"""Tests for frontmatter helpers — frontmatter union, order-preserving dedup, string-level splicing.

These test the pure functions in lib/frontmatter.py (extracted from merge.py Phase 6).
"""

import sys
import os
import pytest

sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..", "lib"))
from frontmatter import (
    REWEAVE_EDGE_FIELDS,
    parse_yaml_frontmatter,
    union_edge_lists,
    serialize_edge_fields,
    serialize_frontmatter,
)


class TestParseYamlFrontmatter:
    def test_basic(self):
        text = "---\ntitle: Test Claim\nsupports:\n- claim-a\n---\nBody text here."
        fm, raw, body = parse_yaml_frontmatter(text)
        assert fm is not None
        assert fm["title"] == "Test Claim"
        assert fm["supports"] == ["claim-a"]
        assert body.startswith("\n---")
        assert "title: Test Claim" in raw

    def test_no_frontmatter(self):
        text = "Just plain text"
        fm, raw, body = parse_yaml_frontmatter(text)
        assert fm is None
        assert raw == ""
        assert body == text

    def test_malformed_yaml(self):
        text = "---\n: invalid: yaml: {{{\n---\nBody"
        fm, raw, body = parse_yaml_frontmatter(text)
        assert fm is None


class TestUnionEdgeLists:
    def test_no_overlap(self):
        main = ["claim-a", "claim-b"]
        branch = ["claim-c", "claim-d"]
        result = union_edge_lists(main, branch)
        assert result == ["claim-a", "claim-b", "claim-c", "claim-d"]

    def test_overlap_preserves_main_order(self):
        main = ["claim-b", "claim-a"]
        branch = ["claim-a", "claim-c"]
        result = union_edge_lists(main, branch)
        assert result == ["claim-b", "claim-a", "claim-c"]

    def test_case_insensitive_dedup(self):
        main = ["Claim A"]
        branch = ["claim a", "Claim B"]
        result = union_edge_lists(main, branch)
        assert len(result) == 2
        assert result[0] == "Claim A"
        assert result[1] == "Claim B"

    def test_empty_main(self):
        result = union_edge_lists([], ["claim-a", "claim-b"])
        assert result == ["claim-a", "claim-b"]

    def test_empty_branch(self):
        result = union_edge_lists(["claim-a"], [])
        assert result == ["claim-a"]

    def test_both_empty(self):
        assert union_edge_lists([], []) == []

    def test_duplicates_within_branch(self):
        main = ["claim-a"]
        branch = ["claim-b", "claim-b"]
        result = union_edge_lists(main, branch)
        assert result == ["claim-a", "claim-b"]


class TestSerializeEdgeFields:
    def test_replaces_existing_field(self):
        raw = "title: Test\nsupports:\n- old-claim"
        merged = {"supports": ["old-claim", "new-claim"]}
        result = serialize_edge_fields(raw, merged)
        assert "- old-claim" in result
        assert "- new-claim" in result
        assert "title: Test" in result

    def test_preserves_non_edge_fields_exactly(self):
        raw = "title: 'Quoted Title'\nconfidence: 0.85\ntype: claim"
        merged = {"related": ["new-claim"]}
        result = serialize_edge_fields(raw, merged)
        assert "title: 'Quoted Title'" in result
        assert "confidence: 0.85" in result
        assert "type: claim" in result
        assert "related:" in result
        assert "- new-claim" in result

    def test_appends_new_field(self):
        raw = "title: Test\ntype: claim"
        merged = {"supports": ["claim-a"]}
        result = serialize_edge_fields(raw, merged)
        assert "title: Test" in result
        assert "supports:" in result
        assert "- claim-a" in result

    def test_empty_edges_removes_field(self):
        raw = "title: Test\nsupports:\n- old-claim\ntype: claim"
        merged = {}  # no edges to write
        result = serialize_edge_fields(raw, merged)
        assert "supports:" not in result
        assert "title: Test" in result
        assert "type: claim" in result

    def test_multiple_edge_fields(self):
        raw = "title: Test\nsupports:\n- a\nchallenges:\n- b"
        merged = {"supports": ["a", "c"], "challenges": ["b", "d"]}
        result = serialize_edge_fields(raw, merged)
        lines = result.split("\n")
        # supports and challenges both present with merged values
        assert "- a" in result
        assert "- c" in result
        assert "- b" in result
        assert "- d" in result


class TestSerializeFrontmatter:
    def test_roundtrip_preserves_formatting(self):
        original = "---\ntitle: 'Quoted Title'\nconfidence: 0.85\nsupports:\n- claim-a\n---\nBody text here."
        fm, raw, body = parse_yaml_frontmatter(original)
        merged_edges = {"supports": ["claim-a", "claim-b"]}
        result = serialize_frontmatter(raw, merged_edges, body)

        # Non-edge fields preserved exactly
        assert "title: 'Quoted Title'" in result
        assert "confidence: 0.85" in result
        # Edge fields updated
        assert "- claim-a" in result
        assert "- claim-b" in result
        # Structure preserved
        assert result.startswith("---\n")
        assert "\n---\n" in result
        assert result.endswith("Body text here.")

