testing.py

from typing import Counter
from trialyzer import *
import timeit
import cProfile
import corpus
import typingdata
import time

# ok what is the chain here
# note vscode debugger makes this roughly 10x slower than straight Python

# layout things
    # layout construction: 0.3 ms
    # calculate_counts: 1798 ms (includes all_nstrokes)
        # just a plain Counter instead (without calculating sums): 1878 ms
        # after making all the nstroke category functions @functools.cache: 
            # calculate_counts: 1266 ms
            # Counter(): 1348 ms
    # ngram to nstroke: 
    # how long does it take to iterate all tristrokes
        # tuple(all_nstrokes()): 1416 ms
            # after caching to_nstroke(): 275 ms!!
        # tuple(itertools all trigrams): 3 ms
        # Very interesting, seems like to_nstroke is the big deal. Cache it?
        # After unwrapping layout.finger() and .coord() from their functions:
            # tuple(all_nstrokes()): 1359 ms
        # After making dedicated dictionaries to replace the functions:
            # tuple(all_nstrokes()): 1216 ms

# trialyzer things
    # stuff that only runs through what is saved
        # load csv data: 3 ms
        # get medians: 125 ms
        # tristroke_category_data: 14 ms
    # runs through all tristrokes
        # summary_tristroke_analysis: 1923 ms
    # load shai: 162 ms
    # summary tristroke rank 3 layouts: 3687 ms
        # after caching layout.to_nstroke(): 1069 ms!!
    # full tristroke rank 3 layouts: 4897 ms
        # after caching layout.to_nstroke(): 1206 ms!!

# nstroke things
    # tristroke_category: 180 ms to go through a precomputed list of qwerty nstrokes

qwerty = layout.Layout("qwerty", False)
# typingdata_ = typingdata.TypingData("tanamr")
# corpus_ = corpus.get_corpus("shai.txt", precision=5000)
# constraintmap_ = constraintmap.get_constraintmap("trad-dead-pinkies")
# n = 3
# keys = ("a", "b", "c")

def stuff():
    # csvdata = load_csv_data("default")
    # for layoutname in ("qwerty", "semimak", "boom"):
    #     lay = layout.get_layout(layoutname)
    #     medians = get_medians_for_layout(csvdata, lay)
    #     tricatdata = tristroke_category_data(medians)
    #     summary_tristroke_analysis(lay, tricatdata, medians)
    # set_1 = {nstroke for nstroke in qwerty.nstrokes_with_any_of(keys, n)} # 116
    # set_2 = {nstroke for nstroke in qwerty.by_brute_force(keys, n)} # 235
    # corpus.Corpus("tr_quotes.txt")
    # for lay, score, swap in steepest_ascent(qwerty, typingdata_, 
    #         corpus_.trigram_counts, constraintmap_, 
    #         pins=qwerty.get_board_keys()[0].values()):
    #     print(f"{swap} gives {score:.3f}")
    #     print(repr(lay))
    corpus_ = corpus.get_corpus("shai.txt", space_key="space_r")

# n_ = 1
# print(timeit.timeit("stuff()", globals=globals(), number=n_)/n_ * 1000)
cProfile.run("corpus.get_corpus('shai.txt', space_key='space_r')", sort="tottime")