Benchmark (November 2018)

Drupal

• Version: 8.6.x @ 5d17da1
• Files: 8401

parser	time (ms)	memory (kb)
bmewburn/php7parser	5789.22	1189471.94
microsoft/tolerant-php-parser	16455.64	1174311.48
nikic/php-parser	25693.36	1784685.80
mattacosta/php-parser (base)	8122.56	268384.77
mattacosta/php-parser (syntax)	--	471899.52

Magento2

• Version: 2.3-develop @ b252ca0
• Files: 21688

parser	time (ms)	memory (kb)
bmewburn/php7parser	12972.62	2440786.40
microsoft/tolerant-php-parser	32648.24	2384537.00
nikic/php-parser	43200.28	3423339.44
mattacosta/php-parser (base)	19071.94	498955.02
mattacosta/php-parser (syntax)	--	882281.16

Methodology

Parse time is determined by reading a PHP file from disk, starting a timer, parsing the file (and saving it to a list), and finally stopping the timer.

Memory usage is determined before and after all files are parsed. For JS implementations, garbage collection is also performed before each measurement.

Notes

Memory requirements

• By default, Node.js applications are limited to 1.5 GB of memory. While the application may not immediately encounter an exception, it will experience a significant slow-down as the garbage collector attempts to reclaim unused memory. The --max-old-space-size flag is used to bypass this restriction.
• All PHP applications immediately terminate once reaching the maximum amount memory a script may consume. This value is determined by the memory_limit ini setting.

mattacosta/php-parser

Parsing a file results in an internal base representation of the syntax tree. This tree is not directly accessible. Instead, syntax nodes are created on-demand and saved for future use. The syntax representation used here is generated by visiting all nodes and tokens, but not trivia (which is unnecessary in most project-wide scenarios).

Packages

parser	version
bmewburn/php7parser	1.0.8
mattacosta/php-parser	1.0.0-beta.4
microsoft/tolerant-php-parser	0.0.15
nikic/php-parser	4.1.0

Runtime environments

runtime	version
Node	8.11.4
PHP	7.3.0RC1