set_sorter_vapicuno

### Written by vapicuno, 2019/05/05
### python 3.5, numpy 1.11.1
### Takes in fin which is in showdown teambuilder format
### Spits out fout which is a list of sets in alphabetical order.
### Within each pokemon, sets are ordered by frequency of use

import numpy as np

fin = 'linear_builder.txt'
fout = 'linear_builder_sets.txt'

setlist = list()
f = open(fin)
line = f.readline()
buffer = line
linestatus = 0; # 0 = team not found, 1 = team found
while line:
    line = f.readline()
    if linestatus == 0: # If importable has not been found
        mark = line.find('Ability:') # Use Ability to find importable set
        if mark == 0: # Find pokemon name
            pos2 = buffer.find(' (F) ')
            if pos2 == -1:
                pos2 = buffer.find(' (M) ')
                if pos2 == -1:
                    pos2 = buffer.find(' @ ')
                    if pos2 == -1:
                        pos2 = len(buffer) - 2 # -2 account for newline
            if buffer[pos2-1] == ')':
                pos2 = pos2 - 1;
                pos1 = pos2 - 1;
                while buffer[pos1] != '(':
                    pos1 = pos1 - 1;
                pos1 = pos1 + 1;
            else:
                pos1 = 0;
            positem = buffer.find(' @ ',pos2)
            mon = buffer[pos1:pos2]
            if positem != -1:
                itemtxt = buffer[positem:]
            else:
                itemtxt = ''
            montxt = mon + itemtxt # Make Name @ Item
            linestatus = 1
        buffer = line
    if linestatus == 1: # If importable has been found
        if line == '\n': # If linebreak has been found
            setlist.append(montxt) # Save as a string
            linestatus = 0
        else:
            montxt = montxt + line # Add remaining importable lines
f.close()
setlist.sort(); # Final sorted list of sets in string format

## Sequentially runs through setlist
## For each mon, arrange sets by frequency of usage
## Writes to txt
setlistsortedreduced = list() # final list
subsetlist = list() # list of strings of full sets
submoveslist = list() # List of sets, because moves are unordered
subsetcount = list() # frequency list
chosenmon = '';
setlistlen = len(setlist)
for n in np.arange(setlistlen):
    currentset = setlist[n]
    currentset = currentset.replace('Shiny: Yes  \n','')
    currentmoves = set()
    nlindex = list() # newline indices
    lenset = len(currentset)
    postemp = lenset
    while len(nlindex) < 5:
        nlindex.append(currentset.rfind('\n',0,postemp)) # Extract move indices
        postemp = nlindex[-1]
    for ii in np.arange(0,4):
        currentmoves.add(currentset[nlindex[4-ii]+4:nlindex[3-ii]]) # Extract moves
    partition = nlindex[4]

    posnewline = currentset.find('\n')
    posspace = currentset.find(' @ ')
    currentmon = currentset[:min(posspace,posnewline)]
    if chosenmon == currentmon:
        # determine set equivalence
        matchindex = -1; # matching set index
        idx = 0; # iterator
        while matchindex < 0 and idx < len(subsetlist):
            if subsetlist[idx][0:partition] == currentset[0:partition]:
                if submoveslist[idx] == currentmoves:
                    matchindex = idx
                    subsetcount[matchindex] += 1
            idx += 1
        if matchindex < 0:
            subsetlist.append(currentset)
            submoveslist.append(currentmoves)
            subsetcount.append(1)
    if chosenmon != currentmon or n == setlistlen-1:
        # all sets for a pokemon completed
        # sort the list by number of replicates
        # append to final list
        subsetsortindex = np.argsort(np.array(subsetcount))
        subsetsortindexrev = subsetsortindex[::-1]
        subsetsortlist = [subsetlist[i] for i in subsetsortindexrev]
        setlistsortedreduced.extend(subsetsortlist)
        # initialize new mon
        subsetlist = list()
        submoveslist = list()
        subsetcount = list()
        chosenmon = currentmon
        # append new mon and set
        subsetlist.append(currentset)
        submoveslist.append(currentmoves)
        subsetcount.append(1)

f = open(fout,'w')
for s in setlistsortedreduced:
    f.write(s)
    f.write('\n')
f.close()