pptxtojson 2.0.3 → 2.0.5

package/src/pptxtojson.js

@@ -7,7 +7,7 @@ import { getVerticalAlign, getTextAutoFit } from './paragraph'
 import { getTextInsets } from './textInsets'
 import { getPosition, getSize } from './position'
 import { genTextBody, getTextNodeValue } from './text'
-import { getCustomShapePath, identifyShape } from './shape'
+import { getCustomShapePath, identifyShape, isStrokeOnlyCustomGeometry } from './shape'
 import { extractFileExtension, getTextByPathList, angleToDegrees, isVideoLink, escapeHtml, hasValidText, numberToFixed } from './utils'
 import { getShadow } from './shadow'
 import { getTableBorders, getTableCellParams, getTableRowParams } from './table'
@@ -22,6 +22,7 @@ export async function parse(file, options = {}) {
   const loadedImages = {}
   const loadedVideos = {}
   const loadedAudios = {}
+  const xmlCache = {}
   const parseOptions = {
     ...options,
     imageMode: options.imageMode || 'base64',
@@ -37,7 +38,7 @@ export async function parse(file, options = {}) {
   const usedFonts = await getUsedFonts(zip)
 
   for (const filename of filesInfo.slides) {
-    const singleSlide = await processSingleSlide(zip, filename, themeContent, defaultTextStyle, loadedImages, loadedVideos, loadedAudios, parseOptions)
+    const singleSlide = await processSingleSlide(zip, filename, themeContent, defaultTextStyle, loadedImages, loadedVideos, loadedAudios, parseOptions, xmlCache)
     slides.push(singleSlide)
   }
 
@@ -143,7 +144,16 @@ async function getTheme(zip) {
   return { themeContent, themeColors }
 }
 
-async function processSingleSlide(zip, sldFileName, themeContent, defaultTextStyle, loadedImages, loadedVideos, loadedAudios, options) {
+async function readXmlFileCached(zip, filename, xmlCache) {
+  if (!filename) return null
+  if (Object.prototype.hasOwnProperty.call(xmlCache, filename)) return xmlCache[filename]
+
+  const content = await readXmlFile(zip, filename)
+  xmlCache[filename] = content
+  return content
+}
+
+async function processSingleSlide(zip, sldFileName, themeContent, defaultTextStyle, loadedImages, loadedVideos, loadedAudios, options, xmlCache) {
   const resName = sldFileName.replace('slides/slide', 'slides/_rels/slide') + '.rels'
   const resContent = await readXmlFile(zip, resName)
   let relationshipArray = resContent['Relationships']['Relationship']
@@ -206,10 +216,10 @@ async function processSingleSlide(zip, sldFileName, themeContent, defaultTextSty
   const slideNotesContent = await readXmlFile(zip, noteFilename)
   const note = getNote(slideNotesContent)
 
-  const slideLayoutContent = await readXmlFile(zip, layoutFilename)
-  const slideLayoutTables = await indexNodes(slideLayoutContent)
+  const slideLayoutContent = await readXmlFileCached(zip, layoutFilename, xmlCache)
+  const slideLayoutTables = indexNodes(slideLayoutContent)
   const slideLayoutResFilename = layoutFilename.replace('slideLayouts/slideLayout', 'slideLayouts/_rels/slideLayout') + '.rels'
-  const slideLayoutResContent = await readXmlFile(zip, slideLayoutResFilename)
+  const slideLayoutResContent = await readXmlFileCached(zip, slideLayoutResFilename, xmlCache)
   relationshipArray = slideLayoutResContent['Relationships']['Relationship']
   if (relationshipArray.constructor !== Array) relationshipArray = [relationshipArray]
 
@@ -231,11 +241,11 @@ async function processSingleSlide(zip, sldFileName, themeContent, defaultTextSty
     }
   }
 
-  const slideMasterContent = await readXmlFile(zip, masterFilename)
+  const slideMasterContent = await readXmlFileCached(zip, masterFilename, xmlCache)
   const slideMasterTextStyles = getTextByPathList(slideMasterContent, ['p:sldMaster', 'p:txStyles'])
   const slideMasterTables = indexNodes(slideMasterContent)
   const slideMasterResFilename = masterFilename.replace('slideMasters/slideMaster', 'slideMasters/_rels/slideMaster') + '.rels'
-  const slideMasterResContent = await readXmlFile(zip, slideMasterResFilename)
+  const slideMasterResContent = await readXmlFileCached(zip, slideMasterResFilename, xmlCache)
   relationshipArray = slideMasterResContent['Relationships']['Relationship']
   if (relationshipArray.constructor !== Array) relationshipArray = [relationshipArray]
 
