Allow Parser pattern size to be parametrizable

[mbenedettini]

When parsing files with big strings allowing a big pattern size can help shave off precious milliseconds making parsing considerably faster since less loops of the parser are needed, leaving the big lift to the Javascript Regexp engine.

[mbenedettini]

this const name is now confusing (compared to DEFAULT_PATTERN_SIZE) but I couldn't find a better name

[uhop]

@mbenedettini do you have any benchmarks showing that the speed of parsing had improved?

I am asking because /.{1,256}/ comes with a penalty for regular expression engines. For example it is less effective
than /.?/, /.*/ or /.+/, especially when the upper number is high. TBH, I had worried that {,256} is too high and
should be reduced.

[KJ7LNW]

@mbenedettini do you have any benchmarks showing that the speed of parsing had improved?

I am asking because /.{1,256}/ comes with a penalty for regular expression engines. For example it is less effective than /.?/, /.*/ or /.+/, especially when the upper number is high. TBH, I had worried that {,256} is too high and should be reduced.

@mbenedettini, if possible always use string indexes (substr, etc) for parsing large volumes of data and then implemented so it scans the input only once instead of doing multiple passes. Also, never do single-character-scanning (ie for i ... str[i]) because that is slow, it should happen in delimited chunks that are as large as possible. also keep in mind that split/join are slow and memory intensive, so avoid these too.

Then your implementation can scan in hundreds of megabytes or even gigabytes per second.