Skip to content

Choose Export Formats, Content, and Layout

After parsing or translation is ready, use the reader's export button to open Export Document Settings (“导出文档配置”). First decide which content you need, which app will open it, and whether page positioning matters, then choose the format and layout. Reader columns do not automatically determine export layout.

This guide covers the workflow and conditions. For Word appearance, see templates. For equation workflows, see Word/MathType and WPS.

Step 1: Choose a Format for Your Use

Interface optionIntended useFile or destination
WordEdit text and equations; table editing depends on actual objects and layout.docx
WPS (Editable Equations)A formula-editing workflow suited to WPS.docx, not .wps
Word MathTypeUse MathType equations in Word.docx; supported equations become MathType, unsupported ones remain Word equations
EduediterContinue editing in Eduediter.wsd
MarkdownKnowledge bases, notes, and Markdown editors.zip for local images; .md for online images
LaTeXContinue typesetting with TeX tools.zip containing .tex and applicable resources
TypstContinue typesetting with Typst.zip containing .typ and applicable resources; image paths and compilation checks
HTMLBrowser reading or further HTML processing.html
Export to MD EditorAdjust content in the app's Markdown editor firstOpens the editor; select it, then click Start Export

WPS, Word MathType, and Eduediter appear only for supported result versions. Advanced exports check access when used. Some formats offer Try Once for Free when eligible; Eduediter requires membership. Follow the actual prompt. A file extension does not establish access, and file-export permissions are separate from clipboard permissions.

Check formula editing, tables, and images in the Word, WPS, or MathType environment you actually use. “Editable equations” does not guarantee lossless conversion of every complex formula.

Choose an output for your next step (workflow diagram)

Choose an output for your next step (workflow diagram). Open the image in a new tab to enlarge it.

Step 2: Choose Word or PDF Handling for Parsed Results

Word Layout

Eligible v3 PDF or structured-image results show Word Layout → Page Layout after selecting Word, WPS, or Word MathType:

  • Reflow (“流式排版”): Reorganizes text for paragraph edits, font changes, and templates.
  • Preserve PDF Layout (“保留 PDF 版式”): For PDF sources, aims to preserve page dimensions and text/image positions.
  • Preserve Image Layout (“保留原图版式”): The corresponding option for structured-image sources.

The app may remember the last submitted layout, so check it every time. Preserved layout supports further text and equation editing but does not guarantee an identical complex page. It hides the reflow Advanced Parsing Configuration, including Word templates.

Same-source example: reflow tables and preserved textboxes

On 2026-10-13, two Word files were downloaded from the same existing parse of a self-authored, two-page Willow record. No new upload, OCR or translation was run, and the native DOCX bytes remain unchanged.

CheckReflow WordPreserve PDF Layout
Native tables2 tables, each 4 rows × 6 columns0
Positioned textboxes063
Document section sizeLetterA4
Local preview pages22
Content distributionBoth source tables on page 1; only the 44.60 total note on page 2Each table remains on its corresponding source page
Source footer sentenceNot retainedRetained, but wraps onto two lines in the local preview

For row and column edits, first confirm that the result contains a native table. Selectable text or visible borders do not establish table structure. For calculations or filtering, see table Excel export; for substantial text edits or source-page positioning, compare the two Word layouts.

Local LibreOffice preview of the actual reflow DOCX, page 1: both native tables are on this page; the total note is on page 2.

Page 1 of the local reflow preview. The full preview also contains page 2; open the image to enlarge it.

Local LibreOffice preview of page 1 from the same-source preserved-layout DOCX; the visible table uses positioned textboxes.

Page 1 of the preserved-layout preview. Similar page positioning does not establish a native Word table.

Download the source PDF, native reflow DOCX, native preserved DOCX, or both DOCX files, full previews and checks. The table-editing guide has more object checks; that guide and the self-authored sample text are in Chinese.

The images and preview PDFs were rendered locally with LibreOfficeDev 26.8 and Arial Unicode MS fallback. They are not Microsoft Word/WPS desktop display or editing tests. The initial local render omitted Chinese even though the DOCX contained the sample words “采样瓶”, “未记录” and “数量冲回”. Process-local font configuration restored the text, changing the reflow preview from one page to two without changing DOCX or system font settings. Locate the content, export and rendering stages before repeating OCR. A preview with missing glyphs cannot validate pagination.

