db.py 2.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990
  1. from pymongo import MongoClient,errors
  2. from bson.objectid import ObjectId
  3. import datetime
  4. import re
  5. class db():
  6. DB="alexis"
  7. def __init__(self):
  8. self.client = MongoClient('localhost', 27017)
  9. self.c_sources=self.client[self.DB]["sources"]
  10. self.c_articles=self.client[self.DB]["articles"]
  11. def jsonable(self, el):
  12. if el is None or "_id" not in el:
  13. return el
  14. el["_id"]=str(el["_id"])
  15. return el
  16. def filter_articles(self, filter, limit=80):
  17. db_filter={}
  18. db_filter["matches"]=True
  19. db_filter["scrap_date"]={}
  20. if "maxdate" in filter and len(filter["maxdate"])>7:
  21. db_filter["scrap_date"]["$lt"]=datetime.datetime.strptime(filter["maxdate"],"%Y-%m-%d")
  22. if "mindate" in filter and len(filter["mindate"])>7:
  23. db_filter["scrap_date"]["$gt"]=datetime.datetime.strptime(filter["mindate"],"%Y-%m-%d")
  24. if db_filter["scrap_date"]=={}:
  25. del db_filter["scrap_date"]
  26. if "minweight" in filter and filter != "":
  27. db_filter["weight"]={"$gt":int(filter["minweight"])}
  28. print(db_filter)
  29. #{'site': 'site', 'mindate': 'mind', 'maxddate': 'maxd', 'minweight': 'minwe'}
  30. return [ self.jsonable(p) for p in self.c_articles.find(db_filter,{"html":0,"text":0}).limit(limit) ]
  31. def articles(self, id=None, limit=100):
  32. if id is None:
  33. return [ self.jsonable(p) for p in self.c_articles.find({"matches":True},{"html":0,"text":0}).limit(limit) ]
  34. ret=self.c_articles.find_one({"_id":ObjectId(id)})
  35. return self.jsonable(ret)
  36. def sources(self,id=None):
  37. if id is None:
  38. return [ self.jsonable(p) for p in self.c_sources.find() ]
  39. ret=self.c_sources.find_one({"_id":ObjectId(id)})
  40. return self.jsonable(ret)
  41. def rss_sources(self):
  42. return [ p for p in self.c_sources.find({"rss":1}) ]
  43. def insert_article(self,o):
  44. try:
  45. self.c_articles.insert_one(o)
  46. return True
  47. except errors.DuplicateKeyError:
  48. print("Dup")
  49. return False
  50. def purge_error(self):
  51. self.c_articles.remove({'error':True})
  52. def article_exists(self,url):
  53. ret=self.c_articles.find_one({'url':url})
  54. if ret is None:
  55. return False
  56. return True
  57. def lower_arr(self,arr):
  58. return [a.lower() for a in arr]
  59. def get_keywords_list(self):
  60. kw=self.client[self.DB]["keywords"]
  61. return [ k["keyword"] for k in kw.find({}) ]
  62. def get_keywords(self,category):
  63. kw=self.client[self.DB]["keywords"] #FIXME, split on "|" or regex
  64. ret = [ {"weight":k["weight"],
  65. "kw": k["keyword"],
  66. "pattern": self.lower_arr(k["keyword"].split("|")) } for k in kw.find({}) ]#FIXME, categories
  67. return ret