import requests from bs4 import BeautifulSoup import urllib import datetime from shutil import copyfile OUTPUTFILE = "../ofu-food.html" # OUTPUTFILE = "/media/data_1/www/html/ofu-food.html" CSSFILE_SRC = "../bootstrap-4.0.0-beta-dist/css/bootstrap.css" CSSFILE = "bootstrap.css" LINK_FEKI_MENSA = "https://www.studentenwerk-wuerzburg.de/bamberg/essen-trinken/speiseplaene.html?tx_thmensamenu_pi2%5Bmensen%5D=3&tx_thmensamenu_pi2%5Baction%5D=show&tx_thmensamenu_pi2%5Bcontroller%5D=Speiseplan&cHash=c3fe5ebb35e5fba3794f01878e798b7c" LINK_AUSTR_MENSA = "https://www.studentenwerk-wuerzburg.de/bamberg/essen-trinken/speiseplaene.html?tx_thmensamenu_pi2%5Bmensen%5D=2&tx_thmensamenu_pi2%5Baction%5D=show&tx_thmensamenu_pi2%5Bcontroller%5D=Speiseplan&cHash=511e047953ee1370c3b82c11a04624bb" LINK_ERBA_CAFETE = "https://www.studentenwerk-wuerzburg.de/bamberg/essen-trinken/sonderspeiseplaene/cafeteria-erba-insel.html" LINK_MARKUS_CAFETE = "https://www.studentenwerk-wuerzburg.de/bamberg/essen-trinken/sonderspeiseplaene/cafeteria-markusplatz.html" LINK_FEKIDE_GUIDE = "https://www.feki.de/happyhour" def getHtmlHeader(): head = '' head += '' head += '' head += '' head += '' head += 'Essen an der OFU' head += '' return head def getExecuteTime(): return "

Last execute: " + datetime.datetime.today().strftime("%d.%m.%Y") + "

" def getFoodHtml(erbaHtml, markusHtml, austrHtml, fekiHtml, happyHourHtml): html = '' html += '' html += getHtmlHeader() html += '' html += getExecuteTime() html += '
' html += str(erbaHtml) html += str(markusHtml) html += str(austrHtml) html += str(fekiHtml) html += str(happyHourHtml) html += '
' html += '' html += '' return html def getFoodDiv(html, divClass): soup = BeautifulSoup(html, "lxml") return soup.find("div", {"class": divClass}) def getHappyHourGuideFood(html): soup = BeautifulSoup(html, "lxml") return soup.find("div", {"id": "food"}) def getWebPages(): erbaCafete = requests.get(LINK_ERBA_CAFETE).content markusCafete = requests.get(LINK_MARKUS_CAFETE).content austrMensa = requests.get(LINK_AUSTR_MENSA).content fekiMensa = requests.get(LINK_FEKI_MENSA).content happyHourGuide = requests.get(LINK_FEKIDE_GUIDE).content return (erbaCafete, markusCafete, austrMensa, fekiMensa, happyHourGuide) def getSpecificParts(): erbaCafete, markusCafete, austrMensa, fekiMensa, happyHourGuide = getWebPages() erbaCafeteFood = getFoodDiv(erbaCafete, "csc-default") markusCafeteFood = getFoodDiv(markusCafete, "csc-default") austrMensaFood = getFoodDiv(austrMensa, "day") fekiMensaFood = getFoodDiv(fekiMensa, "day") happyHourGuideFood = getHappyHourGuideFood(happyHourGuide) return (erbaCafeteFood, markusCafeteFood, austrMensaFood, fekiMensaFood, happyHourGuideFood) def writeHtml(html): with open(OUTPUTFILE, "w") as file: file.write(html) def stripCafeteFood(html): soup = BeautifulSoup(str(html), "lxml") food = soup.find_all("p", {"class": "bodytext"}) return food[(food.__len__() - 2)] def htmlErbaCafete(food): html = '

Cafeteria Erba-Insel

' html += str(food) return html def htmlMarkusCafete(food): html = '

Cafeteria Markusplatz

' html += str(food) return html def htmlHappyHour(food): html = '

Feki.de Happy Hour Guide

' html += str(food) return html def getFoodMensa(austrMensaFood, title): html = '
' html += '

' + title + '

' soup = BeautifulSoup(str(austrMensaFood), "lxml") foodDay = soup.find("div", {"class": "day"}).h5 html += str(foodDay) foods = soup.find_all("article", {"class": "menu"}) for food in foods: soupSingleFood = BeautifulSoup(str(food), "lxml") singleFood = soupSingleFood.find("div", {"class": "title"}) alergenLink = soupSingleFood.find("div", {"class": "additnr"}).a html += str(singleFood) html += str(alergenLink) html += '
' return html def main(): (erbaCafeteFood, markusCafeteFood, austrMensaFood, fekiMensaFood, happyHourGuideFood) = getSpecificParts() erbaCafeteFood = stripCafeteFood(erbaCafeteFood) erbaCafeteFood = htmlErbaCafete(erbaCafeteFood) markusCafeteFood = stripCafeteFood(markusCafeteFood) markusCafeteFood = htmlMarkusCafete(markusCafeteFood) happyHourGuideFood = htmlHappyHour(happyHourGuideFood) austrMensaFood = getFoodMensa(austrMensaFood, "Mensa Austraße") fekiMensaFood = getFoodMensa(fekiMensaFood, "Feki Mensa") html = getFoodHtml(erbaCafeteFood, markusCafeteFood, austrMensaFood, fekiMensaFood, happyHourGuideFood) writeHtml(html) main() copyfile(CSSFILE_SRC, CSSFILE)