@@ -275,7 +285,7 @@ async function processSingleSlide(zip, sldFileName, themeContent, defaultTextSty
     }
   }
 
-  const tableStyles = await readXmlFile(zip, 'ppt/tableStyles.xml')
+  const tableStyles = await readXmlFileCached(zip, 'ppt/tableStyles.xml', xmlCache)
 
   const slideContent = await readXmlFile(zip, sldFileName)
   const nodes = slideContent['p:sld']['p:cSld']['p:spTree']
@@ -398,7 +408,7 @@ function getNote(noteContent) {
           text += `<${listType}>`
           listTypes[listLevel] = listType
         }
-        text += `<li style="text-align:${align};">`
+        text += `<li><p style="text-align:${align};">`
       }
       else {
         while (listTypes.length > 0) {
@@ -416,7 +426,7 @@ function getNote(noteContent) {
         }
       }
 
-      if (listType) text += '</li>'
+      if (listType) text += '</p></li>'
       else text += '</p>'
     }
     while (listTypes.length > 0) {
@@ -559,6 +569,8 @@ async function processMathNode(node, warpObj, source) {
   const { width, height } = getSize(xfrmNode, undefined, undefined)
 
   const oMath = findOMath(choice)[0]
+  if (!oMath) return null
+
   const latex = latexFormart(parseOMath(oMath))
 
   const blipFill = getTextByPathList(fallback, ['p:sp', 'p:spPr', 'a:blipFill'])
@@ -606,8 +618,8 @@ async function processGroupSpNode(node, warpObj, source, parentGroupHierarchy =
   if (rotate) rotate = angleToDegrees(rotate)
 
   // 计算缩放因子
-  const ws = cx / chcx
-  const hs = cy / chcy
+  const ws = chcx === 0 ? 0 : cx / chcx
+  const hs = chcy === 0 ? 0 : cy / chcy
 
   // 构建当前组合层级（将当前组合添加到父级层级中）
   const currentGroupHierarchy = [...parentGroupHierarchy, node]
@@ -626,30 +638,44 @@ async function processGroupSpNode(node, warpObj, source, parentGroupHierarchy =
     }
   }
 
+  const transformGroupedElement = (element, offsetX = 0, offsetY = 0) => {
+    const elementRotate = element.rotate || 0
+    const normalizedRotate = ((elementRotate % 360) + 360) % 360
+    const isUniformScale = Math.abs(ws - hs) < 0.000001
+    const shouldSwapDimensions = normalizedRotate === 90 || normalizedRotate === 270
+    const centerX = element.left + element.width / 2
+    const centerY = element.top + element.height / 2
+    const nextCenterX = (centerX - offsetX) * ws
+    const nextCenterY = (centerY - offsetY) * hs
+    const widthScale = shouldSwapDimensions && !isUniformScale ? hs : ws
+    const heightScale = shouldSwapDimensions && !isUniformScale ? ws : hs
+    const width = element.width * widthScale
+    const height = element.height * heightScale
+
+    const transformed = {
+      ...element,
+      left: numberToFixed(nextCenterX - width / 2),
+      top: numberToFixed(nextCenterY - height / 2),
+      width: numberToFixed(width),
+      height: numberToFixed(height),
+    }
+    return transformed
+  }
+
   const processedElements = elements.map(element => ({
-    ...element,
-    left: numberToFixed((element.left - chx) * ws),
-    top: numberToFixed((element.top - chy) * hs),
-    width: numberToFixed(element.width * ws),
-    height: numberToFixed(element.height * hs),
+    ...transformGroupedElement(element, chx, chy),
     ...(element.type === 'group' && element.elements ? {
-      elements: processNestedGroupElements(element.elements, ws, hs)
+      elements: processNestedGroupElements(element.elements)
     } : {})
   }))
 
