# gizlang:text

Text and byte operations written in Gizlang over the built-in string primitives (slice, comparison, concatenation, push, string()). Strings are UTF-8: offsets count bytes, and a match always starts and ends on character boundaries because both sides are valid UTF-8.

```gizlang
import * as Text from "gizlang:text";
```

## Types

### `UnsignedParse`

```gizlang
type UnsignedParse = | { kind: "ok"; value: u64; } | { kind: "invalid"; at: i64; } | { kind: "overflow"; at: i64; }
```

Exact decimal unsigned parsing, retaining the first invalid/overflow byte.

- `ok`
- `invalid`
- `overflow`

### `IntegerParse`

```gizlang
type IntegerParse = | { kind: "ok"; value: i64 } | { kind: "invalid"; offset: i64 } | { kind: "overflow"; offset: i64 }
```

The result of reading an integer from text: the value, or where it is invalid or overflows.

- `ok`
- `invalid`
- `overflow`

### `VarintUnsigned`

```gizlang
type VarintUnsigned = { value: u64; next: i64; }
```

Bounded LEB128 offsets are byte positions, not characters or buffer slices.

- `value`: `u64`
- `next`: `i64`

### `VarintSigned`

```gizlang
type VarintSigned = { value: i64; next: i64; }
```

A signed varint that was read: its value and the offset after it.

- `value`: `i64`
- `next`: `i64`

### `VarintFailure`

```gizlang
type VarintFailure = { message: string; next: i64; continuation: bool; }
```

A varint that could not be read: why, the offset after it, and whether more bytes could finish it.

- `message`: `string`
- `next`: `i64`
- `continuation`: `bool`

### `Decimal`

```gizlang
type Decimal = { value: f64; end: i64; valid: bool; strict: bool; }
```

The number at data[start..end): a sign, digits, a fraction and an exponent,
each optional, read once into the nearest f64 and where it ends. valid: a
digit before or after the point, and digits after an exponent mark (parseF64,
gizlang:csv); strict, JSON's grammar: also digits on both sides of a point
and no leading zero (gizlang:json).

- `value`: `f64`
- `end`: `i64`
- `valid`: `bool`
- `strict`: `bool`

### `Splitter`

```gizlang
type Splitter = { size: i64; overlap: i64; keep: string; strip: bool; separators: Vec<string>; recursive: bool; }
```

A splitter's settings: TextSplitter's defaults, then the keyword arguments. keep is "", "start" or "end".

- `size`: `i64`
- `overlap`: `i64`
- `keep`: `string`
- `strip`: `bool`
- `separators`: `Vec<string>`
- `recursive`: `bool`

## Functions

### `parseU64`

```gizlang
function parseU64(text: string): UnsignedParse
```

TEXT as a decimal unsigned 64-bit integer, or where it is invalid or overflows.

Effects: fail.

### `appendCodePoint`

```gizlang
function appendCodePoint(out: mut bytes, cp: i64): void requires cp> = 0 requires cp<SCALAR_LIMIT requires cp<SURROGATE_FIRST || cp>SURROGATE_LAST
```

Append one Unicode scalar as UTF-8, preserving the existing byte buffer.

Effects: fail.

### `uvarintSize`

```gizlang
function uvarintSize(v: u64): i64
```

The bytes that V takes as an unsigned LEB128 varint.

Effects: fail.

### `varintSize`

```gizlang
function varintSize(v: i64): i64
```

The bytes that V takes as a zigzag varint.

Effects: fail.

### `writeUVarint`

```gizlang
function writeUVarint(out: mut bytes, v: u64): void
```

Appends V as an unsigned LEB128 varint.

Effects: fail.

### `writeVarint`

```gizlang
function writeVarint(out: mut bytes, v: i64): void
```

Signed integers use zigzag; IEEE data and raw signed words have other encodings.

Effects: fail.

### `readUVarint`

```gizlang
function readUVarint(data: bytes, offset: i64, bits: i64): Result<VarintUnsigned, string>
```

bits is the declared integer width. Invalid/truncated/overflow input never faults or
narrows a value silently. Existing nonminimal in-range LEB128 remains readable.

Effects: fail.

### `readUVarintWithin`

```gizlang
function readUVarintWithin(data: bytes, offset: i64, end: i64, bits: i64): Result<VarintUnsigned, VarintFailure>
```

The same decoder within an explicit byte range. Failures retain the first
unread offset; continuation marks exhaustion of the declared word width.

Effects: fail.

### `readVarint`

```gizlang
function readVarint(data: bytes, offset: i64, bits: i64): Result<VarintSigned, string>
```

The zigzag varint at OFFSET of DATA and the offset after it. It fails when the varint is cut off or has more than
BITS bits.

Effects: fail.

### `join`

```gizlang
function join<T>(parts: Vec<T>, separator: string): string
```

The items as text (`${x}`: numbers and bools too, as TypeScript joins them), SEPARATOR between each two.

### `frameStrings`

```gizlang
function frameStrings(parts: Vec<string>): bytes
```

An unambiguous ordered string sequence: count, then each UTF-8 byte length
and payload, all lengths unsigned 64-bit little endian. Delimiters in a
payload cannot change boundaries. Suitable input to a domain-separated hash.

Effects: fail.

### `frameBytes`

```gizlang
function frameBytes(parts: Vec<bytes>): bytes
```

The same framing engine for binary payloads. Callers admit output size.

Effects: fail.

### `unframeStrings`

```gizlang
function unframeStrings(data: bytes, maxParts: i64, maxBytes: i64): Result<Vec<string>, string>
```

Inverse of frameStrings; bounds are checked before allocating each payload.

Effects: fail.

### `unframeBytes`

```gizlang
function unframeBytes(data: bytes, maxParts: i64, maxBytes: i64): Result<Vec<bytes>, string>
```

The parts that frameBytes joined into DATA. It fails when there are more than MAXPARTS parts or a part has more than
MAXBYTES bytes.

Effects: fail.

### `characters`

```gizlang
function characters(text: string): i64
```

The characters (code points) of a string.

Effects: fail.

### `characterList`

```gizlang
function characterList(text: string): Vec<string>
```

The characters (code points) of a string, each as a string; an incomplete sequence at the end is one.

Effects: fail.

### `grouped`

```gizlang
function grouped(value: i64): string
```

`value` with thousands separators: 12,345.

Effects: fail.

### `clip`

```gizlang
function clip(text: string, most: i64): string
```

TEXT cut to its first MOST characters (code points).

Effects: fail.

### `asciiSpace`

```gizlang
function asciiSpace(byte: u64): bool
```

Whether a byte is ASCII whitespace as C's isspace: space, tab, newline, vertical tab, form feed, return.

### `asciiDigit`

```gizlang
function asciiDigit(byte: u64): bool
```

Whether a byte is an ASCII decimal digit.

### `asciiAlpha`

```gizlang
function asciiAlpha(byte: u64): bool
```

An ASCII letter.

### `asciiWord`

```gizlang
function asciiWord(byte: u64): bool
```

An ASCII letter, digit or underscore: a byte of a word.

### `identifierByte`

```gizlang
function identifierByte(c: u64): bool
```

Whether byte C can be part of a JavaScript identifier: an ASCII letter or digit, `_`, `$`, or a byte of a non-ASCII character.

### `asciiPunct`

```gizlang
function asciiPunct(byte: u64): bool
```

An ASCII punctuation mark: a printable byte that is not a letter, a digit or a space.

