Find Turkish Newspaper for a Date; - Updated. Old site is now has payment wall. I found free

gazete.py

# Example usage python3 gaste.py "23 Nisan 1935"

import concurrent.futures
import sys
import urllib.parse

import requests

gazeteler = [
   
   "aksam",
   "acik%20soz",
   "anadolu",
   "aravelk",
   "aydin",
   "beyoglu",
   "borsa",
   "bugun(siyasi,iktisadi,ictimai,gundelik%20gazete)/bugun(siyasi,iktisadi,ictimai,gundelik%20gazete)",
   "cumhuriyet",
   "dogu",
   "en%20son%20dakika",
   "en%20son%20havadis",
   "haber(aksam%20postasi)",
   "hakikat",
   "hakimiyeti%20milliye",
   "hakkin%20sesi",
   "halkin%20dili",
   "halkin%20sesi",
   "ikdam%20(halk%20gazetesi)",
   "ikdam(cumhuriyet%20icin,%20halk%20icin)",
   "ikdam(sabah%20postasi)",
   "izmir%20postasi",
   "jamanak",
   "kurun",
   "milliyet",
   "munakasa",
   "piyasa%20cetveli",
   "savas",
   "son%20dakika",
   "son%20posta",
    "son%20telgraf",
    "son%20saat",
    "tan",
    "tasviri%20efkar",
    "turk%20sozu",
    "turk%20dili",
    "turkische%20post",
    "ulus",
    "ulus%20sesi",
    "ulusal%20birlik",
    "vakit",
    "vatan",
    "yarin",
    "yeni%20asir",
    "yeni%20mersin",
    "yeni%20sabah",
    "yenigun",
    "yeniyol"
]


# Mapping of Turkish month names to their lowercase values for URL
turkish_months = {
    "Ocak": "ocak",
    "Şubat": "subat",
    "Mart": "mart",
    "Nisan": "nisan",
    "Mayıs": "mayis",
    "Haziran": "haziran",
    "Temmuz": "temmuz",
    "Ağustos": "agustos",
    "Eylül": "eylul",
    "Ekim": "ekim",
    "Kasım": "kasim",
    "Aralık": "aralik",
}



# https://nek.istanbul.edu.tr/ekos/GAZETE/izmir%20postasi/izmir%20postasi_1934/izmir%20postasi_1934_temmuz_/izmir%20postasi_1934_temmuz_2_.pdf
def parse_turkish_date(date_str):
    day, month_name, year = date_str.split()
    month_lower = turkish_months.get(month_name)
    if not month_lower:
        raise ValueError(f"Invalid Turkish month name: {month_name}")
    return {
        'day': int(day),
        'month': month_lower,
        'year': year
    }


def get_date():
    if len(sys.argv) > 1:
        return parse_turkish_date(sys.argv[1])
    else:
        return parse_turkish_date(input("Gün Ay Yıl şeklinde tarih giriniz: "))


# Using the function
tarih = get_date()


def fetch_gazete(gazete):
    
    # URL encode the newspaper name to handle spaces
    gazete_encoded = urllib.parse.quote(gazete)
    
    # Build the new URL format
    url = f"https://nek.istanbul.edu.tr/ekos/GAZETE/{gazete_encoded}//{gazete_encoded}_{tarih['year']}/{gazete_encoded}_{tarih['year']}_{tarih['month']}_/{gazete_encoded}_{tarih['year']}_{tarih['month']}_{tarih['day']}_.pdf"
    

    
    r = requests.head(url, timeout=3)
    print(r)
    if r.status_code == 200:
        print(url)
    else:
        # Optional: print failed attempts forçç√ debugging
        # print(f"Failed to fetch {gazete}: {r.status_code}")
        pass


# Adjust the number of threads in the pool
pool_size = 10

with concurrent.futures.ThreadPoolExecutor(max_workers=pool_size) as executor:
    executor.map(fetch_gazete, gazeteler)

print("Bitti")