"""Refresh the closed semantic OpenAPI catalog from the frozen Rust leaf table.

Run from any directory with Python 3: semantic_schema.py [--check].
This generates documentation only, never runtime selectors or model mutations.
"""
import copy
import json
import re
import sys
from pathlib import Path

ROOT = Path(__file__).resolve().parents[3]
PATH = ROOT / "docs/openapi.json"
spec = json.loads(PATH.read_text(encoding="utf-8"))
schemas = spec["components"]["schemas"]
types = (ROOT / "crates/oa-core/src/semantic/types.rs").read_text(encoding="utf-8")
model = (ROOT / "crates/oa-core/src/model.rs").read_text(encoding="utf-8")


def ref(name):
    return {"$ref": f"#/components/schemas/{name}"}


def obj(properties, required=None):
    return {"type": "object", "properties": properties,
            "required": list(properties) if required is None else required,
            "additionalProperties": False}


def array(item, maximum=None):
    result = {"type": "array", "items": item}
    if maximum is not None:
        result["maxItems"] = maximum
    return result


def optional(item):
    return {"anyOf": [item, {"type": "null"}]}


def enum(names):
    return {"type": "string", "enum": names}


def variant(name, value):
    return obj({name: value})


def typed(name):
    if name.startswith("Option<"):
        return optional(typed(name[7:-1]))
    if name.startswith("Vec<"):
        return array(typed(name[4:-1]))
    primitives = {"String": {"type": "string"}, "bool": {"type": "boolean"},
                  "u32": {"type": "integer", "minimum": 0, "maximum": 4294967295},
                  "i64": {"type": "integer", "minimum": -9223372036854775808,
                          "maximum": 9223372036854775807}, "f64": {"type": "number"}}
    return primitives[name] if name in primitives else ref(name)


def reviewed(name):
    key = "Reviewed" + re.sub(r"[^A-Za-z0-9]", "", name)
    observation = {"oneOf": [enum(["NotApplicable"]), variant("Value", typed(name))]}
    schemas[key] = obj({"oa": observation, "sa": copy.deepcopy(observation)})
    return ref(key)


def keypair(name):
    key = "KeyPair" + name
    schemas[key] = obj({"oa": optional(typed(name)), "sa": optional(typed(name))}, [])
    return ref(key)


for name in ["DataConstructor", "PortConstructor", "OccurrenceConstructor", "EventConstructor",
             "MessageConstructor", "ReferenceCollection", "OrderedCollection", "IncidentEnd",
             "ArtifactFacet", "ComponentPortKind", "ComponentPortOrientation", "PortDirection",
             "FragmentOperator", "StateActionKind", "ArtifactRelation"]:
    source = types if f"pub enum {name}" in types else model
    body = re.search(r"pub enum " + name + r"\s*\{([^}]+)\}", source).group(1)
    body = re.sub(r"//[^\n]*|#\[[^\]]*\]", "", body)
    names = [value.strip() for value in body.split(",") if value.strip()]
    assert all(re.fullmatch(r"[A-Za-z]+", value) for value in names), (name, names)
    schemas[name] = enum(names)