### `lowerAscii`

```gizlang
function lowerAscii(byte: u64): u64
```

One byte's ASCII lower case (headers, globs): other bytes unchanged.

Effects: fail.

### `sameFolded`

```gizlang
function sameFolded(a: bytes, i: i64, b: bytes, j: i64, n: i64): bool
```

Whether `n` bytes of `a` from `i` and of `b` from `j` are equal after ASCII lower-casing (header names, keywords).
Eight bytes are compared at a time; both ranges must lie in their data.

Effects: fail.

### `newlines`

```gizlang
function newlines(data: bytes, from: i64, to: i64): i64
```

The number of newline bytes in DATA from FROM up to TO.

Effects: fail.

### `lowerAsciiBytes`

```gizlang
function lowerAsciiBytes(data: bytes): bytes
```

ASCII folding over arbitrary bytes; non-ASCII and invalid UTF-8 stay intact.

Effects: fail.

### `optionText`

```gizlang
function optionText<T>(value: Option<T>): string
```

The text of a TypeScript T | undefined: its value's, or "undefined".

### `parseInt`

```gizlang
function parseInt(text: string, radix: i64 = 10): Option<i64>
```

TypeScript's parseInt: after leading white space and a sign, the longest prefix of digits in RADIX (2 to 36; 16 also
after `0x`, and a `0x` prefix with no radix given), or none when there is no digit. It does not detect overflow.

Effects: fail.

### `parseFloat`

```gizlang
function parseFloat(text: string): Option<f64>
```

TypeScript's parseFloat: after leading white space, the longest prefix in JSON's number grammar (a leading + too),
or none when there is none.

Effects: fail.

### `parseI64`

```gizlang
function parseI64(text: string): IntegerParse
```

A decimal integer with an optional leading minus sign.

Effects: fail.

### `writeU16LE`

```gizlang
function writeU16LE(out: mut bytes, value: u64): void
```

Appends the low 2 bytes of VALUE, little-endian.

Effects: fail.

### `writeU16BE`

```gizlang
function writeU16BE(out: mut bytes, value: u64): void
```

Appends the low 2 bytes of VALUE, big-endian.

Effects: fail.

### `writeU32LE`

```gizlang
function writeU32LE(out: mut bytes, value: u64): void
```

Appends the low 4 bytes of VALUE, little-endian.

Effects: fail.

### `writeU32BE`

```gizlang
function writeU32BE(out: mut bytes, value: u64): void
```

Appends the low 4 bytes of VALUE, big-endian.

Effects: fail.

### `writeU64LE`

```gizlang
function writeU64LE(out: mut bytes, value: u64): void
```

Appends VALUE as 8 bytes, little-endian.

Effects: fail.

### `writeU64BE`

```gizlang
function writeU64BE(out: mut bytes, value: u64): void
```

Appends VALUE as 8 bytes, big-endian.

Effects: fail.

### `writeF32LE`

```gizlang
function writeF32LE(out: mut bytes, x: f64): void
```

Appends X as a 32-bit float, little-endian.

Effects: fail.

### `writeF32BE`

```gizlang
function writeF32BE(out: mut bytes, x: f64): void
```

Appends X as a 32-bit float, big-endian.

Effects: fail.

### `writeF64LE`

```gizlang
function writeF64LE(out: mut bytes, x: f64): void
```

Appends X as a 64-bit float, little-endian.

Effects: fail.

### `writeF64BE`

```gizlang
function writeF64BE(out: mut bytes, x: f64): void
```

Appends X as a 64-bit float, big-endian.

Effects: fail.

### `signed`

```gizlang
function signed(value: u64, bits: i64): i64 requires bits> = 1 requires bits<= 64 requires bits === 64 || (value>>u64(bits)) === 0u { const half = 1u<<u64(bits-1)
```

The two's-complement value of the low `bits` bits, and back: signed(0xFFFFu, 16)
is -1 and unsigned(-1, 16) is 0xFFFF. Each requires its input to fit.

Effects: fail.

### `unsigned`

```gizlang
function unsigned(value: i64, bits: i64): u64 requires bits> = 1 requires bits<= 64 requires value> = -i64((1u<<u64(bits-1))-1u)-1 requires value<= i64((1u<<u64(bits-1))-1u) { if(value> = 0) { return u64(value)
```

The low BITS bits of VALUE in two's complement.

Effects: fail.

### `packI64s`

```gizlang
function packI64s(out: mut bytes, v: Vec<i64>): void
```

Little-endian arrays, 64 bytes a SIMD step: the same bytes as writeU64LE, writeF64LE, writeU32LE
or writeF32LE of each value, appended to `out`.

Effects: fail.

### `packF64s`

```gizlang
function packF64s(out: mut bytes, v: Vec<f64>): void
```

Appends the values of V as 64-bit little-endian floats.

Effects: fail.

### `packU32s`

```gizlang
function packU32s(out: mut bytes, v: Vec<u32>): void
```

Appends the values of V as 32-bit little-endian integers.

Effects: fail.

### `packF32s`

```gizlang
function packF32s(out: mut bytes, v: Vec<f32>): void
```

Appends the values of V as 32-bit little-endian floats.

Effects: fail.

### `unpackI64s`

```gizlang
function unpackI64s(data: bytes, at: i64, n: i64): Vec<i64>
```

`n` values from data[at..], which must hold them.

Effects: fail.

### `unpackF64s`

```gizlang
function unpackF64s(data: bytes, at: i64, n: i64): Vec<f64>
```

N 64-bit little-endian floats from AT of DATA.

Effects: fail.

### `unpackU32s`

```gizlang
function unpackU32s(data: bytes, at: i64, n: i64): Vec<u32>
```

N 32-bit little-endian integers from AT of DATA.

Effects: fail.

### `unpackF32s`

```gizlang
function unpackF32s(data: bytes, at: i64, n: i64): Vec<f32>
```

N 32-bit little-endian floats from AT of DATA.

Effects: fail.

### `digitValue`

```gizlang
function digitValue(byte: u64, base: i64): i64
```

A byte's value as a digit of `base` (2-36: 0-9, then a-z or A-Z as
10-35), or -1: one rule for parseRadix and every hex reader (gizlang:json,
gizweb, gizlang:http).

Effects: fail.

### `parseRadix`

```gizlang
function parseRadix(text: string, base: i64): IntegerParse requires base> = 2 requires base<= 36
```

A non-negative integer in base 2..36, digits in either case (the inverse of `radix`).

Effects: fail.

### `radix`

```gizlang
function radix(value: u64, base: i64): string requires base> = 2 requires base<= 36
```

`value` in base 2..36 with lowercase digits, as JavaScript's toString(base).

Effects: fail.

### `fixed`

```gizlang
function fixed(x: f64, digits: i64): string requires digits> = 0 requires digits<= 64
```

x with `digits` digits after the point, rounded half to even from x's exact
binary value, as C's printf("%.Nf") and Rust's format!("{:.N}"): checksums
and reports print the same text in every language. Pure Gizlang over a
little big integer (32-bit limbs): the mantissa is found by scaling by 2,
which is exact, so no bit cast is needed.

Effects: fail.

### `formatF64`

```gizlang
function formatF64(x: f64): string
```

x as string(x) prints it, which the compiler lowers to this: writeF64's text.

Effects: fail.

