Struct SchemaAnalysis 

pub struct SchemaAnalysis {
    pub schemas: BTreeMap<String, AnalyzedSchema>,
    pub dependencies: DependencyGraph,
    pub patterns: DetectedPatterns,
    pub operations: BTreeMap<String, OperationInfo>,
    pub operation_responses: BTreeMap<String, BTreeMap<String, OperationResponse>>,
    pub operation_id_aliases: BTreeMap<String, Vec<String>>,
    pub used_type_features: UsedFeatures,
    pub enum_extensions: BTreeMap<String, EnumExtensions>,
    pub validation_context: ValidationContext,
    pub untagged_union_branches: BTreeSet<String>,
}

Fields

schemas: BTreeMap<String, AnalyzedSchema>

All schemas indexed by name

dependencies: DependencyGraph

Dependency graph for generation ordering

patterns: DetectedPatterns

Detected patterns and transformations

operations: BTreeMap<String, OperationInfo>

OpenAPI operations and their request/response schemas

operation_responses: BTreeMap<String, BTreeMap<String, OperationResponse>>

Complete response contracts by emitted operation ID and response key. Unlike OperationInfo::response_schemas, this retains responses with no body as well as their selected JSON media type and SSE declaration.

operation_id_aliases: BTreeMap<String, Vec<String>>

Source operationId to emitted operation IDs. Duplicate or Rust-identifier-colliding IDs are renamed during analysis; retaining this mapping lets selector resolution report ambiguity or renaming.

used_type_features: UsedFeatures

Optional crates the TypeMapper was asked to reference during analysis (e.g. chrono when a format: date-time field became chrono::DateTime<Utc>). The generator reads this to decide which helper modules (e.g. base64_serde) to emit. Complete dependency reporting is collected from retained emitted files so pruned schemas cannot leak stale requirements.

enum_extensions: BTreeMap<String, EnumExtensions>

Q2.6: per-schema vendor enum extensions (x-enum-varnames / x-enum-descriptions). Populated during analysis when a StringEnum / ExtensibleEnum schema declares either extension; the generator uses these to override the default heuristic variant names and emit per-variant doc comments. Indexed by analyzed-schema name. Side-channel so we don’t have to touch every StringEnum constructor.

validation_context: ValidationContext

Raw, unpruned schema material used to build offline server validators. This is deliberately independent of schemas, which model pruning may mutate before server artifacts are emitted.

untagged_union_branches: BTreeSet<String>

Schemas reachable as a branch of an untagged union.

#[serde(untagged)] tries branches in order and takes the first that deserializes, so a branch that accepts more than its schema allows claims values belonging to a later branch. A struct in this set whose document states additionalProperties: false must therefore reject undeclared keys; elsewhere leniency costs nothing and keeps generated clients working when a server adds a field.

Populated at the end of SchemaAnalyzer::analyze, after inline branch schemas exist. Side-channel keyed by analyzed-schema name so no object constructor has to carry it.

Implementations

impl SchemaAnalysis

pub fn untyped_fields(&self) -> Vec<UntypedFinding>

Every generated field that carries serde_json::Value, with the reason.

Derived from the analyzed types rather than recorded as analysis runs, so the count tracks generated output: a schema referenced by fifty properties contributes fifty findings, and a pruned one contributes none.

Trait Implementations

impl Clone for SchemaAnalysis

fn clone(&self) -> SchemaAnalysis

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for SchemaAnalysis

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.