#!/usr/bin/env python3

# http://inamidst.com/saxo/
# Created by Sean B. Palmer

import re
import saxo

pattern_sentence = \
    r"^(.*?(?<!%s)(?:\.(?= [A-Z0-9]|\Z)|\Z))" % ")(?<!".join((
    "cf", "lit", "etc", "Ger", "Du", "Skt", "Rus", "Eng", "Amer.Eng", "Sp",
    "Fr", "N", "E", "S", "W", "L", "Gen", "J.C", "dial", "Gk", "19c",
    "18c", "17c", "16c", "St", "Capt", "obs", "Jan", "Feb", "Mar", "Apr",
    "Jul", "Aug", "Sep", "Oct", "Nov", "Dec", "c", "tr", "e", "g"))
regex_sentence = re.compile(pattern_sentence)
regex_definition = re.compile("(?ims)class=.word[^>]*>(.*?)</section>")
regex_updated = re.compile('<p class="word__date(.*?)</p>')
regex_tag = re.compile(r"<[^>]+>")

def etymology(term, limit=None):
    term = term.rstrip("\x09\x20")
    page = saxo.request(
        "http://etymonline.com/search",
        query={"q": term},
        headers={"Referer": "http://www.etymonline.com/"})

    if not "text" in page:
        raise Exception("No response from %s" % page["url"])

    text = page["text"]
    text = text.replace('<span class="adLabel">Advertisement</span>', '')
    definitions = regex_definition.findall(text)
    if not definitions:
        raise Exception("No definitions found in %s" % page["url"])

    definition = regex_updated.sub(" ", definitions[0])
    definition = regex_tag.sub("", definition)
    definition = definition.replace("\r", "")
    definition = definition.replace("\n", " ")
    definition = definition.replace("&quot;", "'")
    definition = definition.replace("&#x27;", "'")
    limit = limit - len(page["url"]) - 3
    definition = definition[:limit] + "..."
    definition = definition.replace('"', "'")
    return definition, page["url"]

@saxo.pipe
def ety(arg):
    if not arg:
        return "Display the etymology of a term from Etymonline"

    try: sentence, url = etymology(arg, 360)
    except Exception as err:
        link = "http://etymonline.com/search?term=" + arg
        return "Nothing found. Try " + link + (" (%s)" % err)

    if sentence:
        return "\"%s\" - %s" % (sentence, url)
    return "?"