### `writeF64`

```gizlang
function writeF64(out: mut bytes, x: f64): void
```

x as JavaScript's String(x) writes it, appended to out: the shortest digits
that read back as x (Schubfach, as Java's Double.toString: one 126-bit
multiplication by a power of ten from a table, no loop over precisions),
placed as 123.5, 1e+21, 0.000001 or 1e-7; NaN, Infinity, and 0 for -0.

Effects: fail.

### `base64Encode`

```gizlang
function base64Encode(data: bytes): string
```

DATA in base64 (RFC 4648), with padding.

Effects: fail.

### `base64UrlEncode`

```gizlang
function base64UrlEncode(data: bytes): string
```

RFC 4648 URL alphabet, without padding.

Effects: fail.

### `base64UrlDecode`

```gizlang
function base64UrlDecode(text: string): Option<bytes>
```

Strict unpadded RFC 4648 URL text. Reject a noncanonical last sextet.

Effects: fail.

### `hexEncode`

```gizlang
function hexEncode(data: bytes): string
```

Two lowercase hexadecimal digits per byte (digests, tokens, wire hashes): 16 bytes a step, each nibble one
table lookup (pshufb or NEON tbl) and the digits interleaved by shuffles; byte by byte below that.

Effects: fail.

### `hexDecode`

```gizlang
function hexDecode(text: string): Option<bytes>
```

Whole bytes in either hexadecimal case; reject odd widths, whitespace and non-hex.
Validate before allocation so malformed input never leaves a partial decoded buffer.

Effects: fail.

### `hexDecodePrefix`

```gizlang
function hexDecodePrefix(data: bytes): bytes
```

Hexadecimal pairs of DATA up to the first pair with a non-hex digit or a lone last digit, as Node's Buffer.from(s,
'hex') forgives. hexDecode is the strict one.

Effects: fail.

### `base64DecodeLenient`

```gizlang
function base64DecodeLenient(data: bytes): bytes
```

Base64 of DATA in either alphabet, up to the first '=', skipping any other character, as Node's Buffer.from(s,
'base64') forgives. base64Decode is the strict one.

Effects: fail.

### `base64Decode`

```gizlang
function base64Decode(text: string): Option<bytes>
```

The bytes of padded Base64 text, or none for any other character, a bad
length or bad padding.

Effects: fail.

### `decimalPrefix`

```gizlang
function decimalPrefix(data: bytes, start: i64, end: i64): Decimal requires start> = 0 requires start<= end requires end<= len(data) { let i = start
```

The decimal number in DATA from START to END, read into the nearest f64, and where it ends.

Effects: fail.

### `quoteIdentifier`

```gizlang
function quoteIdentifier(name: string): string
```

One SQL delimited identifier, preserving every character and doubling embedded quotes.
Always quote: this is independent of a server's keyword set and needs no network client.

Effects: fail.

### `quoteIdentifierPath`

```gizlang
function quoteIdentifierPath(name: string): string
```

A dotted name with each part quoted: schema.table is "schema"."table".

Effects: fail.

### `quoteLiteral`

```gizlang
function quoteLiteral(text: string): string
```

One single-quoted literal (SQL, YAML): the text between single quotes, each embedded quote doubled.

Effects: fail.

### `quoteShell`

```gizlang
function quoteShell(text: string): string
```

One word for a POSIX shell: the text between single quotes, each embedded quote closed, escaped and reopened.

Effects: fail.

### `shellCommand`

```gizlang
function shellCommand(argv: Vec<string>): string
```

A command line for a POSIX shell: each word of ARGV quoted, joined by spaces.

Effects: fail.

### `parseF64`

```gizlang
function parseF64(text: string): Option<f64>
```

A whole string in JSON's number grammar, also with a leading + or zeros,
as the nearest f64; none for anything else.

Effects: fail.

### `splitter`

```gizlang
function splitter(kind: string): Splitter throws string
```

The splitter of `kind` with TextSplitter's defaults: character, recursive_character, python, markdown or latex.

Effects: fail.

## Methods

### `IntegerParse.valueOr`

```gizlang
function IntegerParse.valueOr(self: IntegerParse, fallback: i64): i64
```

The parsed value, or `fallback` when the text is not an integer or overflows.

### `IntegerParse.value`

```gizlang
function IntegerParse.value(self: IntegerParse): i64
```

The parsed value; a text that is not an integer is a broken invariant (assert).

Effects: fail.

### `bytes.isUtf8`

```gizlang
function bytes.isUtf8(self: bytes, start: i64, end: i64): bool requires start> = 0 requires start<= end requires end<= len(self) { return utf8Valid(self, start, end)
```

Whether bytes start..end are valid UTF-8, so string() of them cannot fail.

Effects: fail.

### `bytes.indexOf`

```gizlang
function bytes.indexOf(self: bytes, needle: bytes, start: i64): i64 requires start> = 0 requires start<= len(self)
```

The first byte offset of needle at or after start, or -1: the offsets
whose first and last bytes match (W. Muła), 128 a step while none does,
each then compared; one byte by one compare a block.

Effects: fail.

### `bytes.indexOfAny`

```gizlang
function bytes.indexOfAny(self: bytes, set: bytes, start: i64): i64 requires start> = 0 requires start<= len(self)
```

The first byte offset at or after start of any byte in set, or -1: up to
four members by comparison, more by a 256-bit map read with two nibble
lookups (each lane's row for its low nibble, and its high nibble's bit).

Effects: fail.

### `bytes.positions`

```gizlang
function bytes.positions(self: bytes, start: i64, end: i64, set: bytes, out: mut Vec<i64>): i64 requires start> = 0 requires start<= end requires end<= len(self) { return self.walkSet(start, end, set, out, true)
```

The structural index of self[start, end): the offset of every byte that is in set, in order, appended to out
(simdjson's stage 1 without its string mask). A parser of delimited records (COPY text and CSV, JSON, HTTP
headers) then walks the hits, not the bytes: 64 bytes a step, up to four members by comparison, more by the
nibble lookups of indexOfAny. Returns the count appended.

Effects: fail.

### `bytes.countAny`

```gizlang
function bytes.countAny(self: bytes, start: i64, end: i64, set: bytes): i64 requires start> = 0 requires start<= end requires end<= len(self) { let none: Vec<i64> = []
```

The count of bytes of self[start, end) that are in set (bytes.positions without its index).

Effects: fail.

### `string.byteLength`

```gizlang
function string.byteLength(self: string): i64
```

The UTF-8 bytes of the text.

### `string.byteAt`

```gizlang
function string.byteAt(self: string, index: i64): u64
```

Byte INDEX of the text's UTF-8 (R001 out of range).

Effects: fail.

### `string.byteSlice`

```gizlang
function string.byteSlice(self: string, start: i64, end: i64): string
```

Bytes START to END of the text (R001 for a bad range or a split UTF-8 character).

Effects: fail.

### `string.byteIndexOf`

```gizlang
function string.byteIndexOf(self: string, needle: string, from: i64 = 0): i64
```

The first byte offset of NEEDLE at or after byte FROM (clamped to the text), or -1.

Effects: fail.

### `string.indexOf`

```gizlang
function string.indexOf(self: string, needle: string, from: i64 = 0): i64
```

The first place of NEEDLE at or after FROM (clamped to the text, as TypeScript does), or -1.

Effects: fail.

