`image_to_pdf`

Convert one or more images into a single PDF file.

Parameters

sourcesarrayone of sources or source_key: Array of source images. Each item may be a path, artifact URI/descriptor, { artifact = ... }, or an explicit { base64 = "..." } object.
source_keystringone of sources or source_key: Context key containing an array of source images (same formats as sources).
output_pathstringrequired: Destination path for generated PDF. Supports ${ctx.key} interpolation.
output_keystringdefault pdf_path: Context key to store generated PDF path.

Paths inside image entries support ${ctx.key} interpolation.

Image entry formats

String

"images/logo.png"

Object

{ path = "images/logo.png" }

{ base64 = "iVBORw0KGgoAAAANSUhEUg..." }

{ artifact_uri = "artifact://sha256/0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef",
  sha256 = "0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef",
  size_bytes = 184322,
  mime_type = "image/png" }

Output wrappers from pdf_to_image are also accepted directly: set source_key = "images" and each { artifact = ... } entry in that context array is resolved without Base64. For one pdf_thumbnail result, place the thumbnail object in a context array before passing its key.

Artifact inputs are opened and SHA-256 verified inside the tracked blocking worker; encoding reads that same rewound handle rather than a resolved store pathname.

Supported source formats are BMP, Farbfeld, GIF, HDR, ICO, JPEG, PNG, PNM, QOI, TGA, TIFF, and WebP. AVIF and OpenEXR are not supported.

Context Output

<output_key> (default pdf_path) — output PDF path.
image_count — number of source images processed.
<output_key>_count — same as image_count.
<output_key>_success — boolean true.

Example

local flow = Flow.new("images_to_pdf")

flow:step("make_pdf", nodes.image_to_pdf({
    sources = {
        "data/images/front.png",
        "data/images/back.png",
    },
    output_path = "output/gallery.pdf",
    output_key = "pdf_file"
}))

flow:step("log", nodes.log({
    message = "Wrote ${ctx.pdf_file} with ${ctx.pdf_file_count} page(s)"
})):depends_on("make_pdf")

return flow

Resource contract

Source objects must select exactly one of path, artifact, base64, or data; ambiguous objects fail instead of silently selecting the first field. The node rejects more than IRONFLOW_MAX_IMAGE_TO_PDF_SOURCES entries (default 100) before parsing them. Each image is bounded by IRONFLOW_MAX_IMAGE_ENCODED_BYTES (50 MiB), IRONFLOW_MAX_IMAGE_PIXELS (25 million), and IRONFLOW_MAX_IMAGE_DECODE_ALLOCATION_BYTES (128 MiB). A whole call is additionally bounded by IRONFLOW_MAX_IMAGE_TO_PDF_ENCODED_BYTES (100 MiB) and IRONFLOW_MAX_IMAGE_TO_PDF_PIXELS (50 million).

Headers, dimensions, decoded byte estimates, and cumulative budgets are checked before pixel decoding or adding a PDF object. Base64 length is admitted against the per-source and cumulative decoded-byte ceilings before IronFlow clones a source string out of workflow context; malformed or oversized later entries do not cause an unbounded pre-worker copy. JPEG bytes can be embedded without decoding. Other formats use the configured image decoder limits and a bounded pixel conversion and compression-buffer preflight. The in-progress lopdf document still retains embedded streams until save, and accepted Base64 input temporarily retains both its bounded context string and decoded bytes; prefer paths or artifacts for large media. Decode, conversion, compression, and save run on a tracked blocking worker with cooperative checkpoints between opaque codec operations.