go-mxl-pattern-generator/internal/generator/text.go

// Static text overlay for v210 frames: text is rasterized with real font
// anti-aliasing onto a black backing box, pre-packed into whole v210 blocks
// once, and stamped per frame with a block-aligned copy (microseconds).
package generator

import (
	"encoding/binary"
	"fmt"
	"image"
	"os"
	"slices"

	"golang.org/x/image/font"
	"golang.org/x/image/font/opentype"
	"golang.org/x/image/math/fixed"
)

type TextOverlay struct {
	frameW int
	x, y   int    // box top-left in frame pixels
	w, h   int    // box size in pixels (w is a multiple of 6)
	cov    []byte // 8-bit anti-alias coverage, row-major w*h
	blocks []byte // pre-packed v210 tile: h rows x (w/6) blocks x 16 bytes
}

var TextPosAvailable []string = []string{"tl", "tc", "tr", "cc", "bl", "bc", "br"}

// LoadFace parses a TTF/OTF file and builds a render-ready face at the given
// pixel size (DPI 72, full hinting for crisp video text). Call once and
// reuse for every overlay.
func LoadFace(path string, size float64) (font.Face, error) {
	data, err := os.ReadFile(path)
	if err != nil {
		return nil, fmt.Errorf("text: %w", err)
	}
	f, err := opentype.Parse(data)
	if err != nil {
		return nil, fmt.Errorf("text: parse %s: %w", path, err)
	}
	face, err := opentype.NewFace(f, &opentype.FaceOptions{
		Size:    size,
		DPI:     72,
		Hinting: font.HintingFull,
	})
	if err != nil {
		return nil, fmt.Errorf("text: face %s: %w", path, err)
	}
	return face, nil
}

// NewTextOverlay rasterizes text (white, anti-aliased) onto a black backing box.
// Box width and x are multiples of 6 so the box
// covers whole v210 blocks and the per-frame copy is word-aligned.
func NewTextOverlay(text string, frameW, frameH, posX, posY int, textPos string, face font.Face) (*TextOverlay, error) {
	if text == "" {
		return nil, fmt.Errorf("text: empty string")
	}
	if frameW <= 0 || frameW%6 != 0 {
		return nil, fmt.Errorf("text: frame width %d not a multiple of 6", frameW)
	}
	pix := func(v fixed.Int26_6) int { return int(v+63) / 64 } // ceil to px
	m := face.Metrics()
	textH := pix(m.Ascent+m.Descent) + 2 // slack for AA/hinting overshoot

	// Box width: sum of per-rune advances (uniform for monospace fonts).
	spaceAdv, ok := face.GlyphAdvance(' ')
	if !ok {
		spaceAdv = m.Height
	}
	var total fixed.Int26_6
	for _, r := range text {
		a, ok := face.GlyphAdvance(r)
		if !ok {
			a = spaceAdv
		}
		total += a
	}
	textW := pix(total)

	padX := pix(m.Ascent) / 3
	padY := padX / 2
	w := 2*padX + textW
	if r := w % 6; r != 0 {
		w += 6 - r
	}
	if w > frameW {
		return nil, fmt.Errorf("text: box %dpx wider than frame %dpx", w, frameW)
	}
	h := 2*padY + textH
	if h > frameH {
		return nil, fmt.Errorf("text: box %dpx higher than frame %dpx", h, frameH)
	}

	if textPos != "" {
		if !slices.Contains(TextPosAvailable, textPos) {
			return nil, fmt.Errorf("text: unknown text position preset '%s'", textPos)
		}
		switch textPos {
		case "tl":
			posX = 0
			posY = 0
		case "tc":
			posX = ((frameW - w) / 2 / 6) * 6
			posY = 0
		case "tr":
			posX = frameW - w
			posY = 0
		case "cc":
			posX = ((frameW - w) / 2 / 6) * 6
			posY = (frameH - h) / 2
		case "bl":
			posX = 0
			posY = frameH - h
		case "bc":
			posX = ((frameW - w) / 2 / 6) * 6
			posY = frameH - h
		case "br":
			posX = frameW - w
			posY = frameH - h
		}
	}
	if posX < 0 || posY < 0 {
		return nil, fmt.Errorf("text: position (%d, %d) must not be negative", posX, posY)
	}
	if posX%6 != 0 {
		return nil, fmt.Errorf("text: x position %d must be divisible by 6 for v210", posX)
	}
	if posX+w > frameW || posY+h > frameH {
		return nil, fmt.Errorf(
			"text: box at (%d, %d), size %dx%d, does not fit frame %dx%d",
			posX, posY, w, h, frameW, frameH,
		)
	}
	o := &TextOverlay{
		frameW: frameW,
		x:      posX,
		y:      posY,
		w:      w,
		h:      h,
		cov:    make([]byte, w*h),
	}

	// Rasterize: drawing white onto *image.Alpha writes pure glyph coverage.
	dst := image.NewAlpha(image.Rect(0, 0, textW, textH))
	(&font.Drawer{
		Dst: dst, Src: image.White, Face: face,
		Dot: fixed.P(0, pix(m.Ascent)),
	}).DrawString(text)
	for row := 0; row < textH; row++ {
		copy(o.cov[(padY+row)*w+padX:], dst.Pix[row*dst.Stride:row*dst.Stride+textW])
	}

	o.pack()
	return o, nil
}

// pack pre-renders the box into v210 blocks: white text (Y=940) blended by
// coverage over a black box (Y=64), neutral chroma (512). Same word layout
// as the kernels, chroma co-sited with luma samples 0/2/4.
func (o *TextOverlay) pack() {
	const neutral = uint32(512)
	o.blocks = make([]byte, o.h*(o.w/6)*16)
	for row := 0; row < o.h; row++ {
		for blk := 0; blk < o.w/6; blk++ {
			var y [6]uint32
			for i := 0; i < 6; i++ {
				c := int(o.cov[row*o.w+blk*6+i])
				y[i] = 64 + uint32((c*876+127)/255)
			}
			off := (row*(o.w/6) + blk) * 16
			binary.LittleEndian.PutUint32(o.blocks[off+0:], neutral|(y[0]<<10)|(neutral<<20))
			binary.LittleEndian.PutUint32(o.blocks[off+4:], y[1]|(neutral<<10)|(y[2]<<20))
			binary.LittleEndian.PutUint32(o.blocks[off+8:], neutral|(y[3]<<10)|(neutral<<20))
			binary.LittleEndian.PutUint32(o.blocks[off+12:], y[4]|(neutral<<10)|(y[5]<<20))
		}
	}
}

// ApplyV210 stamps the pre-packed text tile over a packed v210 frame.
// The tile occupies whole 16-byte blocks, so this is a row-wise copy.
func (o *TextOverlay) ApplyV210(dest []byte) error {
	lastPixel := (o.y+o.h-1)*o.frameW + o.x + o.w - 1
	if need := (lastPixel/6 + 1) * 16; len(dest) < need {
		return fmt.Errorf("text: dest %d bytes too small, need %d", len(dest), need)
	}
	tileStride := (o.w / 6) * 16
	for row := 0; row < o.h; row++ {
		frameBlock := ((o.y+row)*o.frameW + o.x) / 6
		src := o.blocks[row*tileStride : (row+1)*tileStride]
		copy(dest[frameBlock*16:frameBlock*16+tileStride], src)
	}
	return nil
}