pptx-glimpse 0.10.1 → 0.10.3

package/dist/index.cjs

@@ -1692,6 +1692,7 @@ var standardParser = new import_fast_xml_parser.XMLParser({
   attributeNamePrefix: "@_",
   removeNSPrefix: true,
   htmlEntities: true,
+  trimValues: false,
   isArray: (_name, jpath, _isLeafNode, _isAttribute) => {
     const tag = String(jpath).split(".").pop() ?? "";
     return ARRAY_TAGS.has(tag);
@@ -3911,6 +3912,114 @@ function findMatchingPlaceholder(placeholderType, placeholderIdx, styles) {
   }
   return void 0;
 }
+function hasMultipleBulletPPr(p, orderedChildren) {
+  const rawPPr = p.pPr;
+  const pPrList = Array.isArray(rawPPr) ? rawPPr : rawPPr ? [rawPPr] : [];
+  let bulletPPrCount = 0;
+  let pPrCounter = 0;
+  for (const child of orderedChildren) {
+    const tag = Object.keys(child).find((k) => k !== ":@");
+    if (tag === "pPr") {
+      const pPrNode = pPrList[pPrCounter];
+      if (pPrNode && (pPrNode.buChar !== void 0 || pPrNode.buAutoNum !== void 0 || pPrNode.buBlip !== void 0)) {
+        bulletPPrCount++;
+        if (bulletPPrCount >= 2) return true;
+      }
+      pPrCounter++;
+    }
+  }
+  return false;
+}
+function splitInterleavedParagraph(p, orderedPChildren, colorResolver, rels, fontScheme, lstStyle) {
+  const rawPPr = p.pPr;
+  const pPrList = Array.isArray(rawPPr) ? rawPPr : rawPPr ? [rawPPr] : [];
+  const rList = p.r ?? [];
+  const fldList = p.fld ?? [];
+  const brList = p.br ?? [];
+  const groups = [];
+  let currentGroup = null;
+  let pPrCounter = 0;
+  const tagCounters = {};
+  for (const child of orderedPChildren) {
+    const tag = Object.keys(child).find((k) => k !== ":@");
+    if (!tag) continue;
+    if (tag === "pPr") {
+      const pPrNode = pPrList[pPrCounter];
+      const hasBullet = pPrNode && (pPrNode.buChar !== void 0 || pPrNode.buAutoNum !== void 0 || pPrNode.buBlip !== void 0);
+      if (hasBullet || !currentGroup) {
+        if (currentGroup) groups.push(currentGroup);
+        currentGroup = {
+          pPrIdx: pPrCounter,
+          rIndices: [],
+          fldIndices: [],
+          brIndices: [],
+          orderedChildren: [child]
+        };
+      } else {
+        currentGroup.orderedChildren.push(child);
+      }
+      pPrCounter++;
+    } else if (tag === "endParaRPr") {
+    } else {
+      const globalIdx = tagCounters[tag] ?? 0;
+      tagCounters[tag] = globalIdx + 1;
+      if (!currentGroup) {
+        currentGroup = {
+          pPrIdx: -1,
+          rIndices: [],
+          fldIndices: [],
+          brIndices: [],
+          orderedChildren: []
+        };
+      }
+      if (tag === "r") currentGroup.rIndices.push(globalIdx);
+      else if (tag === "fld") currentGroup.fldIndices.push(globalIdx);
+      else if (tag === "br") currentGroup.brIndices.push(globalIdx);
+      currentGroup.orderedChildren.push(child);
+    }
+  }
+  if (currentGroup) groups.push(currentGroup);
+  const results = [];
+  for (let g = 0; g < groups.length; g++) {
+    const group = groups[g];
+    const isLast = g === groups.length - 1;
+    const syntheticP = {};
+    if (group.pPrIdx >= 0 && pPrList[group.pPrIdx]) {
+      syntheticP.pPr = pPrList[group.pPrIdx];
+    }
+    if (group.rIndices.length > 0) {
+      syntheticP.r = group.rIndices.map((i) => rList[i]);
+    }
+    if (group.fldIndices.length > 0) {
+      syntheticP.fld = group.fldIndices.map((i) => fldList[i]);
+    }
+    if (group.brIndices.length > 0) {
+      syntheticP.br = group.brIndices.map((i) => brList[i]);
+    }
+    if (isLast && p.endParaRPr) {
+      syntheticP.endParaRPr = p.endParaRPr;
+    }
+    const paragraph = parseParagraph(
+      syntheticP,
+      colorResolver,
+      rels,
+      fontScheme,
+      lstStyle,
+      group.orderedChildren
+    );
+    if (!isLast && paragraph.runs.length > 0) {
+      const lastRun = paragraph.runs[paragraph.runs.length - 1];
+      if (lastRun.text.endsWith("\n")) {
+        paragraph.runs[paragraph.runs.length - 1] = {
