Class FakeEmbeddingService

experiments/tests/test_reading_report_flow.py:831–859

def test_retrieve_report_evidence_ranks_expected_pdf_chunks(monkeypatch):
    class FakeEmbeddingService:
        descriptor = "fake:test:4"

        def embed_batch(self, texts):
            vectors = []
            for text in texts:
                lowered = text.lower()
                background_score = float(sum(lowered.count(token) for token in ("background", "motivation", "challenge", "dataset shift")))
                method_score = float(sum(lowered.count(token) for token in ("method", "approach", "two-stage planner", "evidence retriever", "gating network")))
                results_score = float(sum(lowered.count(token) for token in ("results", "improves", "12%", "beats the baseline", "benchmark", "evaluation")))
                limitation_score = float(sum(lowered.count(token) for token in ("limitation", "limitations", "small number of domains", "future work")))
                vectors.append(
                    [
                        background_score,
                        method_score,
                        results_score,
                        limitation_score,
                    ]
                )
            return vectors

        @staticmethod
        def cosine_similarity(vector1, vector2):
            dot_product = sum(a * b for a, b in zip(vector1, vector2))
            norm1 = sum(a * a for a in vector1) ** 0.5
            norm2 = sum(b * b for b in vector2) ** 0.5
            if norm1 == 0 or norm2 == 0:
                return 0.0
            return dot_product / (norm1 * norm2)

    class FakeEmbeddingModule:
        @staticmethod