🌐 AI Document Translator & Layout
AI document translation and layout: organise specs into sections and blocks, translate block by block with glossary terms across 8 languages, edit live on A4, and export HTML, Word, PDF, or JSON.
📑 Templates
🗂️ Section outline
📝 Source text & glossary
Source text (extracted original, read-only)
🤖 AI translate & structure
Running extraction again costs 2 credits again; manual edits in the preview are free.
🖼️ Document theme
🎨 Header & footer
🚀 Export
How to use AI Document Translator & Layout
Follow the steps to get it done
Pick a template from the dropdown in the left column: choose a preset (Product Datasheet, Component Datasheet, Generic Technical Document, Blank, Auto Detect) or one of My Templates. The middle column immediately shows the empty structure — adjust section titles and block types before extracting
Open the page and start — no registration or sign-in needed.
Upload a file at the top or paste the source text in the left column: upload a PDF, DOCX, TXT, or Markdown file in the top upload area to extract text, or paste the text directly in the left column; add one “source = translation” pair per line in the glossary when terms must stay consistent. In the right column, pick the source language, target language, and tone, then click “Start AI structuring & translation” (2 credits per run)
Settings can be changed at any time; requests go straight from your browser to the API you configured.
The middle column is preview-as-editor: click text to edit it, hover for block controls (change type, add/remove rows and columns, upload images), and drag the ⠿ handles to reorder sections and blocks; click a section’s “Prompt” button in the left column “Section outline” to write extra instructions for that section only, and add or delete sections there too; pick a preset theme or adjust the brand color, fonts, and heading style in the right column “Document theme” panel, adjust the logo switch and height, then export HTML, Word, print/PDF, JSON, or save to the cloud
Copy or download the result when you are done. Nothing is kept on this site.
Page Actions
In the left column, pick a template from the dropdown, grouped as “Preset templates / My templates”; the middle column immediately shows the empty structure. With source text or edits present, an inline “Switch & clear / Cancel” confirmation appears first.
The button to the right of the dropdown, available when a custom template is selected; preset templates cannot be deleted and the button is disabled. Click once and it turns into “Confirm delete?”, click again to delete locally and in the cloud.
Inline confirmation shown when switching templates with content present: “Switch & clear” clears the text and edits and switches, “Cancel” keeps everything.
The input at the bottom of the left column template panel, next to the 💾 icon button; type a name for a custom template, blank falls back to the current document title.
The 💾 icon button next to the name input; click to save the current section titles, block types, and each section’s prompt as a custom template (structure only, no text or images), stored locally and in the cloud. It turns into ✓ once saved, and back to 💾 when the structure, a section title or prompt, the name, or the template changes.
At the top, click or drop a single PDF, DOCX, TXT, or Markdown file (max 100MB, PDF max 100 pages); the browser extracts text locally into the source box. After upload a file card appears; click it or drop a new file to replace.
Paste or edit the source text directly; the count line below shows characters against the 12,000 limit live, and over-limit text triggers a truncation confirmation when the AI button is clicked.
One “source = translation” pair per line (=> also works), capped at 50 lines — extra lines are not sent. Pairs are injected into the translate prompt to keep proper nouns and model names consistent.
Collapsed by default; expand to view the extracted source text (read-only) and check it against edits made in the input box. Pasted text does not appear here.
In the right column, pick the source language (8 languages, Simplified Chinese by default); when it matches the target language, the translate stage is skipped.
In the right column, pick the translation target (8 languages: English, Simplified Chinese, Spanish, German, French, Japanese, Russian, Arabic; English by default).
In the right column, pick the terminology and tone style: Professional, Marketing, or Concise.
In the right column, calls the site AI gateway (sign-in required, 2 credits per run, BYOK free) to organise the text into structure and then translate block by block; over-limit text shows the truncation confirmation first.
Shown when the source text exceeds 12,000 characters: “Truncate & continue” keeps only the first 12,000 characters and notes the truncation in the status line; “Cancel” does nothing.
Persistent status in the right column showing progress, completion, or failure reasons; a glossary truncated to 50 lines is noted here too.
In the right column “Document theme” panel, pick one of 3 preset themes (Classic Blue, Business Black, Minimal Gray); switching resets the brand color, fonts, heading style, title alignment, and page background to the preset defaults, changing only the look.
The button to the right of the theme dropdown; restores all individually adjusted settings (brand color, fonts, heading style, and so on) to the current preset’s defaults.
In the right column, pick any brand color; section heading decorations, badges, and soft tints follow immediately.
6 preset color dots next to the picker (blue, black, green, orange, purple, teal); clicking one sets the brand color to it and syncs the picker, with the current color ringed in blue.
