llm-ingestion-okf/tests/test_index_policy.py

"""The index policy as configuration, and the reader that judges it (Phase 3).

Three index shapes have to be expressible without either becoming the default:

- **DEFAULT** — the ingest-spec shape. One index at the bundle root, entries
  `- [label](target)`, curated prose alongside them, and no relationship
  between the index and the directory it sits in.
- **STRICT_V1** — `claude-code-llm-wiki`, read out of their code rather than
  recalled: an index at every level (`bundle.py:527-567`), exactly one `# `
  heading, entries `* [Title](link) - description` and nothing else, the root
  index carrying `okf_version`/`bundle_profile`/`okf_spec_commit` while nested
  indexes carry no frontmatter at all, and gate `BUNDLE_INDEX_COMPLETE`
  (`validate.py:1081-1120`, severity ERROR) demanding the index and the
  directory match exactly in both directions.
- **The second-brain shape** — not built here, but it must not be excluded.
  The catalog reported on 2026-07-25 that their index is an *authored* count
  of a directory's children and never a filesystem lookup, that a reader or
  validator enumerating the directory has implemented the wrong contract, and
  that they require prose in the index but no per-entry description.

The two consumers are directly opposed on exactly one point: the wiki mandates
the directory-derived match the catalog calls the wrong contract. That is what
settles the design. Every one of these differences is a policy field, none is
a library invariant, and `violations` never touches the filesystem — a caller
supplies the listing, and a profile whose index is authored refuses to be
handed one at all.
"""

from __future__ import annotations

import re

import pytest

from llm_ingestion_okf.profiles import DEFAULT, STRICT_V1, IndexPolicy

# --- DEFAULT keeps the Phase 1/2 shape (assumption C1) --------------------


def test_default_index_policy_judges_nothing() -> None:
    """Every new field is off under DEFAULT, so the reader has nothing to say.

    Door A emits an index with a curated summary above managed links, Door B
    and Door C append to one that may be empty. A judging default would
    declare invalid the bundles this library itself produces — the same
    reasoning that kept a required key set off DEFAULT's frontmatter schema.
    """
    text = "# Sales bundle\n\nCurated prose the operator wrote.\n\n- [Sales](ingest-sales.md)\n"
    assert DEFAULT.index.violations(text) == ()
    assert DEFAULT.index.violations(text, is_root=True) == ()


def test_default_index_is_root_only_because_upstream_binds_it_there() -> None:
    """Upstream OKF binds `index.md` to the bundle root alone.

    A nested directory without its own index is not an error there, which is
    why per-directory indexing is policy rather than an OKF rule: a DEFAULT
    demanding one per level would declare upstream-conforming bundles invalid.
    """
    assert DEFAULT.index.per_directory is False
    assert DEFAULT.index.required_indexes(["", "concepts", "concepts/hooks"]) == ("index.md",)


def test_default_index_link_is_unchanged_and_needs_no_description() -> None:
    assert DEFAULT.index.requires_description is False
    assert DEFAULT.index.render_link("Sales", "ingest-sales.md") == "- [Sales](ingest-sales.md)"


# --- the description on an entry ------------------------------------------


def test_render_link_carries_a_description_when_the_template_names_one() -> None:
    assert STRICT_V1.index.requires_description is True
    line = STRICT_V1.index.render_link("Hooks", "hooks.md", "How hooks fire.")
    assert line == "* [Hooks](hooks.md) - How hooks fire."


def test_render_link_refuses_to_emit_a_half_written_entry() -> None:
    """A template naming `{description}` and a caller omitting it must not
    silently render `* [T](t) - `, which parses as nothing and reads as an
    entry."""
    with pytest.raises(ValueError, match="description"):
        STRICT_V1.index.render_link("Hooks", "hooks.md")


def test_a_description_offered_to_a_template_without_one_is_refused() -> None:
    """The mirror case: silently dropping it would lose operator bytes."""
    with pytest.raises(ValueError, match="description"):
        DEFAULT.index.render_link("Sales", "ingest-sales.md", "dropped on the floor")


