@a3s-lab/office 0.12.0 → 0.13.0

package/dist/0~work-document-import-worker-client.js

@@ -0,0 +1,237 @@
+import { createLargeSimpleDocxParagraph, createLargeSimpleDocxParseResult, createLargeSimpleDocxTable } from "./0~work-docx-large-document-import.js";
+const DOCUMENT_IMPORT_WORKER_TIMEOUT_MS = 120000;
+function parseLargeSimpleDocxInWorker(xmlBytes, options = {}, signal) {
+    if ("u" < typeof Worker) return Promise.resolve(null);
+    if (signal?.aborted) return Promise.reject(documentImportAbortError());
+    let worker;
+    try {
+        worker = new Worker(new URL('./work-document-import.worker.js', import.meta.url), {
+            name: 'a3s-office-document-import'
+        });
+    } catch  {
+        return Promise.resolve(null);
+    }
+    const startedAt = documentImportNow();
+    return new Promise((resolve, reject)=>{
+        let settled = false;
+        let timeout;
+        const blocks = [];
+        const html = [];
+        let paragraphBatchCount = 0;
+        let rowBatchCount = 0;
+        let streamedAssemblyMs = 0;
+        let tableRows = null;
+        let tableHtml = null;
+        const finish = (result, error)=>{
+            if (settled) return;
+            settled = true;
+            if (void 0 !== timeout) clearTimeout(timeout);
+            signal?.removeEventListener('abort', handleAbort);
+            worker.removeEventListener('error', handleFailure);
+            worker.removeEventListener('message', handleMessage);
+            worker.removeEventListener('messageerror', handleFailure);
+            worker.terminate();
+            if (error) reject(error);
+            else resolve(result);
+        };
+        const handleAbort = ()=>finish(null, documentImportAbortError());
+        const handleFailure = ()=>finish(null);
+        const handleMessage = (event)=>{
+            const response = event.data;
+            if (!response || 'object' != typeof response) return void finish(null);
+            if ('failure' === response.kind) return void finish(null);
+            if ('ineligible' === response.kind) {
+                recordDocumentImportMeasure('a3s-office.document.large-simple-worker-total', startedAt, documentImportNow(), {
+                    accepted: false,
+                    paragraphBatchCount,
+                    rowBatchCount
+                });
+                finish({
+                    status: 'ineligible'
+                });
+                return;
+            }
+            if ('paragraphs' === response.kind) {
+                const batchStartedAt = documentImportNow();
+                if (tableRows || tableHtml || !textBatchIsValid(response.texts)) return void finish(null);
+                paragraphBatchCount += 1;
+                for (const text of response.texts){
+                    const paragraph = createLargeSimpleDocxParagraph(text);
+                    blocks.push(paragraph.node);
+                    html.push(paragraph.html);
+                }
+                streamedAssemblyMs += documentImportNow() - batchStartedAt;
+                return;
+            }
+            if ('table-start' === response.kind) {
+                if (tableRows || tableHtml) return void finish(null);
+                tableRows = [];
+                tableHtml = [];
+                return;
+            }
+            if ('table-rows' === response.kind) {
+                const batchStartedAt = documentImportNow();
+                if (!tableRows || !tableHtml || !compactTableBatchIsValid(response)) return void finish(null);
+                rowBatchCount += 1;
+                let cellIndex = 0;
+                let textIndex = 0;
+                for (const cellCount of response.rowCellCounts){
+                    const row = createCompactTableRow(response, cellCount, cellIndex, textIndex);
+                    tableRows.push(row.node);
+                    tableHtml.push(row.html);
+                    cellIndex = row.cellIndex;
+                    textIndex = row.textIndex;
+                }
+                streamedAssemblyMs += documentImportNow() - batchStartedAt;
+                return;
+            }
+            if ('table-end' === response.kind) {
+                if (!tableRows?.length || !tableHtml?.length) return void finish(null);
