* Handle some nullable cases
Refactor MysteryGift into a second abstract class (backed by a byte array, or fake data)
Make some classes have explicit constructors instead of { } initialization
* Handle bits more obviously without null
* Make SaveFile.BAK explicitly readonly again
* merge constructor methods to have readonly fields
* Inline some properties
* More nullable handling
* Rearrange box actions
define straightforward classes to not have any null properties
* Make extrabyte reference array immutable
* Move tooltip creation to designer
* Rearrange some logic to reduce nesting
* Cache generated fonts
* Split mystery gift album purpose
* Handle more tooltips
* Disallow null setters
* Don't capture RNG object, only type enum
* Unify learnset objects
Now have readonly properties which are never null
don't new() empty learnsets (>800 Learnset objects no longer created,
total of 2400 objects since we also new() a move & level array)
optimize g1/2 reader for early abort case
* Access rewrite
Initialize blocks in a separate object, and get via that object
removes a couple hundred "might be null" warnings since blocks are now readonly getters
some block references have been relocated, but interfaces should expose all that's needed
put HoF6 controls in a groupbox, and disable
* Readonly personal data
* IVs non nullable for mystery gift
* Explicitly initialize forced encounter moves
* Make shadow objects readonly & non-null
Put murkrow fix in binary data resource, instead of on startup
* Assign dex form fetch on constructor
Fixes legality parsing edge cases
also handle cxd parse for valid; exit before exception is thrown in FrameGenerator
* Remove unnecessary null checks
* Keep empty value until init
SetPouch sets the value to an actual one during load, but whatever
* Readonly team lock data
* Readonly locks
Put locked encounters at bottom (favor unlocked)
* Mail readonly data / offset
Rearrange some call flow and pass defaults
Add fake classes for SaveDataEditor mocking
Always party size, no need to check twice in stat editor
use a fake save file as initial data for savedata editor, and for
gamedata (wow i found a usage)
constrain eventwork editor to struct variable types (uint, int, etc),
thus preventing null assignment errors
New in c# 8
Assert.True(check for null) doesn't give a hint that the obj isnt null
afterwards.
Assert.NotNull does have the compiler hint attribute
could probably use fluentexpression syntax, but resharper doesn't pick
up on the nonnullable hint like vs does.
GeneratorTests: swap FirstOrDefault to First, can keep the fluent style
and resharper gets the hint.
tl;dr : resharper doesn't look at external lib for hints, gotta use
Assert.NotNull or something that doesn't return null
remove unnecessary %25 check (only applicable for gen4 encounters),
unneeded for gen5 method.
add a test case to generate a haxorus for all natures, verify shininess
Closes#2336
Pass them as Legal for wild encounters similar to M1/M2/M4.
Add test cases & test pkm validation
Add pid generator support
I decided against checking these PIDIVs after all others; inlined
methods.
Closes#2328 , thanks @pokecal !
use 'u' to denote uint values, so that less IL ops are required
~(21->17) in the case of IVs
no functional change, just a fun exercise looking at generated IL and
finding ways to use less ops
hide setters for LegalInfo outside assembly (shouldn't overwrite the
stored values)
update GenNumber to Gen* if appropriate; GenNumber checks for first Gen*
to match, is a little more explicit and quicker than calling twice when
checking a range
SetPINGA{S}(pk, criteria) -> pid,iv, nature, gender, ability (stats
future?)
memes aside, this is where criteria data is forced
Split N encounterstatic (has fixed PID and special flag) from
EncounterStaticPID
24 bytes per object -> 16 bytes
2,624 objects are made for legality checking; reduces from 73KB to 41KB
removing unnecessary clone recreation -> count from 2624 -> 414 (41KB to
6.6KB)
yay 10x reduction; not huge in the big picture but a fun exercise
remove linq for characteristic calc (get max IV) and others
remove some reliance on new[] for iv/ev
pk5 replace bitmagic with property get->set copy for cleanliness
the ereader mons are nature/gender locked too, so unroll a little prior
to rechecking the overall team (with ereader mon included). Have to do
it this way as a lock can pop if the ereader data matches a prior spread
before the prior teammate can be generated.
Unused after assignment, but if I rework the validation check to return the teamlock result instead of a bool I can spit out the Species-PID generated list as extra info. Helps identify the as-generated conditions if there are multiple teamlocks possible for an encounter (multiple shadow mons, seen vs unseen).
Now accounts for the shiny locks & associated restrictions, abusing the
recursion by setting & unsetting the required CPU Trainer's shiny value.
Repoint all test methods to use the new obj methods
A lot of internal information is nice to have saved; create a copy of
the static class's logic and pepper in some state variables (Required
CPU Shiny Value) to account for the weird edge cases.
Code is essentially the same as the static methods, except for more
fleshed out logic in the AllLock interrupt checking and the eventual
Trainer checks.
