wordlistgen.py

# usage: wordlistgen.py [words.txt] [answers.txt] [wordlist.asm]
# words.txt should be sorted by word frequency
# since later entries can be removed if there's not enough space

import sys

ATOZ = "abcdefghijklmnopqrstuvwxyz"
BANK_FREES =[0xa3c, 0xfe4, 0xfe4]
NUM_BANKS = len(BANK_FREES)

fulllist = [i.strip().lower() for i in open(sys.argv[1], "r")]
anslist = [i.strip().lower() for i in open(sys.argv[2], "r")]
outfile = open(sys.argv[3], "w")

output_asm = """
    ; generated by wordlistgen.py
"""

section_format = """
    .section wordlist_data_{0}
{1}
    .send

"""

section_1_format = """
_ptrs := ({0})
wordlist_1_ptrs_lo  .byte <(_ptrs)
wordlist_1_ptrs_hi  .byte >(_ptrs)

wordlist_2_ofs
"""

sizes = None
banks = None
bins = {}
remaining = []
for i in ATOZ:
    for j in ATOZ:
        bins[i+j] = []

# first iteration: fit banks
for word in fulllist:
    key = word[:2]
    bins[key].append(word)
    # try fitting
    try_sizes = [0 for i in range(NUM_BANKS)]
    try_banks = [[] for i in range(NUM_BANKS)]
    cur_bank = 0
    exit = False
    for i in ATOZ:
        group_size = sum([len(bins[i+j]) for j in ATOZ])
        while (try_sizes[cur_bank] + group_size) > (BANK_FREES[cur_bank] // 2):
            cur_bank += 1
            if cur_bank >= NUM_BANKS:
                exit = True
                break
        if exit:
            break
        try_sizes[cur_bank] += group_size
        try_banks[cur_bank].append(i)
    if exit:
        remaining.append(word)
        bins[key].remove(word)
    else:
        sizes = try_sizes
        banks = try_banks

# second iteration: fill gaps
for word in remaining:
    for i in range(NUM_BANKS):
        if word[0] not in banks[i]:
            continue
        if (sizes[i] + 2) <= (BANK_FREES[i] // 2):
            print(word)
            bins[word[:2]].append(word)
            sizes[i] += 2

print(sum([len(bins[i]) for i in bins]))

# write results
for i in range(NUM_BANKS):
    stxt = ""
    if i == 0:
        stxt += section_1_format.format(", ".join(["wordlist_3_"+i for i in ATOZ]))
        for j in ATOZ:
            stxt += "    .byte {}\n".format(", ".join(["{:3}".format(len(bins[j+k])*2) for k in ATOZ]))
        stxt += "\n"
    for j in banks[i]:
        stxt += "wordlist_3_"+j
        linectr = 0
        for k in ATOZ:
            # reverse the list to aid searching, since the game checks from the last member first
            for word in bins[j+k][::-1]:
                if linectr % 16 == 0:
                    stxt += "\n    .word "
                else:
                    stxt += ", "
                c2 = (ord(word[2]) - ord("a") + 1)
                c3 = (ord(word[3]) - ord("a") + 1)
                c4 = (ord(word[4]) - ord("a") + 1)
                il = 0x8000 if word in anslist else 0
                stxt += "${:04x}".format(c2 | (c3 << 5) | (c4 << 10) | il)
                linectr += 1
        stxt += "\n"
    output_asm += section_format.format(i + 1, stxt)

outfile.write(output_asm)