In the right column, pick the document heading font (System default, Alibaba PuHuiTi, Inter, Montserrat, Oswald, Playfair Display); Latin fonts load online and fall back to system fonts if loading fails. Only affects the document itself.
In the right column, pick the document body font, with the same options as the heading font; only affects the document itself, not the site interface.
In the right column, pick one of 4 section heading decorations: pill badge + rule line, left accent bar, underline, or none.
In the right column, align the document title and subtitle left or center.
In the right column, pick the document page background: white, light gray, or light blue.
Click the subtitle under the header title (for the model number) and type; the change is written back when the field loses focus. Empty subtitles show a placeholder in the editor and are omitted from exports. Unchecking “Show subtitle” in the right column “Header & footer” panel hides the row.
Click either side of the footer at the bottom of the A4 page (left for copyright, right for model and website) and type; the change is written back when the field loses focus. The footer row is omitted from exports when both sides are empty. Unchecking “Show footer” in the right column “Header & footer” panel hides the footer row.
Checkbox in the right column, on by default; turning it off hides the logo in the header, exports, and print.
Checkbox in the right column “Header & footer” panel, on by default; unchecking hides the subtitle row in the middle column, exports, and print, while the subtitle text stays in the document.
Checkbox in the right column “Header & footer” panel, on by default; unchecking hides the footer row in the middle column, exports, and print, while the footer text stays in the document.
Number input in the right column, 24–80 px, 40 by default; changes apply to the middle column header immediately.
In the right column, downloads a self-contained styled A4 HTML file that opens in any browser.
In the right column, downloads a Word (.doc) document keeping tables and text styles for further editing.
In the right column, opens the clean A4 page in a new window and brings up the browser print panel to save as PDF; a blocked popup is reported in the status line.
In the right column, downloads the current document as JSON (with a version number), including the full section and block structure.
In the right column, uploads the standalone HTML to member cloud storage once signed in (within the 100MB quota); the status line reports the result, and a full quota prompts clearing space on the account page.
In the right column, copies the clean export HTML to the clipboard, without editor controls.
Persistent status in the right column reporting export, print, and cloud-save results or failure reasons.
The left column “Section outline” lists all sections in document order, each row showing its number and title; clicking a row title scrolls the middle column to that section and briefly highlights it. Titles edited in the middle column sync back to the outline; with no sections only a short empty message shows.
Button at the bottom of the left column “Section outline” panel; appends a “New Section” (with one empty paragraph block) at the end of the document — edit its title directly in the preview.
Zoom out, percentage reset, and zoom in buttons in the middle column toolbar, 50%–150% in 10% steps; clicking the percentage returns to 100%. On first load the view shrinks once if the middle column is narrower than A4.
The “Pure preview” button in the middle column toolbar hides all editing handles and placeholders so you can check the exported look; click “Back to editing” to return. The add-section button is disabled in pure preview.
The ⠿ handles on section title rows and block hover toolbars; drag to reorder sections or blocks (including moving a block to another section), with a drop line showing where it lands. The new order applies on release.
Persistent hint in the middle column toolbar: edit directly in the preview, click text to change it, hover for controls.
Click the document title in the A4 header and type; the change is written back when the field loses focus.
Click a section title and type; the change is written back when the field loses focus.
The always-visible “Prompt” button on each row of the left column “Section outline”: click to expand the prompt input box under the row, click again to collapse; only one section can be expanded at a time; sections with a prompt show the button highlighted with a dot.
Input box under an outline row for extra instructions to the AI that apply to this section only (max 500 characters), saved on blur; injected into the organise prompt on the next AI run; not expanded in pure preview; saved with “My template”.
The “Add block” button shown on hovering a section title row; appends an empty paragraph block to that section.
The always-visible ✕ button on each row of the left column “Section outline”; one click deletes the whole section.
Dropdown in a block hover toolbar; converts between the 9 block types, keeping content where possible. An image block converted to a type without an image slot keeps its image as a separate block.
The ✕ button in a block hover toolbar; deletes that content block.
Click text inside paragraph, subheading, or callout blocks and type; the change is written back when the field loses focus.
Click an item to edit it; the ✕ after each item deletes it; “Add item” appends an empty item; the “Single / Double column” button switches between one- and two-column layout.
Click field and value cells to edit them; the ✕ at a row end deletes the row; “Add row” appends an empty row.
Click header and body cells to edit them; “Add row” appends an empty row, “Add column” appends an empty column.
Click group names and parameter names/values to edit them; the ✕ at a parameter row deletes that parameter; “Add parameter” adds a row to the group; “Add group” adds an empty group.
Click an image placeholder to pick an image; with an image present, “Replace” picks a new one and “Remove image” clears it; the width dropdown offers 50%, 75%, and 100%.
Click the text area to edit text, click the image area to upload or replace the image; the “Image on left / right” button swaps sides (text 7, image 5).