def test_strict_v1_link_template_and_pattern_describe_the_same_shape() -> None:
    """Rendering and recognising are separate operations; this is the pair's
    round-trip, the same test DEFAULT's shape already carries."""
    for label, target, description in (
        ("Hooks", "hooks.md", "How hooks fire."),
        ("Concepts", "concepts/index.md", "Machine-generated index of concepts."),
        ("", "a.md", "-"),
    ):
        line = STRICT_V1.index.render_link(label, target, description)
        match = STRICT_V1.index.link_pattern.match(line)
        assert match is not None
        assert match.group("label") == label
        assert match.group("target") == target
        assert match.group("description") == description


def test_strict_v1_link_pattern_is_anchored_to_the_whole_line() -> None:
    assert STRICT_V1.index.link_pattern.match("see * [Hooks](hooks.md) - x below") is None


# --- per-directory indexes ------------------------------------------------


def test_per_directory_policy_names_one_index_per_level() -> None:
    """The caller supplies the directories; the policy never enumerates them.

    This is the catalog's construction rule honoured at the type level: the
    library cannot list a directory because it is never given one to list.
    """
    assert STRICT_V1.index.per_directory is True
    assert STRICT_V1.index.required_indexes(["", "concepts", "concepts/hooks"]) == (
        "concepts/hooks/index.md",
        "concepts/index.md",
        "index.md",
    )


# --- the reader: headings -------------------------------------------------


ROOT_HEAD = "---\nokf_version: 0.1\nbundle_profile: strict-v1\nokf_spec_commit: d44368c\n---\n"


def test_strict_index_wants_exactly_one_h1() -> None:
    body = "\n* [Hooks](hooks.md) - How hooks fire.\n"
    missing = STRICT_V1.index.violations(
        ROOT_HEAD + body, is_root=True, expected_targets={"hooks.md"}
    )
    assert [violation.code for violation in missing] == ["index_heading_missing"]

    two = ROOT_HEAD + "\n# Wiki\n\n# Again\n" + body
    extra = STRICT_V1.index.violations(two, is_root=True, expected_targets={"hooks.md"})
    assert [violation.code for violation in extra] == ["index_heading_extra"]
    assert extra[0].subject == "# Again"


def test_a_second_level_heading_is_prose_not_a_heading() -> None:
    """Only `# ` is the heading; `## ` is body text the strict shape forbids."""
    text = ROOT_HEAD + "\n# Wiki\n\n## Section\n"
    codes = [v.code for v in STRICT_V1.index.violations(text, is_root=True, expected_targets=set())]
    assert codes == ["index_prose_not_allowed"]


# --- the reader: prose ----------------------------------------------------


def test_strict_index_admits_nothing_but_the_heading_and_the_entries() -> None:
    text = ROOT_HEAD + "\n# Wiki\n\nA sentence the generator would never write.\n"
    found = STRICT_V1.index.violations(text, is_root=True, expected_targets=set())
    assert [v.code for v in found] == ["index_prose_not_allowed"]
    assert found[0].subject == "A sentence the generator would never write."


def test_prose_is_the_catalog_shape_and_stays_expressible() -> None:
    """Their index requires progressive-disclosure prose and no per-entry
    description — the opposite of the wiki on both counts, and both have to be
    constructible."""
    second_brain = IndexPolicy(
        name="index.md",
        link_template="- [{label}]({target})",
        link_pattern=DEFAULT.index.link_pattern,
        per_directory=True,
        heading_required=True,
        allows_prose=True,
        root_frontmatter=("okf_version", "okf_layout"),
    )
    text = "# Notes\n\nWhat lives here and why.\n\n- [A note](a.md)\n"
    assert second_brain.violations(text) == ()
    assert second_brain.entries_match_directory is False


