@boxpdf/html-writer 0.1.21 → 0.1.22

package/dist/index.cjs

@@ -372,17 +372,20 @@ function number(value) {
 // src/visual-font.ts
 function visualFontAliases(pageNumber, fonts) {
   return new Map(
-    fonts.filter((font) => font.format === "truetype" && !/(?:courier|^TTE)/i.test(font.family ?? "")).map((font) => [font.id, `boxpdf-${pageNumber}-${font.id}`])
+    fonts.filter(
+      (font) => (font.format === "truetype" || font.format === "opentype") && !/(?:courier|^TTE)/i.test(font.family ?? "")
+    ).map((font) => [font.id, `boxpdf-${pageNumber}-${font.id}`])
   );
 }
 function visualFontFace(font, aliases) {
-  if (font.format !== "truetype") return "";
+  if (font.format !== "truetype" && font.format !== "opentype") return "";
   const alias = aliases.get(font.id);
   if (!alias) return "";
   const styles2 = visualFontStyles(font.family, alias).filter(
     (style) => !style.startsWith("font-family:")
   );
-  return `@font-face{font-family:${alias};src:url(data:font/ttf;base64,${base64(font.data)}) format("truetype");${styles2.join(";")}}`;
+  const mime = font.format === "opentype" ? "font/otf" : "font/ttf";
+  return `@font-face{font-family:${alias};src:url(data:${mime};base64,${base64(font.data)}) format("${font.format}");${styles2.join(";")}}`;
 }
 function visualFontStyles(fontFamily, alias) {
   const normalized = fontFamily?.toLowerCase() ?? "";
@@ -464,7 +467,9 @@ function vectorMedia(page, imageOptions) {
   );
   const aliases = visualFontAliases(page.number, page.fonts ?? []);
   const visualCodeFonts = new Set(
-    (page.fonts ?? []).filter((font) => font.format === "truetype" && font.visualCodeMapping).map((font) => font.id)
+    (page.fonts ?? []).filter(
+      (font) => (font.format === "truetype" || font.format === "opentype") && font.visualCodeMapping
+    ).map((font) => font.id)
   );
   return components.map((component, componentIndex) => {
     const bounds2 = component.bounds;
@@ -1317,7 +1322,8 @@ async function writeHtmlDocument(pages, write, options = {}) {
     );
     options.onSemanticStats?.(stats);
   } else {
-    for await (const page of pages) await writePage(page, write, options);
+    const documentFonts = { entries: [] };
+    for await (const page of pages) await writePositionedPage(page, write, options, documentFonts);
   }
   await write("</main>");
   if (includeDocument) await write("</body></html>");
@@ -1360,7 +1366,7 @@ async function pageToHtml(page, options = {}) {
   );
   return output;
 }
-async function writePositionedPage(page, write, options) {
+async function writePositionedPage(page, write, options, documentFonts) {
   const imageOptions = resolveImageOptions("visual", options);
   const visualImages = await prepareVisualImages(page, imageOptions, options.onImage);
   const visualSpans = coalesceVisualSpans(page.visualSpans ?? page.spans);
@@ -1371,14 +1377,18 @@ async function writePositionedPage(page, write, options) {
   await write(
     `<section class="pdf-page pdf-page--visual pdf-page--positioned" data-page="${page.number}" data-rotate="${page.rotate}" style="width:${number3(displayWidth)}pt;height:${number3(displayHeight)}pt">`
   );
-  const fontAliases = visualFontAliases(page.number, page.fonts ?? []);
+  const { aliases: fontAliases, fontsToEmit } = visualPageFonts(
+    page.number,
+    page.fonts ?? [],
+    documentFonts
+  );
   const type3Fonts = new Map(
     (page.fonts ?? []).filter((font) => font.format === "type3").map((font) => [font.id, font])
   );
   const textClasses = options.includeStyles ?? true ? visualTextClasses(page.number, visualSpans, fontAliases) : void 0;
-  if ((options.includeStyles ?? true) && page.fonts?.length) {
+  if ((options.includeStyles ?? true) && fontsToEmit.length) {
     await write(
-      `<style>${page.fonts.map((font) => visualFontFace(font, fontAliases)).join("")}</style>`
+      `<style>${fontsToEmit.map((font) => visualFontFace(font, fontAliases)).join("")}</style>`
     );
   }
   if (textClasses?.css) await write(`<style>${textClasses.css}</style>`);
@@ -1445,6 +1455,30 @@ async function writePositionedPage(page, write, options) {
   }
   await write("</div></section>");
 }
+function visualPageFonts(pageNumber, fonts, documentFonts) {
+  const aliases = visualFontAliases(pageNumber, fonts);
+  if (!documentFonts) return { aliases, fontsToEmit: fonts };
+  const fontsToEmit = [];
+  for (const font of fonts) {
+    if (!aliases.has(font.id) || font.format === "type3") continue;