Level dropdown on subheading blocks, switching between H3 and H4.
Tone dropdown on callout blocks, switching between Note, Warning, and Remark.
When no logo is uploaded, the header shows a “+ Logo” placeholder button — click it to upload an image; when a logo is present, click the logo image in the header to replace it.
How to Let AI Use It
Follow the steps to operate the browser page — no API key needed
There is no MCP tool for this one. Let an AI assistant follow these steps to operate the browser page (button IDs and labels match the workbench above):
- Open https://tools.waoww.com/tools/ai-datasheet-translator/
- In the left column, pick a template from the dropdown (#ds-template-select) (e.g. “Product Datasheet”); the middle column shows the empty structure. Adjust section titles in the middle column and block types via the block type dropdowns before extracting
- Upload a PDF, DOCX, or TXT file in the top upload area (#ds-upload-dropzone) to extract text, or paste the source text into #ds-raw-text; write one “source = translation” pair per line in the glossary (#ds-glossary)
- In the right column, pick the source language (#ds-source-lang), target language (#ds-target-lang), and tone (#ds-tone), then click “Start AI structuring & translation” (#ds-btn-ai-submit). For text over 12,000 characters, click “Truncate & continue” (#ds-btn-truncate) in the confirmation row first
- Edit text directly in the middle preview (#ds-preview-container), adjust structure with the hover block controls; in the left column “Section outline” (#ds-outline) click a row title to jump to a section, click “Prompt” to write extra instructions, click ✕ to delete a section, and click “Add Section” (#ds-btn-add-section) for a new section; use #ds-zoom-out, #ds-zoom-reset, and #ds-zoom-in for zoom
- In the right export area, click “HTML” (#ds-btn-export-html), “Word” (#ds-btn-export-word), “Print / PDF” (#ds-btn-print), “JSON” (#ds-btn-export-json), or “Save to cloud” (#ds-btn-save-cloud); the logo switch is #ds-logo-enabled and the height input is #ds-logo-height
Example prompt (send it to your AI assistant as-is)
「Open tools.waoww.com/tools/ai-datasheet-translator/, choose the Product Datasheet template, paste product parameters into the source box, set the target language to English, run AI structuring & translation, then export as an HTML file.」
How It Works & Highlights
How it works, and where it fits
The document model is “sections + blocks”: a section has an editable title, and a block is one of 9 single-type semantic blocks (paragraph, bullet list, key-value table, table, grouped specs, image, text + image, subheading, callout); text + image blocks split 7:5. The AI goes through one gateway feature in two stages: the organise stage splits the source text into blocks following the template structure (keeping the source language), and the translate stage translates block by block with glossary pairs injected; the translate stage is skipped when source and target languages match. One A4 stylesheet drives the on-screen preview, printing, and exports: the preview outputs editable text and hover controls, exports output a clean document. PDF extraction groups text by line (rows by y, left-to-right by x); DOCX extraction reads document.xml preserving paragraphs and table structure — both run locally in the browser.
Highlights
- ✓ 9 semantic blocks: paragraph, list, key-value, table, grouped specs, image, text + image, subheading, callout — converting types keeps content where possible
- ✓ Two-stage AI pipeline: organise into structure first, then translate block by block with glossary enforcement; same-language runs skip translation; failures refunded
- ✓ Preview is the editor: adjust the template structure before extraction, keep editing after; click text to edit, block controls on hover, drag handles to reorder
- ✓ True A4 layout: 210mm × 297mm, header logo switch and height controls, HTML / Word / print PDF / JSON exports
- ✓ Local extraction: line-preserving PDF, paragraph- and table-preserving DOCX; original files never uploaded; read-only source fold for checking
- ✓ My Templates stored twice: browser local storage + a dedicated cloud D1 table outside the storage quota; 20-template cap, merged by update time
Tips
- 💡 Write parameters as “Name: value + unit” lines so the organise stage can map them into key-value tables or tables
- 💡 Put fixed model names and proper nouns in the glossary (source = translation, one pair per line) and the translate stage follows the table
- 💡 Arrange the template structure in the middle column before running the AI: the organise stage splits text along the current structure, so a good structure means less rework
Notes
- ⚠️ PDF parsing runs locally in the browser; scanned image-only PDFs yield no text — run OCR first, then paste.
- ⚠️ Each AI run costs 2 credits (organise + translate count as one call, failures refunded); BYOK runs are free of credits; running again costs again.
- ⚠️ Source text is capped at 12,000 characters and needs truncation confirmation when over; files are capped at 100MB, PDFs at 100 pages.
- ⚠️ Custom templates store structure only, no content; cloud sync needs sign-in, with a 20-template cap and 64KB per template.
FAQ
Privacy, format limits, and usage tips