officeparser 7.4.0 → 7.5.1

package/dist/parsers/PowerPointParser.js

@@ -29,6 +29,7 @@ const astUtils_js_1 = require("../utils/astUtils.js");
 const chartUtils_js_1 = require("../utils/chartUtils.js");
 const errorUtils_js_1 = require("../utils/errorUtils.js");
 const imageUtils_js_1 = require("../utils/imageUtils.js");
+const mathUtils_js_1 = require("../utils/mathUtils.js");
 const ocrUtils_js_1 = require("../utils/ocrUtils.js");
 const xmlUtils_js_1 = require("../utils/xmlUtils.js");
 const zipUtils_js_1 = require("../utils/zipUtils.js");
@@ -56,6 +57,7 @@ const parsePowerPoint = async (buffer, config) => {
     const commentsFileRegex = /ppt\/comments\/comment\d+\.xml/;
     const commentAuthorsRegex = /ppt\/commentAuthors\.xml/;
     const slideMastersRegex = /ppt\/slideMasters\/slideMaster\d+\.xml/;
+    const presentationFileRegex = /ppt\/presentation\.xml/;
     const files = await (0, zipUtils_js_1.extractFiles)(buffer, x => !!x.match(config.ignoreNotes ? slidesRegex : allFilesRegex) ||
         !!x.match(corePropsFileRegex) ||
         !!x.match(customPropsFileRegex) ||
@@ -63,7 +65,14 @@ const parsePowerPoint = async (buffer, config) => {
         !!x.match(slideRelsRegex) ||
         (!config.ignoreComments && (!!x.match(commentsFileRegex) || !!x.match(commentAuthorsRegex))) ||
         (!config.ignoreSlideMasters && !!x.match(slideMastersRegex)) ||
-        (!!config.extractAttachments && (!!x.match(mediaFileRegex) || !!x.match(chartFileRegex))), config.decompressionLimits);
+        !!x.match(presentationFileRegex) ||
+        (!!config.extractAttachments && (!!x.match(mediaFileRegex) || !!x.match(chartFileRegex))), config.decompressionLimits, config);
+    // ppt/presentation.xml is the part that makes an archive a presentation, and unlike the
+    // slides it is always present: PowerPoint can save a deck with no slides at all, so an
+    // empty ppt/slides/ is a warning rather than a failure.
+    (0, zipUtils_js_1.findRequiredPart)(files, path => !!path.match(presentationFileRegex), config, { fileType: 'pptx', part: 'ppt/presentation.xml' });
+    if (!files.some(file => !!file.path.match(slidesRegex)))
+        (0, errorUtils_js_1.logWarning)(types_js_1.OfficeWarningType.NO_SLIDES_FOUND, config);
     // Extract metadata
     const corePropsFile = files.find(f => f.path.match(corePropsFileRegex));
     const metadata = corePropsFile ? (0, xmlUtils_js_1.parseOfficeMetadata)(corePropsFile.content.toString()) : {};
@@ -88,7 +97,6 @@ const parsePowerPoint = async (buffer, config) => {
         return aNum - bNum;
     });
     const content = [];
-    const rawContents = [];
     const slideRelsMap = {};
     const authorMap = {};
     if (!config.ignoreComments) {
@@ -576,6 +584,30 @@ const parsePowerPoint = async (buffer, config) => {
                             activeNode.children?.push({ type: 'text', text: "\n" });
                         }
                     }
+                    else {
+                        // Equations. This loop dispatches on `a:r`/`a:fld`, so an `m:oMath` -
+                        // which is a sibling of the runs, not one of them - was never visited at
+                        // all and the formula vanished from the slide without a warning.
+                        //
+                        // PowerPoint writes the equation either directly in the paragraph or
+                        // wrapped in `mc:AlternateContent`/`a14:m` for pre-2010 readers, so take
+                        // the element itself when it is the equation and search inside it
+                        // otherwise. `getElementsByTagName` returns document order, which is the
+                        // order the equations are read in.
