pptxtojson 2.1.0 → 2.2.0

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "pptxtojson",
-	"version": "2.1.0",
+	"version": "2.2.0",
 	"description": "A javascript tool for parsing .pptx file",
 	"type": "module",
 	"main": "./dist/index.umd.js",

package/src/diagram.js

@@ -67,10 +67,12 @@ export async function getDiagramNodeContext(node, warpObj) {
       let relationshipArray = digramResContent['Relationships']['Relationship']
       if (relationshipArray && relationshipArray.constructor !== Array) relationshipArray = [relationshipArray]
       if (relationshipArray) {
-        for (const relationshipArrayItem of relationshipArray) {
-          let relTarget = relationshipArrayItem['attrs']['Target']
-          if (relTarget.indexOf('../') !== -1) relTarget = relTarget.replace('../', 'ppt/')
-          else relTarget = drawingTarget.replace(drawingName, '') + relTarget
+        for (const relationshipArrayItem of relationshipArray) {
+          let relTarget = relationshipArrayItem['attrs']['Target']
+          relTarget = relTarget.replace(/\\/g, '/')
+          if (relTarget.indexOf('/ppt/') === 0) relTarget = relTarget.substr(1)
+          else if (relTarget.indexOf('../') !== -1) relTarget = relTarget.replace('../', 'ppt/')
+          else relTarget = drawingTarget.replace(drawingName, '') + relTarget
 
           diagramResObj[relationshipArrayItem['attrs']['Id']] = {
             type: relationshipArrayItem['attrs']['Type'].replace('http://schemas.openxmlformats.org/officeDocument/2006/relationships/', ''),
@@ -129,4 +131,4 @@ export function getSmartArtTextData(dataContent) {
   }
 
   return result
-}
+}

package/src/paragraph.js

@@ -177,12 +177,16 @@ function getParagraphStyleNodes(pNode, textBodyNode, slideLayoutSpNode, slideMas
   return styleNodes
 }
 
-function getLineSpacingValue(spacingNode) {
-  const spcPct = getTextByPathList(spacingNode, ['a:spcPct', 'attrs', 'val'])
-  const spcPts = getTextByPathList(spacingNode, ['a:spcPts', 'attrs', 'val'])
-
-  if (spcPct) return parseInt(spcPct) / 1000 / 100
-  if (spcPts) return parseInt(spcPts) / 100 + 'pt'
+function getLineSpacingValue(spacingNode, singleLineSpacingFactor = 1) {
+  const spcPct = getTextByPathList(spacingNode, ['a:spcPct', 'attrs', 'val'])
+  const spcPts = getTextByPathList(spacingNode, ['a:spcPts', 'attrs', 'val'])
+
+  if (spcPct !== undefined) {
+    const value = String(spcPct).trim()
+    const percentage = value.endsWith('%') ? Number.parseFloat(value) / 100 : Number.parseInt(value, 10) / 100000
+    if (Number.isFinite(percentage)) return percentage * singleLineSpacingFactor
+  }
+  if (spcPts) return parseInt(spcPts) / 100 + 'pt'
 
   return undefined
 }
@@ -205,16 +209,17 @@ function getParagraphIndentValue(styleNode, attrName) {
   return undefined
 }
 
-export function getParagraphSpacing(pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, warpObj) {
-  const styleNodes = getParagraphStyleNodes(pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, warpObj)
-  if (!styleNodes) return null
-
-  const spacing = {}
-
-  for (const styleNode of styleNodes) {
-    if (spacing.lineSpacing === undefined) {
-      const lineSpacing = getLineSpacingValue(styleNode['a:lnSpc'])
-      if (lineSpacing !== undefined) spacing.lineSpacing = lineSpacing
+export function getParagraphSpacing(pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, warpObj) {
+  const styleNodes = getParagraphStyleNodes(pNode, textBodyNode, slideLayoutSpNode, slideMasterSpNode, type, slideMasterTextStyles, warpObj)
+  if (!styleNodes) return null
+
+  const spacing = {}
+  const singleLineSpacingFactor = getTextByPathList(warpObj, ['options', 'singleLineSpacingFactor'])
+
+  for (const styleNode of styleNodes) {
+    if (spacing.lineSpacing === undefined) {
+      const lineSpacing = getLineSpacingValue(styleNode['a:lnSpc'], singleLineSpacingFactor)
+      if (lineSpacing !== undefined) spacing.lineSpacing = lineSpacing
     }
 
     if (spacing.spaceBefore === undefined) {

package/src/pptxtojson.js

@@ -25,10 +25,11 @@ export async function parse(file, options = {}) {
   const xmlCache = {}
   const parseOptions = {
     ...options,
-    imageMode: options.imageMode || 'base64',
-    videoMode: options.videoMode || 'none',
-    audioMode: options.audioMode || 'none',
-  }
+    imageMode: options.imageMode || 'base64',
+    videoMode: options.videoMode || 'none',
+    audioMode: options.audioMode || 'none',
+    singleLineSpacingFactor: Number.isFinite(options.singleLineSpacingFactor) && options.singleLineSpacingFactor > 0 ? options.singleLineSpacingFactor : 1,
+  }
   
   const zip = await JSZip.loadAsync(file)
 