### `string.includes`

```gizlang
function string.includes(self: string, needle: string): bool
```

Whether NEEDLE occurs in the string.

Effects: fail.

### `string.startsWith`

```gizlang
function string.startsWith(self: string, prefix: string, position: i64 = 0): bool
```

Whether prefix is at byte POSITION (0 by default; clamped to the text, as in JavaScript), bytewise, without a copy:
a slice could split a UTF-8 character (R001).

Effects: fail.

### `bytes.startsWith`

```gizlang
function bytes.startsWith(self: bytes, prefix: bytes, position: i64 = 0): bool
```

Whether prefix is at POSITION (0 by default; clamped to self, as in JavaScript), without a copy.

Effects: fail.

### `string.endsWith`

```gizlang
function string.endsWith(self: string, suffix: string): bool
```

Whether the string ends with SUFFIX.

Effects: fail.

### `string.split`

```gizlang
function string.split(self: string, separator: string): Vec<string> requires separator.byteLength>0 { const text = utf8(self)
```

The pieces between separators; an empty separator is not allowed.

Effects: fail.

### `string.replaceAll`

```gizlang
function string.replaceAll(self: string, from: string, to: string): string requires from.byteLength>0 { return join(self.split(from), to)
```

The string with each FROM replaced by TO.

Effects: fail.

### `string.repeat`

```gizlang
function string.repeat(self: string, count: i64): string requires count> = 0 { let text = ""
```

The string COUNT times.

Effects: fail.

### `string.trim`

```gizlang
function string.trim(self: string): string
```

Without leading and trailing ASCII whitespace.

Effects: fail.

### `string.trimStart`

```gizlang
function string.trimStart(self: string): string
```

The string without the ASCII whitespace at its start.

Effects: fail.

### `string.trimEnd`

```gizlang
function string.trimEnd(self: string): string
```

The string without the ASCII whitespace at its end.

Effects: fail.

### `string.lastIndexOf`

```gizlang
function string.lastIndexOf(self: string, needle: string): i64
```

The last byte offset of needle, or -1; "" is found at the end. Bytewise,
as startsWith: a slice at each offset would split a UTF-8 character (R001).

Effects: fail.

### `string.byteLastIndexOf`

```gizlang
function string.byteLastIndexOf(self: string, needle: string): i64
```

The last byte offset of NEEDLE, or -1; "" is found at the end.

Effects: fail.

### `string.padStart`

```gizlang
function string.padStart(self: string, length: i64, fill: string): string
```

Padded with fill (repeated, the last copy cut) to `length` bytes; a longer string is unchanged.

Effects: fail.

### `string.padEnd`

```gizlang
function string.padEnd(self: string, length: i64, fill: string): string
```

The string followed by copies of FILL up to LENGTH bytes.

Effects: fail.

### `string.toUpperCase`

```gizlang
function string.toUpperCase(self: string): string
```

ASCII case mapping; other characters are unchanged.

Effects: fail.

### `string.toLowerCase`

```gizlang
function string.toLowerCase(self: string): string
```

The string with A to Z in lower case (ASCII only).

Effects: fail.

### `string.localeCompare`

```gizlang
function string.localeCompare(self: string, other: string): i64
```

-1, 0 or 1 as SELF comes before, with or after OTHER in code point order (UTF-8 byte order): TypeScript's
localeCompare without a locale's collation, so `xs.toSorted((a, b) => a.localeCompare(b))` sorts.

Effects: fail.

### `string.charAt`

```gizlang
function string.charAt(self: string, index: i64): string
```

The UTF-16 unit at INDEX as a string, or "" out of range.

Effects: fail.

### `string.substring`

```gizlang
function string.substring(self: string, start: i64, end?: i64): string
```

TypeScript's substring: the text from START to END (the length when left out), each clamped to 0..length, in either
order.

Effects: fail.

### `f64.toFixed`

```gizlang
function f64.toFixed(self: f64, digits: i64): string
```

The value with DIGITS digits after the point (0 to 20), rounded half away from zero: TypeScript's toFixed for
values under 1e21.

Effects: fail.

### `i64.toFixed`

```gizlang
function i64.toFixed(self: i64, digits: i64): string
```

The integer with DIGITS zeros after the point.

Effects: fail.

### `bytes.readU32LE`

```gizlang
function bytes.readU32LE(self: bytes, offset: i64): u64 requires offset> = 0 requires offset<= len(self)-4
```

The unsigned 32-bit little-endian integer at OFFSET.

Effects: fail.

### `bytes.readU64LE`

```gizlang
function bytes.readU64LE(self: bytes, offset: i64): u64
```

One bounds check and one load; out of range faults as indexing does.

Effects: fail.

### `bytes.copyFrom`

```gizlang
function bytes.copyFrom(self: mut bytes, at: i64, source: bytes, from: i64, size: i64): void
```

Writes SIZE bytes of SOURCE from FROM over this buffer from AT, in place, as one copy: a loop of `self[i] = x` through
a field path made the buffer its own again at each byte. A range outside its buffer faults as indexing does.

Effects: fail.

### `bytes.copyWithin`

```gizlang
function bytes.copyWithin(self: mut bytes, at: i64, from: i64, size: i64): void
```

Copies SIZE bytes of this buffer from FROM to AT, in place; the ranges may overlap (bytes.copyFrom).

Effects: fail.

### `bytes.readU32BE`

```gizlang
function bytes.readU32BE(self: bytes, offset: i64): u64 requires offset> = 0 requires offset<= len(self)-4
```

The unsigned 32-bit big-endian integer at OFFSET.

Effects: fail.

### `bytes.readU16LE`

```gizlang
function bytes.readU16LE(self: bytes, offset: i64): u64
```

The unsigned 16-bit little-endian integer at OFFSET.

Effects: fail.

### `bytes.readU16BE`

```gizlang
function bytes.readU16BE(self: bytes, offset: i64): u64
```

The unsigned 16-bit big-endian integer at OFFSET.

Effects: fail.

### `bytes.readU64BE`

```gizlang
function bytes.readU64BE(self: bytes, offset: i64): u64
```

The unsigned 64-bit big-endian integer at OFFSET.

Effects: fail.

### `bytes.readF32LE`

```gizlang
function bytes.readF32LE(self: bytes, offset: i64): f64
```

The 32-bit little-endian float at OFFSET.

Effects: fail.

### `bytes.readF32BE`

```gizlang
function bytes.readF32BE(self: bytes, offset: i64): f64
```

The 32-bit big-endian float at OFFSET.

Effects: fail.

### `bytes.readF64LE`

```gizlang
function bytes.readF64LE(self: bytes, offset: i64): f64
```

The 64-bit little-endian float at OFFSET.

Effects: fail.

### `bytes.readF64BE`

```gizlang
function bytes.readF64BE(self: bytes, offset: i64): f64
```

The 64-bit big-endian float at OFFSET.

Effects: fail.

### `bytes.setU16LE`

```gizlang
function bytes.setU16LE(self: mut bytes, offset: i64, value: u64): void
```

Writes the low 2, 4 or 8 bytes of value at offset, little-endian, over the bytes there.

Effects: fail.

### `bytes.setU32LE`

```gizlang
function bytes.setU32LE(self: mut bytes, offset: i64, value: u64): void
```

Writes the low 4 bytes of VALUE at OFFSET, little-endian.

Effects: fail.

