tests/test_extractive.py

import unittest
from models.extractive.tfidf import CustomTFIDF
from models.extractive.textrank import TextRankSummarizer
from models.extractive.lsa import LSASummarizer
from models.extractive.hybrid_scorer import HybridSummarizer

class TestExtractiveSummarization(unittest.TestCase):

    def setUp(self):
        self.ar_text = (
            "تعد معالجة اللغات الطبيعية فرعا أساسيا من فروع الذكاء الاصطناعي الحديث. "
            "تقوم أنظمة التلخيص التلقائي باستخراج وتوليد ملخصات دقيقة للنصوص الطويلة. "
            "يعتمد التلخيص الاستخراجي على تحديد أهم الجمل من النص الأصلي بالاعتماد على خوارزميات إحصائية. "
            "أما التلخيص التجريدي فيعتمد على شبكات التعلم العميق لتوليد صياغات جديدة للأفكار الرئيسية."
        )

        self.en_text = (
            "Natural language processing is an essential domain in modern artificial intelligence. "
            "Automatic text summarization condenses long documents into concise overviews. "
            "Extractive summarization extracts the most important sentences directly from the source text. "
            "Abstractive summarization creates novel sentences using deep neural networks and attention."
        )

    def test_tfidf_cosine_similarity(self):
        docs = [
            ["natural", "language", "processing"],
            ["natural", "language", "understanding"],
            ["computer", "vision", "images"]
        ]
        tfidf = CustomTFIDF()
        vectors = tfidf.fit_transform(docs)
        self.assertEqual(len(vectors), 3)

        sim_0_1 = CustomTFIDF.cosine_similarity(vectors[0], vectors[1])
        sim_0_2 = CustomTFIDF.cosine_similarity(vectors[0], vectors[2])
        self.assertGreater(sim_0_1, sim_0_2)

    def test_textrank_arabic_and_english(self):
        tr = TextRankSummarizer()
        
        ar_res = tr.summarize(self.ar_text, num_sentences=2, lang="ar")
        self.assertIn("summary", ar_res)
        self.assertEqual(len(ar_res["selected_indices"]), 2)
        self.assertGreater(len(ar_res["summary"]), 10)

        en_res = tr.summarize(self.en_text, num_sentences=2, lang="en")
        self.assertIn("summary", en_res)
        self.assertEqual(len(en_res["selected_indices"]), 2)
        self.assertGreater(len(en_res["summary"]), 10)

    def test_lsa_summarizer(self):
        lsa = LSASummarizer()
        res = lsa.summarize(self.en_text, num_sentences=2, lang="en")
        self.assertEqual(len(res["selected_indices"]), 2)
        self.assertGreater(len(res["summary"]), 10)

    def test_hybrid_summarizer(self):
        hybrid = HybridSummarizer()
        res = hybrid.summarize(self.ar_text, num_sentences=2, lang="ar")
        self.assertEqual(len(res["selected_indices"]), 2)
        self.assertGreater(len(res["summary"]), 10)

if __name__ == "__main__":
    unittest.main()