flash64/biogpt-testing
1
1from bs4 import BeautifulSoup2import requests3 4 5lang_ids = {6 "Afrikaans": "af",7 "Amharic": "am",8 "Arabic": "ar",9 "Asturian": "ast",10 "Azerbaijani": "az",11 "Bashkir": "ba",12 "Belarusian": "be",13 "Bulgarian": "bg",14 "Bengali": "bn",15 "Breton": "br",16 "Bosnian": "bs",17 "Catalan": "ca",18 "Cebuano": "ceb",19 "Czech": "cs",20 "Welsh": "cy",21 "Danish": "da",22 "German": "de",23 "Greeek": "el",24 "English": "en",25 "Spanish": "es",26 "Estonian": "et",27 "Persian": "fa",28 "Fulah": "ff",29 "Finnish": "fi",30 "French": "fr",31 "Western Frisian": "fy",32 "Irish": "ga",33 "Gaelic": "gd",34 "Galician": "gl",35 "Gujarati": "gu",36 "Hausa": "ha",37 "Hebrew": "he",38 "Hindi": "hi",39 "Croatian": "hr",40 "Haitian": "ht",41 "Hungarian": "hu",42 "Armenian": "hy",43 "Indonesian": "id",44 "Igbo": "ig",45 "Iloko": "ilo",46 "Icelandic": "is",47 "Italian": "it",48 "Japanese": "ja",49 "Javanese": "jv",50 "Georgian": "ka",51 "Kazakh": "kk",52 "Central Khmer": "km",53 "Kannada": "kn",54 "Korean": "ko",55 "Luxembourgish": "lb",56 "Ganda": "lg",57 "Lingala": "ln",58 "Lao": "lo",59 "Lithuanian": "lt",60 "Latvian": "lv",61 "Malagasy": "mg",62 "Macedonian": "mk",63 "Malayalam": "ml",64 "Mongolian": "mn",65 "Marathi": "mr",66 "Malay": "ms",67 "Burmese": "my",68 "Nepali": "ne",69 "Dutch": "nl",70 "Norwegian": "no",71 "Northern Sotho": "ns",72 "Occitan": "oc",73 "Oriya": "or",74 "Panjabi": "pa",75 "Polish": "pl",76 "Pushto": "ps",77 "Portuguese": "pt",78 "Romanian": "ro",79 "Russian": "ru",80 "Sindhi": "sd",81 "Sinhala": "si",82 "Slovak": "sk",83 "Slovenian": "sl",84 "Somali": "so",85 "Albanian": "sq",86 "Serbian": "sr",87 "Swati": "ss",88 "Sundanese": "su",89 "Swedish": "sv",90 "Swahili": "sw",91 "Tamil": "ta",92 "Thai": "th",93 "Tagalog": "tl",94 "Tswana": "tn",95 "Turkish": "tr",96 "Ukrainian": "uk",97 "Urdu": "ur",98 "Uzbek": "uz",99 "Vietnamese": "vi",100 "Wolof": "wo",101 "Xhosa": "xh",102 "Yiddish": "yi",103 "Yoruba": "yo",104 "Chinese": "zh",105 "Zulu": "zu",106}107 