@voxgig/apidef 8.10.0 → 8.12.0

package/src/transform/field.ts

@@ -6,6 +6,7 @@ import type { TransformResult, Transform } from '../transform'
 
 import {
   validator, canonizeField, inferFieldType, normalizeFieldName, envelopeProp,
+  canonizeCmpName,
   scanUntaggedUnion, firstSentence,
 } from '../utility'
 
@@ -49,24 +50,6 @@ const fieldTransform: Transform = async function(
         const mpoints = mop.points
 
         for (let mpoint of mpoints) {
-          // ACTION POINTS CONTRIBUTE NO FIELDS, as `identityParams` and
-          // `responseCandidates` below already assume: an action is a VERB
-          // dispatched by `$action`, so its request body is that verb's
-          // arguments and its response is that verb's result. Neither says
-          // anything about what a record of this entity carries.
-          //
-          // A custom action lands under `create`, so every point of it was
-          // harvested along with the plain create's. solar's planet — four
-          // properties in the spec — came out with ten fields, the extra six
-          // being `{start, stop}` and `{forbid, why}` from the two action
-          // bodies and `{ok, state}` from their shared response envelope.
-          // Those reached the generated `Planet` type, its create and update
-          // data types, and the per-entity field table in the generated
-          // reference, none of which a planet has ever carried.
-          if (null != mpoint?.select?.['$action']) {
-            continue
-          }
-
           const opfields = resolveOpFields(ment, mop, mpoint, def)
 
           for (let opfield of opfields) {
@@ -86,42 +69,11 @@ const fieldTransform: Transform = async function(
       return a.name < b.name ? -1 : a.name > b.name ? 1 : 0
     })
 
-    // Mark the entity as having an id only when the spec actually declares one.
-    // Downstream (test generators, fixture builders) gate id-specific code on
-    // this presence so that public read-only APIs without ids don't get
-    // bogus id assertions.
-    // COMPOSITE FIRST, because a compound key need not come with an `id`.
-    //
-    // An entity addressed by `{owner}/{repo}` whose response carries only
-    // `owner` and `name` has no field literally named `id`, and its adjacent
-    // placeholders are left unrenamed so `addressedById` is false too.
-    // Neither branch below then ran, so the entity got NO id descriptor and
-    // even an explicit `guide.entity.<name>.id.parts` was silently ignored —
-    // while the Go port, which initialises a descriptor unconditionally,
-    // emitted the composite. The ports disagreed on exactly the shape this
-    // feature exists for.
     const gent = guide?.entity?.[ment.name]
     const composite = compositeId(ment, gent, def)
 
     const idField = fields.find((f: ModelField) => 'id' === f.name)
 
-    // A COMPOSITE ID IS A STRING, whatever the API's own `id` field is —
-    // AND THE API'S OWN id IS KEPT.
-    //
-    // github's repo declares `id` as an integer, its global database id,
-    // while the composite identity is `owner/repo`. Two facts have to
-    // survive: `id` must hold a string, because that is what the joined
-    // value is and what every generated type has to store; and the spec's
-    // numeric property must not be silently reinterpreted, because a
-    // consumer that wants the database id is entitled to it with its own
-    // type and format intact.
-    //
-    // So the API's field MOVES to `<api>_id` rather than being rewritten in
-    // place, carrying its type, format and per-op overrides with it, and the
-    // entity's `alias.field` map records where it went. Retyping in place
-    // (the first attempt) claimed the server's numeric id was a string;
-    // leaving it alone made `id.field` name a declaration the runtime value
-    // cannot satisfy. Moving it is the only option that lies about neither.
     if (null != composite.parts && null != idField && !scalarStringField(idField)) {
       const idf: any = idField
       const apiname = String((model as any)?.name || 'api')
@@ -187,12 +139,6 @@ const fieldTransform: Transform = async function(
       ment.id = { name: 'id', field: 'id', ...composite }
     }
     else if (addressedById(ment)) {
-      // The FIELD as well as the descriptor. An entity addressed by id has an
-      // id at runtime — the test fixture seeds one, and the SDK sends it — so
-      // a model that declares the descriptor without the field makes the
-      // generated TYPE disagree with the generated TEST: trello's Option,
-      // Reaction and Sticker compiled to `TS2339: Property 'id' does not
-      // exist` the moment the test started assigning data.id.
       fields.push({
         name: 'id',
         type: '`$STRING`',
@@ -201,26 +147,6 @@ const fieldTransform: Transform = async function(
       fields.sort((a: ModelField, b: ModelField) =>
         a.name < b.name ? -1 : a.name > b.name ? 1 : 0)
 