+                const table = createLargeSimpleDocxTable(tableRows, tableHtml);
+                blocks.push(table.node);
+                html.push(table.html);
+                tableRows = null;
+                tableHtml = null;
+                return;
+            }
+            if ('success' !== response.kind || tableRows || tableHtml || !blocks.length || !streamResultIsValid(response.streamed)) return void finish(null);
+            const finalizationStartedAt = documentImportNow();
+            let result;
+            try {
+                result = createLargeSimpleDocxParseResult(response.streamed, blocks, html, options);
+            } catch  {
+                finish(null);
+                return;
+            }
+            const finishedAt = documentImportNow();
+            recordDocumentImportDuration('a3s-office.document.large-simple-worker-parse', response.timings.parseMs, {
+                contentMs: response.timings.contentMs,
+                eligibilityMs: response.timings.eligibilityMs,
+                envelopeMs: response.timings.envelopeMs,
+                xmlMs: response.timings.xmlMs
+            });
+            recordDocumentImportMeasure('a3s-office.document.large-simple-worker-finalize', finalizationStartedAt, finishedAt, {
+                paragraphBatchCount,
+                rowBatchCount,
+                streamedAssemblyMs
+            });
+            recordDocumentImportMeasure('a3s-office.document.large-simple-worker-total', startedAt, finishedAt, {
+                accepted: true,
+                paragraphBatchCount,
+                rowBatchCount,
+                streamedAssemblyMs
+            });
+            finish({
+                result,
+                status: 'accepted'
+            });
+        };
+        timeout = setTimeout(handleFailure, DOCUMENT_IMPORT_WORKER_TIMEOUT_MS);
+        signal?.addEventListener('abort', handleAbort, {
+            once: true
+        });
+        worker.addEventListener('error', handleFailure, {
+            once: true
+        });
+        worker.addEventListener('message', handleMessage);
+        worker.addEventListener('messageerror', handleFailure, {
+            once: true
+        });
+        if (signal?.aborted) return void handleAbort();
+        try {
+            worker.postMessage({
+                kind: 'parse',
+                options,
+                xmlBytes
+            }, {
+                transfer: [
+                    xmlBytes
+                ]
+            });
+        } catch  {
+            handleFailure();
+        }
+    });
+}
+function textBatchIsValid(texts) {
+    return Array.isArray(texts) && texts.length > 0 && texts.every((value)=>'string' == typeof value);
+}
+function compactTableBatchIsValid(response) {
+    if (!(response.rowCellCounts instanceof Uint32Array) || !response.rowCellCounts.length || !(response.cellParagraphCounts instanceof Uint32Array) || !textBatchIsValid(response.texts)) return false;
+    let cellCount = 0;
+    for (const count of response.rowCellCounts){
+        if (!count) return false;
+        cellCount += count;
+    }
+    if (cellCount !== response.cellParagraphCounts.length) return false;
+    let paragraphCount = 0;
+    for (const count of response.cellParagraphCounts){
+        if (!count) return false;
+        paragraphCount += count;
+    }
+    return paragraphCount === response.texts.length;
+}
+function createCompactTableRow(response, cellCount, initialCellIndex, initialTextIndex) {
+    const cellNodes = [];
+    const cellHtml = [];
+    let cellIndex = initialCellIndex;
+    let textIndex = initialTextIndex;
+    for(let cell = 0; cell < cellCount; cell += 1){
+        const paragraphCount = response.cellParagraphCounts[cellIndex] ?? 0;
+        const paragraphNodes = [];
+        let paragraphsHtml = '';
+        for(let paragraph = 0; paragraph < paragraphCount; paragraph += 1){
+            const value = createLargeSimpleDocxParagraph(response.texts[textIndex] ?? '');
+            paragraphNodes.push(value.node);
+            paragraphsHtml += value.html;
+            textIndex += 1;
+        }
+        cellNodes.push({
+            type: 'tableCell',
