| 1 | // Package svg writes a paint list out as a picture. |
| 2 | // |
| 3 | // It is the second half of the one rule (spec §0.1, §4.2): internal/ |
| 4 | // slides/layout measures a slide into a flat list of rectangles, |
| 5 | // ellipses, lines, images and positioned runs of text, and this turns |
| 6 | // that list into SVG 1.1. Nothing here reads a deck — the paint list is |
| 7 | // the whole input — which is what keeps a thumbnail on the editor strip |
| 8 | // and a page in an exported PDF from being two opinions about one |
| 9 | // slide. |
| 10 | // |
| 11 | // Two consumers, one serialiser: |
| 12 | // |
| 13 | // - A thumbnail, served from every surface that can show a deck |
| 14 | // (internal/slides/thumbs.go). It goes into an `<img src>`, where a |
| 15 | // browser renders the SVG in a restricted mode that fetches |
| 16 | // NOTHING: no stylesheet, no font file, no external picture. So a |
| 17 | // thumbnail inlines the images it can (Options.InlineImages) and |
| 18 | // sets type in whatever the reader's machine has under the same |
| 19 | // stack slides.css names — which is the one place a thumbnail is |
| 20 | // knowingly not the slide, and why it is a thumbnail and not the |
| 21 | // export. |
| 22 | // - The PDF, where the same bytes go through krilla-svg. A WebP asset |
| 23 | // is decoded and re-encoded as a PNG on its way in (webpAsPNG), so a |
| 24 | // picture a person uploaded in the format their phone writes is in |
| 25 | // both the thumbnail and the PDF rather than in neither. |
| 26 | // |
| 27 | // # What comes out |
| 28 | // |
| 29 | // Well-formed XML, always. Every string that reaches the output goes |
| 30 | // through esc, which escapes the five predefined entities and DROPS the |
| 31 | // characters XML 1.0 cannot carry — a document nothing can parse is |
| 32 | // worse than a document missing a control character nobody can see. |
| 33 | // svg_test.go parses every output it makes with encoding/xml, and |
| 34 | // internal/slides/thumbs_test.go does the same for every slide of the |
| 35 | // sample deck. |
| 36 | // |
| 37 | // The output is deliberately plain: no transforms, no groups, no |
| 38 | // stylesheet. Every coordinate is already absolute in the 1920 by 1080 |
| 39 | // frame, so a reader of the file can put a number in it beside a number |
| 40 | // in a paint list, and usvg has nothing to resolve before it draws. |
| 41 | package svg |
| 42 | |
| 43 | import ( |
| 44 | "bytes" |
| 45 | "encoding/base64" |
| 46 | "encoding/xml" |
| 47 | "errors" |
| 48 | "fmt" |
| 49 | "image/png" |
| 50 | "io" |
| 51 | "math" |
| 52 | "net/url" |
| 53 | "slices" |
| 54 | "strconv" |
| 55 | "strings" |
| 56 | |
| 57 | "golang.org/x/image/webp" |
| 58 | |
| 59 | "amadan.net/oficina/slides/internal/slides/fonts" |
| 60 | "amadan.net/oficina/slides/internal/slides/layout" |
| 61 | ) |
| 62 | |
| 63 | // Options is what the surface writing a picture knows that the paint |
| 64 | // list does not: what to call the slide, and how to turn an asset id |
| 65 | // into bytes or into an address. |
| 66 | type Options struct { |
| 67 | // Title is the slide's label, written as the document's <title>. |
| 68 | // With role="img" on the root it is what a screen reader announces |
| 69 | // for the whole picture, and it is what an `<img alt>` around it |
| 70 | // should say too. |
| 71 | Title string |
| 72 | |