-      // ADDRESSABLE BY ID WITHOUT DECLARING ONE AS A FIELD.
-      //
-      // The rule above reads the RESPONSE schema, and plenty of real entities
-      // are addressed by an id their response never repeats. github's
-      // private_registry is one: PATCH /orgs/{org}/private-registries/{secret_name}
-      // renames secret_name to id, so the entity is addressed by id on every
-      // one of its own routes, while its schema declares only created_at, key,
-      // name, url and friends.
-      //
-      // Downstream that absence is not cosmetic. TestEntity gates
-      // `data.id = <created>.id` on THIS descriptor, so the generated update
-      // carried no id at all, the test mock's selector fell back to whatever
-      // else was in reqdata (org_id), matched no single record, and the flow
-      // failed with a 404 that named nothing to do with ids.
-      //
-      // An entity whose own points take an `id` param IS addressable by id;
-      // that is the property the downstream generators actually want. Entities
-      // with neither a field nor an id param — the read-only public APIs the
-      // rule above was written for — still get no descriptor, so they still
-      // get no id assertions.
       ment.id = { name: 'id', field: 'id', ...composite }
     }
 
@@ -232,14 +158,6 @@ const fieldTransform: Transform = async function(
 
 
 
-// The separator that joins a composite id into one string.
-//
-// A forward slash cannot occur inside a single path segment — a raw `/`
-// would end the segment, and a value that legitimately contains one arrives
-// percent-encoded as `%2F` — so joining on it can never be ambiguous, and
-// splitting on it can never over-split. That is what makes the composite id
-// safe to carry as a single opaque string, which is the property the SDK and
-// Seneca entities are built on.
 const ID_SEP = '/'
 
 // Subfields that conventionally carry the identifying value of a nested
@@ -255,36 +173,6 @@ const NESTED_ID_KEYS = ['login', 'slug', 'name', 'key', 'id']
 const ID_OPS = ['load', 'update', 'patch', 'remove']
 
 
-// The parameters that TOGETHER name one record: the trailing run of
-// ADJACENT variable segments on the addressing route.
-//
-// ADJACENCY IS THE WHOLE TEST, and it is what separates a compound key from
-// ordinary parent/child nesting:
-//
-//   /repos/{owner}/{repo}                     -> owner, repo   COMPOSITE
-//   /api/planet/{planet_id}/moon/{moon_id}    -> moon_id       single
-//   /repos/{owner}/{repo}/pulls/{pull_number} -> pull_number   single
-//
-// A literal segment between two variables names a SUB-COLLECTION, so the
-// earlier variable scopes the later one — `planet_id` says which planet's
-// moons, and `moon_id` alone identifies the moon. Two variables with nothing
-// between them address no sub-collection: neither value names anything on
-// its own, and only the pair identifies a repository.
-//
-// Taking every variable on the path instead was tried first and is wrong on
-// most real specs — it made `moon` (planet_id + moon_id), petstore's `order`,
-// `pet` and `user`, and taxonomy's `domain` and `kingdom` all falsely
-// composite, which the apidef-validate goldens caught immediately. Nested
-// resources are the common shape; compound keys are the exception, and
-// adjacency is the thing that actually distinguishes them.
-//
-// Read from the op that names a single record, never from `list`: a
-// collection route's path params are the entity's parents. A point ending in
-// a literal is a verb ON the record (`.../{number}/merge`) and carries the
-// same variables, so it is a fallback rather than a different answer.
-// Walk back from a point's end, collecting variables until a literal stops
-// the run. That literal is the sub-collection boundary; anything before it
-// scopes this record rather than naming it.
 function trailingVars(point: any): string[] {
   const segs = ((point?.segments || []) as any[]).filter((s: any) => null != s)
   const run: string[] = []
@@ -301,15 +189,6 @@ function trailingVars(point: any): string[] {
 
