[font] resolve the initial viewport of SVG glyphs

font/export_test.go

@@ -0,0 +1,6 @@
+// SPDX-License-Identifier: Unlicense OR BSD-3-Clause
+
+package font
+
+// SVGDocViewBox exposes [svgViewBox] for tests.
+func SVGDocViewBox(doc []byte, upem uint16) SVGViewBox { return svgViewBox(doc, upem) }

font/renderer.go

@@ -65,6 +65,13 @@ type GlyphSVG struct {
 	// and several glyphs may share the same Source
 	Source []byte
 
+	// ViewBox is the initial viewport of the SVG document :
+	// the rectangle in SVG user space mapped to the em square
+	// when rendering.
+	// It is resolved from the root <svg> element attributes,
+	// defaulting to (0, 0, upem, upem) as required by the specification.
+	ViewBox SVGViewBox
+
 	// According to the specification, a fallback outline
 	// should be specified for each SVG glyphs
 	Outline GlyphOutline
@@ -179,7 +186,7 @@ func (bt bitmap) glyphData(gid gID, xPpem, yPpem uint16) (GlyphBitmap, error) {
 	return out, nil
 }
 
-func (s svg) glyphData(gid gID) (GlyphSVG, bool) {
+func (s svg) glyphData(gid gID, upem uint16) (GlyphSVG, bool) {
 	data, ok := s.rawGlyphData(gid)
 	if !ok {
 		return GlyphSVG{}, false
@@ -193,7 +200,7 @@ func (s svg) glyphData(gid gID) (GlyphSVG, bool) {
 		}
 	}
 
-	return GlyphSVG{Source: data}, true
+	return GlyphSVG{Source: data, ViewBox: svgViewBox(data, upem)}, true
 }
 
 // this file converts from font format for glyph outlines to
@@ -447,7 +454,7 @@ func (f *Face) GlyphDataOutline(gid GID) (GlyphOutline, bool) {
 
 // GlyphDataSVG looks for glyph data in the 'SVG ' table.
 func (f *Face) GlyphDataSVG(gid GID) (GlyphSVG, bool) {
-	outS, ok := f.svg.glyphData(gID(gid))
+	outS, ok := f.svg.glyphData(gID(gid), f.upem)
 	if !ok {
 		return GlyphSVG{}, false
 	}

font/svg.go

@@ -3,7 +3,10 @@
 package font
 
 import (
+	"bytes"
 	"fmt"
+	"strconv"
+	"strings"
 
