]}}';
$json =
'{
a : {
b : "c",
"d.e.f": "g"
}
}';
?>
If your content needs to have newlines, do this:
<?php
$string = "This
Text
Has
Newlines";
$json = '{withnewlines:'.json_encode($string).'}';
?>
Note: This does not fix trailing commas or single quotes.
[EDIT BY danbrown AT php DOT net: Contains a bugfix provided by (sskaje AT gmail DOT com) on 05-DEC-2012 with the following note.]
Old regexp failed when json like
{aaa:[{a:1},{a:2}]}
First of all, since JSON is not native PHP format or even native JavaScript format, everyone who wants to use JSON wisely should carefuly read official documentation. There is a link to it here, in "Introduction" section. Many questions like "it doesn't recognize my strings" and those like previous one (about zip codes) will drop if you will be attentive!
And second. I've found that there is no good, real working example of how to validate string if it is a JSON or not.
There are two ways to make this: parse input string for yourself using regular expressions or anything else, use json_decode to do it for you.
Parsing for yourself is like writing your own compiler, too difficult.
Just testing result of json_decode is not enough because you should test it with NULL, but valid JSON could be like this 'null' and it will evaluate to NULL. So you should use another function - json_last_error. It will return error code of the last encode/decode operation. If no error occured it will be JSON_ERROR_NONE. So here is the function you should use for testing:
<?php
function isValidJson($strJson) {
json_decode($strJson);
return (json_last_error() === JSON_ERROR_NONE);
}
?>
It's so simple, that there is no need to use it and slow down your script with extra delay for function call. Just do it manualy in you code while working with input data:
<?php
//here is my initial string
$sJson = $_POST['json'];
//try to decode it
$json = json_decode($sJson);
if (json_last_error() === JSON_ERROR_NONE) {
//do something with $json. It's ready to use
} else {
//yep, it's not JSON. Log error or alert someone or do nothing
}
?>
Consider that JSON can differ between int and string. So
<?php
var_dump(json_decode('{"foo": 12}'));
// array(1) { ["foo"]=> int(12) }
var_dump(json_decode('{"foo": "12"}'));
// array(1) { ["foo"]=> string(12) }
?>
Numbers that cannot be handled by integer seems to become float casted. This can be a problem if you transfer big numbers like facebook ids over JSON. Either you avoid numbers by cast everything to string before JSON.stringify or you have to use number_format if the value bacome a float value.
<?php
// test
$x = json_decode('{"foo": 123456789012345}');
echo sprintf('%1$f', $x->foo).PHP_EOL;
echo sprintf('%1$u', $x->foo).PHP_EOL;
echo sprintf('%1$s', $x->foo).PHP_EOL;
echo strval($x->foo).PHP_EOL;
echo (string) $x->foo.PHP_EOL;
echo number_format($x->foo, 0, '', '').PHP_EOL;
// output
123456789012345.000000 // printf %f
2249056121 // printf %u
1.2345678901234E+14 // printf %s
1.2345678901234E+14 // strval()
1.2345678901234E+14 // cast (string)
2249056121 // cast (int)
123456789012345 // number_format()
?>
This function will remove trailing commas and encode in utf8, which might solve many people's problems. Someone might want to expand it to also change single quotes to double quotes, and fix other kinds of json breakage.
<?php
function mjson_decode($json)
{
return json_decode(removeTrailingCommas(utf8_encode($json)));
}
function removeTrailingCommas($json)
{
$json=preg_replace('/,\s*([\]}])/m', '$1', $json);
return $json;
}
?>
json_decode()'s handling of invalid JSON is very flaky, and it is very hard to reliably determine if the decoding succeeded or not. Observe the following examples, none of which contain valid JSON:
The following each returns NULL, as you might expect:
<?php
var_dump(json_decode('[')); // unmatched bracket
var_dump(json_decode('{')); // unmatched brace
var_dump(json_decode('{}}')); // unmatched brace
var_dump(json_decode('{error error}')); // invalid object key/value
notation
var_dump(json_decode('["\"]')); // unclosed string
var_dump(json_decode('[" \x "]')); // invalid escape code
Yet the following each returns the literal string you passed to it:
var_dump(json_decode(' [')); // unmatched bracket
var_dump(json_decode(' {')); // unmatched brace
var_dump(json_decode(' {}}')); // unmatched brace
var_dump(json_decode(' {error error}')); // invalid object key/value notation
var_dump(json_decode('"\"')); // unclosed string
var_dump(json_decode('" \x "')); // invalid escape code
?>
(this is on PHP 5.2.6)
Reported as a bug, but oddly enough, it was closed as not a bug.