 
 function identityParams(ment: ModelEntity): string[] {
-  // EVERY ID-BEARING OP AT ONCE, not the first one that offers a candidate.
-  //
-  // These four ops all address a single record, so all four describe the
-  // same identity — but they do not all carry the same routes. gitlab's
-  // `project` has `/api/v4/projects/{id}` under `remove` alone, while its
-  // `load` carries only sub-resources like
-  // `/api/v4/projects/{id}/uploads/{secret}/{filename}`. Returning on the
-  // first op with any candidate therefore made a PROJECT identified by
-  // `secret/filename`. The op order is now only a tie-break.
   const cands: any[] = []
 
   for (let o = 0; o < ID_OPS.length; o++) {
@@ -331,22 +210,6 @@ function identityParams(ment: ModelEntity): string[] {
         run,
         // Segments BEFORE the run: how much parent scope the route needs.
         scope: ((pt.segments || []).length - run.length),
-        // DOES THE RUN END IN THE RECORD'S OWN KEY? Then it is the
-        // record's address and nothing further is needed.
-        //
-        // This transform RENAMES that parameter to `id`, so a run ending in
-        // it is this port's own statement of what identifies the record —
-        // and the composite inference must not contradict it.
-        // `/gists/{gist_id}` becomes `/gists/{id}` and is a gist;
-        // `/gists/{gist_id}/{sha}` is a REVISION of one, and won on key
-        // length alone, so a gist came out keyed `gist_id/sha` while the
-        // generated SDK's own load match takes the single parameter. The
-        // same contradiction gave cloudsmith's repo and vulnerability
-        // compound keys their SDKs never address them by.
-        //
-        // Deliberately narrow: exactly `id` or an unrenamed `<entity>_id`,
-        // never any `*_id`. `actor_type/actor_id` IS a compound key, and a
-        // looser test breaks it.
         own: 'id' === run[run.length - 1] ||
           (ment as any).name + '_id' === run[run.length - 1],
         order: o,
@@ -354,34 +217,6 @@ function identityParams(ment: ModelEntity): string[] {
     }
   }
 
-  // WHICH ROUTE IS THE RECORD'S OWN ADDRESS.
-  //
-  // An entity gathers every route that reads it, and in a large
-  // specification most of those are sub-resources. Three earlier rules were
-  // measured against the validation corpus, and each is wrong:
-  //
-  //   The FIRST route listed gave github's `repo` the single part
-  //   `subject_digest`, from
-  //   `/repos/{owner}/{repo}/attestations/{subject_digest}` — no compound
-  //   key at all, for the entity this feature exists for. Invisible on a
-  //   small spec, where the first item route IS the record's own.
-  //
-  //   The SHORTEST route ending in a variable took cloudsmith's
-  //   `/vulnerabilities/{owner}/` — a LIST of an owner's vulnerabilities —
-  //   and cut a four-part key down to `owner`, dropping three more
-  //   composites. Ending in a variable does not make a route an address.
-  //
-  //   The LONGEST trailing run took
-  //   `/orgs/{org}/teams/{team_slug}/repos/{owner}/{repo}` and made a TEAM
-  //   identified by `owner/repo`. A deep sub-resource can carry more
-  //   adjacent variables than the record's own route does.
-  //
-  // What separates them is PARENT SCOPE: the record's own route is the
-  // least-qualified one that names it, and among equally-qualified routes
-  // the one carrying the fullest key. `/repos/{owner}/{repo}` is qualified
-  // by one segment and the attestations route by four; `/teams/{team_id}`
-  // by one and the org-team-repo route by five; cloudsmith's vulnerability
-  // routes are all qualified by one, so the fullest of them wins.
   const best = cands.reduce((b: any, c: any) => {
     if (null == b) {
       return c
@@ -401,34 +236,6 @@ function identityParams(ment: ModelEntity): string[] {
   return null == best ? [] : best.run
 }
 