| 73 | // InlineImages resolves an asset id to its bytes and its type, for |
| 74 | // a surface that can read them. ok false means "not this one" — a |
| 75 | // missing asset, a store this deployment does not have, a type |
| 76 | // this cannot use — and the picture falls back to AssetURL. |
| 77 | // |
| 78 | // PNG, JPEG, GIF and a sanitised SVG go in as they are. A WebP is |
| 79 | // decoded here and re-encoded as a PNG (golang.org/x/image/webp). |
| 80 | // Anything else falls back to AssetURL. |
| 81 | InlineImages func(asset string) (data []byte, mime string, ok bool) |
| 82 | |
| 83 | // AssetURL is where an asset's bytes are served from on THIS |
| 84 | // surface — render.Options.AssetBase plus the ref. It is the |
| 85 | // fallback, and it only draws where the SVG is fetched as a |
| 86 | // document; inside an `<img>` a browser will not load it. |
| 87 | AssetURL func(asset string) string |
| 88 | } |
| 89 | |
| 90 | // Encode writes p as an SVG document. |
| 91 | // |
| 92 | // It never fails. Everything a paint list can hold serialises as |
| 93 | // something, and the two things that could not — a link to a scheme |
| 94 | // nobody should follow, an embedded picture that will not parse — are |
| 95 | // dropped, with the words they carried kept. |
| 96 | func Encode(p layout.Paint, o Options) []byte { |
| 97 | w := &writer{o: o} |
| 98 | w.body(p) |
| 99 | |
| 100 | var out bytes.Buffer |
| 101 | out.WriteString(xml.Header) |
| 102 | fmt.Fprintf(&out, `<svg xmlns="http://www.w3.org/2000/svg" version="1.1" `+ |
| 103 | `width="%s" height="%s" viewBox="0 0 %s %s" role="img">`+"\n", |
| 104 | num(p.W), num(p.H), num(p.W), num(p.H)) |
| 105 | // The title first, because an assistive technology reads the first |
| 106 | // <title> child of the root as the name of the whole picture. |
| 107 | out.WriteString("<title>" + esc(o.Title) + "</title>\n") |
| 108 | if w.defs.Len() > 0 { |
| 109 | out.WriteString("<defs>\n") |
| 110 | out.Write(w.defs.Bytes()) |
| 111 | out.WriteString("</defs>\n") |
| 112 | } |
| 113 | out.Write(w.out.Bytes()) |
| 114 | out.WriteString("</svg>\n") |
| 115 | return out.Bytes() |
| 116 | } |
| 117 | |
| 118 | // writer is one Encode: the picture so far, the definitions it needs |
| 119 | // at the top, and the ids it has handed out. |
| 120 | type writer struct { |
| 121 | o Options |
| 122 | out bytes.Buffer |
| 123 | defs bytes.Buffer |
| 124 | // arrows maps a stroke colour to the marker defined for it, so a |
| 125 | // slide of arrows in one colour defines one arrowhead. |
| 126 | arrows map[layout.Fill]string |
| 127 | n int |
| 128 | } |
| 129 | |
| 130 | // id mints a document-unique id under this package's own prefix, so |
| 131 | // nothing it defines can collide with an id inside an embedded |
| 132 | // picture. |
| 133 | func (w *writer) id(kind string) string { |
| 134 | w.n++ |
| 135 | return "slides-" + kind + "-" + strconv.Itoa(w.n) |
| 136 | } |
| 137 | |
| 138 | func (w *writer) body(p layout.Paint) { |
| 139 | // The background, and only when there is one: a slide with no |
| 140 | // background colour is transparent, and painting it black would be |
| 141 | // a decision this package has no business making. |
| 142 | if p.Background != "" { |
| 143 | fmt.Fprintf(&w.out, `<rect x="0" y="0" width="%s" height="%s" fill="%s"/>`+"\n", |
| 144 | num(p.W), num(p.H), esc(string(p.Background))) |
| 145 | } |
| 146 | for _, op := range p.Ops { |
