96 lines
2.3 KiB
Python
96 lines
2.3 KiB
Python
|
#!/usr/bin/env python3
|
||
|
|
"""
|
||
|
|
Pull definitions from Wikitionary.org.
|
||
|
|
"""
|
||
|
|
import re
|
||
|
|
import requests
|
||
|
|
|
||
|
|
from module import commands, example
|
||
|
|
|
||
|
|
uri = 'http://en.wiktionary.org/w/index.php?title={}&printable=yes'
|
||
|
|
r_tag = re.compile(r'<[^>]+>')
|
||
|
|
r_ul = re.compile(r'(?ims)<ul>.*?</ul>')
|
||
|
|
|
||
|
|
|
||
|
|
def text(html):
|
||
|
|
text = r_tag.sub('', html).strip()
|
||
|
|
text = text.replace('\n', ' ')
|
||
|
|
text = text.replace('\r', '')
|
||
|
|
text = text.replace('(intransitive', '(intr.')
|
||
|
|
text = text.replace('(transitive', '(trans.')
|
||
|
|
return text
|
||
|
|
|
||
|
|
|
||
|
|
def wikt(word):
|
||
|
|
res = requests.get(uri.format(word))
|
||
|
|
# bytes = r_ul.sub('', bytes)
|
||
|
|
res = res.text
|
||
|
|
|
||
|
|
mode = None
|
||
|
|
etymology = None
|
||
|
|
definitions = {}
|
||
|
|
for line in res.splitlines():
|
||
|
|
if 'id="Etymology"' in line:
|
||
|
|
mode = 'etymology'
|
||
|
|
elif 'id="Noun"' in line:
|
||
|
|
mode = 'noun'
|
||
|
|
elif 'id="Verb"' in line:
|
||
|
|
mode = 'verb'
|
||
|
|
elif 'id="Adjective"' in line:
|
||
|
|
mode = 'adjective'
|
||
|
|
elif 'id="Adverb"' in line:
|
||
|
|
mode = 'adverb'
|
||
|
|
elif 'id="Interjection"' in line:
|
||
|
|
mode = 'interjection'
|
||
|
|
elif 'id="Particle"' in line:
|
||
|
|
mode = 'particle'
|
||
|
|
elif 'id="Preposition"' in line:
|
||
|
|
mode = 'preposition'
|
||
|
|
elif 'id="' in line:
|
||
|
|
mode = None
|
||
|
|
|
||
|
|
elif (mode == 'etmyology') and ('<p>' in line):
|
||
|
|
etymology = text(line)
|
||
|
|
elif (mode is not None) and ('<li>' in line):
|
||
|
|
definitions.setdefault(mode, []).append(text(line))
|
||
|
|
|
||
|
|
if '<hr' in line:
|
||
|
|
break
|
||
|
|
return etymology, definitions
|
||
|
|
|
||
|
|
parts = ('preposition', 'particle', 'noun', 'verb',
|
||
|
|
'adjective', 'adverb', 'interjection')
|
||
|
|
|
||
|
|
|
||
|
|
def format(result, definitions, number=2):
|
||
|
|
for part in parts:
|
||
|
|
if part in definitions:
|
||
|
|
defs = definitions[part][:number]
|
||
|
|
result += u' — {}: '.format(part)
|
||
|
|
n = ['%s. %s' % (i + 1, e.strip(' .')) for i, e in enumerate(defs)]
|
||
|
|
result += ', '.join(n)
|
||
|
|
return result.strip(' .,')
|
||
|
|
|
||
|
|
|
||
|
|
@commands('define', 'dict')
|
||
|
|
@example('.dict bailiwick')
|
||
|
|
def wiktionary(bot, trigger):
|
||
|
|
"""Look up a word on Wiktionary."""
|
||
|
|
word = trigger.group(2)
|
||
|
|
if word is None:
|
||
|
|
return bot.reply('You must tell me what to look up!')
|
||
|
|
|
||
|
|
_, definitions = wikt(word)
|
||
|
|
if not definitions:
|
||
|
|
return bot.say(f"Couldn't get any definitions for {word}.")
|
||
|
|
|
||
|
|
result = format(word, definitions)
|
||
|
|
if len(result) < 150:
|
||
|
|
result = format(word, definitions, 3)
|
||
|
|
if len(result) < 150:
|
||
|
|
result = format(word, definitions, 5)
|
||
|
|
|
||
|
|
if len(result) > 300:
|
||
|
|
result = result[:295] + '[...]'
|
||
|
|
bot.say(result)
|