Improve parsers efficiency

[phenpessoa]

TLDR

(Original TLDR at the end of this comment)

While I was working on removing regexes from parsers, I noticed that the speed gain and reduced usage was not as much as I expected it to be. So I decided to rework the parsers completely, instead of just removing regexes.

For more complex parsers, such as the character one, I'm gonna start by removing the regexes and later try to rewrite it completely. As this is a far more complex task and removing the regexes is already a huge performance boost for such parsers.

But for boostable bosses, for example, while removing regexes gave a 50% speed boost and reduced allocations by 4% rewriting the parser was relatively simple and gave a 98% speed boost while reducing allocations by almost 100%

Files to remove regex from or rewrite parsers:

TibiaBoostableBossesOverview

• Rewrite boostable bosses parser #233

TibiaCharactersCharacter

• TibiaCharactersCharacter: remove title regex #223
• TibiaCharactersCharacter: remove summon regex #224
• TibiaCharactersCharacter: remove death regex #225
• TibiaCharactersCharacter: remove char info regex #226
• TibiaCharactersCharacter: remove account badges regex #227
• TibiaCharactersCharacter: remove account achievements regex #228

TibiaCreaturesCreature

• CreatureDataRegex
• CreatureHitpointsRegex
• CreatureImmuneRegex
• CreatureStrongRegex
• CreatureWeakRegex
• CreatureHealedRegex
• CreatureManaRequiredRegex
• CreatureLootRegex

TibiaCreaturesOverview

• BoostedCreatureNameAndRaceRegex
• BoostedCreatureImageRegex
• CreatureInformationRegex

TibiaDataUtils

• TibiaCharactersCharacter: remove death regex #225
• removeUrlRegex

TibiaFansites

• FansiteInformationRegex
• FansiteImgTagRegex
• FansiteLanguagesRegex
• FansiteAnchorRegex

TibiaGuildsGuild

• GuildLogoRegex
• GuildWorldAndFoundationRegex
• GuildHomepageRegex
• GuildhallRegex
• GuildDisbaneRegex
• GuildMemberInformationRegex
• GuildMemberInvitesInformationRegex

TibiaHighscores

• HighscoresAgeRegex
• HighscoresPageRegex
• SevenColumnRegex
• SixColumnRegex

TibiaHousesHouse

• houseDataRegex
• housePassingRegex
• moveOutRegex
• paidUntilRegex
• houseAuctionedRegex

TibiaHousesOverview

• houseOverviewDataRegex
• houseOverviewAuctionedRegex

TibiaNews

• martelRegex

TibiaSpellsSpell

• SpellDataRowRegex
• SpellNameAndImageRegex
• SpellCooldownRegex
• SpellDescriptionRegex

TibiaWorldsOverview

• worldPlayerRecordRegex
• worldInformationRegex
• worldBattlEyeProtectedSinceRegex

TibiaWorldsWorld

• WorldDataRowRegex
• WorldRecordInformationRegex
• BattlEyeProtectedSinceRegex
• OnlinePlayerRegex

tibia

• characterNameRegex
• creatureAndSpellNameRegex
• guildNameRegex

Original TLDR

I decided to open this issue in order to track the progress of removing all regex from the system.
Reducing the amount of regex is beneficial for two main reasons:

Speed


When parsing without regex, it is quicker


Heap Allocations


Even though we already made the regexes sentinel value, not having to compile them on every iteration, using the compiled regex still make heap allocations. So, by removing regexes completely we reduce a lot of heap allocations, thus reducing the GC load.

[phenpessoa]

Benchmark TibiaCharactersCharacter

old.txt was run on commit f42e70fb178df7b48c065e2a41cfca06cc99a874
new.txt was run with all PRs for the TibiaCharactersCharacter merged

           │   old.txt   │               new.txt               │
           │   sec/op    │   sec/op     vs base                │
Number1-16   1.739m ± 1%   1.345m ± 1%  -22.62% (p=0.000 n=10)
Number2-16   41.03m ± 2%   10.72m ± 1%  -73.87% (p=0.000 n=10)
geomean      8.446m        3.798m       -55.03%

           │    old.txt    │               new.txt                │
           │     B/op      │     B/op      vs base                │
Number1-16    727.8Ki ± 0%   717.6Ki ± 0%   -1.41% (p=0.000 n=10)
Number2-16   14.163Mi ± 0%   7.061Mi ± 0%  -50.15% (p=0.000 n=10)
geomean       3.173Mi        2.224Mi       -29.89%

