parse_info.py

from imports import *

punc = list(string.punctuation)
def parse_string(inp: str, rep=" ", punc=punc, excp=[]) -> str:
    try:
        for i in excp:
            punc.remove(i)
    except:
        pass
    inp = inp.lower()
    inp = re.sub(r"won\'t", "will not", inp)
    inp = re.sub(r"can\'t", "can not", inp)
    inp = re.sub(r"\'re", " are", inp)
    inp = re.sub(r"\'s", " of", inp)
    inp = re.sub(r"\'d", " would", inp)
    inp = re.sub(r"\'ll", " will", inp)
    inp = re.sub(r"\'t", " not", inp)
    inp = re.sub(r"\'ve", " have", inp)
    inp = re.sub(r"\'m", " am", inp)  
    for i in punc:
        inp = inp.replace(i,rep)
    return " ".join(inp.split())

def parse_time(inp: List):
    duration = 0
    for i, _ in enumerate(inp):
        inp[i] = inp[i].lower()
        now = datetime.utcnow().strftime("%d/%m/%Y")
        _ = ["đến", " to "] # list that split 2 time point word
        __ = ["now", "hiện tại", " nay", " đến nay", "present"] # end time point
        for j in _:
            inp[i] = inp[i].replace(j," - ")
        for j in __:
            inp[i] = inp[i].replace(j,now)
        for j in inp[i]:
            if j.isalpha():
                inp[i] = inp[i].replace(j,"").strip()
        inp[i] = parse_string(" ".join(inp[i].split(" ")), rep="", excp=["/","-"])

        time_point = inp[i].split("-") # split to 2 time point
        if len(time_point) != 2: # must be splitted to 2 time point
            continue
        try:
            d1 = parser.parse(time_point[0]).strftime("%d-%m-%Y")
            d2 = parser.parse(time_point[1]).strftime("%d-%m-%Y")
            duration += (datetime.strptime(d2, "%d-%m-%Y") - datetime.strptime(d1, "%d-%m-%Y")).days
        except:
            continue
    return "{:.1f} năm".format(duration/365)

filename = "./skills.csv"
detected = chardet.detect(Path(filename).read_bytes()) # "ISO-8859-1"
skill_list = pd.read_csv(filename, encoding=detected["encoding"])
skill_list = [i.replace("\n","") for i in skill_list["Skill"].to_list()]
def parse_skill(inp: List) -> list:
    res = []
    for i, _ in enumerate(inp):
        inp[i] = parse_string(inp[i])
    for ngram in Counter(map(' '.join, everygrams(" ".join(inp).split(), 1, 3))).keys():
        if ngram in skill_list:
            res.append(ngram)
    return ". ".join([i.capitalize() for i in list(set(res))])

def parse_gender(inp: List) -> str:
    inp = " ".join([parse_string(i) for i in inp])
    gender = ["nam", "nữ", "female", "male", "bisexual", "asexual", "heterosexual", "homosexual", "lgbt"]
    for gen in gender:
        if gen in inp:
            return gen
    return ""

def parse_address(inp: List) -> str:
    inp = [parse_string(i, excp=",") for i in inp]
    for i, _ in enumerate(inp):
        inp[i] = " ".join([j.capitalize() for j in inp[i].split()])
    return ". ".join(inp)

def parse_designation(inp: List) -> str:
    inp = list(set([parse_string(i) for i in inp]))
    for i, _ in enumerate(inp):
        inp[i] = " ".join([j.capitalize() for j in inp[i].split()])
    return ". ".join(inp)    

def parse_email(inp: List) -> str:
    inp = list(set([parse_string(i, rep="", excp=["@","."]) for i in inp]))
    return " ".join(inp)

def decontracted(phrase) -> str:
    phrase = re.sub(r"â€|™|“|”|;|ü|\xad|\xa0|\u200b|·|∙|�|●|�|§|•|!|▪|©|\?|\]|\[|\)|\(", "", phrase)
    phrase = phrase.strip()
    phrase = unicodedata.normalize("NFC", phrase)
    if "   " in phrase or "  " in phrase: # check space character 
        phrase = phrase.replace("   ","_").replace("  ","_").replace(" ","").replace("_"," ")
    tmp = phrase.split(" ")
    check_parse = True
    for i in tmp:
        if len(i) > 1:
            check_parse = False
            break
    if check_parse:
        phrase = phrase.replace(" ","")
    # phrase = phrase.replace("   "," ").replace("  "," ")
    return phrase.replace("\n"," ")

def normalize_bbox(bbox, size): # must normalize bbox to [0;1000]
    return [int(1000 * bbox[0] / size[0]),
            int(1000 * bbox[1] / size[1]),
            int(1000 * bbox[2] / size[0]),
            int(1000 * bbox[3] / size[1])]