extract_scripts.py

import os
import re
import shutil
import zipfile

EXTRACT_FILES = False

if EXTRACT_FILES:
    if os.path.isdir("tmp"):
        shutil.rmtree("tmp")
    os.mkdir("tmp")

    for path in os.listdir():
        if path.endswith(".gro"):
            gro = zipfile.ZipFile(path)
            for zipPath in gro.namelist():
                if zipPath.endswith(".dlg") or zipPath.endswith(".lua"):
                    # This overwrites duplicates in the same way the game does
                    file = open("tmp/" + zipPath.split("/")[-1], "wb")
                    file.write(gro.read(zipPath))
                    file.close()


allFlags = {}
def addFlag(flag, source):
    flag = flag.decode("utf-8")
    if flag in allFlags:
        if source not in allFlags[flag]:
            allFlags[flag].append(source)
    else:
        allFlags[flag] = [source]


for path in os.listdir("tmp"):
    file = open("tmp/" + path, "rb")
    if path.endswith(".lua"):
        for match in re.finditer(b":(Set|Get|Is)(Var|Code)(Value|Set)?\((?P<args>.*?)\)", file.read(), flags=re.I):
            # Need to do this in two parts to make sure quotes match
            argMatch = re.match(b"\s*\"([^\"]+)\"\s*?,?.*?", match.group("args"))
            argMatch2 = re.match(b"\s*'([^']+)'\s*?,?.*?", match.group("args"))
            if argMatch:
                addFlag(argMatch.group(1), path)
            elif argMatch2:
                addFlag(argMatch2.group(1), path)
            else:
                addFlag(b"<code>", path)

    elif path.endswith(".dlg"):
        for match in re.finditer(b"set:\s?(.*?)\s", file.read(), flags=re.I):
            addFlag(match.group(1), path)

    file.close()


file = open("out.csv", "w")
file.write("Flag,Source File(s)\n")
for flag in sorted(allFlags.keys()):
    file.write(flag + "," + ",".join(allFlags[flag]) + "\n")
file.close()