-  function processNestedGroupElements(elements, ws, hs, depth = 0) {
+  function processNestedGroupElements(elements, depth = 0) {
     if (depth > 10) return elements
 
     return elements.map(element => {
-      const processed = {
-        ...element,
-        left: numberToFixed(element.left * ws),
-        top: numberToFixed(element.top * hs),
-        width: numberToFixed(element.width * ws),
-        height: numberToFixed(element.height * hs),
-      }
+      const processed = transformGroupedElement(element)
       if (element.type === 'group' && element.elements) {
-        processed.elements = processNestedGroupElements(element.elements, ws, hs, depth + 1)
+        processed.elements = processNestedGroupElements(element.elements, depth + 1)
       }
       return processed
     })
@@ -767,7 +793,11 @@ async function genShape(node, slideLayoutSpNode, slideMasterSpNode, name, type,
   if (node['p:txBody']) content = genTextBody(node['p:txBody'], node, slideLayoutSpNode, slideMasterSpNode, type, warpObj)
 
   const { borderColor, borderWidth, borderType, strokeDasharray } = getBorder(node, type, warpObj)
-  const fill = await getShapeFill(node, warpObj, source, groupHierarchy)
+  const fill = await getShapeFill(node, warpObj, source, {
+    groupHierarchy,
+    slideLayoutSpNode,
+    slideMasterSpNode,
+  })
 
   let shadow
   const outerShdwNode = getTextByPathList(node, ['p:spPr', 'a:effectLst', 'a:outerShdw'])
@@ -811,12 +841,15 @@ async function genShape(node, slideLayoutSpNode, slideMasterSpNode, name, type,
     const d = getCustomShapePath(custShapType, w, h)
     if (!isHasValidText) data.content = ''
 
-    return {
+    const customShapeData = {
       ...data,
       type: 'shape',
       shapType: 'custom',
       path: d,
     }
+    if (isStrokeOnlyCustomGeometry(custShapType)) customShapeData.strokeOnly = true
+
+    return customShapeData
   }
 
   let shapePath = ''

package/src/shape.js

@@ -223,6 +223,15 @@ export function getCustomShapePath(custShapType, w, h) {
   return d
 }
 
+export function isStrokeOnlyCustomGeometry(custShapType) {
+  const pathNodes = getTextByPathList(custShapType, ['a:pathLst', 'a:path'])
+  if (!pathNodes) return false
+
+  const pathNodeList = Array.isArray(pathNodes) ? pathNodes : [pathNodes]
+
+  return pathNodeList.length === 1 && getTextByPathList(pathNodeList[0], ['attrs', 'fill']) === 'none'
+}
+
 export function identifyShape(shapeData) {
   const pathLst = shapeData['a:pathLst']
   if (!pathLst || !pathLst['a:path']) return 'custom'

package/src/text.js

@@ -1,4 +1,4 @@
-import { getHorizontalAlign, getParagraphSpacing } from './paragraph'
+import { getHorizontalAlign, getParagraphSpacing, getParagraphIndent } from './paragraph'
 import { getTextByPathList } from './utils'
 
 import {
@@ -25,9 +25,9 @@ export function genTextBody(textBodyNode, spNode, slideLayoutSpNode, slideMaster
 
   let text = ''
 
-  const pFontStyle = getTextByPathList(spNode, ['p:style', 'a:fontRef'])
-  const slideMasterTextStyles = spNode && spNode['a:tcPr'] ? undefined : warpObj['slideMasterTextStyles']
-  const defaultTextStyle = spNode && spNode['a:tcPr'] ? warpObj['defaultTextStyle'] : undefined
+  const pFontStyle = getTextByPathList(spNode, ['p:style', 'a:fontRef'])
+  const slideMasterTextStyles = spNode && spNode['a:tcPr'] ? undefined : warpObj['slideMasterTextStyles']
+  const defaultTextStyle = spNode && spNode['a:tcPr'] ? warpObj['defaultTextStyle'] : undefined
 
   const pNode = textBodyNode['a:p']
   const pNodes = pNode.constructor === Array ? pNode : [pNode]
@@ -59,7 +59,10 @@ export function genTextBody(textBodyNode, spNode, slideLayoutSpNode, slideMaster
     }
 
     const align = getHorizontalAlign(pNode, spNode, type, slideLayoutSpNode, slideMasterSpNode, warpObj)
-    const spacing = getParagraphSpacing(pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, warpObj)
+    const spacing = getParagraphSpacing(pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, warpObj)
+    const indent = getParagraphIndent(pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, warpObj)
+    const listType = getListType(pNode)
+    const listLevel = getListLevel(pNode)
 
     let styleText = `text-align: ${align};`
     if (spacing) {
@@ -67,9 +70,10 @@ export function genTextBody(textBodyNode, spNode, slideLayoutSpNode, slideMaster
       if (spacing.spaceBefore) styleText += `margin-top: ${spacing.spaceBefore};`
       if (spacing.spaceAfter) styleText += `margin-bottom: ${spacing.spaceAfter};`
     }
-
-    const listType = getListType(pNode)
-    const listLevel = getListLevel(pNode)
+    if (indent) {
+      if (!listType && indent.marginLeft) styleText += `margin-left: ${indent.marginLeft};`
+      if (!listType && indent.textIndent) styleText += `text-indent: ${indent.textIndent};`
+    }
 
