function_words.py

interjections = ["oh", "wow", "tut-tut", "tsk-tsk", "ugh", "whew", "phew", "yeah", "yea", "shh", "oops", "ouch", "aha",
                 "yikes"]
generalAdjectivesAdverbs = ["new", "newer", "newest", "old", "older", "oldest", "previous", "good", "well", "better", "best",
                            "big", "bigger", "biggest", "easy", "easier", "easiest", "fast", "faster", "fastest", "far", "hard", "harder", "hardest",
                            "least", "own", "large", "larger", "largest", "long", "longer", "longest", "low", "lower", "lowest", "high", "higher",
                            "highest", "regular", "simple", "simpler", "simplest", "small", "smaller", "smallest", "tiny", "tinier", "tiniest",
                            "short", "shorter", "shortest", "main", "actual", "nice", "nicer", "nicest", "real", "same", "able", "certain", "usual",
                            "so-called", "mainly", "mostly", "recent", "anymore", "complete", "lately", "possible", "commonly", "constantly",
                            "continually", "directly", "easily", "nearly", "slightly", "somewhere", "estimated", "latest", "different", "similar",
                            "widely", "bad", "worse", "worst", "great", "specific",  "available", "average", "awful", "awesome", "basic", "beautiful",
                            "busy", "current", "entire", "everywhere", "important", "major", "multiple", "normal", "necessary", "obvious", "partly",
                            "special", "last", "early", "earlier", "earliest", "young", "younger", "youngest", ""]
delexicalizedVerbs = ["seem", "seems", "seemed", "let", "let's", "lets", "make", "makes", "made", "want", "showed", "shown",
                      "go", "goes", "went", "gone", "take", "takes", "took", "taken",	"put", "puts", "use", "used", "try", "tries", "tried", "mean",
                      "means", "meant", "called", "based", "add", "adds", "added", "contain", "contains", "contained", "consist", "consists",
                      "consisted", "ensure", "ensures", "ensured"]
intensifiers = ["highly", "very", "really", "extremely", "absolutely", "completely", "totally", "utterly", "quite",
                "somewhat", "seriously", "fairly", "fully", "amazingly"]

additionalTransitionWords = ["again", "definitely", "eternally", "expressively", "instead",
                             "expressly", "immediately", "including", "instantly", "namely", "naturally", "next", "notably", "now", "nowadays",
                             "ordinarily", "positively", "truly", "ultimately", "uniquely", "usually", "almost", "maybe",
                             "probably", "granted", "initially", "too", "actually", "already", "e.g", "i.e", "often", "regularly", "simply",
                             "optionally", "perhaps", "sometimes", "likely", "never", "ever", "else", "inasmuch", "provided", "currently", "incidentally",
                             "elsewhere", "particular", "recently", "relatively", "f.i", "clearly", "apparently"]
interviewVerbs = ["say", "says", "said", "claimed", "ask", "asks", "asked", "stated", "explain", "explains", "explained",
                  "think", "thinks", "talks", "talked", "announces", "announced", "tells", "told", "discusses", "discussed", "suggests",
                  "suggested", "understands", "understood"]
correlativeConjunctions = ["sooner", "just", "only"]
subordinatingConjunctions = ["if", "even"]
coordinatingConjunctions = ["and", "or", "and/or", "yet"]
prepositionalAdverbs = ["back", "within", "forward", "backward", "ahead"]
prepositions = ["in", "from", "with", "under", "throughout", "atop", "for", "on", "of", "to", "aboard", "about",
                "above", "abreast", "absent", "across", "adjacent", "after", "against", "along", "alongside", "amid", "mid",
                "among", "apropos", "apud", "around", "as", "astride", "at", "ontop", "afore", "tofore", "behind", "ahind",
                "below", "ablow", "beneath", "neath", "beside", "between", "atween", "beyond", "ayond", "by", "chez",
                "circa", "spite", "down", "except", "into", "less", "like", "minus", "near", "nearer", "nearest", "anear", "notwithstanding",
                "off", "onto", "opposite", "out", "outen", "over", "past", "per", "pre", "qua", "sans", "sauf", "sithence", "through",
                "thru", "truout", "toward", "underneath", "up", "upon", "upside", "versus", "via", "vis-à-vis", "without", "ago",
                "apart", "aside", "aslant", "away", "withal", "towards", "amidst", "amongst", "midst", "whilst"]
continuousVerbs = ["doing", "daring", "having", "appearing", "becoming", "coming", "keeping", "remaining", "staying",
                   "saying", "asking", "stating", "seeming", "letting", "making", "setting", "showing", "putting", "adding", "going", "using",
                   "trying", "containing"]
copula = ["appear", "appears", "appeared", "become", "becomes", "became", "come", "comes", "came", "keep", "keeps", "kept",
          "remain", "remains", "remained", "stay", "stays", "stayed", "turn", "turns", "turned"]
otherAuxiliaries = ["can", "cannot", "can't", "could", "couldn't", "could've", "dare", "dares", "dared", "do",
                    "don't", "does", "doesn't", "did", "didn't", "done", "have", "haven't", "had", "hadn't", "has", "hasn't",
                    "i've", "you've", "we've", "they've", "i'd", "you'd", "he'd", "she'd", "it'd", "we'd", "they'd", "would", "wouldn't",
                    "would've", "may", "might", "must", "need", "needn't", "needs", "ought", "shall", "shalln't", "shan't", "should",
                    "shouldn't", "will", "won't", "i'll", "you'll", "he'll", "she'll", "it'll", "we'll", "they'll", "there's", "there're",
                    "there'll", "here's", "here're", "there'll"]
adverbialGenitives = ["always", "once", "twice", "thrice"]
locativeAdverbs = ["there", "here", "whither", "thither", "hither", "whence", "thence"]
pronominalAdverbs = ["therefor", "therein", "hereby", "hereto",
                     "wherein", "therewith", "herewith", "wherewith", "thereby"]

