smalljs$molSearch⌘ KDocsPlaygroundCompareEcosystemAboutEN
Data SchemasGetting StartedIntroductionMental modelGetting StartedFrom TypeScript to view.treeProject StructureToolingEssentialsInstallationViewsState & ReactivityRoutingRenderingTestingDeploymentTroubleshootingDataData FetchingData SchemasGiper BazaMoreShowcaseFrom React, Vue & SvelteCookbookAdvancedPluginsModule metadataOfflineGhost viewsAboutFAQTeamReleasesAPI$mol_button_major$mol_button_minor$mol_string$mol_number$mol_text$mol_paragraph$mol_list$mol_row$mol_link$mol_check$mol_switch$mol_select$mol_scroll$mol_page$mol_pick

Data Schemas

Data that comes back from a network request is any — TypeScript trusts your cast, but the server might send something else. $mol ships two small runtime-schema libraries that turn untrusted JSON into a typed, validated value and fail loudly, with a readable path, when the shape is wrong. Use them right where data enters the app, most often on a fetch response.

Two libraries

$mol_data — concise, functional parsers (zod-like). You compose small parser functions and call the result on a value.$mol_schema — class-based schemas with defaults. You extend a record class and get .guard(), .cast(), .check(), and a .default.Both validate at runtime and infer the static type for you. Reach for $mol_data for quick DTOs and (de)serialization; reach for $mol_schema when you want named, reusable schema classes with default values and relaxed casting.

$mol_data

Describe the shape as a record of field parsers:1const UserDTO = $mol_data_record({2 name: $mol_data_string,3 age: $mol_data_optional( $mol_data_integer ),4 mail: $mol_data_email,5})Call it on the raw value. Valid data passes through, fully typed; bad data throws a $mol_data_error naming the exact path that failed:1const user = UserDTO( json )2// user: { readonly name: string; readonly age?: number; readonly mail: string }3 4// If json.mail is "</script>", it throws:5// ["mail"] </script> is not a /.+@.+/Reuse the inferred type anywhere with typeof UserDTO.Value:1function greet( user: typeof UserDTO.Value ) {2 return `Hello, ${ user.name }`3}The building blocks include $mol_data_string, $mol_data_number, $mol_data_integer, $mol_data_boolean, $mol_data_email, $mol_data_optional, $mol_data_nullable, $mol_data_variant (one of several types), $mol_data_array, $mol_data_dict, and $mol_data_record. $mol_data_pipe feeds a parsed value into a transform, for example an ISO string into a $mol_time_moment, which doubles as (de)serialization.

$mol_schema

Define a schema as a class that extends a record:1export class $my_user extends $mol_schema_record({2 name: $mol_schema_string,3 age: $mol_schema_natural,4}) {}You then have three ways to apply it, plus a ready-made default:1const safe = $my_user.guard( input ) // strict: throws on wrong data2const relaxed = $my_user.cast( input ) // fills defaults instead of throwing3if ( $my_user.check( input ) ) { /* input is $my_user in here */ }4const blank = $my_user.default // { name: '', age: 0 }Leaf schemas include $mol_schema_string, $mol_schema_integer, $mol_schema_natural, $mol_schema_float, $mol_schema_boolean, $mol_schema_enum([ ... ]), and $mol_schema_pattern( /re/ ). Compose them with $mol_schema_list( Item ), $mol_schema_dict([ Key, Val ]), $mol_schema_maybe( S ) (value, null, or undefined), $mol_schema_some([ ... ]) (a union), and $mol_schema_partial({ ... }). Spread another record's fields with ...Base.Fields:Edit this page on GitHubWas this helpful?YesNoPreviousData FetchingNextGiper BazaOn this pageTwo libraries$mol_data$mol_schemaValidating a fetch responseNext
Type to search the documentation.