Parse a string as Float. On failure, call on_error with the original string to produce the error value.

Lenient. Accepts every shape gleam/float.parse accepts (e.g. "3.14", "-0.5") and additionally accepts:

• integer literals — "5" parses as 5.0 (asymmetric strictness between parse.int and parse.float was a UX trap when the raw input is a user-typed numeric value).
• scientific notation — "1e3", "1.5e-2", "5E3" are all accepted; the exponent must itself be a valid integer.

Locale ambiguity warning. gleam/float.parse is documented as lenient: it returns the parsed prefix when the input has trailing non-numeric bytes. Concretely, parse.float("3,000", ...) returns Valid(3.0) because the parse stops at the comma. This silently truncates locale-formatted thousand-separated input (de_DE / fr_FR / ja_JP users typing "3,000" mean three thousand, not three) and a 1000× wrong amount can flow through the rest of the pipeline. For amount fields, address fields, and any other input where partial-parse is a bug rather than a feature, use float_strict/2 instead. (#67)

Example: parse.float(raw, fn(s) { NotAFloat(s) })

Parse a string as Float, rejecting any input the strict-float grammar does not accept end-to-end.

Strict counterpart of float/2. The lenient float/2 delegates to gleam/float.parse, which silently truncates inputs like "3,000" (returns 3.0, parse stops at the comma). For form fields where partial-parse is a bug, use this variant.

Accepts: optional leading -, then digits ("42") or digits-dot-digits ("3.14"), optionally followed by a scientific suffix ([eE]-?\d+, e.g. "1.5e-2", "5E3"). Anything else — commas, spaces, trailing letters, leading dots, multiple dots — is rejected via on_error.

Example: parse.float_strict(“3,000”, fn(s) { NotAFloat(s) }) // Invalid([NotAFloat(“3,000”)]) – not Valid(3.0)

Equivalent to float/2 for inputs the strict grammar accepts; callers can opt into strictness without rewriting their error type.

Parse a string as Int. On failure, call on_error with the original string to produce the error value.

Example: parse.int(raw, fn(s) { NotAnInteger(s) })