@@ -129,7 +130,12 @@ async function getTheme(zip) {
     themeURI = relationshipArray['attrs']['Target']
   }
 
-  const themeContent = await readXmlFile(zip, 'ppt/' + themeURI)
+  let themeContent = null
+  if (themeURI) {
+    themeURI = themeURI.replace(/\\/g, '/')
+    const themeFilename = themeURI.indexOf('/ppt/') === 0 ? themeURI.substr(1) : 'ppt/' + themeURI
+    themeContent = await readXmlFile(zip, themeFilename)
+  }
 
   const themeColors = []
   const clrScheme = getTextByPathList(themeContent, ['a:theme', 'a:themeElements', 'a:clrScheme'])
@@ -173,7 +179,9 @@ async function processSingleSlide(zip, sldFileName, themeContent, defaultTextSty
     let relTarget = relationshipArrayItem['attrs']['Target']
     const isExternal = relationshipArrayItem['attrs']['TargetMode'] === 'External'
     if (!isExternal) {
-      if (relTarget.indexOf('../') !== -1) relTarget = relTarget.replace('../', 'ppt/')
+      relTarget = relTarget.replace(/\\/g, '/')
+      if (relTarget.indexOf('/ppt/') === 0) relTarget = relTarget.substr(1)
+      else if (relTarget.indexOf('../') !== -1) relTarget = relTarget.replace('../', 'ppt/')
       else relTarget = 'ppt/slides/' + relTarget
     }
 
@@ -226,7 +234,9 @@ async function processSingleSlide(zip, sldFileName, themeContent, defaultTextSty
   for (const relationshipArrayItem of relationshipArray) {
     const relType = relationshipArrayItem['attrs']['Type'].replace('http://schemas.openxmlformats.org/officeDocument/2006/relationships/', '')
     let relTarget = relationshipArrayItem['attrs']['Target']
-    if (relTarget.indexOf('../') !== -1) relTarget = relTarget.replace('../', 'ppt/')
+    relTarget = relTarget.replace(/\\/g, '/')
+    if (relTarget.indexOf('/ppt/') === 0) relTarget = relTarget.substr(1)
+    else if (relTarget.indexOf('../') !== -1) relTarget = relTarget.replace('../', 'ppt/')
     else relTarget = 'ppt/slideLayouts/' + relTarget
 
     switch (relationshipArrayItem['attrs']['Type']) {
@@ -252,7 +262,9 @@ async function processSingleSlide(zip, sldFileName, themeContent, defaultTextSty
   for (const relationshipArrayItem of relationshipArray) {
     const relType = relationshipArrayItem['attrs']['Type'].replace('http://schemas.openxmlformats.org/officeDocument/2006/relationships/', '')
     let relTarget = relationshipArrayItem['attrs']['Target']
-    if (relTarget.indexOf('../') !== -1) relTarget = relTarget.replace('../', 'ppt/')
+    relTarget = relTarget.replace(/\\/g, '/')
+    if (relTarget.indexOf('/ppt/') === 0) relTarget = relTarget.substr(1)
+    else if (relTarget.indexOf('../') !== -1) relTarget = relTarget.replace('../', 'ppt/')
     else relTarget = 'ppt/slideMasters/' + relTarget
 
     switch (relationshipArrayItem['attrs']['Type']) {
@@ -267,10 +279,10 @@ async function processSingleSlide(zip, sldFileName, themeContent, defaultTextSty
     }
   }
 