+    const existing = documentFonts.entries.find(
+      (entry) => entry.format === font.format && equalBytes(entry.data, font.data)
+    );
+    if (existing) {
+      aliases.set(font.id, existing.alias);
+      continue;
+    }
+    const alias = `boxpdf-document-font-${documentFonts.entries.length + 1}`;
+    aliases.set(font.id, alias);
+    documentFonts.entries.push({ alias, data: font.data, format: font.format });
+    fontsToEmit.push(font);
+  }
+  return { aliases, fontsToEmit };
+}
+function equalBytes(left, right) {
+  if (left.length !== right.length) return false;
+  return left.every((value, index) => value === right[index]);
+}
 function coalesceVisualSpans(spans) {
   const output = [];
   for (const span of spans) {
@@ -1470,11 +1504,14 @@ function canCoalesceVisualSpans(left, right) {
   if (left.direction !== "ltr" || right.direction !== "ltr") return false;
   if (/guardian/i.test(left.fontFamily ?? "")) return false;
   if (left.glyphCodes || right.glyphCodes) return false;
+  if (left.fontAssetId || right.fontAssetId) return false;
   if (!sameVisualTextState(left, right)) return false;
   const tolerance = Math.max(0.02, left.fontSize * 0.015);
   if (Math.abs(left.bounds.y - right.bounds.y) > tolerance) return false;
   const gap = right.bounds.x - (left.bounds.x + left.bounds.width);
-  return !right.hasLeadingSpace && gap >= -tolerance && gap <= tolerance;
+  if (right.hasLeadingSpace) return false;
+  if (gap >= -tolerance && gap <= tolerance) return true;
+  return right.textAdjustmentBefore !== void 0 && right.textAdjustmentBefore < 0 && Math.abs(right.textAdjustmentBefore - gap) <= 1e-3;
 }
 function sameVisualTextState(left, right) {
   return Math.abs(left.fontSize - right.fontSize) <= 1e-3 && left.fontName === right.fontName && left.fontFamily === right.fontFamily && left.fontAssetId === right.fontAssetId && left.color === right.color && left.fillOpacity === right.fillOpacity && left.strokeColor === right.strokeColor && left.strokeWidth === right.strokeWidth && left.strokeOpacity === right.strokeOpacity && left.renderingMode === right.renderingMode && sameTransform(left.transform, right.transform);
@@ -1515,6 +1552,13 @@ function visualTextLine(spans, startIndex, styleClasses, pageHeight, fontAliases
   if (endIndex === startIndex) return void 0;
   const baseline = pageHeight - first.bounds.y;
   const lineSpans = spans.slice(startIndex, endIndex + 1);
+  const dxRun = visualDxTextRun(lineSpans, fontAliases);
+  if (dxRun) {
+    return {
+      html: `<text class="${className}" x="${number3(first.bounds.x)}" y="${number3(baseline)}" dx="${dxRun.offsets.map(number3).join(" ")}">${escapeHtml4(dxRun.text)}</text>`,
+      endIndex
+    };
+  }
   const content = lineSpans.map(
     (span, index) => visualTextTspan(span, index > 0 ? textSpanGap(lineSpans[index - 1], span) : void 0)
   ).join("");
@@ -1523,6 +1567,32 @@ function visualTextLine(spans, startIndex, styleClasses, pageHeight, fontAliases
     endIndex
   };
 }
+function visualDxTextRun(spans, fontAliases) {
+  const first = spans[0];
+  if (!first?.fontAssetId || !fontAliases.has(first.fontAssetId) || spans.length < 2)
+    return void 0;
+  let text = first.text;
+  const offsets = characterOffsets(first, 0);
+  for (let index = 1; index < spans.length; index += 1) {
+    const previous = spans[index - 1];
+    const current = spans[index];
+    if (!previous || !current || current.fontAssetId !== first.fontAssetId) return void 0;
+    const characters = [...current.text];
+    if (characters.length === 0 || previous.naturalWidth === void 0) return void 0;
+    const gap = current.bounds.x - (previous.bounds.x + previous.bounds.width);
+    const trailingSpacing = (previous.characterSpacing ?? 0) + (previous.text.endsWith(" ") ? previous.wordSpacing ?? 0 : 0);
+    offsets.push(...characterOffsets(current, gap + trailingSpacing));
+    text += current.text;
+  }
+  while (offsets.at(-1) === 0) offsets.pop();
+  return offsets.length > 0 ? { text, offsets } : void 0;
+}
+function characterOffsets(span, first) {
+  const characters = [...span.text];
+  return characters.map(
+    (_, index) => index === 0 ? first : (span.characterSpacing ?? 0) + (characters[index - 1] === " " ? span.wordSpacing ?? 0 : 0)
+  );
+}
 function visualTextTspan(span, dx) {
   const extent = span.bounds.width;
   const offset = dx === void 0 || number3(dx) === "0" ? "" : ` dx="${number3(dx)}"`;