teleo-infrastructure/tests/test_enrichment_dedup.py

"""Tests for enrichment idempotency — dedup at insertion and post-rebase."""

import os
import tempfile

import pytest


# ─── Unit tests for dedup_evidence_blocks ────────────────────────────────


from lib.dedup import dedup_evidence_blocks


class TestDedupEvidenceBlocks:
    """Test the post-rebase evidence block deduplication."""

    def test_no_blocks_unchanged(self):
        content = "---\ntype: claim\n---\n\nSome claim body.\n"
        assert dedup_evidence_blocks(content) == content

    def test_single_block_unchanged(self):
        content = (
            "---\ntype: claim\n---\n\nClaim body.\n\n"
            "### Additional Evidence (extend)\n"
            "*Source: [[some-source-2026-03-19]] | Added: 2026-03-19*\n\n"
            "Evidence text here.\n"
        )
        assert dedup_evidence_blocks(content) == content

    def test_duplicate_blocks_removed(self):
        """Two evidence blocks from the same source — second is removed."""
        block = (
            "\n\n### Additional Evidence (extend)\n"
            "*Source: [[interlune-he3-quantum-demand]] | Added: 2026-03-19*\n\n"
            "Some evidence text.\n"
        )
        content = f"---\ntype: claim\n---\n\nClaim body.{block}{block}\nRelevant Notes:\n"
        result = dedup_evidence_blocks(content)
        # Should contain exactly one occurrence
        assert result.count("[[interlune-he3-quantum-demand]]") == 1
        assert "Relevant Notes:" in result

    def test_different_sources_kept(self):
        """Two evidence blocks from different sources — both kept."""
        block1 = (
            "\n\n### Additional Evidence (extend)\n"
            "*Source: [[source-a]] | Added: 2026-03-19*\n\n"
            "Evidence A.\n"
        )
        block2 = (
            "\n\n### Additional Evidence (challenge)\n"
            "*Source: [[source-b]] | Added: 2026-03-20*\n\n"
            "Evidence B.\n"
        )
        content = f"---\ntype: claim\n---\n\nClaim body.{block1}{block2}"
        result = dedup_evidence_blocks(content)
        assert "[[source-a]]" in result
        assert "[[source-b]]" in result

    def test_auto_enrichment_dedup(self):
        """Duplicate auto-enrichment blocks from substantive fixer."""
        block = (
            "\n\n### Auto-enrichment (near-duplicate conversion, similarity=0.92)\n"
            "*Source: PR #1234 — \"Some duplicate claim\"*\n\n"
            "Converted evidence.\n"
        )
        content = f"---\ntype: claim\n---\n\nBody.{block}{block}"
        result = dedup_evidence_blocks(content)
        assert result.count("PR #1234") == 1

    def test_mixed_types_dedup(self):
        """Same source appears in both Additional Evidence and Auto-enrichment."""
        block1 = (
            "\n\n### Additional Evidence (extend)\n"
            "*Source: [[my-source]] | Added: 2026-03-19*\n\n"
            "First version of evidence.\n"
        )
        block2 = (
            "\n\n### Additional Evidence (extend)\n"
            "*Source: [[my-source]] | Added: 2026-03-19*\n\n"
            "Second version of evidence (rebase duplicate).\n"
        )
        content = f"---\ntype: claim\n---\n\nBody.{block1}{block2}"
        result = dedup_evidence_blocks(content)
        assert result.count("[[my-source]]") == 1
        # First occurrence kept
        assert "First version" in result
        assert "Second version" not in result

    def test_real_pr1751_pattern(self):
        """Reproduce the actual PR #1751 duplicate pattern from space-development."""
        content = (
            "---\ntype: claim\ndomain: space-development\n---\n\n"
            "Claim about SpaceX vertical integration.\n\n"
            "### Additional Evidence (extend)\n"
            "*Source: [[2026-03-00-commercial-stations-haven1-slip-orbital-reef-delays]] | Added: 2026-03-19*\n\n"
            "Orbital Reef multi-party structure experiencing delays.\n\n"
            "### Additional Evidence (extend)\n"
            "*Source: [[2026-03-00-commercial-stations-haven1-slip-orbital-reef-delays]] | Added: 2026-03-19*\n\n"
            "Orbital Reef multi-party structure experiencing delays (duplicate from rebase).\n\n"
            "---\n\n"
            "### Additional Evidence (extend)\n"
            "*Source: [[2026-03-19-space-com-starship-v3-first-static-fire]] | Added: 2026-03-24*\n\n"
            "V3 Starship static fire completed.\n"
        )
        result = dedup_evidence_blocks(content)
        assert result.count("[[2026-03-00-commercial-stations-haven1-slip-orbital-reef-delays]]") == 1
        assert result.count("[[2026-03-19-space-com-starship-v3-first-static-fire]]") == 1


# ─── Insertion-time dedup tests ──────────────────────────────────────────


class TestInsertionDedup:
    """Test that enrichment insertion skips already-enriched claims."""

    def test_entity_batch_dedup(self):
        """_apply_claim_enrichment skips if PR already enriched the claim."""
        from lib.entity_batch import _apply_claim_enrichment

        with tempfile.NamedTemporaryFile(mode="w", suffix=".md", delete=False) as f:
            f.write(
                "---\ntype: claim\n---\n\nClaim body.\n\n"
                "### Auto-enrichment (near-duplicate conversion, similarity=0.90)\n"
                "*Source: PR #100 — \"Existing enrichment\"*\n\n"
                "Already enriched evidence.\n"
            )
            f.flush()
            path = f.name

        try:
            ok, msg = _apply_claim_enrichment(path, "New evidence", 100, "Duplicate", 0.91)
            assert not ok
            assert "already enriched" in msg

            # Different PR should succeed
            ok2, msg2 = _apply_claim_enrichment(path, "New evidence", 200, "Different PR", 0.88)
            assert ok2
        finally:
            os.unlink(path)

    def test_entity_batch_first_enrichment_succeeds(self):
        """First enrichment of a claim by a PR should succeed."""
        from lib.entity_batch import _apply_claim_enrichment

        with tempfile.NamedTemporaryFile(mode="w", suffix=".md", delete=False) as f:
            f.write("---\ntype: claim\n---\n\nClaim body with no enrichments yet.\n")
            f.flush()
            path = f.name

        try:
            ok, msg = _apply_claim_enrichment(path, "New evidence", 500, "First enrichment", 0.92)
            assert ok
            with open(path) as rf:
                content = rf.read()
            assert "PR #500" in content
        finally:
            os.unlink(path)