alexis2/db.py

from pymongo import MongoClient,errors
from bson.objectid import ObjectId
import re

class db():
    DB="alexis"
    def __init__(self):
        self.client = MongoClient('localhost', 27017)
        self.c_sources=self.client[self.DB]["sources"]
        self.c_articles=self.client[self.DB]["articles"]

    def jsonable(self, el):
        if el is None or "_id" not in el:
            return el

        el["_id"]=str(el["_id"])
        return el


    def articles(self, id=None):
        if id is None:
            return [ self.jsonable(p) for p in self.c_articles.find() ]

        ret=self.c_articles.find_one({"_id":ObjectId(id)})
        return self.jsonable(ret)

    def sources(self,id=None):
        if id is None:
            return [ self.jsonable(p) for p in self.c_sources.find() ]

        ret=self.c_sources.find_one({"_id":ObjectId(id)})
        return self.jsonable(ret)

    def rss_sources(self):
        return [ p for p in self.c_sources.find({"rss":1}) ]


    def insert_article(self,o):
        try:
            self.c_articles.insert_one(o)
            return True
        except errors.DuplicateKeyError:
            print("Dup")
            return False

    def article_exists(self,url):
        ret=self.c_articles.find_one({'url':url})
        if ret is None:
            return False
        return True 

    def lower_arr(self,arr):
        return [a.lower() for a in arr]
    def get_keywords(self,category):
        kw=self.client[self.DB]["keywords"]
        #FIXME, split on "|" or regex
        ret = [ {"weight":k["weight"],
                "kw": k["keyword"],
                "pattern": self.lower_arr(k["keyword"].split("|")) } for k in  kw.find({}) ]#FIXME, categories
        return ret