awaforge

pdfShared

AGPL-3.0-onlyRelease 2L4

PDF byte-level constants and stateless helpers, shared across @awacloud/pdf factories.

Module pdfShared | Source packages/front/office/pdf/src/_shared/index.js | Deps (none) | Worker-safe yes

Magic bytes, ASCII byte values, the ISO 32000-2 §7.2 character classes, a hex lookup table, text codec instances and a few byte helpers. pdfTokenizer and the top-level pdf declare it as a dependency; other modules (the filters, pdfWriter, pdfSerializer) still carry their own inline copies of some of these helpers.

Resolve

const shared = runtime.resolve('pdfShared');
// → { HEADER_PREFIX, EOF_MARKER, BINARY_MARKER,
//     ASCII,
//     isWs, isEol, isDigit, isHex, isDelim, isRegular, hexNibble,
//     HEX_LO,
//     te, tdUtf8, tdUtf8Lenient, tdLatin1,
//     encodeAscii, decodeUtf8, decodeUtf8Lenient, decodeLatin1,
//     pad10, hexLit, bytesEqual, concatBytes }

pdfShared is exported by name from the package root (import { pdfShared } from '@awacloud/pdf'); it has no sub-path of its own.

API

Member Type Description
HEADER_PREFIX Uint8Array The 5 bytes %PDF-.
EOF_MARKER Uint8Array The 5 bytes %%EOF.
BINARY_MARKER Uint8Array The binary-marker comment line written after the header (§7.5.2): bytes 0x25 0xE2 0xE3 0xCF 0xD3 0x0A.
ASCII object<string, number> Frozen byte values: NUL, HT, LF, FF, CR, SP, HASH, PERCENT, LPAREN, RPAREN, PLUS, MINUS, DOT, SLASH, ZERO, NINE, LANGLE, RANGLE, A_UP, F_UP, Z_UP, LBRACK, BACKSLASH, RBRACK, A_LO, F_LO, Z_LO, LBRACE, RBRACE.
isWs(b) (number) => boolean PDF white-space byte (NUL, HT, LF, FF, CR, SP), §7.2.3.
isEol(b) (number) => boolean LF or CR.
isDigit(b) (number) => boolean 0–9.
isHex(b) (number) => boolean 0–9, A–F, a–f.
isDelim(b) (number) => boolean A delimiter: ( ) < > [ ] { } / %.
isRegular(b) (number) => boolean Neither white space nor a delimiter.
hexNibble(b) (number) => number The value of one hex digit byte, -1 when b is not one.
HEX_LO Int8Array (128 entries) Lookup table with the same result as hexNibble, indexed by ASCII byte.
te TextEncoder One instance per factory call.
tdUtf8 / tdUtf8Lenient TextDecoder UTF-8 decoders. Both are non-fatal (fatal: false, the TextDecoder default): a malformed sequence decodes to U+FFFD in either.
tdLatin1 TextDecoder latin1 decoder.
encodeAscii(s) (string) => Uint8Array te.encode(s) — UTF-8, so it equals ASCII only for an ASCII string.
decodeUtf8(bytes) / decodeUtf8Lenient(bytes) (Uint8Array) => string Wrappers over tdUtf8 / tdUtf8Lenient.
decodeLatin1(bytes) (Uint8Array) => string Wrapper over tdLatin1.
pad10(n) (number) => string n left-padded with zeros to 10 digits (a cross-reference entry offset).
hexLit(bytes) (Uint8Array) => string The PDF hex-string literal <…>, upper-case digits.
bytesEqual(a, b, n?) (Uint8Array, Uint8Array, number?) => boolean Equality of two byte arrays without an early exit. Without n, both lengths must match; with n, compares the first n bytes and requires both arrays to be at least that long. false when either array is missing.
concatBytes(arrays) (Uint8Array[]) => Uint8Array One new array holding every input in order.

Examples

const shared = runtime.resolve('pdfShared');

shared.decodeLatin1(shared.HEADER_PREFIX);              // '%PDF-'
shared.isWs(shared.ASCII.SP);                           // true
shared.hexNibble(0x61);                                 // 10 ('a')
shared.hexLit(Uint8Array.of(0xde, 0xad));               // '<DEAD>'
shared.pad10(42);                                       // '0000000042'
shared.bytesEqual(Uint8Array.of(1, 2, 3), Uint8Array.of(1, 2, 4), 2);   // true
shared.concatBytes([Uint8Array.of(1), Uint8Array.of(2, 3)]);            // Uint8Array [1, 2, 3]

Notes

  • Worker-safe: every member is a constant, a pure function, or a codec instance created inside the factory body, so factory.toString() is a self-contained closure.
  • No mutable module state: ASCII is frozen; each factory() call creates its own codec instances and tables.

See also