-// THE PROPERTY MAPS A RESPONSE COULD BE DESCRIBING, best first.
-//
-// BOTH SPEC DIALECTS. An OpenAPI 3 response carries its schema under
-// `content['application/json']`; a SWAGGER 2 response carries it directly as
-// `schema`. Reading only the first resolved nothing for every Swagger 2 spec
-// in the validation corpus.
-//
-// JSON ONLY, where there is a choice. An operation may declare several media
-// types with different schemas, and field extraction uses the JSON one — so
-// picking whichever came first in source order could infer a path from an XML
-// or binary schema that the actual JSON record does not have.
-//
-// `allOf` IS EXPANDED, because a response that composes its entity that way
-// has neither `properties` nor `items` of its own. field extraction expands
-// it; not doing so here meant the fields were present while the id could not
-// be reconstructed.
-//
-// ONLY THE ENVELOPE IS DESCENDED, via the same `envelopeProp` rule field
-// extraction uses. Descending every object-valued property instead treats an
-// ordinary nested object as a whole record: for `{ slug, metadata: { tenant } }`
-// addressed by `{tenant}/{slug}`, `tenant` resolved to `tenant` rather than
-// `metadata.tenant` — a confidently wrong path, which is worse than no
-// mapping at all.
-//
-// ACTION POINTS ARE SKIPPED, as `identityParams` skips them: an action's
-// response is a verb's result, not a representation of the entity, so a field
-// that happens to appear there says nothing about what a returned record
-// carries.
 function responseCandidates(ment: ModelEntity, def: any): any[] {
   const out: any[] = []
   const seen = new Set<any>()
@@ -509,7 +316,6 @@ function responseCandidates(ment: ModelEntity, def: any): any[] {
           }
         }
 
-        // Swagger 2 puts it here.
         add(resdef.schema, opname)
       }
     }
@@ -519,6 +325,19 @@ function responseCandidates(ment: ModelEntity, def: any): any[] {
 }
 
 
+
+function namesEntity(schema: any, ment: ModelEntity): boolean {
+  const xref = schema?.['x-ref']
+
+  if ('string' !== typeof xref) {
+    return false
+  }
+
+  const cmp = xref.slice(xref.lastIndexOf('/') + 1)
+
+  return canonizeCmpName(cmp) === ment.name
+}
+
 // A `$ref` followed one hop, or the schema itself. apidef resolves most refs
 // before this stage; this covers the ones that survive on a nested property.
 function resolveRef(schema: any, def: any): any {
@@ -583,12 +402,6 @@ function partAliases(ment: ModelEntity, part: string): string[] {
 }
 
 
-// Where one part is carried in a given property map, or null.
-//
-// The four rules, in order, each a fact the spec states: a scalar property of
-// that name; the part naming this entity, resolved to `name`; a scalar
-// `<part>_name` / `_login` / `_slug`; or an object property's conventional
-// identifying subfield.
 function resolvePart(
   ment: ModelEntity,
   part: string,
@@ -637,52 +450,15 @@ function resolvePart(
 }
 
 
-// WHERE EACH COMPOSITE PART'S VALUE LIVES IN A RESPONSE.
-//
-// The parts are PATH PARAMETER names; a response names its fields whatever it
-// likes. Resolving one to the other is what lets an SDK put an id on a record
-// the API returned, rather than only address a record whose id it was given.
-//
-// The rules, in order, and each of them is a fact about the spec rather than
-// a guess:
-//
-//   1. a scalar field of exactly that name             -> itself
-//   2. the part names this entity, and there is a `name` -> `name`
-//      (`/repos/{owner}/{repo}` on entity `repo`, whose response calls the
-//      repository `name`)
-//   3. a scalar `<part>_name` / `<part>_login` / `<part>_slug`
-//   4. an OBJECT field of that name         -> `<part>.<conventional key>`
-//      (`owner` is a user object; the value is `owner.login`)
-//
-// A part none of these resolve is left OUT. Downstream then knows the id
-// cannot be rebuilt for that entity and can say so, which is better than a
-// confidently wrong id on a real record. guide.aon can state it instead.
 function identityFrom(
   ment: ModelEntity,
   parts: string[],
   def: any,
 ): Record<string, string> {
-  // THE RESPONSE SCHEMA IS THE AUTHORITY, not `ment.fields`.
-  //
-  // `ment.fields` is merged across load, create, update and list, so a part
-  // that exists only in a REQUEST BODY appears there too. Resolving against
-  // it recorded such a part in `from` as though a returned record carried it,
-  // and a consumer then rebuilt an id from a property the response never
-  // sends — worse than leaving the part unresolved, which at least says so.
-  //
-  // Candidate property maps, in order: the response's own properties, then
-  // one level into an envelope. A response that wraps the record
-  // (`{ item: {...} }`, `{ data: [ {...} ] }`) states the record's fields one
-  // level in, and searching only the wrapper found nothing.
   const candidates = responseCandidates(ment, def)
   const out: Record<string, string> = {}
 