[NOTE BY danbrown AT php DOT net: This was later re-evaluated and it was determined that an issue did in fact exist, and was patched by members of the Development Team. See http://bugs.php.net/bug.php?id=45989 for details.]
If var_dump produces NULL, you may be experiencing JSONP aka JSON with padding, here's a quick fix...
<?php
//remove padding
$body=preg_replace('/.+?({.+}).+/','$1',$body);
// now, process the JSON string
$result = json_decode($body);
var_dump($result);
?>
You can't transport Objects or serialize Classes, json_* replace it bei stdClass!
<?php
$dom = new DomDocument( '1.0', 'utf-8' );
$body = $dom->appendChild( $dom->createElement( "body" ) );
$body->appendChild( $dom->createElement( "forward", "Hallo" ) );
$JSON_STRING = json_encode(
array(
"aArray" => range( "a", "z" ),
"bArray" => range( 1, 50 ),
"cArray" => range( 1, 50, 5 ),
"String" => "Value",
"stdClass" => $dom,
"XML" => $dom->saveXML()
)
);
unset( $dom );
$Search = "XML";
$MyStdClass = json_decode( $JSON_STRING );
// var_dump( "<pre>" , $MyStdClass , "</pre>" );
try {
throw new Exception( "$Search isn't a Instance of 'stdClass' Class by json_decode()." );
if ( $MyStdClass->$Search instanceof $MyStdClass )
var_dump( "<pre>instanceof:" , $MyStdClass->$Search , "</pre>" );
} catch( Exception $ErrorHandle ) {
echo $ErrorHandle->getMessage();
if ( property_exists( $MyStdClass, $Search ) ) {
$dom = new DomDocument( "1.0", "utf-8" );
$dom->loadXML( $MyStdClass->$Search );
$body = $dom->getElementsByTagName( "body" )->item(0);
$body->appendChild( $dom->createElement( "rewind", "Nice" ) );
var_dump( htmlentities( $dom->saveXML(), ENT_QUOTES, 'utf-8' ) );
}
}
?>
In PHP <= 5.1.6 trying to decode an integer value that's > PHP_INT_MAX will result in an intger of PHP_INT_MAX.
In PHP 5.2+ decoding an integer > PHP_INT_MAX will cause a conversion to a float.
Neither behaviour is perfect, capping at PHP_INT_MAX is marginally worse, but the float conversion loses precision.
If you expect to deal with large numbers at all, let alone in JSON, ensure you're using a 64-bit system.
Be aware, when decoding JSON strings, where an empty string is a key, this library replaces the empty string with "_empty_".
So the following code gives an unexpected result:
<?php
var_dump(json_decode('{"":"arbitrary"}'));
?>
The result is as follows:
object(stdClass)#1 (1) {
["_empty_"]=>
string(6) "arbitrary"
}
Any subsequent key named "_empty_" (or "" [the empty string] again) will overwrite the value.
Noted in a comment below is that this function will return NULL when given a simple string.
This is new behavior - see the result in PHP 5.2.4 :
php > var_dump(json_decode('this is a simple string'));
string(23) "this is a simple string"
in PHP 5.3.2 :
php > var_dump(json_decode('this is a simple string'));
NULL
I had several functions that relied on checking the value of a purported JSON string if it didn't decode into an object/array. If you do too, be sure to be aware of this when upgrading to PHP 5.3.
I found that even though json_decode() takes UTF-8 strings, it won't parse if a Byte-Order-Mark (BOM) is present.
Also, if you allow your users to edit JSON files, you end up with a mix of ASCII, ISO-8859-1, UTF-8 with and without BOM.
This function prepares a string for the JSON parser by converting it to UTF-8 and removing the BOM if present.
<?php
function prepareJSON($input) {
//This will convert ASCII/ISO-8859-1 to UTF-8.
//Be careful with the third parameter (encoding detect list), because
//if set wrong, some input encodings will get garbled (including UTF-8!)