| 147 | switch v := op.(type) { |
| 148 | case layout.Rect: |
| 149 | w.rect(v) |
| 150 | case layout.Ellipse: |
| 151 | w.ellipse(v) |
| 152 | case layout.Line: |
| 153 | w.line(v) |
| 154 | case layout.Image: |
| 155 | w.image(v) |
| 156 | case layout.Text: |
| 157 | w.text(v) |
| 158 | } |
| 159 | } |
| 160 | } |
| 161 | |
| 162 | func (w *writer) rect(r layout.Rect) { |
| 163 | attrs := []string{ |
| 164 | attr("x", num(r.X)), attr("y", num(r.Y)), |
| 165 | attr("width", num(r.W)), attr("height", num(r.H)), |
| 166 | } |
| 167 | if r.Radius > 0 { |
| 168 | attrs = append(attrs, attr("rx", num(r.Radius))) |
| 169 | } |
| 170 | attrs = append(attrs, w.paint(r.Fill, r.Stroke, r.StrokeW)...) |
| 171 | w.out.WriteString("<rect " + strings.Join(attrs, " ") + "/>\n") |
| 172 | } |
| 173 | |
| 174 | func (w *writer) ellipse(e layout.Ellipse) { |
| 175 | // The paint list carries a box; SVG wants a centre and two radii. |
| 176 | attrs := []string{ |
| 177 | attr("cx", num(e.X+e.W/2)), attr("cy", num(e.Y+e.H/2)), |
| 178 | attr("rx", num(e.W/2)), attr("ry", num(e.H/2)), |
| 179 | } |
| 180 | attrs = append(attrs, w.paint(e.Fill, e.Stroke, e.StrokeW)...) |
| 181 | w.out.WriteString("<ellipse " + strings.Join(attrs, " ") + "/>\n") |
| 182 | } |
| 183 | |
| 184 | // paint is the fill and stroke every shape carries. `fill="none"` is |
| 185 | // written out loud rather than left off, because SVG's initial fill is |
| 186 | // black and an outline would come out solid. |
| 187 | func (w *writer) paint(fill, stroke layout.Fill, strokeW float64) []string { |
| 188 | out := []string{attr("fill", fillOr(fill, "none"))} |
| 189 | if stroke != "" && strokeW > 0 { |
| 190 | out = append(out, attr("stroke", esc(string(stroke))), attr("stroke-width", num(strokeW))) |
| 191 | } |
| 192 | return out |
| 193 | } |
| 194 | |
| 195 | func (w *writer) line(l layout.Line) { |
| 196 | attrs := []string{ |
| 197 | attr("x1", num(l.X1)), attr("y1", num(l.Y1)), |
| 198 | attr("x2", num(l.X2)), attr("y2", num(l.Y2)), |
| 199 | attr("stroke", fillOr(l.Stroke, "none")), |
| 200 | attr("stroke-width", num(l.W)), |
| 201 | attr("stroke-linecap", "round"), |
| 202 | } |
| 203 | if l.Arrow && l.Stroke != "" { |
| 204 | attrs = append(attrs, attr("marker-end", "url(#"+w.arrowhead(l.Stroke)+")")) |
| 205 | } |
| 206 | w.out.WriteString("<line " + strings.Join(attrs, " ") + "/>\n") |
| 207 | } |
| 208 | |
| 209 | // arrowhead defines a marker for one stroke colour, once, and returns |
| 210 | // its id. A marker rather than a computed triangle because the marker |
| 211 | // turns with the line and a triangle would have to be rotated here — |
| 212 | // the same picture, with the trigonometry in the file instead of in the |
| 213 | // renderer. |
| 214 | // |
| 215 | // markerUnits is strokeWidth, so a thick line gets a proportionate |
| 216 | // head; orient auto turns it along the line. refX sits the point at the |
| 217 | // line's end rather than a head-length past it. |
| 218 | func (w *writer) arrowhead(stroke layout.Fill) string { |
| 219 | if id, ok := w.arrows[stroke]; ok { |
| 220 | return id |
| 221 | } |
| 222 | if w.arrows == nil { |
| 223 | w.arrows = map[layout.Fill]string{} |
| 224 | } |
| 225 | id := w.id("arrow") |
| 226 | w.arrows[stroke] = id |
| 227 | fmt.Fprintf(&w.defs, `<marker id="%s" viewBox="0 0 10 10" refX="9" refY="5" `+ |
| 228 | `markerWidth="5" markerHeight="5" markerUnits="strokeWidth" orient="auto">`+ |
