Upload 11 files

Dockerfile.txt

@@ -0,0 +1,17 @@
+# Dockerfile
+FROM python:3.10-slim
+
+WORKDIR /app
+
+# Install dependencies
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy the app files
+COPY . .
+
+# Expose the port FastAPI will run on
+EXPOSE 7860
+
+# Run the application
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -0,0 +1,34 @@
+import os
+from fastapi import FastAPI, Body, File, UploadFile, Request
+from fastapi.responses import HTMLResponse, RedirectResponse
+from fastapi.staticfiles import StaticFiles
+from fastapi.templating import Jinja2Templates
+
+from models.Flashcardd import Flashcard
+from translations.translate import load_model_and_tokenizer, translate
+
+
+app = FastAPI()
+
+
+@app.get("/")
+async def home():
+ return {"message": "hola-mondo"}
+
+
+@app.post("/flashcards/create_flashcard")
+async def create_flashcard(new_flashcard=Body()) -> Flashcard:
+ print(f"new flashcard: {new_flashcard}")
+ from_lang = new_flashcard["from"]
+ to_lang = new_flashcard["to"]
+ model, tokenizer = load_model_and_tokenizer(from_lang=from_lang, to_lang=to_lang)
+ translation = translate(new_flashcard["word"], model, tokenizer)
+
+ return Flashcard(
+ name=new_flashcard["word"],
+ translation=translation,
+ # sample_sentence=f"{new_flashcard['word']}",
+ sample_sentence="notes..",
+ # "translation": f"translation: {new_flashcard['word']}",
+ # "translation": translation,
+ )

models/Flashcardd.py

@@ -0,0 +1,10 @@
+from pydantic import BaseModel, constr
+from typing import Union, Optional
+
+
+class Flashcard(BaseModel):
+ name: constr(max_length=100, min_length=5) # type: ignore
+ translation: Optional[str] = ""
+ sample_sentence: Optional[str] = ""
+
+ 

requirements.txt

@@ -0,0 +1,9 @@
+gradio
+fastapi==0.112.0
+sentencepiece==0.2.0
+numpy==1.23.1
+sacremoses==0.1.1
+tokenizers==0.19.1
+transformers==4.43.3
+huggingface-hub==0.24.5
+torch

translations/model_name_mapping.py

@@ -0,0 +1,47 @@
+from dataclasses import dataclass, field
+from typing import Dict, Literal, Tuple
+from pydantic import BaseModel, ValidationError
+from enum import Enum
+
+
+class ModelNameMapping(BaseModel):
+ data: Dict[Tuple, str] = field(default_factory=dict)
+
+ def __getitem__(self, key: Tuple) -> str:
+ return self.data[key]
+
+ def get(self, key: Tuple) -> str:
+ return self.data.get(key, "")
+
+
+class LanguagesEnum(Enum):
+ es = "es"
+ en = "en"
+ ca = "ca"
+ pt = "pt"
+
+
+es_to_en: Tuple = (LanguagesEnum.es.value, LanguagesEnum.en.value)
+es_to_ca: Tuple = (LanguagesEnum.es.value, LanguagesEnum.ca.value)
+es_to_po: Tuple = (LanguagesEnum.es.value, LanguagesEnum.pt.value)
+
+en_to_es: Tuple = (LanguagesEnum.en.value, LanguagesEnum.es.value)
+en_to_po: Tuple = (LanguagesEnum.en.value, LanguagesEnum.pt.value)
+
+ca_to_es: Tuple = (LanguagesEnum.ca.value, LanguagesEnum.es.value)
+
+po_to_en: Tuple = (LanguagesEnum.pt.value, LanguagesEnum.en.value)
+po_to_es: Tuple = (LanguagesEnum.pt.value, LanguagesEnum.es.value)
+
+models = {
+ es_to_en: "Helsinki-NLP/opus-mt-es-en",
+ es_to_ca: "Helsinki-NLP/opus-mt-es-ca",
+ en_to_es: "Helsinki-NLP/opus-mt-en-es",
+ en_to_po: "Helsinki-NLP/opus-mt-tc-big-en-pt",
+ ca_to_es: "Helsinki-NLP/opus-mt-ca-es",
+ po_to_en: "Helsinki-NLP/opus-mt-pt-en",
+ po_to_es: "Helsinki-NLP/opus-mt-pt-es",
+}
+
+
+MODEL_NAME_MAPPING = ModelNameMapping(data=models)

translations/translate.py

@@ -0,0 +1,45 @@
+from transformers import MarianMTModel, MarianTokenizer
+from translations.model_name_mapping import MODEL_NAME_MAPPING
+
+
+def load_model_and_tokenizer(from_lang: str, to_lang: str):
+ print(f"load_model_and_tokenizer from: {from_lang}, to: {to_lang}")
+
+ model_name = MODEL_NAME_MAPPING.get((from_lang, to_lang))
+ print(f"model_name: {model_name}")
+ tokenizer = MarianTokenizer.from_pretrained(model_name)
+ model = MarianMTModel.from_pretrained(model_name)
+ return model, tokenizer
+
+
+def translate(text, model, tokenizer):
+ translated = model.generate(**tokenizer(text, return_tensors="pt", padding=True))
+ translated_text = [
+ tokenizer.decode(t, skip_special_tokens=True) for t in translated
+ ]
+ return translated_text[0]
+
+
+
+def test_translations(from_lang, to_lang, text_to_translate):
+ model, tokenizer = load_model_and_tokenizer(from_lang=from_lang, to_lang=to_lang)
+ translation = translate(text_to_translate, model, tokenizer)
+ print(f"Translated text from: {from_lang}, to: {to_lang}, translation: {translation}")
+
+
+if __name__ == "__main__":
+ # text_to_translate = "hola amigos, tengo hambre"
+
+ # test_translations(from_lang="es", to_lang="en", text_to_translate=text_to_translate)
+ # test_translations(from_lang="es", to_lang="ca", text_to_translate=text_to_translate)
+ # test_translations(from_lang="es", to_lang="po", text_to_translate=text_to_translate)
+ 
+ text_to_translate = "hello friends, who's hungry?"
+ test_translations(from_lang="en", to_lang="es", text_to_translate=text_to_translate)
+ # test_translations(from_lang="en", to_lang="ca", text_to_translate=text_to_translate)
+ test_translations(from_lang="en", to_lang="pt", text_to_translate=text_to_translate)
+ # test_translations(from_lang="es", to_lang="po", text_to_translate=text_to_translate)
+ 
+ 
+ text_to_translate = "un cafè sense sucre i amb llet, si us plau"
+ test_translations(from_lang="ca", to_lang="es", text_to_translate=text_to_translate)