naming

Identifier naming shared by every Gleam generator.

This existed as five separate copies, one per generator, which only happened to agree. They do not have independent behaviour to have: a schema name becomes a type in types.gleam and a cross-module call like types.<name>_decoder() in client.gleam, so any drift between two copies produces a call to a function that was never generated.

Values

to_pascal_case

</>

pub fn to_pascal_case(name: String) -> String

Convert a snake_case or camelCase name to PascalCase.

gleam_routes emits the Route variants and gleam_middleware matches on them, so this has the same must-agree property as to_snake_case.

to_snake_case

</>

pub fn to_snake_case(name: String) -> String

Convert a PascalCase, camelCase, hyphenated, or dotted name to snake_case.

Names arrive straight from the spec, where “X-Request-Id”, “api.key” and “Order_Item” are all legal; every character that cannot appear in a Gleam identifier becomes an underscore.

Runs of underscores are left alone. Collapsing them would be prettier but buys nothing, and every caller has to agree character for character.

to_type_name

</>

pub fn to_type_name(name: String) -> String

Turn a schema name into a Gleam type name.

Schema names are [a-zA-Z0-9._-]+ in OpenAPI, but a Gleam type name is letters and digits only and must start with an uppercase letter. “Order_Item” is a legal schema name that generates pub type Order_Item, which does not parse.

Every reference to the type has to route through here too, or the definition and its uses end up with different names.