python: elhuyar-i hitzak itzultzeko eskatzeko script-a

elhuyar.py

#!/usr/bin/python3
# -*- coding: utf-8 -*-

import sys
import requests
import re
from lxml import html

def printError(search):
        print("Errorea gertatu da, ziur aski sartu duzun hitza ez da hiztegian agertzen")
        if search[-1] == 'a':
                print("Bide batez, beharbada sartu duzun hitzak ez du a itsatsia...")

url="https://hiztegiak.elhuyar.eus"

try:
        search=sys.argv[1]

except:
        print("Erabilera: python3 elhuyar.py {hitza} {sartutako hitzaren hizkuntza: defektuz eu}")
        print("Adibidez:")
        print("\tpython3 elhuyar.py txori")
        print("\tpython3 elhuyar.py txori eu")
        print("\tpython3 elhuyar.py pajaro es")
        sys.exit(1)

try:
        lang=sys.argv[2]
except:
        lang = "eu"

if lang not in ["eu", "es"]:
        lang = "eu"

if lang == 'eu':
        url_lang = 'eu_es'
else:
        url_lang = 'es_eu'

try:
        headers = {
            "User-Agent": "Mozilla/5.0 (X11; Linux x86_64; rv:145.0) Gecko/20100101 Firefox/145.0"
        }
        url        = f"{url}/{url_lang}/{search}"
        response = requests.get(url, headers=headers)
        dom      = html.fromstring(response.content)
        uls      = dom.xpath("//ul[contains(@class, 'hizkuntzaren_arabera')]")

        result  = ""
        i       = 0
        for ul in uls:
                ori    = dom.xpath("//div[contains(@class, 'emaitza-lerroa')]/h1/span")[i].text_content()
                result = result + f"{ori}:\n"
                lis    = ul.xpath('li')
                for li in lis:
                        t = li.xpath("p[@class='lehena']")[0].text_content().strip()
                        t = re.sub(r"\s+", ' ', t)
                        result = result + f"\t{t}\n"
                i = i + 1
        if result != '':
                result = f"Bilatutakoa: {search}\n"+result
                print(result)
                sys.exit(0)
        else:
                printError(search)
                sys.exit(1)

except SystemExit as sysexit:
    sys.exit(sysexit.code)

Ideia originala: https://framagit.org/-/snippets/7189

adibidez:

python3 elhuyar.py txori

Bilatutakoa: txori
1 txori:
	1 iz. pájaro
	2 iz. tupé, copete
	3 iz. lazo, cinta anudada de adorno
2 txori:
	iz. (Med.) lobanillo, tumor enquistado

python3 elhuyar.py pajaro es

Bilatutakoa: pajaro
pájaro:
	1 s.m. txori
	2 s.m. (fig.) [persona] maltzur, azeri, zakur; ale, elementu