   for (const part of parts) {
-    // THE WIRE NAME AS WELL AS THE MODEL NAME. `identityParams` reads the
-    // RENAMED parameter off the path segments, while a response keeps its own
-    // casing — so a `tenantKey` renamed to `tenant_key` was looked up under a
-    // name the response does not use, and the mapping was dropped for every
-    // camel-cased or depluralized parameter.
     const aliases = partAliases(ment, part)
 
     let found: string | null = null
@@ -710,15 +486,6 @@ function scalarStringField(f: any): boolean {
 }
 
 
-// WHICH PARAMETER IS THE RECORD'S OWN KEY, among several that looked
-// adjacent. The same shape apidef's id handling recognises everywhere else:
-//
-//   1. one named exactly `id`
-//   2. `<entity>_id` — the entity's own id, however the path spells it
-//   3. any `*_id` — an id by name
-//   4. failing all that, the terminal parameter
-//
-// Position is the LAST resort, not the first.
 function singleKeyOf(ment: ModelEntity, parts: string[]): string | undefined {
   if (0 === parts.length) {
     return undefined
@@ -731,12 +498,6 @@ function singleKeyOf(ment: ModelEntity, parts: string[]): string | undefined {
 }
 
 
-// The composite half of the id descriptor, or `{}` for the ordinary case.
-//
-// Emitted ONLY for a genuinely composite id (two or more addressing
-// parameters). A single-parameter entity already round-trips through one
-// `id` and gains nothing from carrying a one-element `parts`, so its
-// descriptor is left exactly as it was — no existing model output moves.
 function compositeId(
   ment: ModelEntity,
   gent?: any,
@@ -749,17 +510,6 @@ function compositeId(
   // empty `parts`, because aontu resolves an empty list to nothing and the
   // key would arrive absent — indistinguishable from never having been set.
   if (null != gid && false === gid.composite) {
-    // DISABLING COMPOSITE MUST NOT DISABLE THE ID. The correction says these
-    // adjacent parameters are not a compound key; it does not say the record
-    // has no key. Returning a bare `{}` left an entity whose response has no
-    // literal `id` with no descriptor at all — the false positive removed and
-    // nothing identifying the real key.
-    //
-    // WHICH of the adjacent parameters is that key is decided by the same
-    // id-finding rules apidef uses elsewhere, not by position. Taking the
-    // terminal one picked `archive_format` for
-    // `/artifacts/{artifact_id}/{archive_format}` — the modifier, precisely
-    // the false positive the correction exists to undo.
     return { single: singleKeyOf(ment, identityParams(ment)) } as any
   }
 
@@ -834,19 +584,6 @@ function resolveOpFields(
       req: !!fielddef.required,
       op: {},
     }
-    // Carry the spec's own words for the field, when it has any.
-    //
-    // Every generated per-entity table has a Description column and every cell
-    // was blank, because nothing ever read the property `description` the spec
-    // supplies. Trimmed, and only when it is a non-empty string: a whitespace
-    // or non-string value would put a meaningless cell where an empty one is
-    // honest.
-    // ONE LINE, not the whole description. Every generated Readme drops this
-    // straight into a markdown table cell, where a raw newline ends the row
-    // and orphans the rest of the table — and specs put bullet lists, fenced
-    // examples and multi-paragraph notes in `description`. firstSentence is
-    // the same reduction the API summary uses, so `short` means the same
-    // thing wherever it appears.
     const fdesc = (fielddef as any).description
     if ('string' === typeof fdesc && '' !== fdesc.trim()) {
       const short = firstSentence(fdesc)
@@ -855,20 +592,6 @@ function resolveOpFields(
       }
     }
 
