Example #1
0
class SWDS:
    def __init__(self):
        self.db_Manager = DbManager()

    def get_json(url):
        resp = requests.get(url)
        if resp.status_code != 200:
            return None
        return json.loads(resp.text)

    def get_entity(url, obj):
        json_data = get_json(url)

        entity = obj()
        entity.parse_json(json_data)

        self.db_Manager.save(entity)

    def populate(obj, next_url):
        while next_url:
            group_data = get_json(next_url)
            results = group_data['results']

            for row in results:
                url = row['url']
                results = self.db_Manager.query(obj).filter(
                    obj.url == url).all()

                if len(results) == 0:
                    save_entity(url, obj)
                    # print(url)

            next_url = group_data['next']
Example #2
0
class DCK():
    def __init__(self):
        self.db_Manager = DbManager()

    def get_json(self, url):
        resp = requests.get(url)
        if resp.status_code!=200:
            return None
        return json.loads(resp.text)

    def get_entity(self, obj, url):
        results = self.db_Manager.open().query(obj).filter(obj.api == url).all()
        
        json_data= self.get_json(url)
        entity = obj()
        entity.parse_json(json_data)
        
        if len(results)==0:
            self.db_Manager.save(entity)
        
        return entity

    def find_end(self, endChar, url):
        for index, c in enumerate(url[::-1]):
            if c==endChar:
                return len(url)-index-1
        
        return None

    def find_base_url(self, url):
        index=self.find_end('/', url)

        if index:
            return url[: index+1]
        else:
            return None

    def populate_table(self, obj, next_url):
        base_url=self.find_base_url(next_url)

        if base_url:
            group_data=self.get_json(next_url)
            page_id=1
            
            while group_data:
                print(next_url)
                for result in group_data:
                    single_url = result['api']
                    print(single_url)

                    self.get_entity(obj, single_url)
                
                page_id=page_id+1
                next_url=base_url+str(page_id)
                group_data=self.get_json(next_url)

                # group_data=None
Example #3
0
class CDC():
    def __init__(self):
        self.db_Manager = DbManager()

    def get_data(self, url):
        resp = requests.get(url, headers=DEFAULT_HEADER)
        if resp.status_code != 200:
            print(resp.status_code)
            return None
        return json.loads(resp.text)

    def get_entity(self, obj, url):
        json_data = self.get_data(url)
        entity = obj()

        try:
            entity.parse_json(json_data)
        except:
            entity = None
            print("bad link: {}".format(url))
            return entity

        results = self.db_Manager.open().query(obj).filter(
            obj.url == url).all()
        if len(results) == 0:
            self.db_Manager.save(entity)
            print("new add: {}".format(entity.url))
        else:
            print("already added: {}".format(entity.url))

        return entity

    def populate_committee(self, congress, chamber):

        chamber = chamber.lower()
        url = API_BASE + "{}/{}/committees.json".format(congress, chamber)

        json_data = self.get_data(url)

        try:
            results = json_data['results'][0]
        except:
            print("Bad Link")
            return "Bad Link"

        committee_list = [c['api_uri'] for c in results['committees']]
        subcommittee_list = [
            c['subcommittees'] for c in results['committees']
            if c.get('subcommittees')
        ]
        sc_url_list = [sc for sc in subcommittee_list if len(sc) > 0]

        for sc_url in sc_url_list:
            for sc in sc_url:
                pprint(sc)
                self.get_entity(SubCommittee, sc['api_uri'])

        for com_url in committee_list:
            self.get_entity(Committee, com_url)

        return "Good Link"

    def populate_members(self, congress, chamber):

        chamber = chamber.lower()
        url = API_BASE + "{}/{}/members.json".format(congress, chamber)

        json_data = self.get_data(url)

        try:
            results = json_data['results'][0]
        except:
            return "Bad Link"

        member_list = [m['id'] for m in results['members']]

        for mid in member_list:
            member_url = API_BASE + "members/{}.json".format(mid)
            self.get_entity(Member, member_url)

        return "Good Link"

    def populate_house(self, congress):
        url = API_BASE + "{}/{}/members.json".format(congress, "house")
        self.get_entity(Congress, url)

    def populate_senate(self, congress):
        url = API_BASE + "{}/{}/members.json".format(congress, "senate")
        self.get_entity(Congress, url)
Example #4
0
class SWDS:
    def __init__(self):
        self.db = DbManager()

    def get_json(self, url):
        print("GET\t<{}>".format(url))
        response = requests.get(url)
        return json.loads(response.text)

    def get_person(self, i):
        person_url = 'https://swapi.co/api/people/{}/'.format(i)
        results = self.db.open().query(Person).filter(
            Person.url == person_url).all()
        if len(results) == 0:

            json_data = self.get_json(person_url)

            person = Person()
            if 'detail' not in json_data:
                person.parse_person(json_data)

                return self.db.save(person)
        else:
            return results[0]

    def get_planet(self, i):
        planet_url = 'https://swapi.co/api/planets/{}/'.format(i)
        results = self.db.open().query(Planet).filter(
            Planet.url == planet_url).all()
        if len(results) == 0:

            json_data = self.get_json(planet_url)

            planet = Planet()
            if 'detail' not in json_data:
                planet.parse_planet(json_data)

                self.db.save(planet)
        else:
            return results[0]

    def get_vehicle(self, i):
        vehicle_url = 'https://swapi.co/api/vehicles/{}/'.format(i)
        results = self.db.open().query(Vehicle).filter(
            Vehicle.url == vehicle_url).all()
        if len(results) == 0:

            json_data = self.get_json(vehicle_url)

            vehicle = Vehicle()
            if 'detail' not in json_data:
                vehicle.parse_vehicle(json_data)

                self.db.save(vehicle)

        else:
            return results[0]

    def get_starship(self, i):
        starship_url = 'https://swapi.co/api/starships/{}/'.format(i)
        results = self.db.open().query(Starship).filter(
            Starship.url == starship_url).all()
        if len(results) == 0:

            json_data = self.get_json(starship_url)

            starship = Starship()
            if 'detail' not in json_data:
                starship.parse_starship(json_data)

                self.db.save(starship)

        else:
            return results[0]

    def get_species(self, i):
        species_url = 'https://swapi.co/api/species/{}/'.format(i)
        results = self.db.open().query(Species).filter(
            Species.url == species_url).all()
        if len(results) == 0:

            json_data = self.get_json(species_url)

            species = Species()
            if 'detail' not in json_data:
                species.parse_species(json_data)

                self.db.save(species)

        else:
            return results[0]