Now gets count=2 tests working
yield return new SeedFrame { FrameID = ctr + (l.Seen ? 5 : 7), PID = pid
};
used to be ctr + 6, needs to alternate even-odd
all 2 lock cases pass, but didn't make lock3's pass. Something else is
still amiss :)
break out some classes for the solidified lock objects
Check max case for level first, then iterate upwards (eliminates inner
if)
change wc* nature to sbyte (update comparisons, 0xFF too magic-y)
use Rand.Next(x) instead of (0, x)
commented out test calls = not working
will have to debug in more detail later; the first lock is always
working at least.
data sourced from
https://github.com/ijuintekka/Eligor/blob/master/Eligor/Spread.cs , not
sure if it's perfect.
can't cleanly condense deviating team appearances as double-shadow leads
can't be selectively encountered. didn't feel like modding lockfinder's
recursive algo to be smarter.
Store 4g giratina held item for enc->pkm legality
revise usages of GetSaneGender
Rearrange some logic
Better handle impossible version encounters (gen4)
gen 1-5 done, stuck on 659 - bunnelby egg...?
Fixes Krabby / etc.
Fix genderless encounterstatic gender set (Voltorb)
Now at 103, Exeggutor
pretty cool that over 100 species can have every single encounter
generate a legal pkm ;)
might be useful to add metrics to count how many encounters are
generated in a session
Fix/add missing gameversion references (BU instead of BW,
pokewalker=HGSS)
increase reuse of pidgenerator & add more generating methods.
seems like it's generating pkm fine besides the PID edge cases, which is
nice
Gen5 does not follow the same convention, 0 = non, 1 = rnd, 2 = always;
not gonna bother updating for just that one
bool? occupies 2 bytes; enum:byte is 1 byte.
should probably move validity checking logic into the IEncounterable
objects instead...
tested for 2 locks, later commits with legality check implementation
will feature more lengthy lock checks (test case automation).
https://pastebin.com/VdbjWaqL
(species doesn't matter since most are 50/50 ratio)
arbitrary access for RNG frame results, only generating frames when the
frame is unavailable. all calculated frames are cached.
to be used in RNG searching where caching seeds is necessary (repeat
checks)
dynamic level -> calc level
no sweet scent -> proc available (rock smash/fish)
change frame checking in encountergenerator to save which frame it was
found on (unused but possibly later can capture this for output)
Turns out there was no documentation anywhere on encounterslot numbers
for the individual forms; some sites had overall rates.
Assuming a human filled out the encounter slots, forms would repeat left
to right until a new form takes over. Below slot numbers match the rates
when summed up.
https://pastebin.com/raw/bQt3ZMbq
since there's 300 calls to rand to shuffle, I wouldn't be surprised if
there are vblanks messing up the frames
judging from my data 6 years ago (wow, still wrong, just look at the
frame offsets)
http://www.smogon.com/forums/threads/past-gen-rng-research.61090/page-35#post-4021415
looks like it happens quite erratically. may have to just override this
to return true for all gen3 safari mons as it's entirely unpredictable.
via pret/pokeemerald (and pokeruby); rate is set at 50%
swarm pokemon have the ability to have 4 fixed moves, so derive from
EncounterSlot to have moves, and remove special handling from
VerifyCurrentMoves (now picked up by IMoveset pattern
allows hgss safari mons regardless of frame call (table is currently all
Slot=0).
may need separate handling down the road:
* switch slot number to an enum for slot codes (>0 = slot number, less
than -> do something else).
not something im interested in right now
fix roamer3 not setting IVs
fix usum z items rightmost pixel getting chopped off
use Gen* instead of GenNumber for specific cases (faster)
add WasGiftEgg location case for Gen7
remove some unnecessary array allocations
force colo starters to be male only
use some c# lang features for EncounterType flags
add edge case PIDIV detection for starter gender/shiny lock scenario
adds a little documentation for those unfamiliar with the NPC PKM
generation quirks.
Colosseum Starters will now be recognized with a different PIDIV type
which is specific to them & them only.
check the middle bits for signs of pokewalker origins
can either be regular or adjusted underflow style
saves a little bit of computation time by not computing compatibility
for obvious non-matches
refer to stackexchange links, makes XDRNG PID/IV search so quick (2^3
instead of 2^8 or 2^16; as fast as a gen3 shiny pokemon (2^13=8192 times
as fast as last release)
throw in some auto-inlining to eliminate some overhead in repetitive
search calls
Only usable for searching Method 4 IV spreads -> seeds;
1,3's search uses the same approach as the 1,2 search
I took the 1,2 search derivation to iterate for the next step, which
allows us to not know anything about the middle rand bits.
optimize a little bit more, move the pre-loop add to the initialization
stage; moving to the precomputed section pays off after 256 calls to the
method
Meet in the middle attack trades some RAM (2^16 flag/byte array) to
reduce future searches by a factor of 1:2^8
profiling yielded >100x speed improvements, even a 2x would have been
impressive ;)
knocks the biggest cpu hog (when searching db) out of the race!