# --- the reader: directory match, both directions -------------------------


def test_the_match_is_bidirectional_and_names_each_side() -> None:
    text = ROOT_HEAD + "\n# Wiki\n\n* [Hooks](hooks.md) - How hooks fire.\n"
    found = STRICT_V1.index.violations(
        text, is_root=True, expected_targets={"skills.md", "concepts/index.md"}
    )
    assert [(v.subject, v.code) for v in found] == [
        ("concepts/index.md", "index_entry_missing"),
        ("hooks.md", "index_entry_unexpected"),
        ("skills.md", "index_entry_missing"),
    ]


def test_a_derived_index_cannot_be_judged_without_the_listing() -> None:
    """Skipping the check silently is the failure mode that matters: a gate
    that quietly passes when its input is absent is worse than one that
    raises."""
    text = ROOT_HEAD + "\n# Wiki\n"
    with pytest.raises(ValueError, match="listing"):
        STRICT_V1.index.violations(text, is_root=True)


def test_an_authored_index_refuses_a_listing() -> None:
    """The catalog's rule, enforced rather than documented: a profile whose
    index is authored cannot be asked to check it against the directory, so
    code written to the wrong contract fails at the call instead of passing
    every test anyone would think to write."""
    with pytest.raises(ValueError, match="authored"):
        DEFAULT.index.violations("- [Sales](ingest-sales.md)\n", expected_targets={"a.md"})


# --- the reader: root vs nested frontmatter -------------------------------


def test_the_root_index_pins_its_keys_in_order() -> None:
    text = ROOT_HEAD + "\n# Wiki\n"
    assert STRICT_V1.index.violations(text, is_root=True, expected_targets=set()) == ()


def test_a_root_index_missing_a_pinned_key_is_reported_per_key() -> None:
    text = "---\nokf_version: 0.1\n---\n\n# Wiki\n"
    found = STRICT_V1.index.violations(text, is_root=True, expected_targets=set())
    assert [(v.subject, v.code) for v in found] == [
        ("bundle_profile", "index_root_key_missing"),
        ("okf_spec_commit", "index_root_key_missing"),
    ]


def test_the_pinned_keys_are_ordered_not_merely_present() -> None:
    """Their `ROOT_INDEX_KEY_ORDER` is what makes a regeneration byte-identical,
    so an out-of-order root index is a finding even with every key present."""
    text = "---\nbundle_profile: strict-v1\nokf_version: 0.1\nokf_spec_commit: d44368c\n---\n\n# Wiki\n"
    found = STRICT_V1.index.violations(text, is_root=True, expected_targets=set())
    assert [v.code for v in found] == ["index_root_key_order"]


def test_a_nested_index_carries_no_frontmatter_at_all() -> None:
    """Confirmed independently in two consumers: the wiki writes frontmatter
    only at the bundle root (`bundle.py:544-557`), and the catalog reports
    their `okf_version`/`okf_layout` markers as root-exclusive. The asymmetry
    is shape, not one repo's preference."""
    found = STRICT_V1.index.violations(ROOT_HEAD + "\n# Concepts\n", expected_targets=set())
    assert [(v.subject, v.code) for v in found] == [
        ("okf_version", "index_root_frontmatter_unexpected")
    ]


def test_a_nested_index_without_frontmatter_is_clean() -> None:
    assert STRICT_V1.index.violations("# Concepts\n", expected_targets=set()) == ()


# --- determinism ----------------------------------------------------------


def test_violations_are_sorted_so_two_runs_agree() -> None:
    text = "---\nokf_spec_commit: d\nokf_version: 0.1\n---\n\nz prose\na prose\n"
    found = STRICT_V1.index.violations(text, is_root=True, expected_targets={"m.md"})
    assert list(found) == sorted(found, key=lambda v: (v.subject, v.code))
    assert found == STRICT_V1.index.violations(text, is_root=True, expected_targets={"m.md"})


def test_index_policy_stays_frozen() -> None:
    with pytest.raises(Exception):
        STRICT_V1.index.per_directory = False  # type: ignore[misc]


def test_the_pattern_group_names_the_template_needs_are_present() -> None:
    """A template naming `{description}` whose pattern cannot recover one is a
    pair that round-trips in only one direction."""
    with pytest.raises(ValueError, match="description"):
        IndexPolicy(
            name="index.md",
            link_template="* [{label}]({target}) - {description}",
            link_pattern=re.compile(r"^\* \[(?P<label>[^\]]*)\]\((?P<target>[^)]+)\)$"),
        )