+          ...lastRun,
+          text: lastRun.text.slice(0, -1)
+        };
+      }
+    }
+    results.push(paragraph);
+  }
+  return results;
+}
 function parseTextBody(txBody, colorResolver, rels, fontScheme, lstStyleOverride, orderedTxBody) {
   if (!txBody) return null;
   const bodyPr = txBody.bodyPr;
@@ -3955,9 +4064,23 @@ function parseTextBody(txBody, colorResolver, rels, fontScheme, lstStyleOverride
   const paragraphs = [];
   const pList = txBody.p ?? [];
   for (let i = 0; i < pList.length; i++) {
-    paragraphs.push(
-      parseParagraph(pList[i], colorResolver, rels, fontScheme, lstStyle, orderedParagraphs[i])
-    );
+    const orderedChildren = orderedParagraphs[i];
+    if (orderedChildren && hasMultipleBulletPPr(pList[i], orderedChildren)) {
+      paragraphs.push(
+        ...splitInterleavedParagraph(
+          pList[i],
+          orderedChildren,
+          colorResolver,
+          rels,
+          fontScheme,
+          lstStyle
+        )
+      );
+    } else {
+      paragraphs.push(
+        parseParagraph(pList[i], colorResolver, rels, fontScheme, lstStyle, orderedChildren)
+      );
+    }
   }
   if (paragraphs.length === 0) return null;
   return { paragraphs, bodyProperties };
@@ -7938,7 +8061,7 @@ function renderTextBody(textBody, transform) {
   const firstParaFontSizePt = getParagraphFontSize(paragraphs[0], defaultFontSize) * fontScale;
   const firstLineBaselineOffsetPt = firstParaFontSizePt * defaultAscenderRatio;
   yStart += firstLineBaselineOffsetPt * PX_PER_PT3;
-  const textElement = `<text x="0" y="${yStart}">${tspans.join("")}</text>`;
+  const textElement = `<text x="0" y="${yStart}" xml:space="preserve">${tspans.join("")}</text>`;
   if (isVerticalText(bodyProperties.vert)) {
     return `<g transform="translate(${originalWidth}, 0) rotate(90)">${textElement}</g>`;
   }

package/dist/index.js

@@ -1655,6 +1655,7 @@ var standardParser = new XMLParser({
   attributeNamePrefix: "@_",
   removeNSPrefix: true,
   htmlEntities: true,
+  trimValues: false,
   isArray: (_name, jpath, _isLeafNode, _isAttribute) => {
     const tag = String(jpath).split(".").pop() ?? "";
     return ARRAY_TAGS.has(tag);
@@ -3873,6 +3874,114 @@ function findMatchingPlaceholder(placeholderType, placeholderIdx, styles) {
   }
   return void 0;
 }
+function hasMultipleBulletPPr(p, orderedChildren) {
+  const rawPPr = p.pPr;
+  const pPrList = Array.isArray(rawPPr) ? rawPPr : rawPPr ? [rawPPr] : [];
+  let bulletPPrCount = 0;
+  let pPrCounter = 0;
+  for (const child of orderedChildren) {
+    const tag = Object.keys(child).find((k) => k !== ":@");
+    if (tag === "pPr") {
+      const pPrNode = pPrList[pPrCounter];
+      if (pPrNode && (pPrNode.buChar !== void 0 || pPrNode.buAutoNum !== void 0 || pPrNode.buBlip !== void 0)) {
+        bulletPPrCount++;
+        if (bulletPPrCount >= 2) return true;
+      }
+      pPrCounter++;
+    }
+  }
+  return false;
+}
+function splitInterleavedParagraph(p, orderedPChildren, colorResolver, rels, fontScheme, lstStyle) {
+  const rawPPr = p.pPr;
+  const pPrList = Array.isArray(rawPPr) ? rawPPr : rawPPr ? [rawPPr] : [];
+  const rList = p.r ?? [];
+  const fldList = p.fld ?? [];
+  const brList = p.br ?? [];
+  const groups = [];
+  let currentGroup = null;
+  let pPrCounter = 0;
+  const tagCounters = {};
+  for (const child of orderedPChildren) {
+    const tag = Object.keys(child).find((k) => k !== ":@");
+    if (!tag) continue;
+    if (tag === "pPr") {
+      const pPrNode = pPrList[pPrCounter];
+      const hasBullet = pPrNode && (pPrNode.buChar !== void 0 || pPrNode.buAutoNum !== void 0 || pPrNode.buBlip !== void 0);
+      if (hasBullet || !currentGroup) {
+        if (currentGroup) groups.push(currentGroup);
+        currentGroup = {
+          pPrIdx: pPrCounter,
+          rIndices: [],
+          fldIndices: [],
+          brIndices: [],
+          orderedChildren: [child]
+        };
+      } else {