+            content: paragraphNodes
+        });
+        cellHtml.push(`<td>${paragraphsHtml}</td>`);
+        cellIndex += 1;
+    }
+    return {
+        cellIndex,
+        html: `<tr>${cellHtml.join('')}</tr>`,
+        node: {
+            type: 'tableRow',
+            content: cellNodes
+        },
+        textIndex
+    };
+}
+function streamResultIsValid(value) {
+    return Boolean(value?.layout) && Number.isSafeInteger(value.logicalBlockCount) && value.logicalBlockCount > 0 && Number.isSafeInteger(value.paragraphCount) && value.paragraphCount > 0 && Number.isSafeInteger(value.tableRowCount) && value.tableRowCount >= 0;
+}
+function documentImportNow() {
+    return globalThis.performance?.now?.() ?? Date.now();
+}
+function recordDocumentImportMeasure(name, start, end, detail) {
+    try {
+        globalThis.performance?.measure(name, {
+            detail,
+            end,
+            start
+        });
+    } catch  {}
+}
+function recordDocumentImportDuration(name, duration, detail) {
+    if (!Number.isFinite(duration) || duration < 0) return;
+    try {
+        globalThis.performance?.measure(name, {
+            detail,
+            duration,
+            start: 0
+        });
+    } catch  {}
+}
+function documentImportAbortError() {
+    if ("u" > typeof DOMException) return new DOMException('Document import was cancelled.', 'AbortError');
+    const error = new Error('Document import was cancelled.');
+    error.name = 'AbortError';
+    return error;
+}
+export { parseLargeSimpleDocxInWorker };

package/dist/0~work-docx-export.js

@@ -1,5 +1,5 @@
 import jszip from "jszip";
-import { normalizeDocumentBookmarkName, applyDocumentImageIdentityToElement, DOCUMENT_PAGE_BORDER_EDGES, createDocumentImageIdentityRegistry, normalizeDocumentColumns, normalizeDocumentTableVerticalAlign, uniqueDocumentParagraphIdentity, documentTableColumnPercentagesFromElement, documentTableCellMarginOverridesFromElement, applyDocumentTableRowIdentityToElement, normalizeDocumentBookmarksHtml, documentImageIdentityFromElement, documentTableGeometryFromElement, normalizeDocumentTableRowHeightRule, documentNoteKey, normalizeDocumentParagraphBorders, normalizeDocumentTableBorderWidth, normalizeDocumentParagraphId, normalizeDocumentTableBorderStyle, normalizeDocumentBookmarkReferencesHtml, documentCitationInstruction, parseDocumentParagraphBordersElement, documentImageWrapContourFromElement, documentImagePositionFromElement, normalizeDocumentPageChrome, documentPageMarginsForLayout, normalizeDocumentImageLayer, documentBookmarkReferenceInstruction, documentImageCropFromElement, normalizeDocumentPageBorders, documentImageLayerFromElement, parseDocxThemeReference, documentCaptionLabel, documentCaptionKind, applyDocumentParagraphIdentityToElement, isContourImageLayout, normalizeDocumentImageIdentity, normalizeTableColor, documentTableBordersFromElement, documentSections, createDocumentParagraphIdentityRegistry, documentCitationTags, docxDocumentFieldKind, wrapsBesideImage, DocxThemePatchCollector, defaultDocumentImageWrapContour, normalizeDocumentParagraphIdentity, normalizeDocumentPaperSource, documentEquationFromElement, DOCUMENT_PARAGRAPH_BORDER_EDGES, parseDocumentCharacterFormatting, patchDocxThemeReferences, parseDocumentParagraphShadingElement, documentParagraphIdentityFromElement, normalizeDocumentEquation, parseDocumentParagraphFormatting, documentTableRowIdentityFromElement, normalizeDocumentBookmarkNativeId, collectDocumentNotes, DOCUMENT_PAGE_MARGIN_KEYS, uniqueDocumentImageIdentity, documentImageLayoutFromElement, documentPageHorizontalMarginTwips, documentPageGeometryForLayout, normalizeDocumentPageMargins, documentNoteKind } from "./8928.js";
