Hugging Face's logo

Spaces:
junaid17
/
cortex-translator
Sleeping

App Files Community
cortex-translator
File size: 6,936 Bytes
7a60e77
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
 
# translator.py

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import torch
from langdetect import detect

MODEL_NAME = "facebook/nllb-200-distilled-600M"

# ===============================
# LANGUAGE MAP
# ===============================
LANGUAGE_MAP = {
    # ===== EXISTING =====
    "english": "eng_Latn", "french": "fra_Latn", "german": "deu_Latn", "spanish": "spa_Latn",
    "hindi": "hin_Deva", "marathi": "mar_Deva", "tamil": "tam_Taml", "telugu": "tel_Telu",
    "kannada": "kan_Knda", "bengali": "ben_Beng", "urdu": "urd_Arab", "arabic": "arb_Arab",
    "persian": "pes_Arab", "japanese": "jpn_Jpan", "chinese": "zho_Hans", "korean": "kor_Hang",
    "russian": "rus_Cyrl", "italian": "ita_Latn", "portuguese": "por_Latn", "dutch": "nld_Latn",
    "swedish": "swe_Latn", "norwegian": "nob_Latn", "danish": "dan_Latn", "finnish": "fin_Latn",
    "polish": "pol_Latn", "czech": "ces_Latn", "slovak": "slk_Latn", "hungarian": "hun_Latn",
    "romanian": "ron_Latn", "bulgarian": "bul_Cyrl", "ukrainian": "ukr_Cyrl", "greek": "ell_Grek",
    "gujarati": "guj_Gujr", "punjabi": "pan_Guru", "malayalam": "mal_Mlym",
    "thai": "tha_Thai", "vietnamese": "vie_Latn", "indonesian": "ind_Latn",
    "turkish": "tur_Latn", "hebrew": "heb_Hebr",

    # ===== KURDISH (IMPORTANT) =====
    "kurdish_kurmanji": "kmr_Latn",   # Northern Kurdish (Latin)
    "kurdish_sorani": "ckb_Arab",     # Central Kurdish (Arabic)

    # ===== SOUTH ASIAN =====
    "nepali": "npi_Deva",
    "sinhala": "sin_Sinh",
    "odia": "ory_Orya",
    "assamese": "asm_Beng",
    "maithili": "mai_Deva",
    "santali": "sat_Olck",

    # ===== SOUTHEAST ASIA =====
    "malay": "zsm_Latn",
    "filipino": "tgl_Latn",
    "khmer": "khm_Khmr",
    "lao": "lao_Laoo",
    "burmese": "mya_Mymr",

    # ===== EAST ASIA =====
    "traditional_chinese": "zho_Hant",
    "mongolian": "mon_Cyrl",

    # ===== CENTRAL ASIA =====
    "kazakh": "kaz_Cyrl",
    "uzbek": "uzn_Latn",
    "tajik": "tgk_Cyrl",
    "kyrgyz": "kir_Cyrl",
    "turkmen": "tuk_Latn",

    # ===== MIDDLE EAST =====
    "pashto": "pbt_Arab",
    "sindhi": "snd_Arab",

    # ===== AFRICAN =====
    "swahili": "swh_Latn",
    "amharic": "amh_Ethi",
    "yoruba": "yor_Latn",
    "igbo": "ibo_Latn",
    "hausa": "hau_Latn",
    "zulu": "zul_Latn",
    "xhosa": "xho_Latn",
    "somali": "som_Latn",
    "afrikaans": "afr_Latn",

    # ===== EUROPE EXTRA =====
    "estonian": "est_Latn",
    "latvian": "lav_Latn",
    "lithuanian": "lit_Latn",
    "icelandic": "isl_Latn",
    "irish": "gle_Latn",
    "welsh": "cym_Latn",
    "albanian": "sqi_Latn",
    "serbian": "srp_Cyrl",
    "croatian": "hrv_Latn",
    "slovenian": "slv_Latn",