Two Different PDF Actions

ActionWhat it doesConditions
Add Text Layer to Scanned PDF (“扫描 PDF 添加文本层”)Preserves the scan appearance and adds searchable, selectable, copyable textSource content, not a translated PDF; appears for PDF or eligible structured-image sources and checks export access
Generate PDF After Editing (“编辑后生成 PDF”)Sends parsed content to the Markdown editor and opens PDF export guidanceCurrently shown only for structured-image results in the parsing dialog; not a direct download preserving the original page

To reflow ordinary PDF parsing results into a PDF, choose Export to MD Editor, follow the Markdown editor tutorial to review and adjust the content, then use its PDF export workflow. For translated PDFs, use the next section rather than Add Text Layer.

Step 3: Choose Translation Content and Page Effect

Formats depend on results that the task actually produced and has ready. The following describes the result reader; historical records, older versions, or incomplete outputs may offer fewer choices.

Translation modeAvailable workflow
BilingualWord, Markdown, LaTeX, Typst, HTML, the internal Markdown editor, and other eligible reflow formats; PDF uses Continuous Reading
Preserve LayoutPDF offers translation only or side-by-side pages. When translated Word layout capability and access are available, export preserved-layout Word directly with Translation Only. Reflow Word, bilingual content, Markdown, and similar formats require the corresponding result through More Formats · Create Bilingual Translation
Bilingual + Preserve LayoutWhen both outputs are available, use both reflow formats and fixed-layout PDF; eligible translations also offer Word preserving the PDF layout

The combined mode is not limited to PDF. Nor does selecting that mode make every export available at every stage: the relevant content, PDF, and layout capability must be ready.

Translation Content and Format Matrix

Format and layoutExport contentConditions
Reflow Word, Markdown, LaTeX, Typst, HTML, Export to MD EditorTranslation Only / Bilingual / Source OnlyUsable bilingual content
Reflow WPS / Word MathType, EduediterTranslation Only / Bilingual / Source OnlyBilingual content, supported v3 result, and applicable access
PDF → Continuous ReadingTranslation Only / Bilingual / Source OnlyBilingual content; repaginates rather than preserving source pages
PDF → Preserve Original · Translation OnlyTranslation onlyFixed-layout PDF ready
PDF → Side by Side · Translation on LeftTranslation left, source rightFixed-layout PDF and original source available
PDF → Side by Side · Source on LeftSource left, translation rightFixed-layout PDF and original source available
Word / WPS / Word MathType → Preserve PDF LayoutTranslation onlySupported v3 task, fixed-layout result, saved translated Word layout capability, and access

Follow this order:

  1. Choose Export Format. For non-PDF reflow formats, select Translation Only, Bilingual, or Source Only under Export Content.
  2. For PDF, choose Page Effect first. Only Continuous Reading offers those three additional content choices. The other three fixed effects already determine content and left/right order.
  3. If Word Layout appears, choose from the available options. A pure layout task may offer only Preserve PDF Layout, with Reflow unavailable; export Translation Only directly in that case. Reflow content requires bilingual results. Preserve PDF Layout supports Translation Only, including in a task that also has bilingual content.
  4. If told the historical translation has no saved Word layout, use reflow when bilingual content is available. A layout-only record can create a separate bilingual task for reflow files. If a newly generated layout is needed, follow the prompt to translate again and check the available options and usage conditions.
  5. Fixed-layout PDFs offer More PDF Settings → PDF Bookmarks. Side-by-side output also offers Online Merge Export; enable that before Optimize PDF Size can be used.

In the translation-only reader for a pure Preserve Layout task, a message may say that only online export or translation-only export is supported. Enable Online Merge Export for a side-by-side PDF there, or export translation only. See New Translation for reflow or bilingual Word, Markdown, and other content formats. An available preserved-layout translated Word file can be exported directly; see the actual pure-layout Word example.

Confirm the translation content scope (workflow diagram)

Confirm the translation content scope (workflow diagram). Open the image in a new tab to enlarge it.

Step 4: Adjust Settings Offered for the Format

Advanced Parsing Configuration (“高级解析配置”) starts collapsed, with Recommended Settings or a count of adjusted items. Only applicable options appear. If an option is absent, check the format and whether preserved layout is selected.

Format/stateMain available settings
Markdown · Local PackageCross-page tables, image references, formula delimiters; formula fallback for v3
Markdown · Online ImagesImage references, formula delimiters, remove comments, code indentation compatibility; no cross-page tables or formula fallback
LaTeXCross-page tables, formula delimiters; formula fallback for v3
TypstCross-page tables; formula fallback for v3
Word / WPS / Word MathType · ReflowCross-page tables; formula fallback and Word templates for eligible v3 results
Preserved-layout Word variants or scanned PDF text layerNo reflow advanced configuration
HTML, Eduediter, internal MD editor, and similar destinationsDo not expect every advanced option above

Keep Images for Long-Term Use

Markdown's Image Resource References offers:

  • Local Package (“本地打包”): Extract the ZIP and keep Markdown and image folders in their relative positions. Suitable for long-term storage, sharing, and offline use.
  • Online Images (Valid for 30 Days) (“在线图床(30 天有效)”): Markdown references online images whose links expire after 30 days. For long-term use, choose Local Package rather than retaining only the online .md file.