-    // SPEC FACTS ABOUT THE FIELD, carried through verbatim.
-    //
-    // These four are declared by OpenAPI on the property and were being
-    // dropped on the floor. `readOnly` is the one that matters most: it is
-    // the difference between a field a client MAY send and one it may not,
-    // and nothing else in the model says which — so every generator has been
-    // putting server-assigned fields into the type a caller fills in.
-    //
-    // ONLY WHEN THE SPEC SAYS SO, and for the booleans only when TRUE. Each
-    // defaults to false in OpenAPI, so an absent key and an explicit `false`
-    // carry the same information; emitting the false ones would add a key to
-    // every field of every model and say nothing. Same discipline as
-    // `short`: absent means "the spec did not say", never "apidef dropped
-    // it".
     for (const flag of ['readOnly', 'writeOnly', 'deprecated'] as const) {
       if (true === (fielddef as any)[flag]) {
         mfield[flag] = true
@@ -918,7 +641,6 @@ function findGraphqlFieldDefs(
 
   const out: SchemaDef[] = []
 
-  // Sorted by construction in parse/graphql.ts, so output stays byte-stable.
   for (const fname of Object.keys(gtype.fields)) {
     const f = gtype.fields[fname]
 
@@ -941,17 +663,10 @@ function findGraphqlFieldDefs(
         // custom scalars left unconstrained.
         type: 'ENUM' === kind ? 'string' : gqlFieldType(f.type),
         required: f.reqd,
-        // GraphQL puts the field's own words on GqlField.desc (see
-        // parse/graphql.ts). resolveOpFields reads `description`, the OpenAPI
-        // spelling, so name it that here rather than teaching the reader two.
         description: f.desc,
       } as any)
     }
     else if (('OBJECT' === kind || 'INTERFACE' === kind) && !f.list) {
-      // To-one relation. The default fragment selects `team { id }`, so the
-      // response carries a nested stub object — declare it as such. Naming a
-      // flat `team_id` here would advertise a field the wire never returns,
-      // since nothing flattens the response.
       const idField = ftype.fields?.id
       if (null != idField) {
         out.push({
@@ -968,12 +683,6 @@ function findGraphqlFieldDefs(
 }
 
 
-// GraphQL named type -> the type names the field typing understands.
-//
-// Built-ins only: a custom scalar (JSON, JSONObject, Upload, ...) can hold
-// any JSON value, so advertising it as a string would misdescribe the data
-// and make generated validation reject values the schema accepts. Enums are
-// mapped by the caller, which knows they are strings.
 function gqlFieldType(typeName: string): string | undefined {
   return 'Int' === typeName ? 'integer' :
     'Float' === typeName ? 'number' :
@@ -984,20 +693,24 @@ function gqlFieldType(typeName: string): string | undefined {
 
 
 function findFieldDefs(
-  _ment: ModelEntity,
+  ment: ModelEntity,
   mop: ModelOp,
   mpoint: ModelPoint,
   def: any
 ): SchemaDef[] {
   if ('graphql' === mpoint.kind) {
-    return findGraphqlFieldDefs(_ment, mpoint, def)
+    return findGraphqlFieldDefs(ment, mpoint, def)
   }
 
+  // A verb, rather than an address: see the call site in the transform.
+  const isAction = null != (mpoint as any)?.select?.['$action']
+
   const fielddefs: SchemaDef[] = []
+
   const pathdef = def.paths[mpoint.orig]
 
   const method = mpoint.method.toLowerCase()
-  const opdef: any = pathdef[method]
+  const opdef: any = (pathdef as any)?.[method]
 
   if (opdef) {
     const responses = opdef.responses
@@ -1009,13 +722,6 @@ function findFieldDefs(
       fieldSets = getx(responses, '200 content "application/json" schema') ??
         getx(responses, '200 schema')
       if ('list' == mop.name) {
-        // List responses commonly come in three shapes:
-        //   1. direct array — { type: array, items: { ...item } }
-        //   2. wrapper object — { properties: { items: [Item], page, ... } }
-        //      (a single array-of-object property inside an object schema)
-        //   3. legacy "list of created items" under 201
-        // Resolve to the inner item schema when we can identify one
-        // unambiguously; otherwise fall through to the 200 schema as-is.
         const unwrapped = unwrapArrayWrapper(fieldSets)
         if (unwrapped) {
           fieldSets = unwrapped
@@ -1031,14 +737,6 @@ function findFieldDefs(
           getx(responses, '201 schema')
       }
 