 	"github.com/go-text/typesetting/font/opentype/tables"
 )
@@ -52,3 +55,196 @@ func (s svg) rawGlyphData(gid gID) ([]byte, bool) {
 	}
 	return nil, false
 }
+
+// SVGViewBox is a rectangle in the "user" coordinate space
+// of an SVG document.
+type SVGViewBox struct {
+	MinX, MinY, Width, Height float32
+}
+
+// svgViewBox returns the initial viewport of the SVG document [doc],
+// resolved from the root <svg> element attributes, defaulting to the
+// em square given by [upem], as required by the specification.
+func svgViewBox(doc []byte, upem uint16) SVGViewBox {
+	viewBox, width, height, ok := svgRootAttributes(doc)
+	if ok {
+		if vb, ok := parseSVGViewBox(viewBox); ok {
+			return vb
+		}
+		if w, okW := parseSVGLength(width); okW {
+			if h, okH := parseSVGLength(height); okH {
+				return SVGViewBox{0, 0, w, h}
+			}
+		}
+	}
+	em := float32(upem)
+	return SVGViewBox{0, 0, em, em}
+}
+
+// parseSVGViewBox parses the value of a viewBox attribute:
+// four numbers separated by whitespace and/or commas.
+func parseSVGViewBox(s string) (SVGViewBox, bool) {
+	fields := strings.FieldsFunc(s, func(r rune) bool {
+		return r == ',' || r == ' ' || r == '\t' || r == '\r' || r == '\n'
+	})
+	if len(fields) != 4 {
+		return SVGViewBox{}, false
+	}
+	var nums [4]float32
+	for i, field := range fields {
+		v, err := strconv.ParseFloat(field, 32)
+		if err != nil {
+			return SVGViewBox{}, false
+		}
+		nums[i] = float32(v)
+	}
+	if nums[2] <= 0 || nums[3] <= 0 {
+		return SVGViewBox{}, false
+	}
+	return SVGViewBox{nums[0], nums[1], nums[2], nums[3]}, true
+}
+
+// parseSVGLength parses a positive width or height attribute value
+// expressed in user units, such as "12" or "12px".
+func parseSVGLength(s string) (float32, bool) {
+	s = strings.TrimSpace(s)
+	s = strings.TrimSuffix(s, "px")
+	v, err := strconv.ParseFloat(s, 32)
+	if err != nil || v <= 0 {
+		return 0, false
+	}
+	return float32(v), true
+}
+
+func isXMLSpace(c byte) bool { return c == ' ' || c == '\t' || c == '\r' || c == '\n' }
+
+// svgRootAttributes scans the prolog of the XML document [doc] and returns
+// the raw viewBox, width and height attribute values of the root <svg> element.
+//
+// A dedicated scanner is used instead of [encoding/xml], which would add
+// a significant binary size overhead to this core package for the simple
+// task of reading the root element attributes.
+func svgRootAttributes(doc []byte) (viewBox, width, height string, ok bool) {
+	// skip an optional byte order mark
+	doc = bytes.TrimPrefix(doc, []byte{0xEF, 0xBB, 0xBF})
+	for len(doc) != 0 {
+		if isXMLSpace(doc[0]) {
+			doc = doc[1:]
+			continue
+		}
+		switch {
+		case bytes.HasPrefix(doc, []byte("<?")): // XML declaration or processing instruction
+			end := bytes.Index(doc, []byte("?>"))
+			if end == -1 {
+				return "", "", "", false
+			}
+			doc = doc[end+2:]
+		case bytes.HasPrefix(doc, []byte("<!--")): // comment
+			end := bytes.Index(doc, []byte("-->"))
+			if end == -1 {
+				return "", "", "", false
+			}
+			doc = doc[end+3:]
+		case bytes.HasPrefix(doc, []byte("<!")): // doctype declaration
+			end := svgDoctypeEnd(doc)
+			if end == -1 {
+				return "", "", "", false
+			}
+			doc = doc[end:]
+		case doc[0] == '<': // root start tag
+			return parseSVGStartTag(doc)
+		default:
+			return "", "", "", false
+		}
+	}
+	return "", "", "", false
+}
+
+// svgDoctypeEnd returns the index just after the closing '>' of the
+// doctype declaration starting [doc], or -1.
+func svgDoctypeEnd(doc []byte) int {
+	depth := 0 // nesting in the internal subset brackets
+	for i := 2; i < len(doc); i++ {
+		switch doc[i] {
+		case '[':
+			depth++
+		case ']':
+			depth--
+		case '>':
+			if depth <= 0 {
+				return i + 1
+			}
+		}
+	}
+	return -1
+}
+
+// parseSVGStartTag parses the start tag beginning [doc] and returns
+// the raw viewBox, width and height attribute values, or false if the
+// element is not an <svg> element or is malformed.
+func parseSVGStartTag(doc []byte) (viewBox, width, height string, ok bool) {
+	doc = doc[1:] // '<'
+	i := 0
+	for i < len(doc) && !isXMLSpace(doc[i]) && doc[i] != '>' && doc[i] != '/' {
+		i++
+	}
+	name := doc[:i]
+	if j := bytes.LastIndexByte(name, ':'); j != -1 { // namespace prefix
+		name = name[j+1:]
+	}
+	// The element is matched by its local name only: the namespace binding
+	// is not verified, since conforming fonts always use the SVG namespace,
+	// and an svg element bound to another namespace is too unlikely to be
+	// worth rejecting.
+	if string(name) != "svg" {
+		return "", "", "", false
+	}
+	doc = doc[i:]
+	for {
+		for len(doc) != 0 && isXMLSpace(doc[0]) {
+			doc = doc[1:]
+		}
+		if len(doc) == 0 {
+			return "", "", "", false
+		}
+		if doc[0] == '>' || doc[0] == '/' { // end of the start tag
+			return viewBox, width, height, true
+		}
+		// attribute name
+		i = 0
+		for i < len(doc) && doc[i] != '=' && !isXMLSpace(doc[i]) && doc[i] != '>' && doc[i] != '/' {
+			i++
+		}
+		attr := doc[:i]
+		doc = doc[i:]
+		for len(doc) != 0 && isXMLSpace(doc[0]) {
+			doc = doc[1:]
+		}
+		if len(doc) == 0 || doc[0] != '=' {
+			return "", "", "", false
+		}
+		doc = doc[1:]
+		for len(doc) != 0 && isXMLSpace(doc[0]) {
+			doc = doc[1:]
+		}
+		if len(doc) == 0 || (doc[0] != '"' && doc[0] != '\'') {
+			return "", "", "", false
+		}
+		quote := doc[0]
+		doc = doc[1:]
+		end := bytes.IndexByte(doc, quote)
+		if end == -1 {
+			return "", "", "", false
+		}
+		value := string(doc[:end])
+		doc = doc[end+1:]
+		switch string(attr) {
+		case "viewBox":
+			viewBox = value
+		case "width":
+			width = value
+		case "height":
+			height = value
+		}
+	}
+}

