app.py

import os
import random
from statistics import mean
from typing import Iterator, Union, Any
import fasttext
import gradio as gr
from dotenv import load_dotenv
from huggingface_hub import hf_hub_download
from huggingface_hub.utils import logging
from toolz import concat, groupby, valmap
from pathlib import Path

logger = logging.get_logger(__name__)
load_dotenv()

DEFAULT_FAST_TEXT_MODEL = "laurievb/OpenLID"

def load_model(repo_id: str) -> fasttext.FastText._FastText:
    model_path = hf_hub_download(repo_id, filename="model.bin")
    return fasttext.load_model(model_path)

def yield_clean_rows(rows: Union[list[str], str], min_length: int = 3) -> Iterator[str]:
    for row in rows:
        if isinstance(row, str):
            # split on lines and remove empty lines
            line = row.split("\n")
            for line in line:
                if line:
                    yield line
        elif isinstance(row, list):
            try:
                line = " ".join(row)
                if len(line) < min_length:
                    continue
                else:
                    yield line
            except TypeError:
                continue

FASTTEXT_PREFIX_LENGTH = 9  # fasttext labels are formatted like "__label__eng_Latn"

# Load the model
Path("code/models").mkdir(parents=True, exist_ok=True)
model = fasttext.load_model(
    hf_hub_download(
        "facebook/fasttext-language-identification",
        "model.bin",
        cache_dir="code/models",
        local_dir="code/models",
        local_dir_use_symlinks=False,
    )
)

def model_predict(inputs: str, k=1) -> list[dict[str, float]]:
    predictions = model.predict(inputs, k=k)
    return [
        {"label": label[FASTTEXT_PREFIX_LENGTH:], "score": prob}
        for label, prob in zip(predictions[0], predictions[1])
    ]

def get_label(x):
    return x.get("label")

def get_mean_score(preds):
    return mean([pred.get("score") for pred in preds])

def filter_by_frequency(counts_dict: dict, threshold_percent: float = 0.2):
    """Filter a dict to include items whose value is above `threshold_percent`"""
    total = sum(counts_dict.values())
    threshold = total * threshold_percent
    return {k for k, v in counts_dict.items() if v >= threshold}

def simple_predict(text, num_predictions=3):
    """Simple language detection function for Gradio interface"""
    if not text or not text.strip():
        return {"error": "Please enter some text for language detection."}
    
    try:
        # Clean the text
        cleaned_lines = list(yield_clean_rows([text]))
        if not cleaned_lines:
            return {"error": "No valid text found after cleaning."}
        
        # Get predictions for each line
        all_predictions = []
        for line in cleaned_lines:
            predictions = model_predict(line, k=num_predictions)
            all_predictions.extend(predictions)
        
        if not all_predictions:
            return {"error": "No predictions could be made."}
        
        # Group predictions by language
        predictions_by_lang = groupby(get_label, all_predictions)
        language_counts = valmap(len, predictions_by_lang)
        
        # Calculate average scores for each language
        language_scores = valmap(get_mean_score, predictions_by_lang)
        
        # Format results
        results = {
            "detected_languages": dict(language_scores),
            "language_counts": dict(language_counts),
            "total_predictions": len(all_predictions),
            "text_lines_analyzed": len(cleaned_lines)
        }
        
        return results
        
    except Exception as e:
        return {"error": f"Error during prediction: {str(e)}"}

def batch_predict(text, threshold_percent=0.2):
    """More advanced prediction with filtering"""
    if not text or not text.strip():
        return {"error": "Please enter some text for language detection."}
    
    try:
        # Clean the text
        cleaned_lines = list(yield_clean_rows([text]))
        if not cleaned_lines:
            return {"error": "No valid text found after cleaning."}
        
        # Get predictions
        predictions = [model_predict(line) for line in cleaned_lines]
        predictions = [pred for pred in predictions if pred is not None]
        predictions = list(concat(predictions))
        
        if not predictions:
            return {"error": "No predictions could be made."}
        
        # Group and filter
        predictions_by_lang = groupby(get_label, predictions)
        language_counts = valmap(len, predictions_by_lang)
        keys_to_keep = filter_by_frequency(language_counts, threshold_percent=threshold_percent)
        filtered_dict = {k: v for k, v in predictions_by_lang.items() if k in keys_to_keep}
        
        results = {
            "predictions": dict(valmap(get_mean_score, filtered_dict)),
            "all_language_counts": dict(language_counts),
            "filtered_languages": list(keys_to_keep),
            "threshold_used": threshold_percent
        }
        
        return results
        
    except Exception as e:
        return {"error": f"Error during prediction: {str(e)}"}

def build_demo_interface():
    app_title = "Language Detection Tool" 
    with gr.Blocks(title=app_title) as demo:
        gr.Markdown(f"# {app_title}")
        gr.Markdown("Enter text below to detect the language(s) it contains.")
        
        with gr.Tab("Simple Detection"):
            with gr.Row():
                with gr.Column():
                    text_input1 = gr.Textbox(
                        label="Enter text for language detection",
                        placeholder="Type or paste your text here...",
                        lines=5
                    )
                    num_predictions = gr.Slider(
                        minimum=1, 
                        maximum=10, 
                        value=3, 
                        step=1,
                        label="Number of top predictions per line"
                    )
                    predict_btn1 = gr.Button("Detect Language")
                
                with gr.Column():
                    output1 = gr.JSON(label="Detection Results")
            
            predict_btn1.click(
                simple_predict,
                inputs=[text_input1, num_predictions],
                outputs=output1
            )
        
        with gr.Tab("Advanced Detection"):
            with gr.Row():
                with gr.Column():
                    text_input2 = gr.Textbox(
                        label="Enter text for advanced language detection",
                        placeholder="Type or paste your text here...",
                        lines=5
                    )
                    threshold = gr.Slider(
                        minimum=0.1, 
                        maximum=1.0, 
                        value=0.2, 
                        step=0.1,
                        label="Threshold percentage for filtering"
                    )
                    predict_btn2 = gr.Button("Advanced Detect")
                
                with gr.Column():
                    output2 = gr.JSON(label="Advanced Detection Results")
            
            predict_btn2.click(
                batch_predict,
                inputs=[text_input2, threshold],
                outputs=output2
            )
        
        gr.Markdown("### About")
        gr.Markdown("This tool uses Facebook's FastText language identification model to detect languages in text.")
    
    return demo
    

if __name__ == "__main__":
    demo = build_demo_interface()
    demo.launch(
        server_name="0.0.0.0", 
        server_port=7860,
        share=False
    )