| 229 | `<path d="M 0 0 L 10 5 L 0 10 z" fill="%s"/></marker>`+"\n", |
| 230 | id, esc(string(stroke))) |
| 231 | return id |
| 232 | } |
| 233 | |
| 234 | // ── Text ───────────────────────────────────────────────────────────── |
| 235 | |
| 236 | // text writes one line: a <text> at the baseline, and a <tspan> per |
| 237 | // run. |
| 238 | // |
| 239 | // Every run carries its own x, measured with fonts.Advance — the same |
| 240 | // function the layout engine measured with. Letting the runs flow would |
| 241 | // put each one wherever the renderer's own advances landed, and a |
| 242 | // renderer that kerns (every browser) would drift from the paint list |
| 243 | // within a sentence. Fixing each run's origin means the drift can only |
| 244 | // ever be inside one run. |
| 245 | // |
| 246 | // xml:space="preserve" because a run's trailing space is layout: "the " |
| 247 | // followed by "spec" is a word gap, and an XML processor is otherwise |
| 248 | // entitled to collapse it. |
| 249 | func (w *writer) text(t layout.Text) { |
| 250 | runs := make([]layout.Run, 0, len(t.Runs)) |
| 251 | for _, r := range t.Runs { |
| 252 | if esc(r.S) != "" { |
| 253 | runs = append(runs, r) |
| 254 | } |
| 255 | } |
| 256 | if len(runs) == 0 { |
| 257 | return |
| 258 | } |
| 259 | fmt.Fprintf(&w.out, `<text x="%s" y="%s" xml:space="preserve">`, num(t.X), num(t.Y)) |
| 260 | x := t.X |
| 261 | for _, r := range runs { |
| 262 | attrs := []string{ |
| 263 | attr("x", num(x)), |
| 264 | attr("font-family", esc(r.Face.Family.Stack())), |
| 265 | attr("font-size", num(r.Size)), |
| 266 | } |
| 267 | if r.Face.Bold { |
| 268 | attrs = append(attrs, attr("font-weight", "700")) |
| 269 | } |
| 270 | if r.Face.Italic { |
| 271 | attrs = append(attrs, attr("font-style", "italic")) |
| 272 | } |
| 273 | if r.Fill != "" { |
| 274 | attrs = append(attrs, attr("fill", esc(string(r.Fill)))) |
| 275 | } |
| 276 | span := "<tspan " + strings.Join(attrs, " ") + ">" + esc(r.S) + "</tspan>" |
| 277 | if href := safeHref(r.Href); href != "" { |
| 278 | span = `<a href="` + href + `">` + span + `</a>` |
| 279 | } |
| 280 | w.out.WriteString(span) |
| 281 | adv, _ := fonts.Advance(r.Face, r.Size, r.S) |
| 282 | x += adv |
| 283 | } |
| 284 | w.out.WriteString("</text>\n") |
| 285 | } |
| 286 | |
| 287 | // linkSchemes are the schemes a run's href may carry into a picture. |
| 288 | // The same three render.go allows, for the same reason: an SVG served |
| 289 | // from this app's origin is a document a reader can click through, and |
| 290 | // javascript: and data: are how that becomes a way to run code. |
| 291 | var linkSchemes = []string{"http", "https", "mailto"} |
| 292 | |
| 293 | // safeHref is an escaped href, or "" for a link that is not followed. |
| 294 | // The run's words are drawn either way — losing the sentence because |
| 295 | // somebody typed a scheme nobody should follow would be a strange |
| 296 | // trade. |
| 297 | func safeHref(raw string) string { |
| 298 | if raw == "" { |
| 299 | return "" |
| 300 | } |
| 301 | u, err := url.Parse(strings.TrimSpace(raw)) |
| 302 | if err != nil || !slices.Contains(linkSchemes, strings.ToLower(u.Scheme)) { |
| 303 | return "" |
| 304 | } |
| 305 | return esc(raw) |
| 306 | } |
| 307 | |
| 308 | // ── Images ─────────────────────────────────────────────────────────── |
| 309 | |