     if (listType) {
       while (listTypes.length > listLevel + 1) {
@@ -86,7 +90,7 @@ export function genTextBody(textBodyNode, spNode, slideLayoutSpNode, slideMaster
         text += `<${listType}>`
         listTypes[listLevel] = listType
       }
-      text += `<li style="${styleText}">`
+      text += `<li><p style="${styleText}">`
     }
     else {
       while (listTypes.length > 0) {
@@ -97,14 +101,14 @@ export function genTextBody(textBodyNode, spNode, slideLayoutSpNode, slideMaster
     }
     
     if (!rNode) {
-      text += genSpanElement(pNode, spNode, textBodyNode, pFontStyle, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, defaultTextStyle, warpObj)
+      text += genSpanElement(pNode, spNode, textBodyNode, pFontStyle, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, defaultTextStyle, warpObj)
     } 
     else {
       let prevStyleInfo = null
       let accumulatedText = ''
 
       for (const rNodeItem of rNode) {
-        const styleInfo = getSpanStyleInfo(rNodeItem, pNode, textBodyNode, pFontStyle, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, defaultTextStyle, warpObj)
+        const styleInfo = getSpanStyleInfo(rNodeItem, pNode, textBodyNode, pFontStyle, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, defaultTextStyle, warpObj)
 
         if (!prevStyleInfo || prevStyleInfo.styleText !== styleInfo.styleText || prevStyleInfo.hasLink !== styleInfo.hasLink || styleInfo.hasLink) {
           if (accumulatedText) {
@@ -132,7 +136,7 @@ export function genTextBody(textBodyNode, spNode, slideLayoutSpNode, slideMaster
       }
     }
 
-    if (listType) text += '</li>'
+    if (listType) text += '</p></li>'
     else text += '</p>'
   }
   while (listTypes.length > 0) {
@@ -161,8 +165,8 @@ export function getListLevel(node) {
   return 0
 }
 
-export function genSpanElement(node, pNode, textBodyNode, pFontStyle, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, defaultTextStyle, warpObj) {
-  const { styleText, text, hasLink, linkURL } = getSpanStyleInfo(node, pNode, textBodyNode, pFontStyle, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, defaultTextStyle, warpObj)
+export function genSpanElement(node, pNode, textBodyNode, pFontStyle, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, defaultTextStyle, warpObj) {
+  const { styleText, text, hasLink, linkURL } = getSpanStyleInfo(node, pNode, textBodyNode, pFontStyle, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, defaultTextStyle, warpObj)
   const processedText = text.replace(/\t/g, '&nbsp;&nbsp;&nbsp;&nbsp;').replace(/\s/g, '&nbsp;')
 
   if (hasLink) {
@@ -171,7 +175,7 @@ export function genSpanElement(node, pNode, textBodyNode, pFontStyle, slideLayou
   return `<span style="${styleText}">${processedText}</span>`
 }
 
-export function getSpanStyleInfo(node, pNode, textBodyNode, pFontStyle, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, defaultTextStyle, warpObj) {
+export function getSpanStyleInfo(node, pNode, textBodyNode, pFontStyle, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, defaultTextStyle, warpObj) {
   let lvl = 1
   const pPrNode = pNode['a:pPr']
   const lvlNode = getTextByPathList(pPrNode, ['attrs', 'lvl'])
@@ -183,7 +187,7 @@ export function getSpanStyleInfo(node, pNode, textBodyNode, pFontStyle, slideLay
 
   let styleText = ''
   const fontColor = getFontColor(node, pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, lvl, pFontStyle, warpObj)
-  const fontSize = getFontSize(node, pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, lvl, defaultTextStyle)
+  const fontSize = getFontSize(node, pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, lvl, defaultTextStyle)
   const fontType = getFontType(node, pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, lvl, warpObj)
   const fontBold = getFontBold(node, pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, lvl)
   const fontItalic = getFontItalic(node, pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, lvl)
@@ -221,4 +225,4 @@ export function getSpanStyleInfo(node, pNode, textBodyNode, pFontStyle, slideLay
     hasLink,
     linkURL: hasLink ? warpObj['slideResObj'][linkID]['target'] : null
   }
-}
+}