awaforge

pdfContentStream

AGPL-3.0-onlyRelease 2L4

Content-stream parser — decoded Uint8Array → ordered list of typed operations. ISO 32000-2 §7.8 + §8.2 + Table 60.

Module pdfContentStream | Source packages/front/office/pdf/src/content/stream.js | Deps pdfErrors, pdfParser, pdfContentOps | Worker-safe yes

Reads the bytes of an already decoded content stream (the /Filter chain must have been applied by pdfFilterDispatch upstream) and produces [{ op, args, data? }]. Operands stay typed ({ type, value, … }) — a Tj carries its typed string, ready for ToUnicode resolution; a gs carries its name, usable directly against /Resources /ExtGState. Inline images BI … ID … EI (§8.9.7) are captured as a single synthetic op with data: Uint8Array.

Resolve

const csMod = runtime.resolve('pdfContentStream');
// Returns: { parseContentStream }

API

Method Signature Returns
parseContentStream (bytes: Uint8Array) => Op[] Ordered list of operations.

Op shape

{
    op: string,         // mnemonic (e.g. 'Tj', 'cm', 'BI', '__trailing__')
    args: PdfObject[],  // typed operands, in stack order
    data?: Uint8Array   // present only for 'BI' (raw image data)
}

A synthetic { op: '__trailing__', args } is appended when the stream ends on unconsumed operands — this preserves rewrite fidelity.

Inline images

{
    op: 'BI',
    args: [{ type: 'dict', entries: { Width, Height, ColorSpace, Filter, … } }],
    data: Uint8Array  // raw bytes between `ID\n` and the space before `EI`
}

Examples

Parse a minimal stream

const cs = runtime.resolve('pdfContentStream');
const bytes = new TextEncoder().encode('BT /F1 12 Tf (Hello) Tj ET');
const ops = cs.parseContentStream(bytes);
// → [{op:'BT',args:[]}, {op:'Tf',args:[name,int]}, {op:'Tj',args:[string]}, {op:'ET',args:[]}]

Extract the font names in use

const fonts = new Set();
for (const op of ops) {
    if (op.op === 'Tf') fonts.add(op.args[0].value);  // args[0] = name
}

Inline image

const ops = cs.parseContentStream(streamBytes);
const img = ops.find(o => o.op === 'BI');
img.args[0].entries.W;   // typed int
img.data;                // Uint8Array — feed to pdfFilterDispatch when /F is set

Errors

Code Class When
pdf/content/bad-input ParseError Argument is not a Uint8Array.
pdf/content/unknown-op ParseError Unknown mnemonic (absent from pdfContentOps).
pdf/content/inline-image/no-ID ParseError BI without a matching ID.
pdf/content/inline-image/bad-key ParseError Non-name key in the inline dictionary.
pdf/content/inline-image/no-EI ParseError ID without a matching EI.

It also propagates the pdf/parser/* codes while parsing operands.

See also