Compose Resources

The optional io.github.conamobiledev:pdfkmp-compose-resources module bridges Compose Multiplatform's typed Res.drawable.* references straight into the PdfKmp DSL — no manual byte loading, no VectorImage.parse(...) boilerplate. The module is pure common, so it adds no platform code of its own.

Install it alongside the core dependency:

implementation("io.github.conamobiledev:pdfkmp:1.2.0")
implementation("io.github.conamobiledev:pdfkmp-compose-resources:1.2.0")

Build with pdfAsync { }

Every typed Res.drawable.* overload reads the resource bytes during the suspend preflight pass that pdfAsync { } runs before layout. The synchronous pdf { } entry point has no preflight pass and throws a clear error when it sees a deferred node, so build the document with pdfAsync whenever you use these.

Typed DSL overloads

drawable(...) auto-detects vector vs raster from the file's leading bytes, so you don't have to know whether the asset is <vector> / <svg> XML or a PNG / JPEG:

import com.conamobile.pdfkmp.pdfAsync
import com.conamobile.pdfkmp.composeresources.drawable

suspend fun buildReport(): PdfDocument = pdfAsync {
    page {
        drawable(Res.drawable.logo, width = 64.dp, tint = PdfColor.Blue)
        drawable(Res.drawable.cover_photo, width = 480.dp)
    }
}

tint / strokeMode take effect only when the resource turns out to be a vector; contentScale only when it turns out to be a raster. The full signature:

fun ContainerScope.drawable(
    resource: DrawableResource,
    width: Dp? = null,
    height: Dp? = null,
    tint: PdfColor? = null,                              // vector only
    contentScale: ContentScale = ContentScale.Fit,      // raster only
    strokeMode: VectorStrokeMode = VectorStrokeMode.Inherit, // vector only
    allowDownScale: Boolean = true,                     // raster only
)

When you already know the variant, reach for the type-specific overloads:

import com.conamobile.pdfkmp.composeresources.vector
import com.conamobile.pdfkmp.composeresources.image

vector(Res.drawable.star, width = 48.dp, tint = PdfColor.Red)
image(Res.drawable.banner, width = 480.dp, contentScale = ContentScale.Crop)

Loading bytes yourself

When you'd rather pull the raw bytes (or a parsed VectorImage) out and feed them to the core image(...) / vector(...) DSL — for example to parse a vector once and draw it many times — three suspend extensions on DrawableResource do that:

Extension	Returns	Use for
toBytes()	ByteArray	Raster bytes for image(bytes = …), or raw XML for toVectorImage.
toVectorImage()	VectorImage	Parse a <vector> / <svg> resource once, reuse across call sites.
toPdfDrawable()	PdfDrawable	Format-agnostic — sniffs the leading bytes and returns Vector or Raster.

import com.conamobile.pdfkmp.composeresources.toBytes
import com.conamobile.pdfkmp.composeresources.toVectorImage
import com.conamobile.pdfkmp.composeresources.toPdfDrawable

// Inside a coroutine / LaunchedEffect / Swift Task { } — these are suspend:
val logo = Res.drawable.logo.toVectorImage()   // parse once
val photoBytes = Res.drawable.photo.toBytes()
val either = Res.drawable.icon_or_photo.toPdfDrawable()

val doc = pdf {
    page {
        vector(image = logo, width = 64.dp)        // core DSL, synchronous
        image(bytes = photoBytes, width = 200.dp)
        drawable(either, width = 64.dp)            // eager PdfDrawable overload — no preflight needed
    }
}

Eager drawable(PdfDrawable) works inside pdf { }

Once you've resolved a resource into a PdfDrawable yourself (via toPdfDrawable() from a coroutine), the eager drawable(drawable, …) overload embeds it with no deferred node — so it works inside the synchronous pdf { }. Only the typed Res.drawable.* overloads require pdfAsync.

toVectorImage() throws VectorParseException when the resource exists but isn't a valid Android-Vector / SVG payload.

See also

• Images & vectors — the core image(...) / vector(...) DSL.
• Getting started — pdf { } vs pdfAsync { }.