### `bytes.setU64LE`

```gizlang
function bytes.setU64LE(self: mut bytes, offset: i64, value: u64): void
```

Writes VALUE at OFFSET as 8 bytes, little-endian.

Effects: fail.

### `Splitter.split`

```gizlang
function Splitter.split(self: Splitter, text: string): Vec<string>
```

The chunks of `text`: split_text of the splitter.

Effects: fail.

## Constants

### `SCALAR_LIMIT`

```gizlang
const SCALAR_LIMIT = 1114112
```

The number of Unicode code points: scalar values are below it.

### `SURROGATE_FIRST`

```gizlang
const SURROGATE_FIRST = 55296
```

The first surrogate code point, which is not a scalar value.

### `SURROGATE_LAST`

```gizlang
const SURROGATE_LAST = 57343
```

The last surrogate code point.

### `MAX_VARINT_BYTES`

```gizlang
const MAX_VARINT_BYTES = 10
```

The most bytes of one encoded varint.

gizlang:text

Text and byte operations written in Gizlang over the built-in string primitives (slice, comparison, concatenation, push, string()). Strings are UTF-8: offsets count bytes, and a match always starts and ends on character boundaries because both sides are valid UTF-8.

gizlang

import * as Text from "gizlang:text";

Types

`UnsignedParse`

gizlang

type UnsignedParse = | { kind: "ok"; value: u64; } | { kind: "invalid"; at: i64; } | { kind: "overflow"; at: i64; }

Exact decimal unsigned parsing, retaining the first invalid/overflow byte.

ok
invalid
overflow

`IntegerParse`

gizlang

type IntegerParse = | { kind: "ok"; value: i64 } | { kind: "invalid"; offset: i64 } | { kind: "overflow"; offset: i64 }

The result of reading an integer from text: the value, or where it is invalid or overflows.

ok
invalid
overflow

`VarintUnsigned`

gizlang

type VarintUnsigned = { value: u64; next: i64; }

Bounded LEB128 offsets are byte positions, not characters or buffer slices.

value: u64
next: i64

`VarintSigned`

gizlang

type VarintSigned = { value: i64; next: i64; }

A signed varint that was read: its value and the offset after it.

value: i64
next: i64

`VarintFailure`

gizlang

type VarintFailure = { message: string; next: i64; continuation: bool; }

A varint that could not be read: why, the offset after it, and whether more bytes could finish it.

message: string
next: i64
continuation: bool

`Decimal`

gizlang

type Decimal = { value: f64; end: i64; valid: bool; strict: bool; }

The number at data[start..end): a sign, digits, a fraction and an exponent, each optional, read once into the nearest f64 and where it ends. valid: a digit before or after the point, and digits after an exponent mark (parseF64, gizlang:csv); strict, JSON's grammar: also digits on both sides of a point and no leading zero (gizlang:json).

value: f64
end: i64
valid: bool
strict: bool

`Splitter`

gizlang

type Splitter = { size: i64; overlap: i64; keep: string; strip: bool; separators: Vec<string>; recursive: bool; }

A splitter's settings: TextSplitter's defaults, then the keyword arguments. keep is "", "start" or "end".

size: i64
overlap: i64
keep: string
strip: bool
separators: Vec<string>
recursive: bool

Functions

`parseU64`

gizlang

function parseU64(text: string): UnsignedParse

TEXT as a decimal unsigned 64-bit integer, or where it is invalid or overflows.

Effects: fail.

`appendCodePoint`

gizlang

function appendCodePoint(out: mut bytes, cp: i64): void requires cp> = 0 requires cp<SCALAR_LIMIT requires cp<SURROGATE_FIRST || cp>SURROGATE_LAST

Append one Unicode scalar as UTF-8, preserving the existing byte buffer.

Effects: fail.

`uvarintSize`

gizlang

function uvarintSize(v: u64): i64

The bytes that V takes as an unsigned LEB128 varint.

Effects: fail.

`varintSize`

gizlang

function varintSize(v: i64): i64

The bytes that V takes as a zigzag varint.

Effects: fail.

`writeUVarint`

gizlang

function writeUVarint(out: mut bytes, v: u64): void

Appends V as an unsigned LEB128 varint.

Effects: fail.

`writeVarint`

gizlang

function writeVarint(out: mut bytes, v: i64): void

Signed integers use zigzag; IEEE data and raw signed words have other encodings.

Effects: fail.

`readUVarint`

gizlang

function readUVarint(data: bytes, offset: i64, bits: i64): Result<VarintUnsigned, string>

bits is the declared integer width. Invalid/truncated/overflow input never faults or narrows a value silently. Existing nonminimal in-range LEB128 remains readable.

Effects: fail.

`readUVarintWithin`

gizlang

function readUVarintWithin(data: bytes, offset: i64, end: i64, bits: i64): Result<VarintUnsigned, VarintFailure>

The same decoder within an explicit byte range. Failures retain the first unread offset; continuation marks exhaustion of the declared word width.

Effects: fail.

`readVarint`

gizlang

function readVarint(data: bytes, offset: i64, bits: i64): Result<VarintSigned, string>

The zigzag varint at OFFSET of DATA and the offset after it. It fails when the varint is cut off or has more than BITS bits.

Effects: fail.

`join`

gizlang

function join<T>(parts: Vec<T>, separator: string): string

The items as text (${x}: numbers and bools too, as TypeScript joins them), SEPARATOR between each two.

`frameStrings`

gizlang

function frameStrings(parts: Vec<string>): bytes

An unambiguous ordered string sequence: count, then each UTF-8 byte length and payload, all lengths unsigned 64-bit little endian. Delimiters in a payload cannot change boundaries. Suitable input to a domain-separated hash.

Effects: fail.

`frameBytes`

gizlang

function frameBytes(parts: Vec<bytes>): bytes

The same framing engine for binary payloads. Callers admit output size.

Effects: fail.

`unframeStrings`

gizlang

function unframeStrings(data: bytes, maxParts: i64, maxBytes: i64): Result<Vec<string>, string>

Inverse of frameStrings; bounds are checked before allocating each payload.

Effects: fail.

`unframeBytes`

gizlang

function unframeBytes(data: bytes, maxParts: i64, maxBytes: i64): Result<Vec<bytes>, string>

The parts that frameBytes joined into DATA. It fails when there are more than MAXPARTS parts or a part has more than MAXBYTES bytes.

Effects: fail.

`characters`

gizlang

function characters(text: string): i64

The characters (code points) of a string.

Effects: fail.

`characterList`

gizlang

function characterList(text: string): Vec<string>

The characters (code points) of a string, each as a string; an incomplete sequence at the end is one.

Effects: fail.

`grouped`

gizlang

function grouped(value: i64): string

value with thousands separators: 12,345.

Effects: fail.

`clip`

gizlang

function clip(text: string, most: i64): string

TEXT cut to its first MOST characters (code points).

Effects: fail.

`asciiSpace`

gizlang

function asciiSpace(byte: u64): bool

Whether a byte is ASCII whitespace as C's isspace: space, tab, newline, vertical tab, form feed, return.

`asciiDigit`

gizlang

function asciiDigit(byte: u64): bool

Whether a byte is an ASCII decimal digit.

`asciiAlpha`

gizlang

function asciiAlpha(byte: u64): bool

An ASCII letter.

`asciiWord`

gizlang

function asciiWord(byte: u64): bool

An ASCII letter, digit or underscore: a byte of a word.