+                        const isMath = tag === "m:oMath" || tag === "m:oMathPara";
+                        const equations = isMath ? [element] : (0, xmlUtils_js_1.getElementsByTagName)(element, "m:oMath");
+                        for (const equation of equations) {
+                            const latex = (0, mathUtils_js_1.ommlToLatex)(equation);
+                            if ((0, mathUtils_js_1.isEmptyMath)(latex))
+                                continue;
+                            activeNode.text += latex;
+                            activeNode.children?.push({
+                                type: 'code',
+                                text: latex,
+                                metadata: { math: tag === "m:oMathPara" ? 'block' : 'inline' }
+                            });
+                        }
+                    }
                 }
             }
         }
@@ -623,12 +655,8 @@ const parsePowerPoint = async (buffer, config) => {
             }
             // Case 4: Grouped shape (recursive!)
             else if (tag === "p:grpSp") {
-                // Extract the nested <p:spTree> inside the group
-                const nestedTree = (0, xmlUtils_js_1.getFirstElementByTagName)(element, "p:spTree");
-                // Recurse into the nested tree
-                if (nestedTree) {
-                    nodes.push(...traverseSpTree(nestedTree, slideNumber, xmlContentString));
-                }
+                // Recurse into the group element itself which holds the child shapes
+                nodes.push(...traverseSpTree(element, slideNumber, xmlContentString));
             }
         }
         return nodes;
@@ -731,15 +759,24 @@ const parsePowerPoint = async (buffer, config) => {
             continue;
         if (file.path.match(slideRelsRegex))
             continue;
+        // All three document-property parts are extracted for metadata and were read earlier;
+        // only the core one was skipped here, leaving the other two to be parsed again as if
+        // they might be slides.
         if (file.path.match(corePropsFileRegex))
             continue;
+        if (file.path.match(appPropsFileRegex))
+            continue;
+        if (file.path.match(customPropsFileRegex))
+            continue;
         if (file.path.includes("comment"))
             continue;
+        // This loop treats every remaining file as a slide or note, so the presentation part
+        // has to be skipped explicitly: it carries no slide number and would otherwise be
+        // added to the deck as an empty slide.
+        if (file.path.match(presentationFileRegex))
+            continue;
         const xmlContentString = file.content.toString();
         const xml = (0, xmlUtils_js_1.parseXmlString)(xmlContentString, { locator: config.includeRawContent });
-        if (config.includeRawContent) {
-            rawContents.push(xmlContentString);
-        }
         const slideMatch = file.path.match(slideNumberRegex);
         const slideNumber = slideMatch ? parseInt(slideMatch[1]) : 0;
         const masterMatch = file.path.match(/slideMaster(\d+)\.xml/);

package/dist/parsers/WordParser.js

@@ -66,6 +66,7 @@ const types_js_1 = require("../types.js");
 const astUtils_js_1 = require("../utils/astUtils.js");
 const errorUtils_js_1 = require("../utils/errorUtils.js");
 const imageUtils_js_1 = require("../utils/imageUtils.js");
+const mathUtils_js_1 = require("../utils/mathUtils.js");
 const ocrUtils_js_1 = require("../utils/ocrUtils.js");
 const xmlUtils_js_1 = require("../utils/xmlUtils.js");
 const zipUtils_js_1 = require("../utils/zipUtils.js");
@@ -94,8 +95,12 @@ const parseWord = async (buffer, config) => {
     const footnotesFileRegex = /word\/footnotes[\d+]?.xml/;
     const endnotesFileRegex = /word\/endnotes[\d+]?.xml/;
     const commentsFileRegex = /word\/comments[\d+]?.xml/;
-    const headerFileRegex = /word\/header[\d+]?.xml/;
-    const footerFileRegex = /word\/footer[\d+]?.xml/;
+    // Headers and footers are the only parts a document can have many of: Word writes up to
+    // three per section (default, first page, even pages), so a handful of sections is enough
+    // to reach header10.xml. The single-character form the other parts use stops matching at
+    // nine, which would drop those later files as silently as not extracting them at all.