-      // Single-entity responses get the same treatment the list branch above
-      // already gives collections: a body that is only an envelope around the
-      // entity — `{item: {...}}` — describes the WRAPPER, not the entity, so
-      // its sole property would otherwise be harvested as a field. That is
-      // how an entity `todoitem` ended up with a required `item` field of
-      // type object, which then appeared in the generated create/update data
-      // types. envelopeProp applies the same two rules used to pick the
-      // response transform, so the field list and the transform agree.
       if ('list' != mop.name) {
         const envelope = envelopeProp(fieldSets?.properties, mop.name)
         if (null != envelope) {
@@ -1047,11 +745,16 @@ function findFieldDefs(
       }
     }
 
+    if (isAction && !namesEntity(fieldSets, ment)) {
+      return fielddefs
+    }
+
     // A QUERY (RFC 10008) request body is a filter/query schema, not the
     // entity shape, so it must not contribute entity fields. Fields for a
     // QUERY op come from its response only. Other methods (POST/PUT/PATCH)
-    // carry the entity in the body, so merge as usual.
-    if (requestBody && 'query' !== method) {
+    // carry the entity in the body, so merge as usual -- except for an
+    // action, whose body is the verb's arguments and never the record.
+    if (requestBody && 'query' !== method && !isAction) {
       fieldSets = [
         fieldSets,
         getx(requestBody, 'content "application/json" schema') ??
@@ -1149,7 +852,6 @@ function findExampleObject(opdef: any): any {
   example = getx(resdef, 'examples "application/json"')
   if (null != example && 'object' === typeof example) return unwrapExample(example)
 
-  // Swagger 2.0: schema.example
   example = getx(resdef, 'schema example')
   if (null != example && 'object' === typeof example) return unwrapExample(example)
 
@@ -1166,16 +868,6 @@ function unwrapExample(example: any): any {
 }
 
 
-// unwrapArrayWrapper inspects a list-response schema and, when it is an
-// object with a single array-of-object-schema property (e.g.
-// { boards: [Board] }, { items: [Foo], page, total, ... }), returns the
-// inner item schema so that field resolution sees the actual entity
-// properties rather than the wrapper's bookkeeping.
-//
-// Returns null if the input is not unambiguously such a wrapper:
-//   - schema is already an array → return null (let caller use it directly)
-//   - no array-of-object-schema property → return null
-//   - more than one array-of-object-schema property → ambiguous, return null
 function unwrapArrayWrapper(schema: any): any {
   if (null == schema || 'object' !== typeof schema) return null
   // Direct list shape — caller can resolve from items directly.
@@ -1227,27 +919,10 @@ function mergeField(
     }
   }
 
-  // Field identity is first-writer-wins, but a DESCRIPTION is not part of
-  // identity: the op that first names a field is often not the one that
-  // documents it (a load response referencing a bare component, a create body
-  // referencing the annotated one). Take the first non-empty description in
-  // opFieldPrecedence order and keep it — dropping it left a blank cell in
-  // every generated table while the spec had the words all along.
   if (null == existingField.short && null != newField.short) {
     existingField.short = newField.short
   }
 
-  // The spec facts merge the same way, and for the same reason: one schema
-  // annotates the field and another references it bare, so taking the first
-  // declaration in opFieldPrecedence order is what finds the annotation.
-  //
-  // THE PRECEDENCE ORDER PUTS `load` FIRST, WHICH IS THE SAFE DIRECTION HERE.
-  // A field the response schema marks readOnly and a request body also lists
-  // is a self-contradictory spec — OpenAPI says a client must not send a
-  // readOnly property at all — and this resolves it by believing the
-  // restriction rather than the omission. Marking a writable field readOnly
-  // costs a caller one field; the other way round sends a value the server
-  // rejects.
   for (const flag of ['readOnly', 'writeOnly', 'deprecated', 'format'] as const) {
     if (null == existingField[flag] && null != newField[flag]) {
       (existingField as any)[flag] = newField[flag]