           │   old.txt   │               new.txt               │
           │  allocs/op  │  allocs/op   vs base                │
Number1-16   6.618k ± 0%   6.542k ± 0%   -1.15% (p=0.000 n=10)
Number2-16   97.20k ± 0%   24.43k ± 0%  -74.86% (p=0.000 n=10)
geomean      25.36k        12.64k       -50.15%

old.txt Content

goos: windows
goarch: amd64
pkg: github.com/TibiaData/tibiadata-api-go/src
cpu: Intel(R) Core(TM) i9-9900K CPU @ 3.60GHz
BenchmarkNumber1-16          682           1719844 ns/op          745314 B/op       6618 allocs/op

BenchmarkNumber1-16          682           1722013 ns/op          745414 B/op       6618 allocs/op

BenchmarkNumber1-16          686           1753708 ns/op          745315 B/op       6618 allocs/op

BenchmarkNumber1-16          693           1730242 ns/op          745796 B/op       6618 allocs/op

BenchmarkNumber1-16          696           1736189 ns/op          745641 B/op       6618 allocs/op

BenchmarkNumber1-16          693           1721026 ns/op          744553 B/op       6618 allocs/op

BenchmarkNumber1-16          698           1740915 ns/op          745294 B/op       6618 allocs/op

BenchmarkNumber1-16          675           1760454 ns/op          744884 B/op       6618 allocs/op

BenchmarkNumber1-16          703           1755549 ns/op          745195 B/op       6618 allocs/op

BenchmarkNumber1-16          687           1748659 ns/op          744842 B/op       6618 allocs/op
BenchmarkNumber2-16           27          41264707 ns/op        14864054 B/op      97196 allocs/op

BenchmarkNumber2-16           30          40812157 ns/op        14817490 B/op      97199 allocs/op

BenchmarkNumber2-16           28          41004925 ns/op        14858209 B/op      97196 allocs/op

BenchmarkNumber2-16           28          41051718 ns/op        14850961 B/op      97202 allocs/op

BenchmarkNumber2-16           28          40923389 ns/op        14857031 B/op      97198 allocs/op

BenchmarkNumber2-16           28          41677654 ns/op        14819033 B/op      97202 allocs/op

BenchmarkNumber2-16           28          40630121 ns/op        14851969 B/op      97206 allocs/op

BenchmarkNumber2-16           26          40849527 ns/op        14784642 B/op      97193 allocs/op

BenchmarkNumber2-16           28          41743629 ns/op        14869758 B/op      97205 allocs/op

BenchmarkNumber2-16           25          41680376 ns/op        14771126 B/op      97191 allocs/op

new.txt Content

goos: windows
goarch: amd64
pkg: github.com/TibiaData/tibiadata-api-go/src
cpu: Intel(R) Core(TM) i9-9900K CPU @ 3.60GHz
BenchmarkNumber1-16          903           1367013 ns/op          734800 B/op       6542 allocs/op

BenchmarkNumber1-16          873           1356718 ns/op          734813 B/op       6542 allocs/op

BenchmarkNumber1-16          879           1342934 ns/op          734812 B/op       6542 allocs/op

BenchmarkNumber1-16          896           1337580 ns/op          734802 B/op       6542 allocs/op

BenchmarkNumber1-16          891           1335610 ns/op          734805 B/op       6542 allocs/op

BenchmarkNumber1-16          891           1330908 ns/op          734805 B/op       6542 allocs/op

BenchmarkNumber1-16          903           1347810 ns/op          734800 B/op       6542 allocs/op

BenchmarkNumber1-16          886           1350972 ns/op          734807 B/op       6542 allocs/op

BenchmarkNumber1-16          876           1341077 ns/op          734813 B/op       6542 allocs/op

BenchmarkNumber1-16          882           1357755 ns/op          734810 B/op       6542 allocs/op
BenchmarkNumber2-16          100          10645805 ns/op         7403609 B/op      24434 allocs/op

BenchmarkNumber2-16          100          10783254 ns/op         7403623 B/op      24434 allocs/op

BenchmarkNumber2-16          100          10713614 ns/op         7403627 B/op      24434 allocs/op

BenchmarkNumber2-16          100          10743841 ns/op         7403607 B/op      24434 allocs/op

BenchmarkNumber2-16          100          10745159 ns/op         7403615 B/op      24434 allocs/op

BenchmarkNumber2-16          100          10730293 ns/op         7403605 B/op      24434 allocs/op

BenchmarkNumber2-16          100          10673829 ns/op         7403622 B/op      24434 allocs/op

BenchmarkNumber2-16          100          10848976 ns/op         7403606 B/op      24434 allocs/op

BenchmarkNumber2-16          100          10681710 ns/op         7403608 B/op      24434 allocs/op

BenchmarkNumber2-16          100          10627156 ns/op         7403632 B/op      24434 allocs/op