app.py

# 📚 1️⃣ Import All Required Libraries
import streamlit as st
import pickle
import numpy as np
from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing.sequence import pad_sequences

# 📦 2️⃣ Load the Trained LSTM Model
model = load_model("next_word_LSTM.h5")

# 📄 3️⃣ Load the Tokenizer
with open("tokenizer.pickle", "rb") as handle:
    tokenizer = pickle.load(handle)

# 🤖 4️⃣ Prediction Function
def predict_next_word(model, tokenizer, text, max_sequence_len):
    # 1️⃣ Tokenize the input text
    token_list = tokenizer.texts_to_sequences([text])[0]

    # 2️⃣ Truncate if too long
    if len(token_list) >= max_sequence_len:
        token_list = token_list[-(max_sequence_len-1):]

    # 3️⃣ Pad the sequence
    token_list = pad_sequences([token_list], maxlen=max_sequence_len-1, padding='pre')

    # 4️⃣ Predict the next word
    predicted = model.predict(token_list, verbose=0)
    predicted_word_index = np.argmax(predicted, axis=1)[0]

    # 5️⃣ Map back to word
    for word, index in tokenizer.word_index.items():
        if index == predicted_word_index:
            return word
    return None

# 🌐 5️⃣ Build Streamlit Web App
st.set_page_config(page_title="Next Word Prediction", layout="centered")
st.title("🔮 Next Word Predictor using LSTM")
st.markdown("Type a few words and the model will guess the next word!")

# 📥 6️⃣ Text Input
input_text = st.text_input("Enter a sequence of words:", placeholder="e.g. to be or not to")

# 🔘 7️⃣ Prediction Button
if st.button("Predict Next Word"):
    if input_text.strip() == "":
        st.warning("Please enter some text.")
    else:
        max_sequence_len = model.input_shape[1] + 1  # since input_length = max_len - 1 during training
        next_word = predict_next_word(model, tokenizer, input_text, max_sequence_len)

        if next_word:
            st.success(f"✨ Predicted next word: **{next_word}**")
        else:
            st.error("Couldn't predict the next word.")


# ▶️ 8️⃣ Run the Streamlit App
# streamlit run app.py
# •	💻 Runs the web app in your browser.
# •	🕐 Takes a little time to load the model — this is normal.
# •	⚠️ Some warnings may appear, like tf.get_default_graph is deprecated — you can ignore