| 310 | // inlineTypes are the raster types that go in as bytes: the three every |
| 311 | // renderer in this pipeline reads. A WebP is not one of them and is |
| 312 | // converted to the first of them instead (webpAsPNG). |
| 313 | var inlineTypes = []string{"image/png", "image/jpeg", "image/gif"} |
| 314 | |
| 315 | // webpAsPNG decodes a WebP and writes it back out as a PNG, which is |
| 316 | // what both consumers of this package can actually draw. |
| 317 | // |
| 318 | // Re-encoding rather than refusing, because the alternative is a |
| 319 | // picture that is simply missing: WebP is one of the four types the |
| 320 | // upload path admits (assets.go's assetTypes), and it is what a phone |
| 321 | // and a modern image editor write by default. The cost is a bigger data |
| 322 | // URI — PNG is lossless and WebP usually is not — which is paid once per |
| 323 | // export, against a picture that would otherwise not be on the slide at |
| 324 | // all. |
| 325 | // |
| 326 | // The PDF converter has read WebP since svg2pdf was swapped for |
| 327 | // krilla-svg (2026-09-09, spec §1 ruling 13), so on that side this is |
| 328 | // now belt and braces. It is left in place because removing it is a |
| 329 | // change to what the encoder emits, with its own tests, rather than a |
| 330 | // consequence of changing the converter under it. |
| 331 | // |
| 332 | // A file that does not decode answers ok=false and the caller falls |
| 333 | // back to the URL, exactly as an unknown type does. |
| 334 | func webpAsPNG(data []byte) ([]byte, bool) { |
| 335 | img, err := webp.Decode(bytes.NewReader(data)) |
| 336 | if err != nil { |
| 337 | return nil, false |
| 338 | } |
| 339 | var out bytes.Buffer |
| 340 | if err := png.Encode(&out, img); err != nil { |
| 341 | return nil, false |
| 342 | } |
| 343 | return out.Bytes(), true |
| 344 | } |
| 345 | |
| 346 | func (w *writer) image(im layout.Image) { |
| 347 | // preserveAspectRatio is `object-fit` said in SVG: slice fills the |
| 348 | // box and overflows it, meet fits inside it. |
| 349 | fit := "xMidYMid meet" |
| 350 | if im.Fit == "cover" { |
| 351 | fit = "xMidYMid slice" |
| 352 | } |
| 353 | |
| 354 | var href string |
| 355 | var nested []byte |
| 356 | if w.o.InlineImages != nil { |
| 357 | if data, mime, ok := w.o.InlineImages(im.Asset); ok && len(data) > 0 { |
| 358 | switch { |
| 359 | case slices.Contains(inlineTypes, mime): |
| 360 | href = "data:" + mime + ";base64," + base64.StdEncoding.EncodeToString(data) |
| 361 | case mime == "image/webp": |
| 362 | if pngBytes, converted := webpAsPNG(data); converted { |
| 363 | href = "data:image/png;base64," + base64.StdEncoding.EncodeToString(pngBytes) |
| 364 | } |
| 365 | case mime == "image/svg+xml": |
| 366 | nested = nestableSVG(data) |
| 367 | } |
| 368 | } |
| 369 | } |
| 370 | if href == "" && nested == nil && w.o.AssetURL != nil { |
| 371 | href = esc(w.o.AssetURL(im.Asset)) |
| 372 | } |
| 373 | if href == "" && nested == nil { |
| 374 | // Nothing resolved it. layout already recorded the warning |
| 375 | // (WarnAsset); drawing a grey box here would put a thing on |
| 376 | // the slide that is not in the deck. |
| 377 | return |
| 378 | } |
| 379 | |
| 380 | box := []string{ |
| 381 | attr("x", num(im.X)), attr("y", num(im.Y)), |
| 382 | attr("width", num(im.W)), attr("height", num(im.H)), |
| 383 | } |
| 384 | // A cover picture is bigger than its box by definition, so it is |
