app.py

import gradio as gr
from PyPDF2 import PdfReader
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity
from transformers import pipeline
import matplotlib.pyplot as plt
import pandas as pd
import io

semantic_pipeline = pipeline("feature-extraction", model="sentence-transformers/all-MiniLM-L6-v2")

def extract_text_from_pdf(pdf_file):
    try:
        reader = PdfReader(io.BytesIO(pdf_file))
        full_text = ""
        for page in reader.pages:
            text = page.extract_text()
            if text:
                full_text += text
        return full_text.strip()
    except Exception as e:
        print("Error reading PDF:", e)
        return ""

def tfidf_similarity(text1, text2):
    vec = TfidfVectorizer().fit_transform([text1, text2])
    return cosine_similarity(vec[0:1], vec[1:2])[0][0]

def transformer_similarity(text1, text2):
    emb1 = semantic_pipeline(text1[:512])[0]
    emb2 = semantic_pipeline(text2[:512])[0]
    emb1_avg = [sum(x)/len(x) for x in zip(*emb1)]
    emb2_avg = [sum(x)/len(x) for x in zip(*emb2)]
    return cosine_similarity([emb1_avg], [emb2_avg])[0][0]

def semantic_match(lo_list, content):
    vectorizer = TfidfVectorizer().fit_transform([content] + lo_list)
    vectors = vectorizer.toarray()
    content_vec = vectors[0]
    scores = [cosine_similarity([content_vec], [vec])[0][0] for vec in vectors[1:]]
    return scores

def compare_all(old_pdf, new_pdf, lo_file):
    try:
        los = lo_file.decode("utf-8", errors="ignore").splitlines()
        los = [lo.strip() for lo in los if lo.strip()]
    except:
        return "❌ Could not read learning outcomes file.", None, None, None, None, None

    old_text = extract_text_from_pdf(old_pdf)
    new_text = extract_text_from_pdf(new_pdf)

    if not old_text or not new_text:
        return "❌ Could not extract text from one or both PDFs.", None, None, None, None, None

    old_scores = semantic_match(los, old_text)
    new_scores = semantic_match(los, new_text)

    labels = [f"LO{i+1}" for i in range(len(los))]
    x = range(len(labels))
    fig, ax = plt.subplots()
    ax.bar(x, old_scores, width=0.4, label="Old", align='center')
    ax.bar([i + 0.4 for i in x], new_scores, width=0.4, label="New", align='center')
    ax.set_xticks([i + 0.2 for i in x])
    ax.set_xticklabels(labels, rotation=45)
    ax.set_ylabel("Semantic Match Score")
    ax.set_title("Learning Outcomes Comparison")
    ax.legend()

    df = pd.DataFrame({
        "Learning Outcome": labels,
        "Old Match (%)": [round(s*100, 2) for s in old_scores],
        "New Match (%)": [round(s*100, 2) for s in new_scores],
        "Change (%)": [round((n - o)*100, 2) for o, n in zip(old_scores, new_scores)],
    })

    tfidf_sim = tfidf_similarity(old_text, new_text)
    transformer_sim = transformer_similarity(old_text, new_text)
    text_growth = (len(new_text) - len(old_text)) / len(old_text) * 100

    summary = f"📘 **Summary of Comparison**"
    summary += f"📈 **TF-IDF Content Similarity**: {round(tfidf_sim * 100, 2)}%"
    summary += f"🤖 **Transformer-based Similarity**: {round(transformer_sim * 100, 2)}%"
    summary += f"📝 **Text Growth**: {'+' if text_growth >= 0 else ''}{round(text_growth, 2)}% more content in new handout"
    summary += f"🎯 **LOs Matched (New ≥ 0.5)**: {sum(1 for s in new_scores if s >= 0.5)} of {len(los)}"
    summary += f"📊 **Insight**: New content appears {'more' if sum(new_scores) > sum(old_scores) else 'less'} aligned with outcomes."

    explanation = ("---"
        "📚 **Explanation of Methods**:"
                   
        "- **TF-IDF Similarity** checks how often important words appear in both documents. It gives a quick idea of textual overlap."
                   
        "- **Transformer Similarity** uses AI to understand meaning beyond words. It compares the 'sense' of the documents like a human would."
    )

    return summary + explanation, df, fig, new_text

iface = gr.Interface(
    fn=compare_all,
    inputs=[
        gr.File(label="Old Handout PDF", type='binary'),
        gr.File(label="New Handout PDF", type='binary'),
        gr.File(label="Learning Outcomes (Text File)", type='binary'),
    ],
    outputs=[
        gr.Markdown(label="📘 Summary & Insights"),
        gr.Dataframe(label="📊 LO-wise Comparison Table"),
        gr.Plot(label="📈 Visual LO Change Chart"),
        gr.Textbox(label="📄 New Handout Preview (Full Text)", lines=20, interactive=False),
    ],
    title="📘 Handout Comparison + LO Semantic Analysis",
    description="Upload two handouts (old and new) and a text file of Learning Outcomes (LOs). This tool compares content using TF-IDF and Transformers, visualizes LO changes, and explains results in simple terms.",
)

iface.launch()