teleo-infrastructure/tests/test_reweave.py

"""Tests for reweave.py — orphan detection, entity filtering, same-source detection, frontmatter editing."""

import sys
import tempfile
from pathlib import Path

import pytest

sys.path.insert(0, str(Path(__file__).resolve().parent.parent))

from reweave import (
    _is_entity,
    _same_source,
    _parse_frontmatter,
    _get_edge_targets,
    _claim_name_variants,
    find_all_claims,
    build_reverse_link_index,
    find_orphans,
    write_edge,
    _count_reweave_edges,
    CLASSIFY_PROMPT,
)


@pytest.fixture
def kb_dir(tmp_path):
    """Create a minimal KB structure for testing."""
    domains = tmp_path / "domains" / "ai-alignment"
    domains.mkdir(parents=True)
    entities = tmp_path / "entities" / "ai-alignment"
    entities.mkdir(parents=True)
    return tmp_path


def _write_claim(path: Path, name: str, type_: str = "claim", **extra_fm):
    fm_lines = [f"name: {name}", f"type: {type_}"]
    for k, v in extra_fm.items():
        if isinstance(v, list):
            fm_lines.append(f"{k}:")
            for item in v:
                fm_lines.append(f"  - {item}")
        else:
            fm_lines.append(f"{k}: {v}")
    fm = "\n".join(fm_lines)
    path.write_text(f"---\n{fm}\n---\n\nBody of {name}.\n")


# ─── Entity Detection ──────────────────────────────────────────────────────


class TestEntityDetection:
    def test_entity_detected(self, kb_dir):
        p = kb_dir / "entities" / "ai-alignment" / "anthropic.md"
        _write_claim(p, "Anthropic", type_="entity")
        assert _is_entity(p) is True

    def test_claim_not_entity(self, kb_dir):
        p = kb_dir / "domains" / "ai-alignment" / "rlhf-works.md"
        _write_claim(p, "RLHF works", type_="claim")
        assert _is_entity(p) is False

    def test_no_frontmatter(self, tmp_path):
        p = tmp_path / "bare.md"
        p.write_text("No frontmatter here.")
        assert _is_entity(p) is False


# ─── Same Source Detection ──────────────────────────────────────────────────


class TestSameSourceDetection:
    def test_same_source_field(self, kb_dir):
        d = kb_dir / "domains" / "ai-alignment"
        a = d / "claim-a.md"
        b = d / "claim-b.md"
        _write_claim(a, "Claim A", source="paper-xyz.md")
        _write_claim(b, "Claim B", source="paper-xyz.md")
        assert _same_source(a, b) is True

    def test_different_source(self, kb_dir):
        d = kb_dir / "domains" / "ai-alignment"
        a = d / "claim-a.md"
        b = d / "claim-b.md"
        _write_claim(a, "Claim A", source="paper-xyz.md")
        _write_claim(b, "Claim B", source="paper-abc.md")
        assert _same_source(a, b) is False

    def test_same_source_file_field(self, kb_dir):
        d = kb_dir / "domains" / "ai-alignment"
        a = d / "claim-a.md"
        b = d / "claim-b.md"
        _write_claim(a, "Claim A", source_file="sources/arxiv/1234.md")
        _write_claim(b, "Claim B", source_file="sources/arxiv/1234.md")
        assert _same_source(a, b) is True

    def test_no_source_field(self, kb_dir):
        d = kb_dir / "domains" / "ai-alignment"
        a = d / "claim-a.md"
        b = d / "claim-b.md"
        _write_claim(a, "Claim A")
        _write_claim(b, "Claim B")
        assert _same_source(a, b) is False


# ─── Orphan Detection ──────────────────────────────────────────────────────


class TestOrphanDetection:
    def test_orphan_found(self, kb_dir):
        d = kb_dir / "domains" / "ai-alignment"
        a = d / "connected-claim.md"
        b = d / "orphan-claim.md"
        _write_claim(a, "Connected Claim", related=["orphan-claim"])
        _write_claim(b, "Orphan Claim")
        claims = find_all_claims(kb_dir)
        incoming = build_reverse_link_index(claims)
        orphans = find_orphans(claims, incoming, kb_dir)
        orphan_names = [p.stem for p in orphans]
        assert "connected-claim" not in orphan_names or "orphan-claim" not in orphan_names
        # connected-claim has no incoming either (only outgoing), so both may be orphans
        # but the key point: orphan detection runs without error

    def test_no_orphans_when_connected(self, kb_dir):
        d = kb_dir / "domains" / "ai-alignment"
        a = d / "claim-a.md"
        b = d / "claim-b.md"
        _write_claim(a, "Claim A", related=["claim-b"])
        _write_claim(b, "Claim B", related=["claim-a"])
        claims = find_all_claims(kb_dir)
        incoming = build_reverse_link_index(claims)
        orphans = find_orphans(claims, incoming, kb_dir)
        assert len(orphans) == 0


# ─── Frontmatter Editing ───────────────────────────────────────────────────


class TestWriteEdge:
    def test_write_edge_adds_field(self, kb_dir):
        d = kb_dir / "domains" / "ai-alignment"
        p = d / "neighbor.md"
        _write_claim(p, "Neighbor Claim")
        ok = write_edge(p, "Orphan Title", "related", "2026-03-31")
        assert ok is True
        text = p.read_text()
        assert "Orphan Title" in text
        assert "reweave_edges" in text

    def test_no_duplicate_edges(self, kb_dir):
        d = kb_dir / "domains" / "ai-alignment"
        p = d / "neighbor.md"
        _write_claim(p, "Neighbor Claim", related=["Orphan Title"])
        ok = write_edge(p, "Orphan Title", "related", "2026-03-31")
        assert ok is False  # duplicate detected

    def test_per_file_cap(self, kb_dir):
        d = kb_dir / "domains" / "ai-alignment"
        p = d / "neighbor.md"
        # Create a file with 10 reweave_edges already
        rw = [f"edge-{i}|related|2026-03-31" for i in range(10)]
        _write_claim(p, "Neighbor Claim", reweave_edges=rw)
        ok = write_edge(p, "New Orphan", "related", "2026-03-31")
        assert ok is False  # cap reached

    def test_no_blank_lines_in_frontmatter(self, kb_dir):
        d = kb_dir / "domains" / "ai-alignment"
        p = d / "neighbor.md"
        _write_claim(p, "Neighbor Claim", supports=["existing-claim"])
        write_edge(p, "New Orphan", "related", "2026-03-31")
        text = p.read_text()
        # Find frontmatter section
        start = text.index("---") + 3
        end = text.index("---", start)
        fm_section = text[start:end]
        # No blank lines in frontmatter
        for line in fm_section.strip().split("\n"):
            if line.strip() == "":
                pytest.fail(f"Blank line found in frontmatter: {repr(fm_section)}")


# ─── Prompt Content ─────────────────────────────────────────────────────────


class TestClassifyPrompt:
    def test_challenges_guidance_present(self):
        assert "challenges" in CLASSIFY_PROMPT
        assert "underused" in CLASSIFY_PROMPT.lower()

    def test_related_is_weakest(self):
        assert "WEAKEST" in CLASSIFY_PROMPT


# ─── Name Variants ──────────────────────────────────────────────────────────


class TestNameVariants:
    def test_stem_variants(self, kb_dir):
        p = kb_dir / "domains" / "ai-alignment" / "rlhf-reward-hacking.md"
        _write_claim(p, "RLHF Reward Hacking")
        variants = _claim_name_variants(p, kb_dir)
        assert "rlhf-reward-hacking" in variants
        assert "rlhf reward hacking" in variants