113 lines
4 KiB
Python
113 lines
4 KiB
Python
"""SEGMENTED_V1: the capability that lets one document become many concepts.
|
|
|
|
Additive in the strict sense the upstream-version policy demands. `DEFAULT`,
|
|
`STRICT_V1`, `STRUCTURED_V1` and `OKF_V0_2` keep the bytes they had, because
|
|
the new field is defaulted and every branch downstream is keyed on
|
|
`profile.segmentation is not None`.
|
|
|
|
That toggle is deliberately NOT `IndexPolicy.per_directory`. `STRICT_V1`
|
|
already sets that field True while Door B ignores it, so keying the capability
|
|
there would change `STRICT_V1`'s Door B output and break a byte-stability pin
|
|
that four consumers depend on. The measured trap is the reason the assertion
|
|
below is written as an identity check on all four profiles rather than on one.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
from dataclasses import replace
|
|
|
|
from llm_ingestion_okf.profiles import (
|
|
DEFAULT,
|
|
OKF_LATEST,
|
|
OKF_V0_2,
|
|
SEGMENTED_V1,
|
|
STRICT_V1,
|
|
STRUCTURED_V1,
|
|
BundleProfile,
|
|
SegmentationPolicy,
|
|
)
|
|
|
|
EXISTING = (DEFAULT, STRICT_V1, STRUCTURED_V1, OKF_V0_2)
|
|
|
|
|
|
# --- the capability -------------------------------------------------------
|
|
|
|
|
|
def test_only_the_new_profile_declares_the_segmentation_capability() -> None:
|
|
assert SEGMENTED_V1.segmentation is not None
|
|
for profile in EXISTING:
|
|
assert profile.segmentation is None
|
|
|
|
|
|
def test_the_capability_is_not_keyed_on_per_directory() -> None:
|
|
# STRICT_V1 already sets per_directory True and Door B ignores it. Keying
|
|
# the 1-to-N branch there would silently change STRICT_V1's output.
|
|
assert STRICT_V1.index.per_directory is True
|
|
assert STRICT_V1.segmentation is None
|
|
|
|
|
|
def test_the_new_profile_writes_an_index_per_directory_and_names_bundle_id() -> None:
|
|
assert SEGMENTED_V1.index.per_directory is True
|
|
assert SEGMENTED_V1.index.root_frontmatter == ("bundle_id",)
|
|
|
|
|
|
def test_bundle_id_is_permitted_at_the_root_rather_than_demanded_by_the_policy() -> None:
|
|
# The policy NAMES the key and fixes its position; the caller owns the
|
|
# value (D5). Demanding it here would claim a decision this library does
|
|
# not own -- Door B enforces the plan-requires-an-id rule instead.
|
|
assert "bundle_id" not in SEGMENTED_V1.index.root_frontmatter_required
|
|
|
|
|
|
def test_the_new_profile_keeps_the_structured_facets_and_keys() -> None:
|
|
assert SEGMENTED_V1.index.facets == STRUCTURED_V1.index.facets
|
|
assert SEGMENTED_V1.frontmatter.order == STRUCTURED_V1.frontmatter.order
|
|
assert SEGMENTED_V1.types is STRUCTURED_V1.types
|
|
assert SEGMENTED_V1.paths is STRUCTURED_V1.paths
|
|
|
|
|
|
# --- the additivity proof -------------------------------------------------
|
|
|
|
|
|
def test_the_four_existing_profiles_equal_freshly_constructed_copies() -> None:
|
|
# A defaulted field must not move any of them. `replace()` with no changes
|
|
# reconstructs each profile through the dataclass constructor, so a field
|
|
# that had become required -- or had picked up a non-default value -- shows
|
|
# up here as inequality rather than as golden churn six steps later.
|
|
for profile in EXISTING:
|
|
assert replace(profile) == profile
|
|
|
|
|
|
def test_okf_latest_is_still_default() -> None:
|
|
assert OKF_LATEST is DEFAULT
|
|
|
|
|
|
def test_a_profile_constructs_without_naming_the_new_field() -> None:
|
|
minimal = BundleProfile(
|
|
types=DEFAULT.types,
|
|
frontmatter=DEFAULT.frontmatter,
|
|
paths=DEFAULT.paths,
|
|
index=DEFAULT.index,
|
|
)
|
|
assert minimal.segmentation is None
|
|
|
|
|
|
# --- the policy's own defaults -------------------------------------------
|
|
|
|
|
|
def test_the_policy_names_the_keys_it_governs() -> None:
|
|
policy = SEGMENTED_V1.segmentation
|
|
assert policy is not None
|
|
assert policy.hierarchical_paths is True
|
|
assert policy.bundle_id_key == "bundle_id"
|
|
assert policy.segment_id_key == "segment_id"
|
|
assert policy.offset_key == "source_offset"
|
|
assert policy.nav_label == "index"
|
|
|
|
|
|
def test_the_policy_is_frozen() -> None:
|
|
policy = SegmentationPolicy()
|
|
try:
|
|
policy.bundle_id_key = "other" # type: ignore[misc]
|
|
except Exception:
|
|
return
|
|
raise AssertionError("SegmentationPolicy must be frozen")
|