| 385 | // clipped to it. An SVG viewport clips on its own, but usvg is |
| 386 | // happier told, and it costs four numbers. |
| 387 | if im.Fit == "cover" { |
| 388 | id := w.id("clip") |
| 389 | fmt.Fprintf(&w.defs, `<clipPath id="%s"><rect %s/></clipPath>`+"\n", id, strings.Join(box, " ")) |
| 390 | box = append(box, attr("clip-path", "url(#"+id+")")) |
| 391 | } |
| 392 | |
| 393 | if nested != nil { |
| 394 | // The asset's own document, inside a viewport at the box. Its |
| 395 | // <svg> keeps its viewBox and its own preserveAspectRatio, so |
| 396 | // a nested picture always fits rather than filling — the one |
| 397 | // place `fit` is not honoured, and a vector asset is a diagram |
| 398 | // far more often than it is a photograph. |
| 399 | w.out.WriteString("<svg " + strings.Join(box, " ") + ">\n") |
| 400 | w.out.Write(nested) |
| 401 | w.out.WriteString("\n</svg>\n") |
| 402 | return |
| 403 | } |
| 404 | |
| 405 | attrs := append(box, attr("preserveAspectRatio", fit), attr("href", href)) |
| 406 | if im.Alt == "" { |
| 407 | w.out.WriteString("<image " + strings.Join(attrs, " ") + "/>\n") |
| 408 | return |
| 409 | } |
| 410 | w.out.WriteString("<image " + strings.Join(attrs, " ") + "><title>" + esc(im.Alt) + "</title></image>\n") |
| 411 | } |
| 412 | |
| 413 | // nestableSVG is an SVG asset's document, ready to be written inside |
| 414 | // this one, or nil for one that must not be. |
| 415 | // |
| 416 | // Two refusals, and both of them are about a picture being a document: |
| 417 | // |
| 418 | // - A <style> element. CSS in an SVG is document-wide, so |
| 419 | // `text { fill: red }` inside a logo would repaint the words of the |
| 420 | // slide around it. Nothing this package could scope would fix that, |
| 421 | // so the picture goes back to being its own document at its own |
| 422 | // URL. |
| 423 | // - Anything encoding/xml will not read, or that carries a directive |
| 424 | // or a processing instruction. The output of this package is |
| 425 | // promised to be well-formed XML; bytes that are not cannot be |
| 426 | // spliced into it on the strength of having come from the store. |
| 427 | // |
| 428 | // What it does NOT re-check is safety: an asset is sanitised the moment |
| 429 | // it is registered (internal/slides/svg.go, and the note there about |
| 430 | // what an uploaded SVG can carry), and re-deriving that judgement here |
| 431 | // from a second, weaker parser would be the way the two come to |
| 432 | // disagree. |
| 433 | func nestableSVG(raw []byte) []byte { |
| 434 | i := bytes.Index(raw, []byte("<svg")) |
| 435 | if i < 0 { |
| 436 | return nil |
| 437 | } |
| 438 | frag := raw[i:] |
| 439 | dec := xml.NewDecoder(bytes.NewReader(frag)) |
| 440 | depth := 0 |
| 441 | for { |
| 442 | tok, err := dec.Token() |
| 443 | if errors.Is(err, io.EOF) { |
| 444 | break |
| 445 | } |
| 446 | if err != nil { |
| 447 | return nil |
| 448 | } |
| 449 | switch t := tok.(type) { |
| 450 | case xml.ProcInst, xml.Directive: |
| 451 | return nil |
| 452 | case xml.StartElement: |
| 453 | if strings.EqualFold(t.Name.Local, "style") { |
| 454 | return nil |
| 455 | } |
| 456 | depth++ |
| 457 | case xml.EndElement: |
| 458 | depth-- |
| 459 | } |
| 460 | } |
| 461 | if depth != 0 { |
| 462 | return nil |
| 463 | } |
| 464 | return frag |
| 465 | } |
| 466 | |
| 467 | // ── Writing it down ────────────────────────────────────────────────── |