interrogativeProAdverbs = ["where", "how", "why", "whether", "wherever", "whyever", "wheresoever", "whensoever", "howsoever",
                           "whysoever", "whatsoever", "whereso", "whomso", "whenso", "howso", "whyso", "whoso", "whatso"]
interrogativePronouns = ["who", "whom"]
interrogativeDeterminers = ["which", "what", "whose"]
indefinitePronounsPossessive = ["one's", "nobody's", "everyone's", "everybody's", "someone's", "somebody's", "anyone's",
                                "anybody's", "nothing's", "everything's", "something's", "anything's", "whoever's", "others'", "other's", "another's",
                                "neither's", "either's"]

indefinitePronouns = ["none", "nobody", "everyone", "everybody", "someone", "somebody", "anyone", "anybody", "nothing",
                      "everything", "something", "anything", "each", "other", "whatever", "whichever", "whoever", "whomever",
                      "whomsoever", "whosoever", "others", "neither", "both", "either", "any", "such"]
reflexivePronouns = ["myself", "yourself", "himself", "herself",
                     "itself", "oneself", "ourselves", "yourselves", "themselves"]

quantifiers = ["all", "some", "many", "lot", "lots", "ton", "tons", "bit", "no", "every", "enough", "little",
               "much", "more", "most", "plenty", "several", "few", "fewer", "kind", "kinds"]
possessivePronouns = ["my", "your", "his", "her", "its",
                      "their", "our", "mine", "yours", "hers", "theirs", "ours"]

demonstrativePronouns = ["this", "that", "these", "those"]
personalPronounsAccusative = ["me", "him", "us", "them"]
personalPronounsNominative = ["i", "you", "he", "she", "it", "we", "they"]
ordinalNumerals = ["first", "second", "third", "fourth", "fifth", "sixth", "seventh", "eighth", "ninth", "tenth",
                   "eleventh", "twelfth", "thirteenth", "fourteenth", "fifteenth", "sixteenth", "seventeenth", "eighteenth",
                   "nineteenth", "twentieth"]
cardinalNumerals = ["one", "two", "three", "four", "five", "six", "seven", "eight", "nine", "ten", "eleven", "twelve", "thirteen",
                    "fourteen", "fifteen", "sixteen", "seventeen", "eighteen", "nineteen", "twenty", "hundred", "hundreds", "thousand", "thousands",
                    "million", "millions", "billion", "billions"]
articles = ["the", "an", "a"]

function_words_lists = [
    articles,
    cardinalNumerals,
    ordinalNumerals,
    personalPronounsAccusative,
    personalPronounsNominative,
    demonstrativePronouns,
    possessivePronouns,
    quantifiers,
    reflexivePronouns,
    indefinitePronouns,
    indefinitePronounsPossessive,
    interrogativePronouns,
    interrogativeProAdverbs,
    interrogativeDeterminers,
    pronominalAdverbs,
    locativeAdverbs,
    adverbialGenitives,
    otherAuxiliaries,
    copula,
    continuousVerbs,
    prepositionalAdverbs,
    prepositions,
    correlativeConjunctions,
    coordinatingConjunctions,
    subordinatingConjunctions,
    intensifiers,
    interviewVerbs,
    additionalTransitionWords,
    delexicalizedVerbs,
    generalAdjectivesAdverbs,
    interjections
]

all_words = []
for wl in function_words_lists:
    all_words += wl

fn_word_index = {
    word: i for i, word in enumerate(all_words)
}

if __name__ == "__main__":

    print(len(word_index))