catalog = types.split("macro_rules! scalar_families", 1)[1].split("pub(super) use", 1)[0]
families = re.findall(r"(\w+), (\w+), (\w+) \{([^}]+)\}", catalog)
assert len(families) == 13
selectors, reviews = [], []
leaf_count = 0
for family, selector, review, body in families:
    leaves = re.findall(r"(\w+): ([\w<>]+)", body)
    leaf_count += len(leaves)
    schemas[selector] = enum([leaf for leaf, _ in leaves])
    schemas[review] = {"oneOf": [variant(leaf, reviewed(value)) for leaf, value in leaves]}
    selectors.append(variant(family, ref(selector)))
    reviews.append(variant(family, ref(review)))
assert leaf_count == 84
schemas["AnnotationKey"] = obj({"path": array(ref("Id")), "field": ref("Id")})
id_keys, text_keys, annotation_keys = keypair("Id"), keypair("String"), keypair("AnnotationKey")
artifact_types = {"Presence": "bool", "Name": "String", "Description": "String", "Owner": "Id",
                  "Relation": "ArtifactRelation", "Uri": "String", "Revision": "Option<String>",
                  "Fragment": "Option<String>"}
schemas["ArtifactReview"] = {"oneOf": [variant(k, reviewed(v)) for k, v in artifact_types.items()]}
selectors += [variant("Member", obj({"collection": ref("ReferenceCollection"), "keys": id_keys})),
              variant("Order", ref("OrderedCollection")), variant("Literal", text_keys),
              variant("Annotation", annotation_keys), variant("Binding", id_keys), variant("Child", id_keys),
              variant("Incident", obj({"keys": id_keys, "end": ref("IncidentEnd")})),
              variant("Artifact", obj({"keys": id_keys, "facet": ref("ArtifactFacet")}))]
reviews += [variant("Member", obj({"collection": ref("ReferenceCollection"), "keys": id_keys, "review": reviewed("bool")})),
            variant("Order", obj({"collection": ref("OrderedCollection"), "review": reviewed("Vec<Id>")})),
            variant("Literal", obj({"keys": text_keys, "review": reviewed("bool")})),
            variant("Annotation", obj({"keys": annotation_keys, "review": reviewed("Option<MessageValue>")})),
            variant("Binding", obj({"keys": id_keys, "review": reviewed("Option<LifelineBinding>")})),
            variant("Child", obj({"keys": id_keys, "review": reviewed("bool")})),
            variant("Incident", obj({"keys": id_keys, "end": ref("IncidentEnd"), "review": reviewed("bool")})),
            variant("Artifact", obj({"keys": id_keys, "review": ref("ArtifactReview")}))]
schemas["FacetSelector"] = {"oneOf": selectors}
schemas["FacetReview"] = {"oneOf": reviews}
schemas["SemanticBaseline"] = obj({"facets": array(ref("FacetReview"))})
schemas["TraceBaseline"]["properties"]["semantic"] = {**optional(ref("SemanticBaseline")), "default": None,
    "description": "Absent/null means Unreviewed. Historical typed facts never become live graph references."}
schemas["Project"]["properties"]["schema_version"]["const"] = 5
schemas["MigrationReport"]["properties"]["from_schema"]["enum"] = [1, 2, 3, 4]
schemas["MigrationReport"]["properties"]["to_schema"]["const"] = 5
schemas["MigrationReport"]["description"] = schemas["MigrationReport"]["description"].replace("V1/V2/V3 to V4", "V1/V2/V3/V4 to V5")
spec["info"]["description"] = spec["info"]["description"].replace("schema-4", "schema-5")
spec["paths"]["/api/v1/projects/import"]["post"]["description"] = "Schema-5 native JSON is described below. Schema-1/2/3/4 JSON and source are accepted through explicit frozen readers and migration, also on replacement and candidate validation. Prior prose reviews remain intact; absent semantic review stays Unreviewed. Export emits schema 5 / arcadia 5. See docs/language.rst."
schemas["SourceBundle"]["description"] = schemas["SourceBundle"]["description"].replace("source4 workspace (source3 migrates explicitly)", "source5 workspace (source3/4 migrate explicitly)")
schemas["Drift"] = enum(["Unreviewed", "Unchanged", "OaChanged", "SaChanged", "BothChanged"])
schemas["FacetDrift"] = obj({"selector": ref("FacetSelector"), "current": ref("FacetReview"), "drift": ref("Drift")})
schemas["SemanticDrift"] = obj({"own": ref("Drift"), "linked": ref("Drift"),
    "own_unreviewed": typed("bool"), "linked_unreviewed": typed("bool")})
schemas["LinkedSemanticImpact"] = obj({"objects": id_keys, "traces": array(ref("Id")),
    "drift": ref("Drift"), "unreviewed": typed("bool")})
for name, members in [("SemanticRecord", ["Element", "Relation", "Artifact"]),
                      ("ImpactRecord", ["Element", "Relation", "Artifact", "Diagram", "Trace"])]:
    schemas[name] = {"oneOf": [variant(n, ref("ArtifactLink" if n == "Artifact" else n)) for n in members]}
schemas["SemanticImpact"] = obj({"object": ref("Id"), "before": optional(ref("ImpactRecord")), "after": optional(ref("ImpactRecord"))}, ["object"])
schemas["SemanticDecision"] = enum(["KeepTarget", "TakeSource", "Candidate"])
schemas["FacetResolution"] = obj({"facet": ref("FacetSelector"), "decision": ref("SemanticDecision")})
schemas["SemanticReviewRequest"] = obj({"source": ref("Id"), "resolutions": array(ref("FacetResolution"), 5000),
    "reset": {**array(ref("FacetSelector"), 5000), "default": []}}, ["source", "resolutions"])


def mappings(maximum):
    return {"type": "object", "propertyNames": {"format": "uuid"},
            "additionalProperties": ref("Id"), "maxProperties": maximum, "default": {}}