`identifierByte`

gizlang

function identifierByte(c: u64): bool

Whether byte C can be part of a JavaScript identifier: an ASCII letter or digit, _, $, or a byte of a non-ASCII character.

`asciiPunct`

gizlang

function asciiPunct(byte: u64): bool

An ASCII punctuation mark: a printable byte that is not a letter, a digit or a space.

`lowerAscii`

gizlang

function lowerAscii(byte: u64): u64

One byte's ASCII lower case (headers, globs): other bytes unchanged.

Effects: fail.

`sameFolded`

gizlang

function sameFolded(a: bytes, i: i64, b: bytes, j: i64, n: i64): bool

Whether n bytes of a from i and of b from j are equal after ASCII lower-casing (header names, keywords). Eight bytes are compared at a time; both ranges must lie in their data.

Effects: fail.

`newlines`

gizlang

function newlines(data: bytes, from: i64, to: i64): i64

The number of newline bytes in DATA from FROM up to TO.

Effects: fail.

`lowerAsciiBytes`

gizlang

function lowerAsciiBytes(data: bytes): bytes

ASCII folding over arbitrary bytes; non-ASCII and invalid UTF-8 stay intact.

Effects: fail.

`optionText`

gizlang

function optionText<T>(value: Option<T>): string

The text of a TypeScript T | undefined: its value's, or "undefined".

`parseInt`

gizlang

function parseInt(text: string, radix: i64 = 10): Option<i64>

TypeScript's parseInt: after leading white space and a sign, the longest prefix of digits in RADIX (2 to 36; 16 also after 0x, and a 0x prefix with no radix given), or none when there is no digit. It does not detect overflow.

Effects: fail.

`parseFloat`

gizlang

function parseFloat(text: string): Option<f64>

TypeScript's parseFloat: after leading white space, the longest prefix in JSON's number grammar (a leading + too), or none when there is none.

Effects: fail.

`parseI64`

gizlang

function parseI64(text: string): IntegerParse

A decimal integer with an optional leading minus sign.

Effects: fail.

`writeU16LE`

gizlang

function writeU16LE(out: mut bytes, value: u64): void

Appends the low 2 bytes of VALUE, little-endian.

Effects: fail.

`writeU16BE`

gizlang

function writeU16BE(out: mut bytes, value: u64): void

Appends the low 2 bytes of VALUE, big-endian.

Effects: fail.

`writeU32LE`

gizlang

function writeU32LE(out: mut bytes, value: u64): void

Appends the low 4 bytes of VALUE, little-endian.

Effects: fail.

`writeU32BE`

gizlang

function writeU32BE(out: mut bytes, value: u64): void

Appends the low 4 bytes of VALUE, big-endian.

Effects: fail.

`writeU64LE`

gizlang

function writeU64LE(out: mut bytes, value: u64): void

Appends VALUE as 8 bytes, little-endian.

Effects: fail.

`writeU64BE`

gizlang

function writeU64BE(out: mut bytes, value: u64): void

Appends VALUE as 8 bytes, big-endian.

Effects: fail.

`writeF32LE`

gizlang

function writeF32LE(out: mut bytes, x: f64): void

Appends X as a 32-bit float, little-endian.

Effects: fail.

`writeF32BE`

gizlang

function writeF32BE(out: mut bytes, x: f64): void

Appends X as a 32-bit float, big-endian.

Effects: fail.

`writeF64LE`

gizlang

function writeF64LE(out: mut bytes, x: f64): void

Appends X as a 64-bit float, little-endian.

Effects: fail.

`writeF64BE`

gizlang

function writeF64BE(out: mut bytes, x: f64): void

Appends X as a 64-bit float, big-endian.

Effects: fail.

`signed`

gizlang

function signed(value: u64, bits: i64): i64 requires bits> = 1 requires bits<= 64 requires bits === 64 || (value>>u64(bits)) === 0u { const half = 1u<<u64(bits-1)

The two's-complement value of the low bits bits, and back: signed(0xFFFFu, 16) is -1 and unsigned(-1, 16) is 0xFFFF. Each requires its input to fit.

Effects: fail.

`unsigned`

gizlang

function unsigned(value: i64, bits: i64): u64 requires bits> = 1 requires bits<= 64 requires value> = -i64((1u<<u64(bits-1))-1u)-1 requires value<= i64((1u<<u64(bits-1))-1u) { if(value> = 0) { return u64(value)

The low BITS bits of VALUE in two's complement.

Effects: fail.

`packI64s`

gizlang

function packI64s(out: mut bytes, v: Vec<i64>): void

Little-endian arrays, 64 bytes a SIMD step: the same bytes as writeU64LE, writeF64LE, writeU32LE or writeF32LE of each value, appended to out.

Effects: fail.

`packF64s`

gizlang

function packF64s(out: mut bytes, v: Vec<f64>): void

Appends the values of V as 64-bit little-endian floats.

Effects: fail.

`packU32s`

gizlang

function packU32s(out: mut bytes, v: Vec<u32>): void

Appends the values of V as 32-bit little-endian integers.

Effects: fail.

`packF32s`

gizlang

function packF32s(out: mut bytes, v: Vec<f32>): void

Appends the values of V as 32-bit little-endian floats.

Effects: fail.

`unpackI64s`

gizlang

function unpackI64s(data: bytes, at: i64, n: i64): Vec<i64>

n values from data[at..], which must hold them.

Effects: fail.

`unpackF64s`

gizlang

function unpackF64s(data: bytes, at: i64, n: i64): Vec<f64>

N 64-bit little-endian floats from AT of DATA.

Effects: fail.

`unpackU32s`

gizlang

function unpackU32s(data: bytes, at: i64, n: i64): Vec<u32>

N 32-bit little-endian integers from AT of DATA.

Effects: fail.

`unpackF32s`

gizlang

function unpackF32s(data: bytes, at: i64, n: i64): Vec<f32>

N 32-bit little-endian floats from AT of DATA.

Effects: fail.

`digitValue`

gizlang

function digitValue(byte: u64, base: i64): i64

A byte's value as a digit of base (2-36: 0-9, then a-z or A-Z as 10-35), or -1: one rule for parseRadix and every hex reader (gizlang:json, gizweb, gizlang:http).

Effects: fail.

`parseRadix`

gizlang

function parseRadix(text: string, base: i64): IntegerParse requires base> = 2 requires base<= 36

A non-negative integer in base 2..36, digits in either case (the inverse of radix).

Effects: fail.

`radix`

gizlang

function radix(value: u64, base: i64): string requires base> = 2 requires base<= 36

value in base 2..36 with lowercase digits, as JavaScript's toString(base).

Effects: fail.

`fixed`

gizlang

function fixed(x: f64, digits: i64): string requires digits> = 0 requires digits<= 64

x with digits digits after the point, rounded half to even from x's exact binary value, as C's printf("%.Nf") and Rust's format!("{:.N}"): checksums and reports print the same text in every language. Pure Gizlang over a little big integer (32-bit limbs): the mantissa is found by scaling by 2, which is exact, so no bit cast is needed.

Effects: fail.

`formatF64`

gizlang

function formatF64(x: f64): string

x as string(x) prints it, which the compiler lowers to this: writeF64's text.

Effects: fail.