| 468 | |
| 469 | // attr is one name="value" pair. The value is already escaped by the |
| 470 | // caller; the name is a literal from this file. |
| 471 | func attr(name, value string) string { return name + `="` + value + `"` } |
| 472 | |
| 473 | // fillOr is a colour, or the word for its absence. |
| 474 | func fillOr(f layout.Fill, absent string) string { |
| 475 | if f == "" { |
| 476 | return absent |
| 477 | } |
| 478 | return esc(string(f)) |
| 479 | } |
| 480 | |
| 481 | // num writes a coordinate: four decimals at most, and never in |
| 482 | // exponent form. |
| 483 | // |
| 484 | // The arithmetic upstream runs through ratios of a 16.8px unit, so a |
| 485 | // coordinate that is 126.72 comes out of the floats as |
| 486 | // 126.72000000000001 about as often as not. Nothing can see a |
| 487 | // ten-thousandth of a pixel, and 'f' rather than 'g' because SVG does |
| 488 | // not read "1e-9" as a number at all — it is a parse error, and the |
| 489 | // shape carrying it does not draw. |
| 490 | func num(v float64) string { |
| 491 | if math.IsNaN(v) || math.IsInf(v, 0) { |
| 492 | return "0" |
| 493 | } |
| 494 | return strconv.FormatFloat(math.Round(v*1e4)/1e4, 'f', -1, 64) |
| 495 | } |
| 496 | |
| 497 | // escaper is the five predefined entities. Apostrophe and quote are |
| 498 | // escaped everywhere rather than only in attributes: one function that |
| 499 | // is safe in both places cannot be used in the wrong one. |
| 500 | var escaper = strings.NewReplacer( |
| 501 | "&", "&", "<", "<", ">", ">", `"`, """, "'", "'") |
| 502 | |
| 503 | // esc is every string that reaches the output, escaped — and stripped |
| 504 | // of the characters XML 1.0 has no way to carry. |
| 505 | // |
| 506 | // The stripping is what makes Encode's "well-formed, always" true. A |
| 507 | // deck's text arrives through the CRDT from a browser, and a stray |
| 508 | // control character in it would otherwise produce a file no parser will |
| 509 | // open: not a slide with a wrong glyph, a slide that does not render. |
| 510 | // The lost character is invisible; the lost document is the whole |
| 511 | // picture. |
| 512 | func esc(s string) string { |
| 513 | if strings.IndexFunc(s, badForXML) >= 0 { |
| 514 | s = strings.Map(func(r rune) rune { |
| 515 | if badForXML(r) { |
| 516 | return -1 |
| 517 | } |
| 518 | return r |
| 519 | }, s) |
| 520 | } |
| 521 | return escaper.Replace(s) |
| 522 | } |
| 523 | |
| 524 | // badForXML is the XML 1.0 Char production, inverted: everything below |
| 525 | // space except tab, newline and carriage return, the surrogates, and |
| 526 | // the two permanently unassigned code points at the end of the BMP. |
| 527 | // |
| 528 | // U+FFFD is not among them. It is a legal character, and it is what |
| 529 | // strings.Map produces for every invalid UTF-8 byte it meets — the one |
| 530 | // place something is standing in for a character this cannot know, and |
| 531 | // a visible box in the picture is the honest way to show that. |
| 532 | func badForXML(r rune) bool { |
| 533 | switch { |
| 534 | case r == '\t' || r == '\n' || r == '\r': |
| 535 | return false |
| 536 | case r < 0x20: |
| 537 | return true |
| 538 | case r >= 0xd800 && r <= 0xdfff: |
| 539 | return true |
| 540 | case r == 0xfffe || r == 0xffff: |
| 541 | return true |
| 542 | } |
| 543 | return false |
| 544 | } |
| 545 | |
amadan