+    const headerFileRegex = /word\/header\d*\.xml/;
+    const footerFileRegex = /word\/footer\d*\.xml/;
     const numberingFileRegex = /word\/numbering[\d+]?.xml/;
     const mediaFileRegex = /(word\/)?media\/.*/;
     const corePropsFileRegex = /docProps\/core[\d+]?.xml/;
@@ -243,7 +248,12 @@ const parseWord = async (buffer, config) => {
         !!x.match(appPropsFileRegex) ||
         !!x.match(relsFileRegex) ||
         !!x.match(stylesFileRegex) ||
-        (!!config.extractAttachments && !!x.match(mediaFileRegex)), config.decompressionLimits);
+        (!config.ignoreComments && !!x.match(commentsFileRegex)) ||
+        (!config.ignoreHeadersAndFooters && (!!x.match(headerFileRegex) || !!x.match(footerFileRegex))) ||
+        (!!config.extractAttachments && !!x.match(mediaFileRegex)), config.decompressionLimits, config);
+    // A DOCX without its main document part is not a DOCX. Checked with the same regex the
+    // parse loop below uses to recognize it, so the two cannot fall out of step.
+    (0, zipUtils_js_1.findRequiredPart)(files, path => !!path.match(documentFileRegex), config, { fileType: 'docx', part: 'word/document.xml' });
     // Extract metadata
     const corePropsFile = files.find(f => f.path.match(corePropsFileRegex));
     const metadata = corePropsFile ? (0, xmlUtils_js_1.parseOfficeMetadata)(corePropsFile.content.toString()) : {};
@@ -404,7 +414,6 @@ const parseWord = async (buffer, config) => {
         }
     }
     const content = [];
-    const rawContents = [];
     const numberingState = {};
     const listCounters = {}; // Track item index per listId/level
     // Helper to parse a paragraph node
@@ -751,6 +760,26 @@ const parseWord = async (buffer, config) => {
                     }
                 }
             }
+            else if ((0, xmlUtils_js_1.isElement)(node) && (node.nodeName === 'm:oMath' || node.nodeName === 'oMath'
+                || node.nodeName === 'm:oMathPara' || node.nodeName === 'oMathPara')) {
+                // Equations. Without this branch they reach the generic fallback below, which
+                // recurses into every child and concatenates the `m:t` runs with no separators -
+                // so `<m:num>1</m:num><m:den>2</m:den>` came out as "12". That is worse than
+                // dropping the formula: the result still reads as a number, so nothing downstream
+                // can tell it is wrong.
+                //
+                // `m:oMathPara` is a display equation on its own line; a bare `m:oMath` is inline.
+                const isBlock = node.nodeName === 'm:oMathPara' || node.nodeName === 'oMathPara';
+                const latex = (0, mathUtils_js_1.ommlToLatex)(node);
+                if (!(0, mathUtils_js_1.isEmptyMath)(latex)) {
+                    text += latex;
+                    children.push({
+                        type: 'code',
+                        text: latex,
+                        metadata: { math: isBlock ? 'block' : 'inline' }
+                    });
+                }
+            }
             else if (node.childNodes.length > 0) {
                 // Generic fallback for unknown elements that might contain content
                 for (const child of Array.from(node.childNodes))
@@ -1071,9 +1100,6 @@ const parseWord = async (buffer, config) => {
         if (file.path.match(footerFileRegex))
             continue;
         const documentContent = file.content.toString();
-        if (config.includeRawContent) {
-            rawContents.push(documentContent);
-        }
         const doc = (0, xmlUtils_js_1.parseXmlString)(documentContent, { locator: config.includeRawContent });
         const body = (0, xmlUtils_js_1.getFirstElementByTagName)(doc, "w:body");
         if (body) {