`writeF64`

gizlang

function writeF64(out: mut bytes, x: f64): void

x as JavaScript's String(x) writes it, appended to out: the shortest digits that read back as x (Schubfach, as Java's Double.toString: one 126-bit multiplication by a power of ten from a table, no loop over precisions), placed as 123.5, 1e+21, 0.000001 or 1e-7; NaN, Infinity, and 0 for -0.

Effects: fail.

`base64Encode`

gizlang

function base64Encode(data: bytes): string

DATA in base64 (RFC 4648), with padding.

Effects: fail.

`base64UrlEncode`

gizlang

function base64UrlEncode(data: bytes): string

RFC 4648 URL alphabet, without padding.

Effects: fail.

`base64UrlDecode`

gizlang

function base64UrlDecode(text: string): Option<bytes>

Strict unpadded RFC 4648 URL text. Reject a noncanonical last sextet.

Effects: fail.

`hexEncode`

gizlang

function hexEncode(data: bytes): string

Two lowercase hexadecimal digits per byte (digests, tokens, wire hashes): 16 bytes a step, each nibble one table lookup (pshufb or NEON tbl) and the digits interleaved by shuffles; byte by byte below that.

Effects: fail.

`hexDecode`

gizlang

function hexDecode(text: string): Option<bytes>

Whole bytes in either hexadecimal case; reject odd widths, whitespace and non-hex. Validate before allocation so malformed input never leaves a partial decoded buffer.

Effects: fail.

`hexDecodePrefix`

gizlang

function hexDecodePrefix(data: bytes): bytes

Hexadecimal pairs of DATA up to the first pair with a non-hex digit or a lone last digit, as Node's Buffer.from(s, 'hex') forgives. hexDecode is the strict one.

Effects: fail.

`base64DecodeLenient`

gizlang

function base64DecodeLenient(data: bytes): bytes

Base64 of DATA in either alphabet, up to the first '=', skipping any other character, as Node's Buffer.from(s, 'base64') forgives. base64Decode is the strict one.

Effects: fail.

`base64Decode`

gizlang

function base64Decode(text: string): Option<bytes>

The bytes of padded Base64 text, or none for any other character, a bad length or bad padding.

Effects: fail.

`decimalPrefix`

gizlang

function decimalPrefix(data: bytes, start: i64, end: i64): Decimal requires start> = 0 requires start<= end requires end<= len(data) { let i = start

The decimal number in DATA from START to END, read into the nearest f64, and where it ends.

Effects: fail.

`quoteIdentifier`

gizlang

function quoteIdentifier(name: string): string

One SQL delimited identifier, preserving every character and doubling embedded quotes. Always quote: this is independent of a server's keyword set and needs no network client.

Effects: fail.

`quoteIdentifierPath`

gizlang

function quoteIdentifierPath(name: string): string

A dotted name with each part quoted: schema.table is "schema"."table".

Effects: fail.

`quoteLiteral`

gizlang

function quoteLiteral(text: string): string

One single-quoted literal (SQL, YAML): the text between single quotes, each embedded quote doubled.

Effects: fail.

`quoteShell`

gizlang

function quoteShell(text: string): string

One word for a POSIX shell: the text between single quotes, each embedded quote closed, escaped and reopened.

Effects: fail.

`shellCommand`

gizlang

function shellCommand(argv: Vec<string>): string

A command line for a POSIX shell: each word of ARGV quoted, joined by spaces.

Effects: fail.

`parseF64`

gizlang

function parseF64(text: string): Option<f64>

A whole string in JSON's number grammar, also with a leading + or zeros, as the nearest f64; none for anything else.

Effects: fail.

`splitter`

gizlang

function splitter(kind: string): Splitter throws string

The splitter of kind with TextSplitter's defaults: character, recursive_character, python, markdown or latex.

Effects: fail.

Methods

`IntegerParse.valueOr`

gizlang

function IntegerParse.valueOr(self: IntegerParse, fallback: i64): i64

The parsed value, or fallback when the text is not an integer or overflows.

`IntegerParse.value`

gizlang

function IntegerParse.value(self: IntegerParse): i64

The parsed value; a text that is not an integer is a broken invariant (assert).

Effects: fail.

`bytes.isUtf8`

gizlang

function bytes.isUtf8(self: bytes, start: i64, end: i64): bool requires start> = 0 requires start<= end requires end<= len(self) { return utf8Valid(self, start, end)

Whether bytes start..end are valid UTF-8, so string() of them cannot fail.

Effects: fail.

`bytes.indexOf`

gizlang

function bytes.indexOf(self: bytes, needle: bytes, start: i64): i64 requires start> = 0 requires start<= len(self)

The first byte offset of needle at or after start, or -1: the offsets whose first and last bytes match (W. Muła), 128 a step while none does, each then compared; one byte by one compare a block.

Effects: fail.

`bytes.indexOfAny`

gizlang

function bytes.indexOfAny(self: bytes, set: bytes, start: i64): i64 requires start> = 0 requires start<= len(self)

The first byte offset at or after start of any byte in set, or -1: up to four members by comparison, more by a 256-bit map read with two nibble lookups (each lane's row for its low nibble, and its high nibble's bit).

Effects: fail.

`bytes.positions`

gizlang

function bytes.positions(self: bytes, start: i64, end: i64, set: bytes, out: mut Vec<i64>): i64 requires start> = 0 requires start<= end requires end<= len(self) { return self.walkSet(start, end, set, out, true)

The structural index of self[start, end): the offset of every byte that is in set, in order, appended to out (simdjson's stage 1 without its string mask). A parser of delimited records (COPY text and CSV, JSON, HTTP headers) then walks the hits, not the bytes: 64 bytes a step, up to four members by comparison, more by the nibble lookups of indexOfAny. Returns the count appended.

Effects: fail.

`bytes.countAny`

gizlang

function bytes.countAny(self: bytes, start: i64, end: i64, set: bytes): i64 requires start> = 0 requires start<= end requires end<= len(self) { let none: Vec<i64> = []

The count of bytes of self[start, end) that are in set (bytes.positions without its index).

Effects: fail.

`string.byteLength`

gizlang

function string.byteLength(self: string): i64

The UTF-8 bytes of the text.

`string.byteAt`

gizlang

function string.byteAt(self: string, index: i64): u64

Byte INDEX of the text's UTF-8 (R001 out of range).

Effects: fail.

`string.byteSlice`

gizlang

function string.byteSlice(self: string, start: i64, end: i64): string

Bytes START to END of the text (R001 for a bad range or a split UTF-8 character).

Effects: fail.

`string.byteIndexOf`

gizlang

function string.byteIndexOf(self: string, needle: string, from: i64 = 0): i64

The first byte offset of NEEDLE at or after byte FROM (clamped to the text), or -1.

Effects: fail.

`string.indexOf`

gizlang

function string.indexOf(self: string, needle: string, from: i64 = 0): i64

The first place of NEEDLE at or after FROM (clamped to the text, as TypeScript does), or -1.

Effects: fail.

`string.includes`

gizlang

function string.includes(self: string, needle: string): bool

Whether NEEDLE occurs in the string.

Effects: fail.

`string.startsWith`

gizlang

function string.startsWith(self: string, prefix: string, position: i64 = 0): bool

Whether prefix is at byte POSITION (0 by default; clamped to the text, as in JavaScript), bytewise, without a copy: a slice could split a UTF-8 character (R001).

Effects: fail.

