import re


def check_if_address_valid(address):
    """
    Take input address and scrape date
    :param addr:
    :return:
    """
    x = re.search("^https://www.goodreads.com/book/show/[0-9]+", address)
    if x:
        return True
    else:
        return False


def parse_query_to_url(query):
    elements = re.findall("[0-9A-Za-z_\"><]+", query)
    count = len(elements)
    if count == 3:
        # can only be A.B:C or wrong
        if re.search("^[0-9a-zA-Z_]+\.[0-9a-zA-Z_]:[0-9a-zA-Z_\"]", query):
            return url_safe(elements[0] + "." + elements[1] + "%3A" + elements[2])
        else:
            print("Invalid query.")
            return ""
    elif count == 4:
        # a pair and one of [NOT, >, <].
        if re.search("^[0-9a-zA-Z_]+\.[0-9a-zA-Z_]:\sNOT\s[0-9a-zA-Z_\"]", query):
            return url_safe(elements[0] + "." + elements[1] + "%3A" + "NOT" + elements[2])
        elif re.search("^[0-9a-zA-Z_]+\.[0-9a-zA-Z_]:\s>\s[0-9a-zA-Z_\"]", query):
            return url_safe(elements[0] + "." + elements[1] + "%3A" + ">" + elements[2])
        elif re.search("^[0-9a-zA-Z_]+\.[0-9a-zA-Z_]:\s<\s[0-9a-zA-Z_\"]", query):
            return url_safe(elements[0] + "." + elements[1] + "%3A" + "<" + elements[2])
        else:
            print("Invalid query.")
            return ""
    elif 6 <= count <= 8:
        # AND or OR operator
        if re.search(".*\sAND\s.*", query):
            parts = query.split(" AND ")
            return parse_query_to_url(parts[0]) + "%26AND%26" + parse_query_to_url(parts[1])
        elif re.search(".*\sOR\s.*", query):
            parts = query.split(" OR ")
            return parse_query_to_url(parts[0]) + "%26OR%26" + parse_query_to_url(parts[1])
        else:
            print("Invalid query.")
            return ""


def parse_query_to_json(pair):
    elements = re.findall("[0-9A-Za-z\"._]", pair)
    count = len(elements)
    print(count)
    if count != 3 and count != 4:
        print("Failed to parse query: invalid args number")
        return {"wrong": "True"}            # will never be founded in database
    elif count == 3:
        # can be A.B: C or A.B: "C"
        if re.search("\".*\"", elements[2]):
            return {elements[0] + "." + elements[1]: elements[2]}
        else:
            return {elements[0] + "." + elements[1]: {"$regex": elements[2], "$options": "i"}}
    else:
        # can be NOT, >, or <
        if elements[2] == "NOT":
            return {elements[0] + "." + elements[1]: {"$not": {"$regex": elements[3], "$options": "i"}}}
        elif elements[2] == ">":
            return {elements[0] + "." + elements[1]: {"$gt": float(elements[3])}}
        elif elements[2] == "<":
            return {elements[0] + "." + elements[1]: {"$lt": float(elements[3])}}
        else:
            print("Failed to parse query: unknown operator")
            return {"wrong": "True"}


def url_safe(element):
    return element.replace("\"", "%22").replace(">", "%3E").replace("<", "%3C")