schemas["SemanticTransitionRequest"] = obj({"transition": ref("TransitionRequest"), "traces": mappings(100),
    "references": mappings(5000), "reviews": {**array(ref("SemanticReviewRequest"), 100), "default": []},
    "candidates": {**array(ref("SemanticRecord"), 200), "default": []},
    "accepted_impacts": {**array(ref("SemanticImpact"), 200), "default": []}}, ["transition"])
schemas["SemanticTransitionRequest"]["description"] = "Canonical OA-to-SA references include historical UUIDs. Traces map selected source IDs to trace IDs. Combined resolutions/reset limit: 5,000. Candidates must equal exactly the result of their selected facets; accepted_impacts explicitly acknowledge exact induced changes."
schemas["SemanticReviewPreview"] = obj({"source": ref("Id"), "target": ref("Id"), "facets": array(ref("FacetDrift")),
    "acknowledged": array(ref("FacetSelector")), "retired": array(ref("FacetSelector")), "linked_impacts": array(ref("LinkedSemanticImpact"))})
schemas["SemanticTransitionPlan"] = obj({"base_fingerprint": typed("String"), "request": ref("SemanticTransitionRequest"),
    "transition": ref("TransitionPlan"), "reviews": array(ref("SemanticReviewPreview")), "impacts": array(ref("SemanticImpact")),
    "conflicts": array(typed("String")), "commands": array(ref("Command"), 200), "prepared_fingerprint": optional(typed("String"))})
schemas["SemanticTransitionPlan"]["required"].remove("prepared_fingerprint")
schemas["SemanticTransitionEnvelope"] = obj({"base_revision": ref("Revision"), "plan": ref("SemanticTransitionPlan")})
schemas["SemanticQueryRequest"] = obj({"traces": {**array(ref("Id"), 100), "minItems": 1, "uniqueItems": True},
    "references": mappings(5000), "revision": optional(ref("Revision"))}, ["traces"])
schemas["SemanticQueryItem"] = obj({"trace": ref("Id"), "oa": ref("Id"), "sa": ref("Id"),
    "facets": array(ref("FacetDrift")), "drift": ref("SemanticDrift"), "linked_impacts": array(ref("LinkedSemanticImpact"))})
schemas["SemanticQueryResult"] = obj({"revision": ref("Revision"), "items": array(ref("SemanticQueryItem"), 100)})
for operation, request, result in [("query", "SemanticQueryRequest", "SemanticQueryResult"),
        ("preview", "SemanticTransitionRequest", "SemanticTransitionEnvelope"),
        ("apply", "SemanticTransitionEnvelope", "MutationResult")]:
    endpoint = copy.deepcopy(spec["paths"]["/api/v1/projects/{id}/transitions/" + ("apply" if operation == "apply" else "preview")])
    post = endpoint["post"]
    post["operationId"] = operation + "SemanticTransition"
    post["summary"] = {"query": "Query typed trace drift and linked impacts without mutation",
        "preview": "Preview and source-validate a typed semantic transition without writing",
        "apply": "Conditionally apply a prepared semantic candidate in one model/source revision"}[operation]
    post["requestBody"]["content"]["application/json"]["schema"] = ref(request)
    post["responses"]["200"]["content"]["application/json"]["schema"] = ref(result)
    post["description"] = "Uses the frozen core typed semantic contract. A supplied plan is a client-owned candidate, not a signed receipt; fingerprints check consistency, not original review authorship. Preview validates source writeback when conflict-free. Apply preserves preview-generated IDs and does not replan random identities."
    if operation == "query":
        post["parameters"] = [p for p in post["parameters"] if p["in"] != "header"]
        post["responses"].pop("412", None)
        post["responses"].pop("428", None)
        post["responses"]["409"] = {"$ref": "#/components/responses/Problem"}
        post["description"] = "Read one immutable snapshot. Optional revision mismatch returns409; missing trace returns404; duplicate/empty trace selection returns400. References are canonical OA-to-SA correspondence. No write, baseline refresh or external artifact fetch."
    spec["paths"]["/api/v1/projects/{id}/semantic-transitions/" + operation] = endpoint


def check_refs(value):
    if isinstance(value, dict):
        if "$ref" in value and value["$ref"].startswith("#/components/schemas/"):
            assert value["$ref"].split("/")[-1] in schemas, value
        for child in value.values():
            check_refs(child)
    elif isinstance(value, list):
        for child in value:
            check_refs(child)


check_refs(spec)
rendered = json.dumps(spec, indent=2, ensure_ascii=False) + "\n"
if "--check" in sys.argv:
    assert json.loads(PATH.read_text(encoding="utf-8")) == spec, "Run semantic_schema.py to refresh OpenAPI"
else:
    PATH.write_text(rendered, encoding="utf-8")
print(f"Closed catalog: {len(families)} families, {leaf_count} scalar leaves; {len(schemas)} schemas")