FP 2022-12-12

# AmBotTools.py
import certifi
import ssl
from urllib.request import urlopen, Request
from http.client import HTTPResponse

import bs4
from bs4 import BeautifulSoup

class AmBotTools:

    def __init__(self):
        self.mCAFile = certifi.where()
        self.mSSLContext =\
            ssl.create_default_context(
                cafile=self.mCAFile
            )
    # def __init__

    def consumeUrl(
        self,
        pUrl:str,
        pbPreferBytes:bool=False,
        pStrEncoding="UTF-8"
    ):
        response:HTTPResponse =\
            urlopen(
                url = pUrl,
                context=self.mSSLContext
            )
        theBytes = response.read()
        if(pbPreferBytes):
            return theBytes
        else:
            strResponse = str(
                theBytes,
                pStrEncoding
            )
            return strResponse
        # if-else
    # def consumeUrl

    def getAnchors(
        self,
        pUrl:str
    ):
        listOfFoundAnchors = list()  #

        strConteudo = \
            self.consumeUrl(
                pUrl=pUrl
            )
        if (strConteudo):
            bs = BeautifulSoup(
                strConteudo,
                "html5lib"
            )
            if (bs):
                theAs = bs.findAll("a")

                if (theAs):
                    for anchor in theAs:
                        texto:str = anchor.title
                        bThereIsHref = "href" in anchor.attrs.keys()
                        if (bThereIsHref):
                            href = anchor.attrs["href"]
                        else:
                            href = ""
                        # if-else

                        listOfFoundAnchors.append(
                            {
                            "anchor":texto,
                            "href":href
                            }
                        )
                    # for every anchor
                # if there are anchors
            # if it was possible to get a bs object
        # if there is content

        return listOfFoundAnchors
    # def getAnchors

    """
    escreva um método que permita filtrar
    uma lista de anchors
    incluindo no retorno apenas aquelas
    cujo atributo href
    contenha certa expressão.
    Por exemplo:
    AmBotTools.getFilteredByHrefAnchors(
        theAnchors, # uma lista
        "4cdn.org" # uma frase de filtro
    )
    """
# class AmBotTools


*****

# bot4chan.py
# um bot para consumir a rede social 4chan.org
from AmBotTools import AmBotTools

import bs4
from bs4 import BeautifulSoup

"""
em HTML todo o conteúdo está na forma
<marca a1=v1 a2=v2>conteúdo</marca>
<p class="info">A taxa juro XPTO é 5.6%</p>

<a href="endereço">Texto do âncora</a>
"""

class Bot4Chan:
    def __init__(
        self,
        pStrBoard:str
    ):
        self.mBoard = pStrBoard
    # def __init__

    # https://boards.4channel.org/<nome da board/
    def getUrlForBoard(self):
        strUrl =\
        f"https://boards.4channel.org/{self.mBoard}/"
        return strUrl
    # def getUrlForBoard

    def getBoardAnchors(self):
        bot = AmBotTools()
        listOfDictsEachOneIsAnAnchor =\
            bot.getAnchors(
                pUrl = self.getUrlForBoard()
            )
        return listOfDictsEachOneIsAnAnchor
    # def getBoardAnchors
# class Bot4Chan

bot = Bot4Chan("an")
theAnchors = bot.getBoardAnchors()
print (theAnchors)