✂️ Text Splitter

Split overlong text by length or by Chinese/English sentence boundaries, to feed models with a character cap

English · 简体中文

Text Splitter cuts one overlong piece of text into chunks you can feed in one at a time — useful for ChatGPT, DeepSeek, Claude and other models with a per-message character cap, and equally for splitting documents or long social posts. Four modes — by symbol, by length, Chinese paragraph, English paragraph — with delimiter-aware cut points (escapes like \n and \t supported); download the chunks one by one or as a ZIP. Everything runs entirely in your browser — no servers, no uploads.

👉 Try it online: https://tools.newzone.top/en/text-splitter

Key Features

• Four Splitting Modes: By Symbol, By Length, Chinese Paragraph, and English Paragraph, switched with a Segmented control that only shows the active mode's settings
• Delimiter-Aware Boundaries: in By Length mode the optional alignment-delimiter field snaps cuts to sentence boundaries so text isn't cut mid-sentence; in By Symbol mode the delimiter field itself decides where to cut
• Quick Delimiter Presets: One-click "CJK (Chinese/Japanese/Korean)" (。 ？ ！) and "English/Latin" (. ? !) buttons
• Escape Character Support: \n (newline), \r (carriage return), \t (tab), \s (space), \\ (backslash)
• Batch Export: ≤ 3 segments download individually; > 3 are automatically packaged as a ZIP
• Merged Export: Combine all segments into one file separated by double newlines
• Quick Copy: One-click copy of any segment, with a checkmark confirmation
• Performance Friendly: Preview auto-hides above 500 segments to keep the page responsive (export still works)
• Fully Local: Runs entirely in your browser, so even million-character files stay private

How to Use

1. Paste or upload your long text (drag in a text file).
2. Pick a mode:
   • Split at delimiters (the default) → By Symbol

• Fixed-size chunks (for AI input) → By Length
  • Long Chinese text by paragraph → Chinese Paragraph
  • English text by sentence → English Paragraph

3. To align cuts to delimiters (. ? ! etc.), fill in the alignment-delimiter field in By Length mode; in By Symbol mode the delimiter field decides the cut points.
4. Click Split by symbol (the button label follows the selected mode) — segments download as individual files (≤ 3) or a ZIP bundle (> 3).

Four Splitting Modes

Modes are picked via the Segmented control at the top of the Configuration card. Each mode shows only its own settings, so the panel stays focused.

1. By Symbol

Splits on delimiters alone — every Nth delimiter starts a new segment (default 1 = one segment per separator, 2 = merge every two). The delimiter field is pre-filled per locale (。 ？ ！ for Chinese, . ? ! for Latin) and accepts:

• Manual entry, e.g. 。 ？ ！ (separate multiple delimiters with a space)
• Click CJK to fill 。 ？ ！
• Click Latin to fill . ? !
• Escape sequences: \n, \r, \t, \s, \\

Delimiter splits keep the delimiter at the end of each segment.

2. By Length

Chunks text into blocks of a given character count (default 5000 — the per-request cap of the free Google Translate / DeepL tiers, and a comfortable size for LLM context windows). The optional alignment delimiter field snaps chunk boundaries to the nearest delimiter so sentences aren't cut mid-way; leave it empty and the text is hard-chunked by length. The character count can never be 0 — it falls back to the default.

3. Chinese Paragraph

Detects Chinese paragraph boundaries using punctuation and line-break rules; paragraph interiors are preserved. Designed for Chinese articles, blog posts, and other content where paragraph integrity matters.

4. English Paragraph

Uses the browser's built-in Intl.Segmenter sentence segmentation to detect paragraph boundaries. Good for documents, emails, and papers.

Shared Settings

All modes share these:

• Hide Results: when there are many segments (auto-triggers above 500), enable this to hide the preview and keep the page responsive. Export still works regardless of visibility.
• Export Filename: base name for exported files. Defaults to the uploaded file's name, or split_text if nothing was uploaded. A .txt suffix is added unless you type another.

Export and Copy Functions

• Export Merged Text: combine all segments into a single file, separated by double newlines.
• Batch Export: ≤ 3 segments download individually; > 3 are bundled into a single <filename>_split_files.zip automatically.
• Per-Segment Copy: each segment card has a copy button; the icon flips to a checkmark on success.

Common Patterns

• LLM Context Prep: length 2000–4000 with the alignment delimiter 。 ？ ！ or . ? ! — boundaries auto-snap to sentence ends
• Sentence-Level Split: use By Symbol mode with delimiters 。 ？ ！ (or . ? !) — one split at each delimiter
• Paragraph-Level Split: use Chinese Paragraph or English Paragraph mode
• Code / Markdown Split: By Length mode with the alignment delimiter \n\n (literal \n supported)

FAQ

What's the difference between the four splitting modes? "By Symbol" cuts at delimiters alone; "By Length" cuts into fixed-size blocks (default 5000) and optionally snaps the boundaries to a delimiter; "Chinese Paragraph" / "English Paragraph" run language-aware sentence detection — Chinese uses punctuation rules, English uses the browser's built-in Intl.Segmenter. Pick whichever matches your input language.

Do the two delimiter fields do the same thing? No — each mode has its own field and they are independent. By Symbol uses the delimiter to decide where to cut (every Nth one). By Length uses the optional alignment delimiter only to snap the boundaries of a length-based chunk; leave it empty and the text is hard-chunked.

When does batch export package as ZIP? Above 3 segments. ≤ 3 segments download as individual files; > 3 are bundled into a single <filename>_split_files.zip to avoid spamming your downloads folder.

Why does the preview hide automatically sometimes? When the result exceeds 500 segments, the preview is auto-hidden to keep the page responsive. Click "Show Anyway" to render it, or just use the export buttons (which work regardless of visibility). You can also force-hide via the "Hide results" toggle in the config.

Is anything uploaded? No. The tool runs entirely in your browser — no servers, no uploads. Process even million-character files without bandwidth concerns.

Documentation & Deployment

For detailed usage instructions and deployment guides, see the Official Documentation.

Contributing

Contributions are welcome! Feel free to open issues and pull requests.