    def test_no_blank_line_before_closing_delimiter(self):
        """Ganymede critical: no extra blank line compounds on repeat reweaves."""
        original = "---\ntitle: Test\nsupports:\n- a\n---\nBody."
        fm, raw, body = parse_yaml_frontmatter(original)
        merged_edges = {"supports": ["a", "b"]}
        result = serialize_frontmatter(raw, merged_edges, body)
        # Should NOT have \n\n--- (double newline before closing)
        assert "\n\n---" not in result

    def test_repeated_serialize_no_drift(self):
        """Repeated serialization should be idempotent — no accumulating blank lines."""
        text = "---\ntitle: Test\nsupports:\n- a\n---\nBody."
        merged_edges = {"supports": ["a", "b"]}

        for _ in range(5):
            fm, raw, body = parse_yaml_frontmatter(text)
            text = serialize_frontmatter(raw, merged_edges, body)

        assert text.count("\n\n") == 0  # no double newlines anywhere


class TestSupersetDetection:
    def test_branch_is_superset(self):
        main_edges = {"claim-a", "claim-b"}
        branch_edges = {"claim-a", "claim-b", "claim-c"}
        assert len(main_edges - branch_edges) == 0

    def test_branch_missing_edge(self):
        main_edges = {"claim-a", "claim-b"}
        branch_edges = {"claim-a", "claim-c"}
        assert "claim-b" in (main_edges - branch_edges)

    def test_equal_sets(self):
        main_edges = {"claim-a", "claim-b"}
        branch_edges = {"claim-a", "claim-b"}
        assert len(main_edges - branch_edges) == 0


class TestEdgeFieldsCoverage:
    def test_standard_fields_present(self):
        assert "supports" in REWEAVE_EDGE_FIELDS
        assert "challenges" in REWEAVE_EDGE_FIELDS
        assert "related" in REWEAVE_EDGE_FIELDS
        assert "reweave_edges" in REWEAVE_EDGE_FIELDS
        assert "depends_on" in REWEAVE_EDGE_FIELDS


class TestFullUnionWorkflow:
    def test_main_evolved_branch_stale(self):
        """Main got new edges after branch was created. Union includes both."""
        main_text = (
            "---\ntitle: Test Claim\nconfidence: 0.8\n"
            "supports:\n- claim-a\n- claim-b\n"
            "related:\n- claim-x\n"
            "---\nBody text."
        )
        branch_text = (
            "---\ntitle: Test Claim\nconfidence: 0.8\n"
            "supports:\n- claim-a\n"
            "related:\n- claim-x\n- claim-y\n"
            "reweave_edges:\n- \"claim-y|related|2026-04-04\"\n"
            "---\nBody text."
        )

        main_fm, main_raw, main_body = parse_yaml_frontmatter(main_text)
        branch_fm, _, _ = parse_yaml_frontmatter(branch_text)

        merged_edges = {}
        for field in REWEAVE_EDGE_FIELDS:
            main_list = main_fm.get(field, [])
            branch_list = branch_fm.get(field, [])
            if not isinstance(main_list, list):
                main_list = [main_list] if main_list else []
            if not isinstance(branch_list, list):
                branch_list = [branch_list] if branch_list else []
            if main_list or branch_list:
                merged_edges[field] = union_edge_lists(main_list, branch_list)

        assert merged_edges["supports"] == ["claim-a", "claim-b"]
        assert "claim-x" in merged_edges["related"]
        assert "claim-y" in merged_edges["related"]
        assert len(merged_edges.get("reweave_edges", [])) == 1

        # Verify non-edge fields preserved in serialization
        result = serialize_frontmatter(main_raw, merged_edges, main_body)
        assert "confidence: 0.8" in result
        assert "title: Test Claim" in result

    def test_no_edge_fields_untouched(self):
        """Non-edge fields (title, confidence, type) come from main unchanged."""
        main_text = "---\ntitle: Original\nconfidence: 0.9\ntype: claim\n---\nBody."
        branch_text = "---\ntitle: Original\nconfidence: 0.9\ntype: claim\nrelated:\n- new-claim\n---\nBody."

        main_fm, main_raw, main_body = parse_yaml_frontmatter(main_text)
        branch_fm, _, _ = parse_yaml_frontmatter(branch_text)

        merged_edges = {}
        for field in REWEAVE_EDGE_FIELDS:
            main_list = main_fm.get(field, [])
            branch_list = branch_fm.get(field, [])
            if not isinstance(main_list, list):
                main_list = [main_list] if main_list else []
            if not isinstance(branch_list, list):
                branch_list = [branch_list] if branch_list else []
            if main_list or branch_list:
                merged_edges[field] = union_edge_lists(main_list, branch_list)

        result = serialize_frontmatter(main_raw, merged_edges, main_body)
        assert "title: Original" in result
        assert "confidence: 0.9" in result
        assert "type: claim" in result
        assert "- new-claim" in result

    def test_scalar_edge_field_converted_to_list(self):
        """Edge fields stored as scalars (not lists) are handled gracefully."""
        main_fm = {"supports": "single-claim"}
        branch_fm = {"supports": ["single-claim", "new-claim"]}

        main_list = main_fm.get("supports", [])
        branch_list = branch_fm.get("supports", [])
        if not isinstance(main_list, list):
            main_list = [main_list] if main_list else []
        if not isinstance(branch_list, list):
            branch_list = [branch_list] if branch_list else []

        result = union_edge_lists(main_list, branch_list)
        assert result == ["single-claim", "new-claim"]

    def test_yaml_formatting_preserved_across_reweave(self):
        """The key test: non-edge YAML formatting stays byte-identical."""
        # Use unusual but valid YAML formatting
        main_text = "---\ntitle: 'A \"Quoted\" Title'\nconfidence: 0.85\nsome_custom_field: [1, 2, 3]\nsupports:\n- old-claim\n---\nBody."
        fm, raw, body = parse_yaml_frontmatter(main_text)
        merged_edges = {"supports": ["old-claim", "new-claim"]}
        result = serialize_frontmatter(raw, merged_edges, body)

        # These non-edge fields must be byte-identical to source
        assert "title: 'A \"Quoted\" Title'" in result
        assert "confidence: 0.85" in result
        assert "some_custom_field: [1, 2, 3]" in result