Coverage for src/botadi/mokadi/mokadi_wikipedia.py: 89%

1# -*- coding: utf-8 -*- 

2""" 

3@file 

4@brief Look into the mail box 

5""" 

6import re 

7import wikipedia 

8from .mokadi_exceptions import WikipediaException 

9 

10 

11def definition_wikipedia(query, summary=True, lang="fr"): 

12 """ 

13 Search for the definition of something on wikipedia. 

14 

15 @param query what to search for 

16 @param summary summary or long definition 

17 @param lang language 

18 @return definition 

19 """ 

20 wikipedia.set_lang("fr") 

21 if summary: 

22 res = wikipedia.summary(query, sentences=1, auto_suggest=False) 

23 return res.strip("# ") 

24 else: 

25 res = wikipedia.search(query, results=1) 

26 if len(res) == 0: 

27 raise WikipediaException( 

28 "Unable to find a page for '{0}'".format(query)) 

29 title = res[0] 

30 page = wikipedia.page(title) 

31 if page is None or not hasattr(page, "content"): 

32 raise WikipediaException( 

33 "Unable to find a page for '{0}' (2)".format(page)) 

34 page = page.content 

35 if lang == "fr": 

36 if "== Notes et références ==" in page: 

37 page = page.split("== Notes et références ==")[0] 

38 if "== Voir aussi ==" in page: 

39 page = page.split("== Notes et références ==")[0] 

40 return page 

41 

42 

43def suggestions_wikipedia(query, lang="fr"): 

44 """ 

45 Suggestions for something on wikipedia. 

46 

47 @param query what to search for 

48 @param lang language 

49 @return definition 

50 """ 

51 wikipedia.set_lang("fr") 

52 return wikipedia.search(query, results=10) 

53 

54 

55def synonyms_wiktionary(name, lang="fr"): 

56 """ 

57 Returns the definition from Wiktionary. 

58 

59 @param name name to look for 

60 @param lang language 

61 @return definition 

62 

63 `Wiktionay API <https://en.wiktionary.org/w/api.php>`_. 

64 """ 

65 import wptools 

66 page = wptools.page(name, wiki='{0}.wiktionary.org'.format( 

67 lang), lang=lang, silent=True) 

68 page.get_parse() 

69 text = page.data['wikitext'] 

70 syn = "==== {{S|synonymes}} ====" 

71 if syn not in text: 

72 return None 

73 text = text.split(syn)[1].split("====")[0] 

74 reg = re.compile("[[]{2}(.*?)[]]{2}") 

75 res = reg.findall(text) 

76 return res