font/svg_test.go

@@ -0,0 +1,78 @@
+// SPDX-License-Identifier: Unlicense OR BSD-3-Clause
+
+package font_test
+
+import (
+	"bytes"
+	"testing"
+
+	td "github.com/go-text/typesetting-utils/opentype"
+	"github.com/go-text/typesetting/font"
+	tu "github.com/go-text/typesetting/testutils"
+)
+
+func TestSVGViewBox(t *testing.T) {
+	const upem = 1000
+	for _, test := range []struct {
+		doc      string
+		expected font.SVGViewBox
+	}{
+		// no viewport information: default to the em square
+		{`<svg xmlns="http://www.w3.org/2000/svg" id="glyph1"></svg>`, font.SVGViewBox{0, 0, upem, upem}},
+		{``, font.SVGViewBox{0, 0, upem, upem}},
+		{`not xml`, font.SVGViewBox{0, 0, upem, upem}},
+		{`<html></html>`, font.SVGViewBox{0, 0, upem, upem}},
+		{`<svg`, font.SVGViewBox{0, 0, upem, upem}},
+		// viewBox attribute
+		{`<svg viewBox="0 0 128 128"></svg>`, font.SVGViewBox{0, 0, 128, 128}},
+		{`<svg viewBox="-10,-20 30,40"></svg>`, font.SVGViewBox{-10, -20, 30, 40}},
+		{`<svg viewBox=" 0 , 0 , 1e2 , 50.5 "></svg>`, font.SVGViewBox{0, 0, 100, 50.5}},
+		{`<svg viewBox='0 0 128 64'/>`, font.SVGViewBox{0, 0, 128, 64}},
+		{"<svg\tviewBox\t=\r\n\"0 0 128 128\"></svg>", font.SVGViewBox{0, 0, 128, 128}},
+		// invalid viewBox attributes
+		{`<svg viewBox="0 0 128"></svg>`, font.SVGViewBox{0, 0, upem, upem}},
+		{`<svg viewBox="0 0 0 128"></svg>`, font.SVGViewBox{0, 0, upem, upem}},
+		{`<svg viewBox="0 0 128 -1"></svg>`, font.SVGViewBox{0, 0, upem, upem}},
+		{`<svg viewBox="a b c d"></svg>`, font.SVGViewBox{0, 0, upem, upem}},
+		// width and height attributes
+		{`<svg width="100" height="200"></svg>`, font.SVGViewBox{0, 0, 100, 200}},
+		{`<svg width="100px" height="50px"></svg>`, font.SVGViewBox{0, 0, 100, 50}},
+		{`<svg width="100"></svg>`, font.SVGViewBox{0, 0, upem, upem}},
+		{`<svg width="100%" height="100%"></svg>`, font.SVGViewBox{0, 0, upem, upem}},
+		{`<svg width="12pt" height="12pt"></svg>`, font.SVGViewBox{0, 0, upem, upem}},
+		// viewBox has precedence over width and height
+		{`<svg width="100" height="200" viewBox="0 0 128 128"></svg>`, font.SVGViewBox{0, 0, 128, 128}},
+		{`<svg viewBox="invalid" width="100" height="200"></svg>`, font.SVGViewBox{0, 0, 100, 200}},
+		// prolog, comments and doctype before the root element
+		{"\uFEFF" + `<svg viewBox="0 0 128 128"></svg>`, font.SVGViewBox{0, 0, 128, 128}},
+		{`<?xml version="1.0" encoding="UTF-8"?><svg viewBox="0 0 128 128"></svg>`, font.SVGViewBox{0, 0, 128, 128}},
+		{`<?xml version="1.0"?>
+			<!-- comment with <svg viewBox="0 0 1 1"> inside -->
+			<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+			<svg viewBox="0 0 128 128"></svg>`, font.SVGViewBox{0, 0, 128, 128}},
+		{`<!DOCTYPE svg [ <!ENTITY foo "bar"> ]><svg viewBox="0 0 128 128"></svg>`, font.SVGViewBox{0, 0, 128, 128}},
+		// namespace prefix on the root element
+		{`<svg:svg xmlns:svg="http://www.w3.org/2000/svg" viewBox="0 0 128 128"></svg:svg>`, font.SVGViewBox{0, 0, 128, 128}},
+		// other attributes, possibly with tricky values
+		{`<svg xmlns="http://www.w3.org/2000/svg" title="a>b" viewBox="0 0 128 128" enable-background="new"></svg>`, font.SVGViewBox{0, 0, 128, 128}},
+	} {
+		got := font.SVGDocViewBox([]byte(test.doc), upem)
+		if got != test.expected {
+			t.Fatalf("document %q: expected %v, got %v", test.doc, test.expected, got)
+		}
+	}
+}
+
+func TestGlyphDataSVGViewBox(t *testing.T) {
+	// this font has no viewBox attribute in its SVG document:
+	// the viewport defaults to the em square
+	data, err := td.Files.ReadFile("toys/chromacheck-svg.ttf")
+	tu.AssertNoErr(t, err)
+
+	face, err := font.ParseTTF(bytes.NewReader(data))
+	tu.AssertNoErr(t, err)
+
+	glyph, ok := face.GlyphDataSVG(1)
+	tu.Assert(t, ok)
+	tu.Assert(t, glyph.ViewBox == font.SVGViewBox{0, 0, 1024, 1024})
+}