atproto_codegen

//// Lexicon -> Gleam codegen. Reads `<lexicons_dir>/**/*.json` and writes typed
//// modules into `<out_dir>`, importable as `<out_module>/...`. See
//// `codegen/lexicon` (parsing), `codegen/naming` (names/refs), `codegen/emit`
//// (source emission).

import argv
import atproto_codegen/config.{type Config, Config}
import atproto_codegen/emit
import atproto_codegen/lexicon.{
  type Def, type FieldType, type Lexicon, Object, Record, TArray, TRef, TUnion,
  Unsupported,
}
import atproto_codegen/naming
import gleam/erlang/application
import gleam/io
import gleam/json
import gleam/list
import gleam/option.{None, Some}
import gleam/result
import gleam/string
import simplifile

const generated_banner = "//// Generated by codegen from priv/internal.gleam. Do not edit by hand.\n\n"

const usage = "usage: gleam run -- <lexicons_dir> <out_dir> <out_module> <nsid_prefix>[,<nsid_prefix>...]"

pub fn main() -> Nil {
  case argv.load().arguments {
    [lexicons_dir, out_dir, out_module, prefixes] ->
      run(Config(
        lexicons_dir:,
        out_dir:,
        out_module:,
        nsid_prefixes: string.split(prefixes, ","),
      ))
    _ -> io.println(usage)
  }
}

fn run(cfg: Config) -> Nil {
  let assert Ok(files) = simplifile.get_files(cfg.lexicons_dir)
  let lexicons =
    files
    |> list.filter(string.ends_with(_, ".json"))
    |> list.filter_map(fn(path) {
      case load_lexicon(path) {
        Ok(lex) -> Ok(lex)
        Error(reason) -> {
          io.println("  WARN  skipping " <> path <> ": " <> reason)
          Error(Nil)
        }
      }
    })

  report_unresolved_refs(lexicons)

  let assert Ok(_) = simplifile.create_directory_all(cfg.out_dir)
  let assert Ok(internal) = simplifile.read(internal_template())
  let assert Ok(_) =
    simplifile.write(
      to: cfg.out_dir <> "/internal.gleam",
      contents: generated_banner <> internal,
    )

  list.each(lexicons, fn(lex) {
    list.each(lex.defs, fn(d) {
      case d {
        Unsupported(nsid, name, reason) ->
          io.println("  skip " <> nsid <> "#" <> name <> " (" <> reason <> ")")
        _ -> Nil
      }
    })
    case emit.emit_lexicon(cfg, lex) {
      Some(source) -> {
        let path =
          cfg.out_dir
          <> "/"
          <> naming.module_subpath(cfg.nsid_prefixes, lex.nsid)
          <> ".gleam"
        let assert Ok(_) = simplifile.create_directory_all(parent_dir(path))
        let assert Ok(_) = simplifile.write(to: path, contents: source)
        io.println("  gen  " <> path)
      }
      None -> Nil
    }
  })
  io.println("codegen: done")
}

// Resolved via the app priv dir so the tool also works installed as a dep.
fn internal_template() -> String {
  let priv =
    application.priv_directory("atproto_codegen")
    |> result.unwrap("priv")
  priv <> "/internal.gleam"
}

fn load_lexicon(path: String) -> Result(Lexicon, String) {
  use contents <- result.try(
    simplifile.read(path)
    |> result.map_error(fn(e) { "read failed: " <> string.inspect(e) }),
  )
  json.parse(contents, lexicon.lexicon_decoder())
  |> result.map_error(fn(e) { "parse failed: " <> string.inspect(e) })
}

/// Warn about any `ref` that points at a def we never generate (a different
/// lexicon's def, an external NSID, or an unsupported def). Such a ref would
/// otherwise emit Gleam that doesn't compile, with no signal from codegen.
fn report_unresolved_refs(lexicons: List(Lexicon)) -> Nil {
  let known =
    list.flat_map(lexicons, fn(lex) {
      list.filter_map(lex.defs, fn(d) {
        case d {
          Record(nsid, _) -> Ok(nsid <> "#main")
          Object(nsid, name, _) -> Ok(nsid <> "#" <> name)
          _ -> Error(Nil)
        }
      })
    })
  let refs =
    list.flat_map(lexicons, fn(lex) {
      list.flat_map(lex.defs, fn(d) {
        list.flat_map(def_properties(d), fn(p) {
          field_ref_keys(lex.nsid, p.field_type)
        })
      })
    })
  refs
  |> list.unique
  |> list.filter(fn(key) { !list.contains(known, key) })
  |> list.each(fn(key) {
    io.println(
      "  WARN  unresolved ref " <> key <> " (referencing def won't compile)",
    )
  })
}

fn def_properties(def: Def) -> List(lexicon.Property) {
  case def {
    Record(_, props) -> props
    Object(_, _, props) -> props
    _ -> []
  }
}

fn field_ref_keys(current: String, ft: FieldType) -> List(String) {
  case ft {
    TRef(ref) -> ref_key(current, ref)
    TUnion(refs) -> list.flat_map(refs, ref_key(current, _))
    TArray(inner) -> field_ref_keys(current, inner)
    _ -> []
  }
}

fn ref_key(current: String, ref: String) -> List(String) {
  let r = naming.parse_ref(current, ref)
  [r.nsid <> "#" <> r.def_name]
}

fn parent_dir(path: String) -> String {
  case string.split(path, "/") |> list.reverse {
    [_, ..rest] -> rest |> list.reverse |> string.join("/")
    [] -> "."
  }
}