+import { normalizeDocumentBookmarkName, applyDocumentImageIdentityToElement, DOCUMENT_PAGE_BORDER_EDGES, createDocumentImageIdentityRegistry, normalizeDocumentColumns, normalizeDocumentTableVerticalAlign, uniqueDocumentParagraphIdentity, documentTableColumnPercentagesFromElement, documentTableCellMarginOverridesFromElement, applyDocumentTableRowIdentityToElement, normalizeDocumentBookmarksHtml, documentImageIdentityFromElement, documentTableGeometryFromElement, normalizeDocumentTableRowHeightRule, documentNoteKey, normalizeDocumentParagraphBorders, normalizeDocumentTableBorderWidth, normalizeDocumentParagraphId, normalizeDocumentTableBorderStyle, normalizeDocumentBookmarkReferencesHtml, documentCitationInstruction, parseDocumentParagraphBordersElement, documentImageWrapContourFromElement, documentImagePositionFromElement, normalizeDocumentPageChrome, documentPageMarginsForLayout, normalizeDocumentImageLayer, documentBookmarkReferenceInstruction, documentImageCropFromElement, normalizeDocumentPageBorders, documentImageLayerFromElement, parseDocxThemeReference, documentCaptionLabel, documentCaptionKind, applyDocumentParagraphIdentityToElement, isContourImageLayout, normalizeDocumentImageIdentity, normalizeTableColor, documentTableBordersFromElement, documentSections, createDocumentParagraphIdentityRegistry, documentCitationTags, docxDocumentFieldKind, wrapsBesideImage, DocxThemePatchCollector, defaultDocumentImageWrapContour, normalizeDocumentParagraphIdentity, normalizeDocumentPaperSource, documentEquationFromElement, DOCUMENT_PARAGRAPH_BORDER_EDGES, parseDocumentCharacterFormatting, patchDocxThemeReferences, parseDocumentParagraphShadingElement, documentParagraphIdentityFromElement, normalizeDocumentEquation, parseDocumentParagraphFormatting, documentTableRowIdentityFromElement, normalizeDocumentBookmarkNativeId, collectDocumentNotes, DOCUMENT_PAGE_MARGIN_KEYS, uniqueDocumentImageIdentity, documentImageLayoutFromElement, documentPageHorizontalMarginTwips, documentPageGeometryForLayout, normalizeDocumentPageMargins, documentNoteKind } from "./6282.js";
 import { serializeUtf8Xml, directChildren, xmlNamespacePrefix, parseXml, decodeXmlBytes, descendants, resolvePartTarget, directChild, attribute as work_ooxml_package_attribute } from "./5184.js";
 import { assignDocxNoteIds, patchDocxPageColor, normalizeDocxCommentId, assignDocxCommentThreads } from "./0~4980.js";
 import { paragraphTabStops, mergeDocxIgnorableExtensionsAtPairs, patchDocxBibliography, XMLNS_NAMESPACE as work_docx_settings_xml_XMLNS_NAMESPACE, paragraphAlignment, paragraphSpacingOptions, xmlNamespaceUri, hasNonWhitespaceXmlText, cloneXmlElement, cssColorToHex, DocxParagraphDefaultCollapsedPatchCollector, assertXmlRoot, xmlDeclaredPrefix, paragraphIndent, domDirection, DOCX_WORDPROCESSING_NAMESPACES, paragraphBidirectional, XML_NAMESPACE as work_docx_settings_xml_XML_NAMESPACE, patchDocxParagraphDefaultCollapsed, xmlAttributeNamespace, dataBoolean, mergeDocxIgnorableExtensions, parseDocxColorSchemeMappingElement, paragraphPaginationOptions, DOCX_COLOR_SCHEME_MAPPING_ATTRIBUTES, xmlAttributeLocalName, cssFontSize, paragraphDirectionOptions, cssFontFamily } from "./0~2805.js";

package/dist/0~work-docx-import.js

@@ -1,6 +1,6 @@
 import jszip from "jszip";
 import { contentTypeForPart, directChildren, bytesToDataUrl, OoxmlPackage, firstDescendant, descendants, directChild, attribute } from "./5184.js";
-import { normalizeDocumentBookmarkName, documentInitialSectionLayout, normalizeDocumentFieldsHtml, applyDocumentImageIdentityToElement, renderDocumentTableCellMarginOverrides, documentCaptionLabel, documentTableBordersFromElement, isContourImageLayout, normalizeDocumentCitationsHtml, createDocumentImageIdentityRegistry, normalizeDocumentColumns, normalizeDocumentImageIdentity, normalizeDocumentImageWrapSide, applyDocumentParagraphIdentityToElement, docxDocumentFieldKind, serializeDocxThemeReference, renderDocumentTableBorders, docxBookmarkReferenceTarget, normalizeDocumentImagePosition, wrapsBesideImage, documentPageChromeLegacyFields, normalizeDocumentImageWrapContour, documentParagraphShadingDomAttributes, normalizeDocumentBookmarksHtml, documentParagraphBordersDomAttributes, DOCUMENT_TABLE_ROW_ID_ATTRIBUTE, applyDocumentTableRowIdentityToElement, normalizeDocumentCaptionsHtml, normalizeDocumentParagraphIdentity, DEFAULT_DOCUMENT_TABLE_CELL_MARGINS, DEFAULT_DOCUMENT_TABLE_GEOMETRY, documentSectionDomAttributes, documentPageMarginBody, normalizeDocumentImageAlignment, normalizeDocumentParagraphId, applyDocumentImageLayerToElement, normalizeDocumentBookmarkNativeId, normalizeDocumentTableRowIdentity, uniqueDocumentImageIdentity, normalizeDocumentBookmarkReferencesHtml, documentCitationTagsFromInstruction, applyDocumentImageWrapContourToElement, normalizeDocumentNotesHtml, applyDocumentTableGeometryToElement, normalizeDocumentImageCrop, normalizeDocumentPageChrome, normalizeDocumentImageLayoutOptions, applyDocumentImageCropToElement, documentPageMarginsForLayout, applyDocumentPageGeometry, documentContentLayoutProperties, createDocumentNoteElement, normalizeDocumentImageLayer, DOCUMENT_TABLE_ROW_TEXT_ID_ATTRIBUTE, documentBookmarkReferenceInstruction } from "./8928.js";