Tables, Formulas, and Templates

  • Cross-page Tables: Preserve by Original Page retains page-based organization; Smart Merge attempts to combine tables. Check headers, merged cells, and continuation boundaries afterward.
  • Markdown Table Format: HTML Tables retain merged cells and complex structure. Markdown Tables expand merged cells, keep content in the top-left cell, and simplify complex styles. Choose for your target editor.
  • Formula Delimiters: Choose \[ \] or $ as required by the target tool. This changes formula markers; it does not convert equations to MathType.
  • Formula Fallback Level: Where offered, choose Preserve Formulas, Inline Formulas to Plain Text, All Formulas to Plain Text, or Smart Fallback. Use text conversion cautiously when mathematical editing matters.
  • Word Template: Eligible reflow Word variants offer the default online template, General, Academic, Business, Elegant, Minimal, and Technical. Compare fonts and indentation in template previews.

Image assets and Word templates (workflow diagram)

Image assets and Word templates (workflow diagram). Open the image in a new tab to enlarge it.

Step 5: Name, Save, and Check

  1. If the dialog offers Base Filename, enter a recognizable name. Add Task Date if needed, then check separators and filename preview. The tested pure-layout task offered this editor for both Word and PDF; an entry without it uses the displayed name. Actual downloads may append a content label such as Translation Only or Bilingual. The export method determines the real extension.
  2. Fixed-layout translation PDFs retain the suffixes 仅译文 (translation only), 译文在左 (translation on left), or 原文在左 (source on left) to distinguish versions.
  3. On the web, follow browser download settings and check its downloads list. A folder picker is not guaranteed every time.
  4. On desktop, use Personal Settings → Default File Download Location → Choose Directory. The Settings link next to Save Location in the export dialog opens it too. Check whether that directory is enabled for this export.
  5. Click Start Export, wait for conversion and download, then open the result in the target app. Check complete pages, equations, tables, images, and source/translation order.

HTML mobile reading: image resources and table width

An HTML file that displays its image can still have a table wider than a phone screen. On 2026-10-13, we continued checking an existing export from a two-page English PDF. Its native HTML references an external image. The earlier locally embedded copy contains that same image, but its document width still reaches about 430 CSS pixels in 312- and 390-pixel viewports.

We saved a separate local reading copy, constraining the reading column and putting the original table inside a named, keyboard-focusable horizontal scroll region. This is not a new product export style, upload or recognition result.

File or conditionObserved resultWhat to check
Unchanged native HTMLStill references an external imageResource availability and image completeness
Earlier embedded-image copyContains the same JPEG but still overflows on narrow screensImage loading and page width separately
New reading copy, in the review ZIP belowNo document-level horizontal overflow at 312, 390 or 1280 pixelsScroll within the table to its final column
Keyboard operation at 312 pixelsA 278-pixel region contains a 512-pixel table; scrolling 234 pixels reveals the final columnFocus the region and use arrow keys; do not hide columns

Earlier embedded HTML at 312 CSS pixels: the table makes the document about 430 pixels wide and right-hand columns fall outside the current view.

Actual narrow-screen view of the earlier embedded copy. Embedding the image did not fix the wide table.

Local reading copy at 312 CSS pixels, with the table scrolled right to show time and speed; the document itself remains within the viewport.

Actual reading-copy view with the table scrolled right. Both screenshots use the same browser and viewport; content height changes with the layout.

Download the native HTML, two local copies, source PDF, image and stylesheet. Extract it and open native-export.html, embedded-copy.html and mobile-reading.html separately, keeping the original unchanged.

Check in this order:

  1. Confirm the content is present. The new copy retains the original table markup, three data rows, ten math containers and the same 1061×604 JPEG. It does not restore the already-missing label above the parsed chart.
  2. In a working copy, put the table in a named, focusable region with a bounded width and overflow-x: auto. This example uses a 32rem minimum table width; choose for the actual columns in another document.
  3. Do not hide data with page-wide overflow: hidden or shrink the whole page. Reach the final column, compare every row and check that the body does not overflow.
  4. Reopen and check resources. A cache-disabled loopback HTTP load of this reading copy observed only the document and its embedded data image. This was not a fully disconnected browser, physical-phone touch test or print validation.

The exact stylesheet is reading-layout.css in the ZIP and applies to this example. Layout, resources and OCR content require separate checks. A receiving CMS or email editor may filter tags and styles, so preview again in the intended environment.