`bytes.startsWith`

gizlang

function bytes.startsWith(self: bytes, prefix: bytes, position: i64 = 0): bool

Whether prefix is at POSITION (0 by default; clamped to self, as in JavaScript), without a copy.

Effects: fail.

`string.endsWith`

gizlang

function string.endsWith(self: string, suffix: string): bool

Whether the string ends with SUFFIX.

Effects: fail.

`string.split`

gizlang

function string.split(self: string, separator: string): Vec<string> requires separator.byteLength>0 { const text = utf8(self)

The pieces between separators; an empty separator is not allowed.

Effects: fail.

`string.replaceAll`

gizlang

function string.replaceAll(self: string, from: string, to: string): string requires from.byteLength>0 { return join(self.split(from), to)

The string with each FROM replaced by TO.

Effects: fail.

`string.repeat`

gizlang

function string.repeat(self: string, count: i64): string requires count> = 0 { let text = ""

The string COUNT times.

Effects: fail.

`string.trim`

gizlang

function string.trim(self: string): string

Without leading and trailing ASCII whitespace.

Effects: fail.

`string.trimStart`

gizlang

function string.trimStart(self: string): string

The string without the ASCII whitespace at its start.

Effects: fail.

`string.trimEnd`

gizlang

function string.trimEnd(self: string): string

The string without the ASCII whitespace at its end.

Effects: fail.

`string.lastIndexOf`

gizlang

function string.lastIndexOf(self: string, needle: string): i64

The last byte offset of needle, or -1; "" is found at the end. Bytewise, as startsWith: a slice at each offset would split a UTF-8 character (R001).

Effects: fail.

`string.byteLastIndexOf`

gizlang

function string.byteLastIndexOf(self: string, needle: string): i64

The last byte offset of NEEDLE, or -1; "" is found at the end.

Effects: fail.

`string.padStart`

gizlang

function string.padStart(self: string, length: i64, fill: string): string

Padded with fill (repeated, the last copy cut) to length bytes; a longer string is unchanged.

Effects: fail.

`string.padEnd`

gizlang

function string.padEnd(self: string, length: i64, fill: string): string

The string followed by copies of FILL up to LENGTH bytes.

Effects: fail.

`string.toUpperCase`

gizlang

function string.toUpperCase(self: string): string

ASCII case mapping; other characters are unchanged.

Effects: fail.

`string.toLowerCase`

gizlang

function string.toLowerCase(self: string): string

The string with A to Z in lower case (ASCII only).

Effects: fail.

`string.localeCompare`

gizlang

function string.localeCompare(self: string, other: string): i64

-1, 0 or 1 as SELF comes before, with or after OTHER in code point order (UTF-8 byte order): TypeScript's localeCompare without a locale's collation, so xs.toSorted((a, b) => a.localeCompare(b)) sorts.

Effects: fail.

`string.charAt`

gizlang

function string.charAt(self: string, index: i64): string

The UTF-16 unit at INDEX as a string, or "" out of range.

Effects: fail.

`string.substring`

gizlang

function string.substring(self: string, start: i64, end?: i64): string

TypeScript's substring: the text from START to END (the length when left out), each clamped to 0..length, in either order.

Effects: fail.

`f64.toFixed`

gizlang

function f64.toFixed(self: f64, digits: i64): string

The value with DIGITS digits after the point (0 to 20), rounded half away from zero: TypeScript's toFixed for values under 1e21.

Effects: fail.

`i64.toFixed`

gizlang

function i64.toFixed(self: i64, digits: i64): string

The integer with DIGITS zeros after the point.

Effects: fail.

`bytes.readU32LE`

gizlang

function bytes.readU32LE(self: bytes, offset: i64): u64 requires offset> = 0 requires offset<= len(self)-4

The unsigned 32-bit little-endian integer at OFFSET.

Effects: fail.

`bytes.readU64LE`

gizlang

function bytes.readU64LE(self: bytes, offset: i64): u64

One bounds check and one load; out of range faults as indexing does.

Effects: fail.

`bytes.copyFrom`

gizlang

function bytes.copyFrom(self: mut bytes, at: i64, source: bytes, from: i64, size: i64): void

Writes SIZE bytes of SOURCE from FROM over this buffer from AT, in place, as one copy: a loop of self[i] = x through a field path made the buffer its own again at each byte. A range outside its buffer faults as indexing does.

Effects: fail.

`bytes.copyWithin`

gizlang

function bytes.copyWithin(self: mut bytes, at: i64, from: i64, size: i64): void

Copies SIZE bytes of this buffer from FROM to AT, in place; the ranges may overlap (bytes.copyFrom).

Effects: fail.

`bytes.readU32BE`

gizlang

function bytes.readU32BE(self: bytes, offset: i64): u64 requires offset> = 0 requires offset<= len(self)-4

The unsigned 32-bit big-endian integer at OFFSET.

Effects: fail.

`bytes.readU16LE`

gizlang

function bytes.readU16LE(self: bytes, offset: i64): u64

The unsigned 16-bit little-endian integer at OFFSET.

Effects: fail.

`bytes.readU16BE`

gizlang

function bytes.readU16BE(self: bytes, offset: i64): u64

The unsigned 16-bit big-endian integer at OFFSET.

Effects: fail.

`bytes.readU64BE`

gizlang

function bytes.readU64BE(self: bytes, offset: i64): u64

The unsigned 64-bit big-endian integer at OFFSET.

Effects: fail.

`bytes.readF32LE`

gizlang

function bytes.readF32LE(self: bytes, offset: i64): f64

The 32-bit little-endian float at OFFSET.

Effects: fail.

`bytes.readF32BE`

gizlang

function bytes.readF32BE(self: bytes, offset: i64): f64

The 32-bit big-endian float at OFFSET.

Effects: fail.

`bytes.readF64LE`

gizlang

function bytes.readF64LE(self: bytes, offset: i64): f64

The 64-bit little-endian float at OFFSET.

Effects: fail.

`bytes.readF64BE`

gizlang

function bytes.readF64BE(self: bytes, offset: i64): f64

The 64-bit big-endian float at OFFSET.

Effects: fail.

`bytes.setU16LE`

gizlang

function bytes.setU16LE(self: mut bytes, offset: i64, value: u64): void

Writes the low 2, 4 or 8 bytes of value at offset, little-endian, over the bytes there.

Effects: fail.

`bytes.setU32LE`

gizlang

function bytes.setU32LE(self: mut bytes, offset: i64, value: u64): void

Writes the low 4 bytes of VALUE at OFFSET, little-endian.

Effects: fail.

`bytes.setU64LE`

gizlang

function bytes.setU64LE(self: mut bytes, offset: i64, value: u64): void

Writes VALUE at OFFSET as 8 bytes, little-endian.

Effects: fail.

`Splitter.split`

gizlang

function Splitter.split(self: Splitter, text: string): Vec<string>

The chunks of text: split_text of the splitter.

Effects: fail.

Constants

`SCALAR_LIMIT`

gizlang

const SCALAR_LIMIT = 1114112

The number of Unicode code points: scalar values are below it.

`SURROGATE_FIRST`

gizlang

const SURROGATE_FIRST = 55296

The first surrogate code point, which is not a scalar value.

`SURROGATE_LAST`

gizlang

const SURROGATE_LAST = 57343

The last surrogate code point.

`MAX_VARINT_BYTES`

gizlang

const MAX_VARINT_BYTES = 10

The most bytes of one encoded varint.