+        currentGroup.orderedChildren.push(child);
+      }
+      pPrCounter++;
+    } else if (tag === "endParaRPr") {
+    } else {
+      const globalIdx = tagCounters[tag] ?? 0;
+      tagCounters[tag] = globalIdx + 1;
+      if (!currentGroup) {
+        currentGroup = {
+          pPrIdx: -1,
+          rIndices: [],
+          fldIndices: [],
+          brIndices: [],
+          orderedChildren: []
+        };
+      }
+      if (tag === "r") currentGroup.rIndices.push(globalIdx);
+      else if (tag === "fld") currentGroup.fldIndices.push(globalIdx);
+      else if (tag === "br") currentGroup.brIndices.push(globalIdx);
+      currentGroup.orderedChildren.push(child);
+    }
+  }
+  if (currentGroup) groups.push(currentGroup);
+  const results = [];
+  for (let g = 0; g < groups.length; g++) {
+    const group = groups[g];
+    const isLast = g === groups.length - 1;
+    const syntheticP = {};
+    if (group.pPrIdx >= 0 && pPrList[group.pPrIdx]) {
+      syntheticP.pPr = pPrList[group.pPrIdx];
+    }
+    if (group.rIndices.length > 0) {
+      syntheticP.r = group.rIndices.map((i) => rList[i]);
+    }
+    if (group.fldIndices.length > 0) {
+      syntheticP.fld = group.fldIndices.map((i) => fldList[i]);
+    }
+    if (group.brIndices.length > 0) {
+      syntheticP.br = group.brIndices.map((i) => brList[i]);
+    }
+    if (isLast && p.endParaRPr) {
+      syntheticP.endParaRPr = p.endParaRPr;
+    }
+    const paragraph = parseParagraph(
+      syntheticP,
+      colorResolver,
+      rels,
+      fontScheme,
+      lstStyle,
+      group.orderedChildren
+    );
+    if (!isLast && paragraph.runs.length > 0) {
+      const lastRun = paragraph.runs[paragraph.runs.length - 1];
+      if (lastRun.text.endsWith("\n")) {
+        paragraph.runs[paragraph.runs.length - 1] = {
+          ...lastRun,
+          text: lastRun.text.slice(0, -1)
+        };
+      }
+    }
+    results.push(paragraph);
+  }
+  return results;
+}
 function parseTextBody(txBody, colorResolver, rels, fontScheme, lstStyleOverride, orderedTxBody) {
   if (!txBody) return null;
   const bodyPr = txBody.bodyPr;
@@ -3917,9 +4026,23 @@ function parseTextBody(txBody, colorResolver, rels, fontScheme, lstStyleOverride
   const paragraphs = [];
   const pList = txBody.p ?? [];
   for (let i = 0; i < pList.length; i++) {
-    paragraphs.push(
-      parseParagraph(pList[i], colorResolver, rels, fontScheme, lstStyle, orderedParagraphs[i])
-    );
+    const orderedChildren = orderedParagraphs[i];
+    if (orderedChildren && hasMultipleBulletPPr(pList[i], orderedChildren)) {
+      paragraphs.push(
+        ...splitInterleavedParagraph(
+          pList[i],
+          orderedChildren,
+          colorResolver,
+          rels,
+          fontScheme,
+          lstStyle
+        )
+      );
+    } else {
+      paragraphs.push(
+        parseParagraph(pList[i], colorResolver, rels, fontScheme, lstStyle, orderedChildren)
+      );
+    }
   }
   if (paragraphs.length === 0) return null;
   return { paragraphs, bodyProperties };
@@ -7900,7 +8023,7 @@ function renderTextBody(textBody, transform) {
   const firstParaFontSizePt = getParagraphFontSize(paragraphs[0], defaultFontSize) * fontScale;
   const firstLineBaselineOffsetPt = firstParaFontSizePt * defaultAscenderRatio;
   yStart += firstLineBaselineOffsetPt * PX_PER_PT3;
-  const textElement = `<text x="0" y="${yStart}">${tspans.join("")}</text>`;
+  const textElement = `<text x="0" y="${yStart}" xml:space="preserve">${tspans.join("")}</text>`;
   if (isVerticalText(bodyProperties.vert)) {
     return `<g transform="translate(${originalWidth}, 0) rotate(90)">${textElement}</g>`;
   }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pptx-glimpse",
-  "version": "0.10.1",
+  "version": "0.10.3",
   "description": "A lightweight JavaScript library for rendering PowerPoint (.pptx) files as SVG or PNG in Node.js. No LibreOffice required.",
   "type": "module",
   "main": "./dist/index.cjs",