    # ===== OTHERS =====
    "latin": "lat_Latn",
    "esperanto": "epo_Latn"
}

ISO_TO_LANGUAGE_KEY = {
    # ===== EXISTING =====
    "en": "english", "fr": "french", "de": "german", "es": "spanish", "hi": "hindi",
    "mr": "marathi", "ta": "tamil", "te": "telugu", "kn": "kannada", "bn": "bengali",
    "ur": "urdu", "ar": "arabic", "fa": "persian", "ja": "japanese", "zh": "chinese",
    "ko": "korean", "ru": "russian", "it": "italian", "pt": "portuguese", "nl": "dutch",
    "sv": "swedish", "no": "norwegian", "da": "danish", "fi": "finnish", "pl": "polish",
    "cs": "czech", "sk": "slovak", "hu": "hungarian", "ro": "romanian", "bg": "bulgarian",
    "uk": "ukrainian", "el": "greek", "gu": "gujarati", "pa": "punjabi", "ml": "malayalam",
    "th": "thai", "vi": "vietnamese", "id": "indonesian", "tr": "turkish", "he": "hebrew",

    # ===== KURDISH =====
    "ku": "kurdish_kurmanji",   # default kurdish
    "ckb": "kurdish_sorani",

    # ===== SOUTH ASIAN =====
    "ne": "nepali",
    "si": "sinhala",
    "or": "odia",
    "as": "assamese",
    "mai": "maithili",
    "sat": "santali",

    # ===== SOUTHEAST ASIA =====
    "ms": "malay",
    "tl": "filipino",
    "km": "khmer",
    "lo": "lao",
    "my": "burmese",

    # ===== CENTRAL ASIA =====
    "kk": "kazakh",
    "uz": "uzbek",
    "tg": "tajik",
    "ky": "kyrgyz",
    "tk": "turkmen",

    # ===== MIDDLE EAST =====
    "ps": "pashto",
    "sd": "sindhi",

    # ===== AFRICAN =====
    "sw": "swahili",
    "am": "amharic",
    "yo": "yoruba",
    "ig": "igbo",
    "ha": "hausa",
    "zu": "zulu",
    "xh": "xhosa",
    "so": "somali",
    "af": "afrikaans",

    # ===== EUROPE EXTRA =====
    "et": "estonian",
    "lv": "latvian",
    "lt": "lithuanian",
    "is": "icelandic",
    "ga": "irish",
    "cy": "welsh",
    "sq": "albanian",
    "sr": "serbian",
    "hr": "croatian",
    "sl": "slovenian",

    # ===== OTHERS =====
    "la": "latin",
    "eo": "esperanto"
}


# ===============================
# LOAD MODEL ONCE
# ===============================
print("Loading translation model...")
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_NAME)
print("Model loaded successfully.")

# ===============================
# HELPERS
# ===============================
def get_nllb_code(lang: str):
    lang = lang.lower().strip()

    if lang in LANGUAGE_MAP:
        return LANGUAGE_MAP[lang]

    if lang in ISO_TO_LANGUAGE_KEY:
        key = ISO_TO_LANGUAGE_KEY[lang]
        return LANGUAGE_MAP[key]

    return "eng_Latn"  # fallback


# ===============================
# MANUAL TRANSLATOR
# ===============================
def standard_translator(text: str, src_lang: str, tgt_lang: str) -> str:
    src_code = get_nllb_code(src_lang)
    tgt_code = get_nllb_code(tgt_lang)

    tokenizer.src_lang = src_code
    inputs = tokenizer(text, return_tensors="pt")

    with torch.no_grad():
        output = model.generate(
            **inputs,
            forced_bos_token_id=tokenizer.convert_tokens_to_ids(tgt_code),
            max_length=512
        )

    return tokenizer.decode(output[0], skip_special_tokens=True)


# ===============================
# AUTO TRANSLATOR
# ===============================
def smart_auto_translator(text: str, target_lang: str):
    detected_iso = detect(text)
    detected_lang_key = ISO_TO_LANGUAGE_KEY.get(detected_iso, detected_iso)

    src_code = get_nllb_code(detected_iso)
    tgt_code = get_nllb_code(target_lang)

    tokenizer.src_lang = src_code
    inputs = tokenizer(text, return_tensors="pt")

    with torch.no_grad():
        output = model.generate(
            **inputs,
            forced_bos_token_id=tokenizer.convert_tokens_to_ids(tgt_code),
            max_length=512
        )

    translated_text = tokenizer.decode(output[0], skip_special_tokens=True)

    return {
        "translated_text": translated_text,
        "detected_iso": detected_iso,
        "detected_language": detected_lang_key
    }