+import { normalizeDocumentBookmarkName, documentInitialSectionLayout, normalizeDocumentFieldsHtml, applyDocumentImageIdentityToElement, renderDocumentTableCellMarginOverrides, documentCaptionLabel, documentTableBordersFromElement, isContourImageLayout, normalizeDocumentCitationsHtml, createDocumentImageIdentityRegistry, normalizeDocumentColumns, normalizeDocumentImageIdentity, normalizeDocumentImageWrapSide, applyDocumentParagraphIdentityToElement, docxDocumentFieldKind, serializeDocxThemeReference, renderDocumentTableBorders, docxBookmarkReferenceTarget, normalizeDocumentImagePosition, wrapsBesideImage, documentPageChromeLegacyFields, normalizeDocumentImageWrapContour, documentParagraphShadingDomAttributes, normalizeDocumentBookmarksHtml, documentParagraphBordersDomAttributes, DOCUMENT_TABLE_ROW_ID_ATTRIBUTE, applyDocumentTableRowIdentityToElement, normalizeDocumentCaptionsHtml, normalizeDocumentParagraphIdentity, DEFAULT_DOCUMENT_TABLE_CELL_MARGINS, DEFAULT_DOCUMENT_TABLE_GEOMETRY, documentSectionDomAttributes, documentPageMarginBody, normalizeDocumentImageAlignment, normalizeDocumentParagraphId, applyDocumentImageLayerToElement, normalizeDocumentBookmarkNativeId, normalizeDocumentTableRowIdentity, uniqueDocumentImageIdentity, normalizeDocumentBookmarkReferencesHtml, documentCitationTagsFromInstruction, applyDocumentImageWrapContourToElement, normalizeDocumentNotesHtml, applyDocumentTableGeometryToElement, normalizeDocumentImageCrop, normalizeDocumentPageChrome, normalizeDocumentImageLayoutOptions, applyDocumentImageCropToElement, documentPageMarginsForLayout, applyDocumentPageGeometry, documentContentLayoutProperties, createDocumentNoteElement, normalizeDocumentImageLayer, DOCUMENT_TABLE_ROW_TEXT_ID_ATTRIBUTE, documentBookmarkReferenceInstruction } from "./6282.js";
 import { markDocxParagraphFormattingChanges, markDocxParagraphIndents, parseDocxPageGeometry, parseDocxPageBorders, docxCaptionReferenceTarget, applyImportedDocxEquationMarkers, hasImportedDocxParagraphPaginationMarkers, createDocxTableStyleResolver, applyImportedDocxParagraphSpacingMarkers, hasImportedDocxParagraphShadingMarkers, hasImportedDocxParagraphDirectionMarkers, isDocxEquationLikeRoot, inspectDocxPageMarginSettings, applyImportedDocxParagraphFormattingChangeMarkers, hasImportedDocxParagraphTabStopMarkers, hasImportedDocxRunFormattingMarkers, hasImportedDocxParagraphIndentMarkers, docxCaptionSequenceKind, resolveDocxParagraphBordersForParagraph, applyImportedDocxRunFormattingMarkers, markDocxParagraphAlignments, markDocxParagraphDirections, applyImportedDocxParagraphShadingMarkers, markDocxParagraphSpacing, hasImportedDocxParagraphBorderMarkers, hasImportedDocxParagraphFormattingChangeMarkers, docxCaptionBookmark, docxFieldResultText, createDocxParagraphStyleResolver, docxEquationHtml, docxFieldOccurrences, hasImportedDocxParagraphAlignmentMarkers, markDocxParagraphTabStops, applyImportedDocxParagraphIndentMarkers, resolveDocxTableStyleResolver, applyImportedDocxParagraphDirectionMarkers, docxTableCellStyleLayers, applyImportedDocxParagraphAlignmentMarkers, markDocxParagraphPagination, markDocxParagraphBorders, markDocxRunFormatting, applyImportedDocxParagraphTabStopMarkers, docxFieldOccurrenceIsInlineEditable, applyImportedDocxParagraphBorderMarkers, applyImportedDocxParagraphPaginationMarkers, parseDocxPaperSource, hasImportedDocxParagraphSpacingMarkers, docxTablePropertySources, parseDirectDocxParagraphShading, parseDocxPageMargins, markDocxParagraphShading, markDocxPackageEquations } from "./0~5093.js";
 import { xmlAttributeLocalName, readDocxBibliography, createDocxThemeResolver, docxThemeColor, parseDocxParagraphDefaultCollapsed, resolveDocxThemeResolver, DOCX_WORDPROCESSING_NAMESPACES, xmlAttributeNamespace } from "./0~2805.js";
 import { importDocxPageColor, importedDocumentNoteId } from "./0~4980.js";