Exosphere.Lexicon.Parser (Exosphere v0.4.0)

Copy Markdown View Source

Parses ATProto lexicon JSON files into a normalized schema IR.

Unlike a flat extractor, this walks schema nodes recursively, preserving every node type the lexicon spec defines:

  • Primitives: string (with format/maxLength/maxGraphemes/enum), integer (minimum/maximum), boolean, bytes, unknown
  • Containers: object (properties/required), array (items)
  • Links: ref (#def, nsid, nsid#def), union (refs), cid-link, blob
  • Records: record (key + embedded object)

Nodes are plain maps tagged with :kind. Refs are kept symbolic; the generator resolves them against the full set of parsed lexicons.

Examples

{:ok, lexicons} = Parser.parse_dir("priv/lexicons")
%{id: "app.bsky.feed.post", defs: %{"main" => %{kind: :record, ...}}} =
  lexicons["app.bsky.feed.post"]

Summary

Types

lexicon()
node_kind()
schema_node()

Functions

known_types()
parse(json)

Parse a decoded lexicon JSON map.

parse_dir(dir)

Parse every lexicon JSON file under dir (recursively).

parse_file(path)

Parse a single lexicon JSON file.

parse_schema_node(node)

Parse a single schema node (e.g. an XRPC parameters block or a property) into the IR. Public for the endpoint generator.

Types

lexicon()

@type lexicon() :: %{
  id: String.t(),
  description: String.t() | nil,
  defs: %{required(String.t()) => schema_node()}
}

node_kind()

@type node_kind() ::
  :string
  | :token
  | :integer
  | :boolean
  | :bytes
  | :unknown
  | :cid_link
  | :blob
  | :array
  | :object
  | :record
  | :ref
  | :union
  | :params
  | :query
  | :procedure
  | :permission_set
  | :subscription

schema_node()

@type schema_node() :: %{kind: node_kind()} | %{optional(atom()) => term()}

Functions

known_types()

parse(json)

@spec parse(map()) :: {:ok, lexicon()} | {:error, term()}

Parse a decoded lexicon JSON map.

parse_dir(dir)

@spec parse_dir(Path.t()) ::
  {:ok, %{required(String.t()) => lexicon()}} | {:error, term()}

Parse every lexicon JSON file under dir (recursively).

Returns {:ok, %{nsid => lexicon}} or {:error, {path, reason}} for the first file that fails.

parse_file(path)

@spec parse_file(Path.t()) :: {:ok, lexicon()} | {:error, term()}

Parse a single lexicon JSON file.

parse_schema_node(node)

@spec parse_schema_node(map()) :: {:ok, schema_node()} | {:error, term()}

Parse a single schema node (e.g. an XRPC parameters block or a property) into the IR. Public for the endpoint generator.