awaforge

pdfCcittFaxDecoder

AGPL-3.0-onlyRelease 2L4

CCITTFaxDecode codec (ITU-T T.4 / T.6), full Group 3/Group 4 read + write.

Module pdfCcittFaxDecoder | Source packages/front/office/pdf/src/extra/ccitt-fax-decoder.js | Deps pdfErrors | Worker-safe yes

Implements the CCITT Group 3/Group 4 fax codec referenced by /Filter /CCITTFaxDecode (ISO 32000-2 §7.4.6): K < 0 selects Group 4 (T.6, pure 2D coding), K == 0 selects Group 3 one-dimensional (T.4 baseline, white/black run Huffman), K > 0 selects Group 3 mixed 1D/2D (T.4 with a K-row block and a tag bit). Output is a packed bitstream of Columns bits per row, MSB-first, with polarity following /BlackIs1 (default: white = 1). This is the real decoder that pdfLegacyDeprecatedFilters's ccittFaxDecode delegates to after validating DecodeParms; it is registered directly in src/main.js#extras and can also be resolved and used standalone.

Resolve

const ext = runtime.resolve('pdfCcittFaxDecoder');
// Returns: { decode, encode, _internals }

API

Member Signature Returns
decode (bytes: Uint8Array, parms: { Columns, Rows?, K?, EndOfLine?, EncodedByteAlign?, BlackIs1?, DamagedRowsBeforeError?, EndOfBlock? }) => Uint8Array Packed bitstream, Columns-bits-per-row, MSB-first.
encode (lines: Uint8Array[], parms?: { Columns?, K?, EndOfLine?, EncodedByteAlign?, EndOfBlock? }) => Uint8Array Encodes an array of one-byte-per-pixel rows (0=white, 1=black) into a CCITT bitstream, including RTC (G3) / EOFB (G4) terminators.
_internals object Test/introspection surface: { WHITE_MAP, BLACK_MAP, MODE_CODES, EOL_BITS, codeForRun, encode1DLine, encode2DLine, decode1DLine, decode2DLine, makeBitReader, makeBitWriter, findB1, findNextChange, findChangeFrom }. Not part of the stable public contract.

Examples

Decode a Group 4 stream

const ext = runtime.resolve('pdfCcittFaxDecoder');
const bits = ext.decode(streamBytes, { K: -1, Columns: 1728, Rows: 100 });
// bits.length === Math.ceil(1728 / 8) * 100

Round-trip encode then decode

const rows = [ new Uint8Array(1728), new Uint8Array(1728) ]; // all-white
const encoded = ext.encode(rows, { Columns: 1728, K: -1 });
const decoded = ext.decode(encoded, { Columns: 1728, Rows: rows.length, K: -1 });

Errors

Code Class When
pdf/ccitt/bad-input ParseError bytes is not a Uint8Array.
pdf/ccitt/bad-columns ParseError Columns is <= 0 or > 65535.
pdf/ccitt/bad-runcode ParseError An unrecognized white/black run-length code is read.
pdf/ccitt/bad-2d-mode ParseError An unrecognized 2D mode code is read.
pdf/ccitt/unsupported-2d-mode ParseError A resolved 2D mode has no vertical-offset handling (defensive).
pdf/ccitt/truncated ParseError Stream ends mid-line and DamagedRowsBeforeError does not tolerate it.
pdf/ccitt/decode-failed ParseError Any other decode-time error, wrapped with cause.

encode does not throw a validated-input error of its own; malformed lines/parms produce best-effort output.

Notes

  • The encoder side is production-grade (greedy make-up code choice, EncodedByteAlign padding, RTC/EOFB terminators, full 2D state machine) even though the module is filed as a "legacy read" extra alongside pdfLegacyDeprecatedFilters — it is exercised by that module's ccittFaxDecode/validateCcittParms pair but is independently usable.
  • Rows tolerant of damage (DamagedRowsBeforeError > 0) are filled with a blank (all-white) row instead of throwing.

See also