Read All Region PDF Document using tabula

medium_8547b733ae6a_07_get_all_province_region_data.py

import tabula
pdf_path = "PMDN 72 TH 2019+lampiran.pdf"

output_folder = "csv/"

region_list = [
    # sumatera
    {"file": output_folder + "11-aceh.csv", "pages": "8-274"},
    {"file": output_folder + "12-sumut.csv", "pages": "287-582"},
    {"file": output_folder + "13-sumbar.csv", "pages": "601-653"},
    {"file": output_folder + "14-riau.csv", "pages": "662-786"},
    {"file": output_folder + "15-jambi.csv", "pages": "794-864"},
    {"file": output_folder + "16-sumsel.csv", "pages": "872-1022"},
    {"file": output_folder + "17-bengkulu.csv", "pages": "1034-1106"},
    {"file": output_folder + "18-lampung.csv", "pages": "1114-1233"},
    {"file": output_folder + "19-babel.csv", "pages": "1244-1259"},
    {"file": output_folder + "21-kepri.csv", "pages": "1264-1284"},
    # jawa
    {"file": output_folder + "31-jakarta.csv", "pages": "1290-1299"},
    {"file": output_folder + "32-jabar.csv", "pages": "1304-1524"},
    {"file": output_folder + "33-jateng.csv", "pages": "1547-1823"},
    {"file": output_folder + "34-jogja.csv", "pages": "1843-1858"},
    {"file": output_folder + "35-jatim.csv", "pages": "1864-2144"},
    {"file": output_folder + "36-banten.csv", "pages": "2166-2228"},
    # bali & nusa tenggara
    {"file": output_folder + "51-bali.csv", "pages": "2236-2260"},
    {"file": output_folder + "52-ntb.csv", "pages": "2265-2307"},
    {"file": output_folder + "53-ntt.csv", "pages": "2315-2468"},
    # kalimantan
    {"file": output_folder + "61-kalbar.csv", "pages": "2481-2565"},
    {"file": output_folder + "62-kalteng.csv", "pages": "2574-2642"},
    {"file": output_folder + "63-kalsel.csv", "pages": "2650-2727"},
    {"file": output_folder + "64-kaltim.csv", "pages": "2735-2777"},
    {"file": output_folder + "65-kaltara.csv", "pages": "2784-2801"},
    # sulawesi
    {"file": output_folder + "71-sulut.csv", "pages": "2806-2893"},
    {"file": output_folder + "72-sulteng.csv", "pages": "2902-3004"},
    {"file": output_folder + "73-sulsel.csv", "pages": "3014-3134"},
    {"file": output_folder + "74-sultra.csv", "pages": "3147-3268"},
    {"file": output_folder + "75-gorontalo.csv", "pages": "3280-3315"},
    {"file": output_folder + "76-sulbar.csv", "pages": "3321-3348"},
    # maluku
    {"file": output_folder + "81-maluku.csv", "pages": "3353-3418"},
    {"file": output_folder + "82-malut.csv", "pages": "3426-3486"},
    # papua
    {"file": output_folder + "91-papua.csv", "pages": "3494-3789"},
    {"file": output_folder + "92-pabar.csv", "pages": "3810-3924"},
]

for region in region_list:
    tabula.convert_into(
        input_path=pdf_path,
        output_path=region["file"],
        output_format="csv",
        pages=region["pages"]
    )
    print("generate file " + region["file"] + '.')