-  let currentThemeContent = themeContent
-  if (themeFilename) {
-    currentThemeContent = currentThemeContent || await readXmlFileCached(zip, themeFilename, xmlCache)
-    const themeName = themeFilename.split('/').pop()
+  let currentThemeContent = themeContent
+  if (themeFilename) {
+    currentThemeContent = await readXmlFileCached(zip, themeFilename, xmlCache) || currentThemeContent
+    const themeName = themeFilename.split('/').pop()
     const themeResFileName = themeFilename.replace(themeName, '_rels/' + themeName) + '.rels'
     const themeResContent = await readXmlFile(zip, themeResFileName)
     if (themeResContent) {
@@ -278,9 +290,14 @@ async function processSingleSlide(zip, sldFileName, themeContent, defaultTextSty
       if (relationshipArray) {
         if (relationshipArray.constructor !== Array) relationshipArray = [relationshipArray]
         for (const relationshipArrayItem of relationshipArray) {
+          let relTarget = relationshipArrayItem['attrs']['Target']
+          relTarget = relTarget.replace(/\\/g, '/')
+          if (relTarget.indexOf('/ppt/') === 0) relTarget = relTarget.substr(1)
+          else relTarget = relTarget.replace('../', 'ppt/')
+
           themeResObj[relationshipArrayItem['attrs']['Id']] = {
             'type': relationshipArrayItem['attrs']['Type'].replace('http://schemas.openxmlformats.org/officeDocument/2006/relationships/', ''),
-            'target': relationshipArrayItem['attrs']['Target'].replace('../', 'ppt/')
+            'target': relTarget
           }
         }
       }
@@ -558,6 +575,19 @@ async function processNodesInSlide(nodeKey, nodeValue, warpObj, source, groupHie
     default:
   }
 
+  if (json && typeof json === 'object' && !json.id) {
+    const id = getTextByPathList(nodeValue, ['p:nvSpPr', 'p:cNvPr', 'attrs', 'id']) ||
+      getTextByPathList(nodeValue, ['p:nvPicPr', 'p:cNvPr', 'attrs', 'id']) ||
+      getTextByPathList(nodeValue, ['p:nvCxnSpPr', 'p:cNvPr', 'attrs', 'id']) ||
+      getTextByPathList(nodeValue, ['p:nvGrpSpPr', 'p:cNvPr', 'attrs', 'id']) ||
+      getTextByPathList(nodeValue, ['p:nvGraphicFramePr', 'p:cNvPr', 'attrs', 'id']) ||
+      getTextByPathList(nodeValue, ['mc:Choice', 'p:sp', 'p:nvSpPr', 'p:cNvPr', 'attrs', 'id']) ||
+      getTextByPathList(nodeValue, ['mc:Fallback', 'p:sp', 'p:nvSpPr', 'p:cNvPr', 'attrs', 'id']) ||
+      getTextByPathList(nodeValue, ['mc:Fallback', 'p:nvGrpSpPr', 'p:cNvPr', 'attrs', 'id'])
+
+    json.id = id || ''
+  }
+
   return json
 }
 
@@ -795,7 +825,7 @@ async function genShape(node, slideLayoutSpNode, slideMasterSpNode, name, type,
   let content = ''
   if (node['p:txBody']) content = genTextBody(node['p:txBody'], node, slideLayoutSpNode, slideMasterSpNode, type, warpObj)
 
-  const { borderColor, borderWidth, borderType, strokeDasharray, headEnd, tailEnd } = getBorder(node, type, warpObj)
+  const { borderColor, borderWidth, borderType, strokeDasharray, headEnd, tailEnd } = getBorder(node, type, warpObj)
   const fill = await getShapeFill(node, warpObj, source, {
     groupHierarchy,
     slideLayoutSpNode,
@@ -808,6 +838,7 @@ async function genShape(node, slideLayoutSpNode, slideMasterSpNode, name, type,
 
   const vAlign = getVerticalAlign(node, slideLayoutSpNode, slideMasterSpNode, type)
   const isVertical = getTextByPathList(node, ['p:txBody', 'a:bodyPr', 'attrs', 'vert']) === 'eaVert'
+  const wrap = getTextByPathList(node, ['p:txBody', 'a:bodyPr', 'attrs', 'wrap']) !== 'none'
   const autoFit = getTextAutoFit(node, slideLayoutSpNode, slideMasterSpNode)
   const textInset = getTextInsets(node, slideLayoutSpNode, slideMasterSpNode)
 
@@ -826,16 +857,17 @@ async function genShape(node, slideLayoutSpNode, slideMasterSpNode, name, type,
     isFlipH,
     rotate,
     vAlign,
+    wrap,
     name,
     order,
   }
 
   if (shadow) data.shadow = shadow
-  if (autoFit) data.autoFit = autoFit
-  if (link) data.link = link
-  if (textInset) data.textInset = textInset
-  if (headEnd) data.headEnd = headEnd
-  if (tailEnd) data.tailEnd = tailEnd
+  if (autoFit) data.autoFit = autoFit
+  if (link) data.link = link
+  if (textInset) data.textInset = textInset
+  if (headEnd) data.headEnd = headEnd
+  if (tailEnd) data.tailEnd = tailEnd
 
   const isHasValidText = data.content && hasValidText(data.content)