Copy Full Text and Export Tables

These are independent shortcuts: they run immediately when clicked, without Start Export.

  • Copy Full Markdown: Copies to the clipboard; it does not create a file. If automatic copying is blocked, use Copy Full Text in the Manual Copy Markdown window, or Select All and press Ctrl+C / ⌘C. Prepare and copy again after editing the document.
  • Table Excel: Exports all recognized tables to one .xlsx, with one sheet per table. It does not automatically turn body text or images into spreadsheet cells. If none are found, the app reports that the document contains no tables.

For a particular area, use Export This Paragraph's Tables to Excel or Export This Page's Tables to Excel in paragraph/page actions. Check merged cells, values, and formula characters in the XLSX. Failed image retrieval may leave [图片] placeholders; review the reported missing items.

In unified translation export, both shortcuts require bilingual content and follow the current content choice. Preserved-layout Word uses translation only. If a fixed-layout PDF effect is currently selected, switch to a reflow format such as Word or Markdown first, confirm source/translation/bilingual under Export Content, then use the shortcut. The PDF left/right order does not determine the shortcut content.

Distinguish export settings from shortcuts (workflow diagram)

Distinguish export settings from shortcuts (workflow diagram). Open the image in a new tab to enlarge it.

Same-source tables: actual XLSX and a saved correction

On 2026-10-10, we uploaded a self-authored JPG and a one-page scanned PDF containing only the same image. After parsing, we opened 导出文档 → 表格 Excel (Export Document → Table Excel). This shortcut generates the file immediately; Start Export is unnecessary. Each input contains two tables, and each actual XLSX has 表1 (5 rows × 5 columns) and 表2 (4 rows × 3 columns).

The actual Table Excel shortcut in the export dialog, with the self-authored source and recognized tables behind it.

Download the source scan PDF, its actual XLSX, or the image, PDF, three XLSX files and check notes (ZIP). These are actual interface downloads; the workbook files have not been rewritten.

CheckActual files in this exampleBefore use
Merged cells表1 retains A1:A2, B1:B2, C1:D1, E1:E2, A3:A4Compare heading levels and the A-group row span with the source
Leading zeros and long ID0012, 0013, 0014 and the 18-digit ID are textAvoid forcing these columns to numbers when saving or importing again
Quantities and adjustments12, 18, 24, -3.5, 0, 8.25 are numbersThe source's -3.50 and 0.00 are not fixed two-decimal formatting; set display formatting if needed
Percentages and date12.5%, 100%, 0%, 2026-10-10 are textConfirm meaning and units before converting types or calculating
Mathematical expressionCheck x^2 + 1 as textIt is not an Excel calculation formula; verify recognition too

The first image result misread x^2 + 1 in table 2, row 4, column 2 as \times {72} + 1. In Edit Parsing Result, we selected that table, replaced the cell with the source text, clicked Save Changes, and confirmed saved version 1 before exporting again. Only that cell differs between the two image workbooks. The scan-PDF result already contained the correct text in this run and was not edited.

Saved image-result version 1, with x^2 + 1 restored in the second table before the actual reviewed XLSX download.

Compare the initial image XLSX, containing one recognition error with the reviewed image XLSX. See result review and saving. This single clean-page example does not establish that scanned PDFs always perform better than images. Cross-page continuation, blurred photographs, nested tables and image cells were not tested. We checked workbook values, types and merge ranges; native Excel and WPS appearance and interactive editing were not tested. Open the files in your own spreadsheet software to verify them.

For a PDF table continued onto another page, follow the table review tutorial to check actual worksheets, repeated headers, source pages, blanks, zeros, amounts and CSV. It uses a separate two-page parsing example and distinguishes the native XLSX from files prepared afterwards.

If Export Does Not Finish

Message or situationNext step
Parsing export failed and settings were retainedAdjust settings if necessary and click Retry Export; uploading and parsing again are unnecessary
Combined translation export failedThe dialog retains choices and an error. Resolve the cause, then click Start Export again
A pure Preserve Layout PDF dialog closed but no file appearedCheck errors and browser downloads, then reopen PDF export if needed. Closing alone does not confirm a saved file
File generated, saving incompleteChoose Use System Download (“改用系统下载”) or Not Now. A failed default-directory write does not require reprocessing the source
Certain conversion workflows report conversion complete but download incompleteUse Download Again in that message; not every format has this button
You cancelled system savingThe file was not saved by that action. Export again or use a still-available download entry; cancellation is not a parsing failure
Access or trial quota is insufficientFollow the prompt or choose an available format; avoid repeated clicks
You cannot find the fileCheck browser downloads, the desktop default directory, and the actual extension. Extract ZIPs to find MD/TEX/TYP files and images