$imput = mb_convert_encoding($input, 'UTF-8', 'ASCII,UTF-8,ISO-8859-1');
//Remove UTF-8 BOM if present, json_decode() does not like it.
if(substr($input, 0, 3) == pack("CCC", 0xEF, 0xBB, 0xBF)) $input = substr($input, 3);
return $input;
}
//Usage:
$myFile = file_get_contents('somefile.json');
$myDataArr = json_decode(prepareJSON($myFile), true);
?>
When in php 5.2.3, I found that json_decode 4000000000 float value into integer value resulting int(-294967296) from var_dump function
When in php 5.3.x, php correctly returning float(4000000000)
thank you for people creating & upgrading php
Yohan W.
I thought I would advise you to be careful if using U.S. addresses, particularly five-digit zip codes, with json_encode and json_decode.
The manual states that json_decode($data) will return $data in the appropriate PHP type. But it may not be the type you stored!
In my case, I have some data that might be scalar (e.g. string containing "01234") or might be a json packet (containing multiple zipcodes), and I was calling json_decode on everything.
<?php
$zip = "01234";
$test_zip = json_decode($zip);
echo $zip === $test_zip; // FALSE
?>
Because $test_zip is an integer containing 1234.
I pulled my hair off for hours trying to get rid of strange backslashes in my incoming JSON-data in POST-pool, making it impossible to decode the incoming JSON-data.
For those of you facing the same problem:
just make sure you disable 'magic_quotes_gpc' in your php.ini and the incoming data will not be pseudo-escaped anymore. Now your incoming JSON-data should just be decoded fine.
Maybe this will help.
There is a problem when passing to json_decode a string with the "\" symbol. It seems to identify it as an escape character and trying to follow it. So sometimes it leads to failed parsing.
It seems that just replacing it with "\\" helps.
<?php
print_r(json_decode(str_replace('\\', '\\\\', '{"name":"/\"}')));
?>
where /\ is the string which doesn't worked.
It was quite hard to figure out the allowed Javascript formats. Some extra remarks:
json_decode() doesn't seem to allow single quotes:
<?php
print_r(json_decode('[0,{"a":"a","b":"b"},2,3]')); //works
print_r(json_decode("[0,{'a':'a','b':'b'},2,3]")); //doesn't work
?>
json_decode() doesn't allow an extra comma in a list of entries:
<?php
print_r(json_decode('[0,1 ]')); //works
print_r(json_decode('[0,1,]')); //doesn't work
?>
(I like to write a comma behind every entry when the entries are spread over several lines.)
json_decode() does allow linefeeds in the data!
?>
There seems to be a difference in the way json_decode works between 5.2.9 and 5.2.6. Trying to run json_decode on a URL in 5.2.6 will return the URL, but in 5.2.9 it will return NULL.
5.2.6
var_dump(json_decode("http://www.php.net")); // Displays string(18) "http://www.php.net"
5.2.9
var_dump(json_decode("http://www.php.net")); // Displays NULL
The servers I tested it on both had json version 1.2.1.
/*since json_decode() decodes json, it MUST check if the argument passed in is json, so logically one can assume that json_decode() also functions as a json checker*/
//thus....
if(json_decode($input_json) == NULL)
{
echo("not valid json!");
}else
{
$my_json_object = json_decode($input_json);
}
Re requiring to escape the forward slash:
I think PHP 5.2.1 had that problem, as I remember it occurring here when I posted that comment; but now I'm on 5.2.5 it doesn't, so it has obviously been fixed. The JSON one gets from all the browsers escape the forward slashes anyway.
Please note that this function does NOT convert back to PHP values all strings resulting from a call to json-encode.
Since the json spec says that "A JSON text is a serialized object or array", this function will return NULL when decoding any json string that does not represent either an object or an array.
To successfully encode + decode single php values such as strings, booleans, integers or floats, you will have to wrap them in an array before converting them.
I had a JSON string with multiple commas in it without any quotes, which was causing the json_decode to return false.
Once the double quotes were added, everything worked out fine.
A quick:
while(strpos($contents, ',,') !== false) {
$contents = str_replace(',,', ',"",', $contents);
}
solved the problem, and the function worked correctly. If you have tried everything and are still a FALSE return, I suggest trying the JSON validator in another post.
